Informix Guide to SQL

Tutorial

Informix Guide to SQL: Tutorial

Version 6.0
March 1994
Part No. 000-7598
 THE INFORMIX SOFTWARE AND USER MANUAL ARE PROVIDED ‘‘AS IS’’ WITHOUT WARRANTY
OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE
ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE INFORMIX SOFTWARE AND
USER MANUAL IS WITH YOU. SHOULD THE INFORMIX SOFTWARE AND USER MANUAL PROVE
DEFECTIVE, YOU (AND NOT INFORMIX OR ANY AUTHORIZED REPRESENTATIVE OF INFORMIX)
ASSUME THE ENTIRE COST OF ALL NECESSARY SERVICING, REPAIR, OR CORRECTION. IN NO
EVENT WILL INFORMIX BE LIABLE TO YOU FOR ANY DAMAGES, INCLUDING ANY LOST
PROFITS, LOST SAVINGS, OR OTHER INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
OUT OF THE USE OF OR INABILITY TO USE SUCH INFORMIX SOFTWARE OR USER MANUAL,
EVEN IF INFORMIX OR AN AUTHORIZED REPRESENTATIVE OF INFORMIX HAS BEEN ADVISED
OF THE POSSIBILITY OF SUCH DAMAGES, OR FOR ANY CLAIM BY ANY OTHER PARTY. IN
ADDITION, INFORMIX SHALL NOT BE LIABLE FOR ANY CLAIM ARISING OUT OF THE USE OF OR
INABILITY TO USE SUCH INFORMIX SOFTWARE OR USER MANUAL BASED UPON STRICT
LIABILITY OR INFORMIX’S NEGLIGENCE. SOME STATES DO NOT ALLOW THE EXCLUSION OF
IMPLIED WARRANTIES, SO THE ABOVE EXCLUSION MAY NOT APPLY TO YOU IN WHOLE OR IN
PART. THIS WARRANTY GIVES YOU SPECIFIC LEGAL RIGHTS AND YOU MAY ALSO HAVE OTHER
RIGHTS, WHICH VARY FROM STATE TO STATE.
All rights reserved. No part of this work covered by the copyright hereon may be reproduced or used in
any form or by any means—graphic, electronic, or mechanical, including photocopying, recording, taping,
or information storage and retrieval systems—without permission of the publisher.
Published by: Informix Software, Inc.
4100 Bohannon Drive
Menlo Park, CA 94025
Published by INFORMIX® Press Informix Software, Inc.
4100 Bohannon Drive
Menlo Park, CA 94025
INFORMIX® and C-ISAM® are registered trademarks of Informix Software, Inc.
UNIX® is a registered trademark, licensed exclusively by the X/Open Company Ltd. in the United Kingdom
and other countries.
X/Open® is a registered trademark of X/Open Company Ltd. in the United Kingdom and other countries.
MS® and MS-DOS® are registered trademarks of Microsoft Corporation.
(“DOS” as used herein refers to MS-DOS and/or PC-DOS operating systems.)
PostScript® is a registered trademark of Adobe Systems Incorporated.
IBM® is a registered trademark and DRDA™ is a trademark of International Business Machines Corporation.
Some of the products or services mentioned in this document are provided by companies other than
Informix. These products or services are identified by the trademark or servicemark of the appropriate
company. If you have a question about one of those products or services, please call the company in
question directly.
Documentation Team: Robert Berry, Mary Cole, Sally Cox, Signe Haugen, Geeta Karmarkar, Steven
Klitzing, Judith Sherwood, Rob Weinberg, Chris Willis, Eileen Wollam

RESTRICTED RIGHTS LEGEND

The Informix software and accompanying materials are provided with Restricted Rights. Use, duplication,
or disclosure by the Government is subject to restrictions as set forth in subdivision (c) (1) (ii) of the Rights
in Technical Data and Computer Software Clause at DFARS 252.227-7013 or subparagraphs (c) (1) and (2)
of the Commercial Computer Software-Restricted Rights at 48 CFR 52.227-19, as applicable (and any other
applicable license provisions set forth in the Government contract).

Copyright © 1981-1994 by Informix Software, Inc.

ii
Preface
This book is a tutorial on the Structured Query Language (SQL) as it is
implemented by Informix products. The Informix Guide to SQL: Tutorial and
its companion volumes, the Informix Guide to SQL: Reference and the Informix
Guide to SQL: Syntax, tell you how to create, manage, and use relational
databases with Informix software tools.
Examples in this manual represent a number of Informix products and
operating systems at different release levels.
You must have the following Informix software:
• An INFORMIX-OnLine Dynamic Server database server or an
INFORMIX-SE database server
The database server either must be installed on your machine or on
another machine to which your machine is connected over a network.
• Either an Informix application development tool, such as INFORMIX-4GL;
or an SQL application programming interface (API), such as
INFORMIX-ESQL/C; or the DB-Access database access utility, which is
shipped as part of your database server.
The application development tool, SQL API, or DB-Access enables you to
compose queries, send them to the database server, and view the results
that the database server returns. You can use DB-Access to try out all the
SQL statements described in this guide.

Preface iii
Summary of Chapters
The Informix Guide to SQL: Tutorial includes the following chapters:
• This Preface provides general information about the book and lists
additional reference materials that can help you understand relational
database management.
• The Introduction tells how SQL fits into the Informix family of products
and books, explains how to use this book, introduces the demonstration
database from which the product examples are drawn, and lists the new
features for Version 6.0 of Informix database server products.
Part I – “Using Basic SQL.” This part contains introductory chapters on how
to use SQL. Read these chapters first if you are new to databases and SQL.
• Chapter 1, “Introducing Database Features and SQL,” contains an
overview of database terminology and defines some important terms and
ideas that are used throughout the book.
• Chapter 2, “Composing Simple SELECT Statements,” begins the
exploration of making simple queries to retrieve and display database
data.
• Chapter 3, “Composing Advanced SELECT Statements,” discusses
making advanced queries to retrieve and display database data.
• Chapter 4, “Modifying Data,” describes the statements you use to insert,
delete, or update data, and introduces the concepts of database privileges,
maintaining data integrity, and archiving data.
• Chapter 5, “Programming with SQL,” discusses calling the database
server, retrieving rows, and embedding data.
• Chapter 6, “Modifying Data Through SQL Programs,” provides an
in-depth look at INSERT, DELETE, and UPDATE statements and using
these statements in SQL programs.
• Chapter 7, “Programming for a Multiuser Environment,” provides a
complete discussion on concurrency, isolation levels, and locking.
Part II – “Designing and Managing Databases.” This part contains overview
chapters about designing and managing databases. Read these chapters to
understand the entity-relationship data model for databases and how to
create and tune a database.
• Chapter 8, “Building Your Data Model,” describes the components of a
data model and provides a step-by-step procedure for building one.
• Chapter 9, “Implementing Your Data Model,” tells you how to define
database data types and create a database.

iv Preface
 • Chapter 10, “Tuning Your Data Model,” discusses many of the details that
help you set up an efficient database model, including disk storage, cal-
culating table sizes, managing indexes, and maximizing concurrency.
• Chapter 11, “Granting and Limiting Access to Your Database,” details
how you can ensure data security by granting privileges and using stored
procedures and views.
• Chapter 12, “Understanding Informix Networking,” discusses networks
and how you can best set up a database to work over a network.
Part III – “Using Advanced SQL.” This part contains advanced SQL chapters,
such as how to optimize your queries, how to create and use stored proce-
dures, and how to create and use triggers. Read these chapters if you need
better performance in your database.
• Chapter 13, “Optimizing Your Queries,” defines techniques to refine and
optimize your queries, introduces and explains the query Optimizer, and
discusses the time costs of various operations.
• Chapter 14, “Creating and Using Stored Procedures,” describes how to
create and use stored procedures.
• Chapter 15, “Creating and Using Triggers,” describes how to create and
use triggers.

Informix Welcomes Your Comments

A reader-response card is provided with this manual. Please use this card to
tell us what you like or dislike about this manual. To help us with future ver-
sions of the manual, please tell us about any corrections or clarifications that
you would find useful. Return this card to:
Informix Software, Inc.
Technical Publications Department
4100 Bohannon Drive
Menlo Park, CA 94025
If you prefer to share your comments on-line, address your e-mail to:
[email protected]

Preface v
Related Reading
If you want additional technical information on database management,
consult the following texts by C. J. Date:
• An Introduction to Database Systems, Volume I (Addison-Wesley Publishing,
1990)
• An Introduction to Database Systems, Volume II (Addison-Wesley
Publishing, 1983)
This guide assumes that you are familiar with your computer operating sys-
tem. If you have limited UNIX system experience, you might want to look at
your operating system manual or a good introductory text before you read
this tutorial.
The following list contains some suggested texts about UNIX systems:
• A Practical Guide to the UNIX System, Second Edition, by M. Sobell
(Benjamin/Cummings Publishing, 1989)
• A Practical Guide to UNIX System V by M. Sobell (Benjamin/Cummings
Publishing, 1985)
• Introducing the UNIX System by H. McGilton and R. Morgan (McGraw-Hill
Book Company, 1983)
• UNIX for People by P. Birns, P. Brown, and J. Muster (Prentice-Hall, 1985)

If you are interested in learning more about SQL, consider the following text:
• Using SQL by J. Groff and P. Weinberg (Osborne McGraw-Hill, 1990)

vi Preface
 Table of
Contents

Table of Contents
Introduction
Informix Products That Use SQL 3
Products Included in This Manual 3
Other Useful Documentation 4
How to Use This Manual 5
Typographical Conventions 5
Example Code Conventions 5
Useful On-Line Files 6
ASCII and PostScript Error Message Files 7
The Demonstration Database 7
Creating the Demonstration Database 8
New Features in Informix Version 6.0 Products
That Use SQL 9

Chapter 1 Introducing Database Features and SQL

Chapter Overview 1-3
Databases: What and Why? 1-3
The Data Model 1-3
Concurrent Use and Security 1-8
Centralized Management 1-9
Important Database Terms 1-11
The Relational Model 1-11
Structured Query Language 1-14
Standard SQL 1-14
Informix SQL and ANSI SQL 1-15
ANSI-Compliant Databases 1-15
NLS Databases 1-16
The Database Software 1-16
The Applications 1-16
The Database Server 1-16
Interactive SQL 1-17
 Reports and Forms 1-17
General Programming 1-17
Applications and Database Servers 1-18
Summary 1-18

Chapter 2 Composing Simple SELECT Statements

Chapter Overview 2-3
Introducing the SELECT Statement 2-4
Some Basic Concepts 2-4
The Forms of SELECT 2-10
Special Data Types 2-11
Single-Table SELECT Statements 2-11
Selecting All Columns and Rows 2-12
Selecting Specific Columns 2-18
Using the WHERE Clause 2-28
Creating a Comparison Condition 2-29
Expressions and Derived Values 2-46
Using Functions in SELECT Statements 2-53
Using Stored Procedures in SELECT Statements 2-67
Multiple-Table SELECT Statements 2-69
Creating a Cartesian Product 2-69
Creating a Join 2-71
Some Query Shortcuts 2-79
Summary 2-84

Chapter 3 Composing Advanced SELECT Statements

Chapter Overview 3-3
Using the GROUP BY and HAVING Clauses 3-4
Using the GROUP BY Clause 3-4
Using the HAVING Clause 3-9
Creating Advanced Joins 3-11
Self-Joins 3-11
Outer Joins 3-21
Subqueries in SELECT Statements 3-32
Using ALL 3-33
Using ANY 3-34
Single-Valued Subqueries 3-36
Correlated Subqueries 3-37
Using EXISTS 3-38

viii Table of Contents

 Set Operations 3-42
Union 3-43
Intersection 3-51
Difference 3-53
Summary 3-54

Chapter 4 Modifying Data

Chapter Overview 4-3
Statements That Modify Data 4-3
Deleting Rows 4-4
Deleting a Known Number of Rows 4-4
Inserting Rows 4-6
Updating Rows 4-11
Database Privileges 4-15
Displaying Table Privileges 4-16
Data Integrity 4-17
Entity Integrity 4-18
Semantic Integrity 4-18
Referential Integrity 4-19
Interrupted Modifications 4-22
The Transaction 4-23
Transaction Logging 4-24
Specifying Transactions 4-24
Archives and Logs 4-25
Archiving with INFORMIX-SE 4-26
Archiving with INFORMIX-OnLine Dynamic Server 4-26
Concurrency and Locks 4-27
Data Replication 4-28
INFORMIX-OnLine Dynamic Server Data Replication 4-29
Summary 4-29

Chapter 5 Programming with SQL

Chapter Overview 5-3
SQL in Programs 5-3
Static Embedding 5-5
Dynamic Statements 5-5
Program Variables and Host Variables 5-5
Calling the Database Server 5-8
The SQL Communications Area 5-8
The SQLCODE Field 5-11
The SQLERRD Array 5-12
The SQLAWARN Array 5-12
The SQLSTATE Value 5-12

Table of Contents ix
 Retrieving Single Rows 5-13
Data Type Conversion 5-14
Dealing with Null Data 5-15
Dealing with Errors 5-16
Retrieving Multiple Rows 5-19
Declaring a Cursor 5-20
Opening a Cursor 5-20
Fetching Rows 5-21
Cursor Input Modes 5-23
The Active Set of a Cursor 5-24
Using a Cursor: A Parts Explosion 5-26
Dynamic SQL 5-28
Preparing a Statement 5-29
Executing Prepared SQL 5-31
Dynamic Host Variables 5-33
Freeing Prepared Statements 5-33
Quick Execution 5-34
Embedding Data Definition Statements 5-34
Embedding Grant and Revoke Privileges 5-34
Summary 5-37

Chapter 6 Modifying Data Through SQL Programs

Chapter Overview 6-3
Using DELETE 6-3
Direct Deletions 6-4
Deleting with a Cursor 6-7
Using INSERT 6-8
Using an Insert Cursor 6-8
Rows of Constants 6-11
An Insert Example 6-12
Using UPDATE 6-14
Using an Update Cursor 6-15
Cleaning up a Table 6-16
Summary 6-17

Chapter 7 Programming for a Multiuser Environment

Chapter Overview 7-3
Concurrency and Performance 7-3
Locking and Integrity 7-3
Locking and Performance 7-3
Concurrency Issues 7-4
How Locks Work 7-5
Kinds of Locks 7-6

x Table of Contents
 Lock Scope 7-6
The Duration of a Lock 7-8
Locks While Modifying 7-9
Setting the Isolation Level 7-9
Dirty Read Isolation 7-10
Committed Read Isolation 7-10
Cursor Stability Isolation 7-11
Repeatable Read Isolation 7-12
Setting the Lock Mode 7-13
Waiting for Locks 7-13
Not Waiting for Locks 7-14
Waiting a Limited Time 7-14
Handling a Deadlock 7-14
Handling External Deadlock 7-15
Simple Concurrency 7-15
Locking with Other Database Servers 7-16
Isolation While Reading 7-17
Locking Updated Rows 7-17
Hold Cursors 7-18
Summary 7-19

Chapter 8 Building Your Data Model

Chapter Overview 8-3
Why Build a Data Model 8-3
Entity-Relationship Data Model Overview 8-3
Identifying and Defining Your Principal Data Objects 8-4
Discovering Entities 8-4
Defining the Relationships 8-8
Identifying Attributes 8-14
Diagramming Your Data Objects 8-16
Translating E-R Data Objects Into Relational Constructs 8-20
Rules for Defining Tables, Rows, and Columns 8-20
Determining Keys for Tables 8-22
Resolving Your Relationships 8-26
Normalizing Your Data Model 8-29
Summary 8-33

Chapter 9 Implementing Your Data Model

Chapter Overview 9-3
Defining the Domains 9-3
Data Types 9-4
Default Values 9-19
Check Constraints 9-20

Table of Contents xi
 Creating the Database 9-20
Using CREATE DATABASE 9-21
Using CREATE TABLE 9-24
Using Command Scripts 9-26
Populating the Tables 9-27
Summary 9-29

Chapter 10 Tuning Your Data Model

Chapter Overview 10-3
INFORMIX-OnLine Dynamic Server Disk Storage 10-3
Chunks and Pages 10-4
Dbspaces and Blobspaces 10-4
Disk Mirroring 10-5
Databases 10-5
Tables and Spaces 10-6
Tblspaces 10-7
Extents 10-8
Reorganizing dbspaces and Tables 10-10
Calculating Table Sizes 10-12
Estimating Fixed-Length Rows 10-13
Estimating Variable-Length Rows 10-14
Estimating Index Pages 10-16
Estimating Blobpages 10-17
Locating Blob Data 10-18
Managing Indexes 10-19
Space Costs of Indexes 10-20
Time Costs of Indexes 10-20
Choosing Indexes 10-21
Duplicate Keys Slow Index Modifications 10-22
Dropping Indexes 10-24
Clustered Indexes 10-25
Denormalizing 10-26
Shorter Rows for Faster Queries 10-26
Expelling Long Strings 10-26
Splitting Wide Tables 10-28
Splitting Tall Tables 10-29
Redundant and Derived Data 10-30
Maximizing Concurrency 10-32
Easing Contention 10-32
Rescheduling Modifications 10-33
Isolating and Dispersing Updates 10-35
Summary 10-36

xii Table of Contents

Chapter 11 Granting and Limiting Access to Your Database
Chapter Overview 11-3
Controlling Access to Databases 11-4
Securing Database Files 11-4
Securing Confidential Data 11-5
Granting Privileges 11-5
Database-Level Privileges 11-5
Ownership Rights 11-7
Table-Level Privileges 11-7
Procedure-Level Privileges 11-12
Automating Privileges 11-13
Controlling Access to Data Using Stored Procedures 11-15
Restricting Reads of Data 11-16
Restricting Changes to Data 11-16
Monitoring Changes to Data 11-17
Restricting Object Creation 11-18
Using Views 11-19
Creating Views 11-20
Modifying Through a View 11-23
Privileges and Views 11-27
Privileges When Creating a View 11-27
Privileges When Using a View 11-28
Summary 11-30

Chapter 12 Understanding Informix Networking

Chapter Overview 12-3
What Is a Network? 12-4
Database Management System Configurations 12-4
A Single-User Configuration 12-4
A Local Multiuser Configuration 12-6
A Remote Configuration 12-8
Single-Machine Configuration Using Network Communication 12-9
Distributed Databases 12-10
Distributed Databases Using Multiple Vendor Servers 12-11
Connecting to Data on a UNIX Network 12-12
Example of Client / Server Connections 12-13
Environment Variables 12-14
Connection Information 12-15
SQL Connection Statements 12-16
Accessing Tables 12-16
Using Synonyms with Table Names 12-17
Synonym Chains 12-18
Protecting Your Data in a Networked Environment 12-19

Table of Contents xiii

 Data Protection with INFORMIX-SE 12-19
Data Protection with INFORMIX-OnLine Dynamic Server 12-19
Data Integrity for Distributed Data 12-20
Summary 12-21

Chapter 13 Optimizing Your Queries

Chapter Overview 13-3
Optimizing Techniques 13-4
Verifying the Problem 13-4
Considering the Total System 13-5
Understanding the Application 13-5
Measuring the Application 13-6
Finding the Guilty Functions 13-7
Keeping an Open Mind 13-7
The Query Optimizer 13-8
How the Optimizer Works 13-8
Reading the Plan 13-12
Time Costs of a Query 13-14
Activities in Memory 13-14
Disk-Access Management 13-14
The Cost of Reading a Row 13-16
The Cost of Sequential Access 13-17
The Cost of Nonsequential Access 13-18
The Cost of Rowid Access 13-18
The Cost of Indexed Access 13-18
The Cost of Small Tables 13-19
The Cost of NLS Functionality 13-19
The Cost of Network Access 13-20
The Importance of Table Order 13-22
Making Queries Faster 13-27
Preparing a Test Environment 13-28
Studying the Data Model 13-28
Studying the Query Plan 13-29
Examining the Indexes 13-29
Refining the Column Filters 13-31
Rethinking the Query 13-33
Reviewing the Optimization Level 13-36
Using a Temporary Table to Speed Queries 13-36
Summary 13-41

xiv Table of Contents

Chapter 14 Creating and Using Stored Procedures
Chapter Overview 14-3
Introduction to Stored Procedures and SPL 14-3
What You Can Do with Stored Procedures 14-3
Relationship Between SQL and a Stored Procedure 14-4
Creating and Using Stored Procedures 14-4
Creating a Procedure Using DB-Access 14-5
Creating a Procedure Using an SQL API 14-5
Commenting and Documenting a Procedure 14-6
Diagnosing Compile-Time Errors 14-6
Looking at Compile-Time Warnings 14-8
Generating the Text or Documentation 14-8
Executing a Procedure 14-9
Executing a Stored Procedure Dynamically 14-11
Debugging a Procedure 14-11
Re-creating a Procedure 14-13
Privileges on Stored Procedures 14-13
Privileges Necessary at Creation 14-14
Privileges Necessary at Execution 14-14
Revoking Privileges 14-16
Variables and Expressions 14-16
Variables 14-16
SPL Expressions 14-21
Program Flow Control 14-22
Branching 14-22
Looping 14-23
Function Handling 14-24
Passing Information into and out of a Procedure 14-25
Returning Results 14-25
Exception Handling 14-27
Trapping an Error and Recovering 14-28
Scope of Control of an ON EXCEPTION Statement 14-29
User-Generated Exceptions 14-30
Summary 14-33

Chapter 15 Creating and Using Triggers

Chapter Overview 15-3
When to Use Triggers 15-3
How to Create a Trigger 15-4
Assigning a Trigger Name 15-5
Specifying the Trigger Event 15-5
Defining the Triggered Actions 15-6
A Complete CREATE TRIGGER Statement 15-6

Table of Contents xv
 Using Triggered Actions 15-7
Using BEFORE and AFTER Triggered Actions 15-7
Using FOR EACH ROW Triggered Actions 15-8
Using Stored Procedures as Triggered Actions 15-10
Tracing Triggered Actions 15-12
Generating Error Messages 15-14
Applying a Fixed Error Message 15-15
Generating a Variable Error Message 15-16
Summary 15-17

Index

xvi Table of Contents

 Introduction

Introduction
Informix Products That Use SQL 3
Products Included in This Manual 3
Other Useful Documentation 4
How to Use This Manual 5
Typographical Conventions 5
Example Code Conventions 5
Useful On-Line Files 6
ASCII and PostScript Error Message Files 7
The Demonstration Database 7
Creating the Demonstration Database 8
Compliance with Industry Standards 9
New Features in Informix Version 6.0 Products
That Use SQL 10
2 Introduction
 Structured Query Language (SQL) is an English-like language that you can
use when creating, managing, and using relational databases. The SQL
provided with Informix products is an enhanced version of the industry-
standard query language developed by International Business Machines
Corporation (IBM).

Informix Products That Use SQL

Informix produces many application development tools and SQL application
programming interfaces (API). Application development tools currently
available include products like INFORMIX-SQL, INFORMIX-4GL, and the
Interactive Debugger. SQL APIs currently available include
INFORMIX-ESQL/C and INFORMIX-ESQL/COBOL.

Informix products work with a database server, either INFORMIX-OnLine

Dynamic Server or INFORMIX-SE. The DB-Access database access utility is
shipped as a part of each database server.
If you are running client applications developed with Version 4.1 and 5.0
application development tools, you use INFORMIX-NET to connect the client
to the network.

Products Included in This Manual

All of the information presented in this manual is valid for the following
products:
• INFORMIX-ESQL/C, Version 6.0
• INFORMIX-ESQL/COBOL, Version 6.0
• INFORMIX-SE, Version 6.0
• INFORMIX-OnLine Dynamic Server, Version 6.0
• INFORMIX-OnLine/Optical, Version 6.0
• INFORMIX-TP/XA, Version 6.0

Introduction 3
Other Useful Documentation

An additional product that uses information from this manual is the

INFORMIX-Gateway with DRDA, Version 6.0.

In addition, some examples in this manual are built using Informix products
at earlier release levels.

Other Useful Documentation

You can refer to a number of related Informix documents that complement
this manual.
• A companion volume to the Tutorial, the Informix Guide to SQL: Reference,
provides reference information on the types of Informix databases you
can create, the data types supported in Informix products, system catalog
tables associated with the database, environment variables, and the SQL
utilities. This guide also provides a detailed description of the stores6
demonstration database and contains a glossary.
• An additional companion volume to the Tutorial, the Informix Guide to
SQL: Syntax, provides a detailed description for all of the SQL statements
supported by Informix products. This guide also provides a detailed
description of stored procedure language (SPL) statements.
• The SQL Quick Syntax Guide contains syntax diagrams for all statements
and segments described in this manual.
• You, or whoever installs your Informix products, should refer to the
UNIX Products Installation Guide for your particular release to ensure that
your Informix product is properly set up before you begin to work with
it. A matrix depicting possible client/server configurations is included in
the Installation Guide.
• Depending on the database server you are using, you or your system
administrator need either the INFORMIX-SE Administrator’s Guide or the
INFORMIX-OnLine Dynamic Server Administrator’s Guide and
INFORMIX-OnLine Dynamic Server Archive and Backup Guide.
• The DB-Access User Manual describes how to invoke the utility to access,
modify, and retrieve information from Informix database servers.
• When errors occur, you can look them up by number and learn their cause
and solution in the Informix Error Messages manual. If you prefer, you can
look up the error messages in the on-line message file described in the
section “ASCII and PostScript Error Message Files” later in this
Introduction.

4 Introduction
 How to Use This Manual

How to Use This Manual

This manual assumes that you are using INFORMIX-OnLine Dynamic Server
as your database server. Features and behavior specific to INFORMIX-SE are
noted throughout the manual.
Other conventions used in this manual for typographical format and
example code are described in the following sections.

Typographical Conventions
Informix product manuals use a standard set of conventions to introduce
new terms, illustrate screen displays, describe command syntax, and so forth.
The following typographical conventions are used throughout this manual:
italics New terms, emphasized words, and variables are printed in
italics.
boldface Database names, table names, column names, filenames,
utilities, and other similar terms are printed in boldface.
computer Information that the product displays and information that
you enter are printed in a computer typeface.
KEYWORD All keywords appear in uppercase letters.
This symbol indicates a warning. Warnings provide critical
! information that, if ignored, could cause harm to your
database.
Additionally, when you are instructed to “enter” or “execute” text,
immediately press RETURN after the entry. When you are instructed to “type”
the text or “press” a key, no RETURN is required.

Example Code Conventions

Examples of SQL code occur throughout this manual. Except where noted,
the code is not specific to any single Informix application development tool.
If only SQL statements are listed in the example, they are not delineated by
semicolons. To use this SQL code for a specific product, you must apply the
syntax rules for that product. For example, if you are using the Query-
language option of DB-Access, you must delineate multiple statements with
semicolons. If you are using an SQL API, you must use EXEC SQL and a semi-
colon (or other appropriate delimiters) at the start and end of each statement,
respectively.

Introduction 5
Useful On-Line Files

For example, you might see the following example code:

CONNECT TO stores6
.
.
.
DELETE FROM customer
WHERE customer_num = 121
.
.
.
COMMIT WORK
DISCONNECT CURRENT

For detailed directions on using SQL statements for a particular application

development tool or SQL API, see the manual for your product.
Also note that dots in the example indicate that more code would be added
in a full application, but it is not necessary to show it to describe the concept
being discussed.

Useful On-Line Files

In addition to the Informix set of manuals, the following on-line files, located
in the $INFORMIXDIR/release directory, may supplement the information in
this manual:
Documentation describe features not covered in the manual or that
Notes have been modified since publication. The file containing
the Documentation Notes for this product is called
SQLTDOC_6.0.
Release Notes describe feature differences from earlier versions of Informix
products and how these differences may affect current prod-
ucts. The file containing the Release Notes for Version 6.0 of
Informix database server products is called SERVERS_6.0.
Machine Notes describe any special actions required to configure and use
Informix products on your machine. Machine notes are
named for the product described, for example, the machine
notes file for INFORMIX-OnLine Dynamic Server is
ONLINE_6.0.
Please read these files because they contain vital information about
application and performance issues.

6 Introduction
 ASCII and PostScript Error Message Files

A number of Informix products also provide on-line Help files that walk you
through each menu option. To invoke the Help feature, simply press CTRL-W
wherever you are in your Informix product.

ASCII and PostScript Error Message Files

Informix software products provide ASCII files that contain all the Informix
error messages and their corrective actions. To access the error messages in
the ASCII file, Informix provides scripts that let you display error messages
on the screen (finderr) or print formatted error messages (rofferr). See the
Introduction to the Informix Error Messages manual for a detailed description
of these scripts.
The optional Informix Messages and Corrections product provides
PostScript files that contain the error messages and their corrective actions. If
you have installed this product, you can print the PostScript files on a Post-
Script printer. The PostScript error messages are distributed in a number of
files of the format errmsg1.ps, errmsg2.ps, and so on. These files are located
in the $INFORMIXDIR/msg directory.

The Demonstration Database

The DB-Access utility, which is provided with your Informix database server
products, includes a demonstration database called stores6 that contains
information about a fictitious wholesale sporting-goods distributor. The sam-
ple command files that make up a demonstration application are also
included.
Most of the examples in this manual are based on the stores6 demonstration
database. The stores6 database is described in detail and its contents are
listed in Appendix A of the Informix Guide to SQL: Reference.
The script that you use to install the demonstration database is called
dbaccessdemo6 and is located in the $INFORMIXDIR/bin directory. The data-
base name that you supply is the name given to the demonstration database.
If you do not supply a database name, the name defaults to stores6. Follow
these rules for naming your database:
• Names for databases can be up to 18 characters long for
INFORMIX-OnLine Dynamic Server databases and up to 10 characters
long for INFORMIX-SE databases.
• The first character of a name must be a letter or an underscore (_).

Introduction 7
The Demonstration Database

• You can use letters, characters, and underscores (_) for the rest of the
name.
• DB-Access makes no distinction between uppercase and lowercase
letters.
• The database name should be unique.
When you run dbaccessdemo6, you are, as the creator of the database, the
owner and Database Administrator (DBA) of that database.
If you installed your Informix database server according to the installation
instructions, the files that make up the demonstration database are protected
so you cannot make any changes to the original database.
You can run the dbaccessdemo6 script again whenever you want to work
with a fresh demonstration database. The script prompts you when the cre-
ation of the database is complete, and asks if you would like to copy the sam-
ple command files to the current directory. Enter “N” if you have made
changes to the sample files and do not want them replaced with the original
versions. Enter “Y” if you want to copy over the sample command files.

Creating the Demonstration Database

Use the following steps to create and populate the demonstration database:
1. Set the INFORMIXDIR environment variable so that it contains the name
of the directory in which your Informix products are installed. Set
INFORMIXSERVER to the name of the default database server. The name
of the default database server must exist in the $INFORMIXDIR/etc/sql-
hosts file. (For a full description of environment variables, see Chapter 4
of the Informix Guide to SQL: Reference.) For more information about
sqlhosts, see the INFORMIX-OnLine Dynamic Server Administrator’s Guide
or INFORMIX-SE Administrator’s Guide.
2. Create a new directory for the SQL command files. Create the directory by
entering
mkdir dirname

3. Make the new directory the current directory by entering

cd dirname

8 Introduction
 New Features in Informix Version 6.0 Products That Use SQL

4. Create the demonstration database and copy over the sample command
files by entering the following command:
To create the database without logging enter:
dbaccessdemo6 dbname

To create the demonstration database with logging enter:

dbaccessdemo6 -log dbname

If you are using INFORMIX-OnLine Dynamic Server, by default the data

for the database is put into the root dbspace. If you wish, you can specify
a dbspace for the demonstration database.
To create a demonstration database in a particular dbspace enter:
dbaccessdemo6 dbspacename

If you are using INFORMIX-SE, a subdirectory called dbname.dbs is

created in your current directory and the database files associated with
stores6 are placed there. You will see both data (.dat) and index (.idx) files
in the dbname.dbs directory. (If you specify a dbspace name, it will be
ignored.)
To use the database and the command files that have been copied to your
directory, you must have UNIX read and execute permissions for each
directory in the pathname of the directory from which you ran the
dbaccessdemo6 script.
5. To give someone else the permissions to access the command files in your
directory, use the UNIX chmod command.
6. To give someone else the access to the database that you have created,
grant them the appropriate privileges using the GRANT statement. To
remove privileges, use the REVOKE statement. The GRANT and REVOKE
statements are described in Chapter 1 of the Informix Guide to SQL: Syntax.

New Features in Informix Version 6.0 Products

That Use SQL
The Introduction to each Version 6.0 product manual contains a list of new
features for that product. The Introduction to each manual in the Version 6.0
Informix Guide to SQL series contains a list of new SQL features.
A comprehensive listing of all the new features for Version 6.0 Informix
products is found in the Release Notes file called SERVERS_6.0.

Introduction 9
New Features in Informix Version 6.0 Products That Use SQL

This section highlights the major new features implemented in Version 6.0 of
Informix products that use SQL.
• Native Language Support
Native Language Support (NLS) makes Informix Version 6.0 products
adaptable to various European cultural and language environments
without requiring changes to the application source code.
When appropriate environment variables are set to activate NLS and
specify a locale, Informix products can properly collate strings that con-
tain foreign characters, accept money and decimal input, and print dates
and times in the format required by the locale where the software is run.
The user can
o Create or access database information in any language available on
the system by changing a few environment variables
o Name user-defined objects such as databases, tables, columns, views,
cursors, and files using a foreign character set
o Use the new NCHAR and NVARCHAR data types in place of CHAR and
VARCHAR, respectively, for storing national characters
In addition, by installing one or more language supplements with an
Informix product, the user can view error and warning messages in the
language of the locale.
• Enhanced Database Connections
You can now use three new statements, CONNECT TO, DISCONNECT, and
SET CONNECTION to provide a connection-oriented association between
client and server processes in a networked or a non-networked environ-
ment. These three statements are compliant with X/Open and ANSI/ISO
specifications. Applications with embedded SQL can use these statements
for more uniform and portable syntax when accessing local or remote
data.
• Cascading Deletes
Support for cascading deletes is provided in INFORMIX-OnLine
Dynamic Server as an enhancement to referential integrity. Previously,
when you attempted to delete a row from a parent table without deleting
rows from associated child tables first, the delete was disallowed. Now
you can specify the ON DELETE CASCADE option on either the CREATE
TABLE or ALTER TABLE statements or on DB-Access menus to allow
deletes from a parent table to cause deletes on child tables.

10 Introduction
 New Features in Informix Version 6.0 Products That Use SQL

• Enhanced CREATE INDEX statement

You can now use the FILLFACTOR option on the CREATE INDEX statement
to create indexes that provide for expansion of the index at a later time or
to create compacted indexes.
• Enhanced Arithmetic Functions
You can now use certain trigonometric and algebraic functions in your
SQL statements. The new functions include ABS, MOD, POW, ROOT, SQRT,
COS, SIN, TAN, ACOS, ASIN, ATAN, ATAN2, EXP, LOGN, and LOG10.

• Enhanced Error Handling

You can now use a new statement to retrieve standard and multiple
diagnostic information about your SQL statements. The GET DIAGNOS-
TICS statement is compliant with X/Open and ANSI/ISO specifications
and provides a standard method for detecting and handling error
messages. The GET DIAGNOSTICS statement is used with a new
error-handling status variable called SQLSTATE.
• New DBINFO Function
The DBINFO function allows you to find the following information:
o The dbspace name for a given table
o The last serial value inserted into a table
o The number of rows processed by SELECT, INSERT, DELETE, UPDATE,
and EXECUTE PROCEDURE statements
• New Environment Variables
The following Informix environment variables and X/Open categories,
described in Chapter 4 of the Informix Guide to SQL: Reference, are new in
Version 6.0:
o ARC_DEFAULT
o ARC_KEYPAD
o COLLCHAR
o DBAPICODE
o DBNLS
o DBSPACETEMP
o DBUPSPACE
o ENVIGNORE
o INFORMIXC

Introduction 11
New Features in Informix Version 6.0 Products That Use SQL

o INFORMIXSERVER
o INFORMIXSHMBASE
o INFORMIXSTACKSIZE
o LANG
o LC_COLLATE
o LC_CTYPE
o LC_MONETARY
o LC_NUMERIC
o LC_TIME
• New Data Distribution Features
You can use the UPDATE STATISTICS statement to create data
distributions for each table. The database server uses these data distribu-
tions to improve the choice of execution paths for SELECT statements. The
new data-distribution feature affects application development in the
following three ways:
o The syntax for the UPDATE STATISTICS statement has been expanded.
o A new environment variable, DBUPSPACE, sets the upper limit of disk
space that you want to use when columns are sorted.
o You can use the dbschema utility to print distribution information.
• Introduction of Environment-Configuration Files
The following environment-configuration files can contain default values
of environment variables:
o $INFORMIXDIR/etc/informix.rc
o ~/.informix
These optional files allow the administrator or the user to set
environment variables in much the same way a .login or .profile file is
used, so that the variables do not need to be set at each session.

12 Introduction
Section I: Using Basic SQL
 Chapter

1
Introducing
Database Features
and SQL
Chapter Overview 3
Databases: What and Why? 3
The Data Model 3
Storing Data 6
Querying Data 6
Modifying Data 8
Concurrent Use and Security 8
Centralized Management 9
Group and Private Databases 10
Essential Databases 10
Important Database Terms 11
The Relational Model 11
Tables 11
Columns 12
Rows 12
Tables, Rows, and Columns 12
Operations on Tables 13
Structured Query Language 14
Standard SQL 14
Informix SQL and ANSI SQL 15
ANSI-Compliant Databases 15
NLS Databases 16
The Database Software 16
The Applications 16
The Database Server 16
 Interactive SQL 17
Reports and Forms 17
General Programming 17
Applications and Database Servers 18
Summary 18

1-2 Introducing Database Features and SQL

Chapter Overview
This is a book about databases and about how you can exploit them using
Informix software. Your real use of a database begins with the SELECT state-
ment, which is described in Chapter 2, “Composing Simple SELECT State-
ments.” If you are in a hurry, and if you know at least a little about databases,
turn to it now.
This chapter covers the fundamental concepts of databases and defines some
terms that are used throughout the book, emphasizing the following topics:
• What makes a database different from a collection of files?
• What terms are used to describe the main components of a database?
• What language is used to create, query, and modify a database?
• What are the main parts of the software that manages a database, and
how do these parts work with each other?

Databases: What and Why?

A database is a collection of information—but so is a simple computer file. What
makes a database so special? There must be reasons; otherwise people would
not spend so much money, effort, and computer time on databases. Two fun-
damental differences exist. First, a database comprises not only data but also
a plan, or model, of the data. Second, a database can be a common resource,
used concurrently by many people.

The Data Model

The principal difference between information collected in a database and
information in a file is in the way the data is organized. A flat file is organized
physically; certain items precede or follow other items. But the contents of a
database are organized according to a data model. A data model is a plan, or
map, that defines the units of data and specifies how each unit is related to
the others.

Introducing Database Features and SQL 1-3

Databases: What and Why?

For example, a number can appear in either a file or a database. In a file, it is

simply a number that occurs at a certain point in the file. A number in a data-
base, however, has a role assigned to it by the data model. It may be a price
that is associated with a product that was sold as one item of an order that was
placed by a customer. Each of these things—prices, products, items, orders,
and customers—are also roles specified by the data model. (See Figure 1-1.)
The data model is designed when the database is created; then units of data
are inserted according to the plan laid out in the model. In some books, the
term schema is used instead of data model.

1-4 Introducing Database Features and SQL

 Databases: What and Why?

1014 06/25/93 1 case footballs 960.00

1015 06/27/93 1 case baseball gloves $450.00

1014 06/25/93 1 case football $960.00
1013 06/22/93 1 each tennis racquet $19.80
1012 06/18/93 1 case volleyball $ 840.00
1011 06/18/93 5 each tennis racquet $99.00
1010 06/17/93 1 case tennis ball $36.00

ORDERS

order
1003
05/22/93
order
1001
05/20/93 customer
Anthony Higgins

item order
tennis 1011
racquet order 06/18/93
$19.80 1013
06/22/93

item
2 item
volleyball 1 case
nets tennis
ball

Figure 1-1 Illustration of a data model

Introducing Database Features and SQL 1-5

Databases: What and Why?

Storing Data
Another difference between a database and a file is that the organization of
the database is stored with the database.
A file may have a complex inner structure, but the definition of that structure
is not within the file; it is in the programs that create or use the file. For exam-
ple, a document file as stored by a word-processing program may contain
very detailed structures describing the format of the document. However,
only the word-processing program can decipher the contents of the file
because the structure is defined within the program, not within the file.
A data model, however, is contained in the database it describes. It travels
with the database and is available to any program that uses the database. The
model defines not only the names of the data items but also their data types,
so a program can adapt itself to the database. For example, a program can
find out that, in the current database, a price item is a decimal number with
eight digits, two to the right of the decimal point; then it can allocate storage
for a number of that type. How programs work with databases is the subject
of Chapter 5, “Programming with SQL,” and Chapter 6, “Modifying Data
Through SQL Programs.”

Querying Data
Another difference between a database and a file is the way you can
interrogate them. You can search a file sequentially, looking for particular val-
ues at particular physical locations in each line or record. That is, you might
ask a file, “What records have numbers under 20 in the fifth field?” Figure 1-2
illustrates this type of search.

1-6 Introducing Database Features and SQL

 Databases: What and Why?

1015 06/27/93 1 case baseball gloves$450.00

1013 06/22/93
1014 1 each
06/25/93 tennis
1 case racquet $19.80
footballs 960.00
06/22/93 1 case tennis ball $36.00
06/22/93 1 case tennis ball $48.00
1012 06/18/93 1 case volleyball
06/23/93 $840.00
1011 06/18/93 5 each tennis racquet $99.00
1010 06/17/93 1 case tennis ball $36.00

ORDERS

Figure 1-2 Searching a file sequentially

In contrast, when you query a database you use the terms defined by its
model. To a database, you can pose questions such as, “What orders have been
placed for products made by the Shimara Corporation, by customers in New
Jersey, with ship dates in the third quarter?” Figure 1-3 illustrates this type of
query.
In other words, when you interrogate data stored in a file, you must state
your question in terms of the physical layout of the file. When you query a
database, you can ignore the arcane details of computer storage and state
your query in terms that reflect the real world—at least, to the extent that the
data model reflects the real world.
In this manual, Chapters 2 and 3 discuss the language you use for making
queries. Chapters 8 through 11 discuss designing an accurate, robust data
model for other users to query.

Introducing Database Features and SQL 1-7

Databases: What and Why?

order
1016 state
order manufacturer 06/29/93 New Jersey
1019 Shimara
customer
07/16/93
Cathy
O’Brian
customer
Bob
order
Shorter
1023
07/24/93

Run: Next Restart Exit

Display the next page of query results

--------stores6-----------Press CTRL-W for Help------

1019 Bob Shorter SHM swim cap 07/16/93

Figure 1-3 Querying a database

Modifying Data
The model also makes it possible to modify the contents of the database with
less chance for error. You can query the database with commands such as
“Find every stock item with a manufacturer of Presta or Schraeder, and increase
its price by 13 percent.” You state changes in terms that reflect the meaning of
the data. You do not have to waste time and effort thinking about details of
fields within records in a file, so the chances for error are less.
The statements you use to modify stored data are covered in Chapter 5, “Pro-
gramming with SQL.”

Concurrent Use and Security

A database can be a common resource for many users. Multiple users can
query and modify a database simultaneously. The database server (the pro-
gram that manages the contents of all databases) ensures that the queries and
modifications are done in sequence and without conflict.

1-8 Introducing Database Features and SQL

 Databases: What and Why?

Having concurrent users on a database provides great advantages but also

introduces new problems of security and privacy. Some databases are pri-
vate; individuals set them up for their own use. Other databases contain con-
fidential material that must be shared but among only a select group of
persons. Still other databases provide public access.
Informix database software provides the means to control database use.
When you design a database, you can perform any of the following functions:
• Keep the database completely private.
• Open its entire contents to all users or to selected users.
• Restrict the selection of data that some users can view. (In fact, you can
reveal entirely different selections of data to different groups of users.)
• Allow specified users to view certain items but not modify them.
• Allow specified users to add new data but not modify old data.
• Allow specified users to modify all, or specified items of, existing data.
• Ensure that added or modified data conforms to the data model.
The facilities that make these and other things possible are discussed in
Chapter 11, “Granting and Limiting Access to Your Database.”

Centralized Management
Databases that are used by many people are highly valuable and must be
protected as important business assets. This creates two significant problems:
protecting data and maintaining performance. The INFORMIX-OnLine
Dynamic Server database server allows you to centralize these tasks.

Databases must be guarded against loss or damage. The hazards are many:
failures in software and hardware, and the risks of fire, flood, and other nat-
ural disasters. Losing an important database creates a huge potential for
damage. The damage could include not only the expense and difficulty of re-
creating the lost data but also the loss of productive time by the database
users as well as the loss of business and good will while users cannot work.
A plan for regular archiving of critical databases can help avoid or mitigate
these potential disasters.
A large database used by many people must be maintained and tuned.
Someone must monitor its use of system resources, chart its growth, antici-
pate bottlenecks, and plan for expansion. Users will report problems in the
application programs; someone must diagnose these and correct them. If
rapid response is important, someone must study and analyze the
performance of the system and find the causes of slow responses.

Introducing Database Features and SQL 1-9

Databases: What and Why?

Group and Private Databases

Some Informix database servers are designed to manage relatively small
databases that are used privately by individuals or that are shared among a
small group of users.
These database servers (for example, INFORMIX-SE for the UNIX operating
system) store databases in files managed by the host operating system. These
databases can be archived using the same procedures for backing up files that
work with other computer files; that is, copying the files to another medium
when they are not in use. The only difference is that when a database is
archived its transaction log file must be reset to empty. (The use of transaction
logs is discussed in Chapter 7, “Programming for a Multiuser Environment.”
Chapter 10, “Tuning Your Data Model,” has more information on archiving.)
Performance problems that arise in group and private databases are usually
related to particular queries that take too long. Chapter 13, “Optimizing Your
Queries,” deals in depth with the reasons why a query takes more or less
time. After you understand all the features of the SELECT statement and the
alternative ways of stating a query, as covered in Chapter 2, “Composing
Simple SELECT Statements,” and Chapter 3, “Composing Advanced
SELECT Statements,” you can use the tips in Chapter 13, “Optimizing Your
Queries,” to improve the performance of queries that otherwise might take
excessive amounts of time.

Essential Databases
The INFORMIX-OnLine Dynamic Server database server is designed to man-
age large databases with requirements for high reliability, high availability,
and high performance. Although it supports private and group databases
very well, it is at its best managing the databases that are essential for your
organization to carry out its work.
INFORMIX-OnLine Dynamic Server gives you the ability to make archival
copies while the databases are in use. It also allows incremental archiving
(archiving only modified data), an important feature when making a com-
plete copy could take many reels of tape.
INFORMIX-OnLine Dynamic Server has an interactive monitor program by
which its operator (or any user) can monitor the activities within the database
server to see when bottlenecks are developing. It also comes with utility pro-
grams to analyze its use of disk storage. In addition, OnLine provides the
sysmaster tables that contain information about an entire OnLine database
server, which might manage many databases. For more information about
the sysmaster tables, see the INFORMIX-OnLine Dynamic Server Administra-
tor’s Guide.

1-10 Introducing Database Features and SQL

 Important Database Terms

Chapter 10, “Tuning Your Data Model,” contains an overview of the

INFORMIX-OnLine Dynamic Server disk-storage methods as part of the nec-
essary background information to the main topic. All the details of using and
managing OnLine are contained in the INFORMIX-OnLine Dynamic Server
Administrator’s Guide.

Important Database Terms

You should know two sets of terms before beginning the next chapter. One
set of terms describes the database and the data model; the other set describes
the computer programs that manage the database.

The Relational Model

Informix databases are relational databases. In technical terms, that means
that the data model by which an Informix database is organized is based on
the relational calculus devised by E.F. Codd. In practical terms, it means that
all data is presented in the form of tables comprising rows and columns.

Tables
A database is a collection of information grouped into one or more tables. A
table is an array of data items organized into rows and columns. A demon-
stration database is distributed with every Informix product. A partial table
from the demonstration database is shown in Figure 1-4.

stock Table

stock_num manu_code description unit_price unit unit_descr

1 HRO baseball gloves 250.00 case 10 gloves/case
1 HSK baseball gloves 800.00 case 10 gloves/case
1 SMT baseball gloves 450.00 case 10 gloves/case
2 HRO baseball 126.00 case 24/case
3 HSK baseball bat 240.00 case 12/case
4 HSK football 960.00 case 24/case
4 HRO football 480.00 case 24/case
5 NRG tennis racquet 28.00 each each
... .... ......... ... ... ......
313 ANZ swim cap 60.00 box 12/box

Figure 1-4 The stock table from the demonstration database distributed with all Informix
products

Introducing Database Features and SQL 1-11

Important Database Terms

A table represents all that is known about one entity, one type of thing that
the database describes. The example table, stock, represents all that is known
about the merchandise that is stocked by a sporting-goods store. Other tables
in the demonstration database represent such entities as customer and
orders.
A database can be thought of as a collection of tables. To create a database is
to create a set of tables. The right to query or modify tables can be controlled
on a table-by-table basis, so that some users can view or modify some tables
and not others.

Columns
Each column of a table stands for one attribute, which is one characteristic,
feature, or fact that is true of the subject of the table. The stock table has col-
umns for the following facts about items of merchandise: stock numbers,
manufacturer codes, descriptions, prices, and units of measure.

Rows
Each row of a table stands for one instance of the subject of the table, which is,
one particular example of that entity. Each row of the stock table stands for
one item of merchandise that is sold by the sporting-goods store.

Tables, Rows, and Columns

You now understand that the relational model of data is a very simple way of
organizing data to reflect the world using the following simple correspond-
ing relationships:
table = entity A table represents all that the database knows about
one subject or kind of thing.
column = attribute A column represents one feature, characteristic, or fact
that is true of the table subject.
row = instance A row represents one individual instance of the table
subject.
Some rules apply about how you choose entities and attributes, but they are
important only when you are designing a new database. (Database design is
covered in Chapters 8 through 11.) The data model in an existing database
already is set. To use the database, you need to know only the names of the
tables and columns and how they correspond to the real world.

1-12 Introducing Database Features and SQL

 Important Database Terms

Operations on Tables
Because a database is really a collection of tables, database operations are
operations on tables. The relational model supports three fundamental oper-
ations, two of which are illustrated in Figure 1-5. (All three operations are
defined in more detail, with many examples, in Chapter 2, “Composing Sim-
ple SELECT Statements,” and Chapter 3, “Composing Advanced SELECT
Statements.”)
To select from a table is to choose certain rows, leaving others aside. One
selection that could be made on the stock table is “select all rows in which the
manufacturer code is HRO and the unit price is between 100.00 and 200.00.”
To project from a table is to choose certain columns, leaving others aside. One
projection that can be made from the stock table is “show me the stock_num,
unit_descr, and unit_price columns only.”
A table contains information about only one entity; when you want informa-
tion about multiple entities, you must join their tables. You can join tables in
many ways. (The join operation is the subject of Chapter 3, “Composing
Advanced SELECT Statements.”)

stock Table

stock_num manu_code description unit_price unit unit_descr

1 HRO baseball gloves 250.00 case 10 gloves/case
1 HSK baseball gloves 800.00 case 10 gloves/case
1 SMT baseball gloves 450.00 case 10 gloves/case
2 HRO baseball 126.00 case 24/case SELECTION
3 HSK baseball bat 240.00 case 12/case
4 HSK football 960.00 case 24/case
4 HRO football 480.00 case 24/case

P R O J E C T I O N

Figure 1-5 Illustration of selection and projection

Introducing Database Features and SQL 1-13

Structured Query Language

Structured Query Language

Most computer software has not yet reached a point where you can literally
ask a database, “what orders have been placed by customers in New Jersey
with ship dates in the second quarter?” You must still phrase questions in a
restricted syntax that the software can easily parse. You can pose the same
question to the demonstration database in the following terms:

SELECT * FROM customer, orders

WHERE customer.customer_num = orders.customer_num
AND customer.state = 'NJ'
AND orders.ship_date
BETWEEN DATE('7/1/93') AND DATE('7/30/93')

This question is a sample of Structured Query Language (SQL). It is the

language that you use to direct all operations on the database. SQL is com-
posed of statements, each of which begins with one or two keywords that
specify a function. SQL includes about 73 statements, from ALLOCATE
DESCRIPTOR to WHENEVER.

All the SQL statements are specified in detail in the Informix Guide to SQL:
Syntax. Most of the statements are used infrequently—when setting up or
tuning a database. People generally use three or four statements to query or
update databases.
One statement, SELECT, is in almost constant use. It is the only statement that
you can use to retrieve data from the database. It is also the most complicated
statement, and the next two chapters of this book are devoted to exploring its
many uses.

Standard SQL
SQL and the relational model were invented and developed at IBM in the
early and middle 1970s. Once IBM proved that it was possible to implement
practical relational databases and that SQL was a usable language for manip-
ulating them, other vendors began to provide similar products for non-IBM
computers.
For reasons of performance or competitive advantage, or to take advantage
of local hardware or software features, each of these SQL implementations
differed in small ways from each other and from the IBM version of the lan-
guage. To ensure that the differences remained small, a standards committee
was formed in the early 1980s.

1-14 Introducing Database Features and SQL

 Structured Query Language

Committee X3H2, sponsored by the American National Standards Institute

(ANSI), issued the SQL1 standard in 1986. This standard defines a core set of
SQL features and the syntax of statements such as SELECT.

Informix SQL and ANSI SQL

The SQL version that is supported by Informix products is highly compatible
with standard SQL (it is also compatible with the IBM version of the lan-
guage). However, it does contain extensions to the standard; that is, extra
options or features for certain statements, and looser rules for others. Most of
the differences occur in the statements that are not in everyday use. For exam-
ple, few differences occur in the SELECT statement, which accounts for 90
percent of the SQL use for a typical person.
However, the extensions do exist, and that creates a conflict. Thousands of
Informix customers have embedded Informix-style SQL in programs and
stored queries. They rely on Informix to keep its language the same. Other
customers require the ability to use databases in a way that conforms exactly
to the ANSI standard. They rely on Informix to change its language to
conform.
Informix resolved the conflict with the following compromise:
• The Informix version of SQL, with its extensions to the standard, is avail-
able by default.
• You can ask any Informix SQL language processor to check your use of
SQL and post a warning flag whenever you use an Informix extension.

This resolution is fair but makes the SQL documentation more complicated.
Wherever a difference exists between Informix and ANSI SQL, the Informix
Guide to SQL: Syntax describes both versions. Because you probably intend to
use only one version, you will have to ignore the version you do not need.

ANSI-Compliant Databases
You can designate a database as ANSI-compliant by using the MODE ANSI
keywords when you create it. Within such a database, certain characteristics
of the ANSI standard apply. For example, all actions that modify data auto-
matically take place within a transaction, which means that the changes are
made in their entirety or not at all. Differences in the behavior of ANSI-
compliant databases are noted where appropriate in the Informix Guide to
SQL: Syntax.

Introducing Database Features and SQL 1-15

The Database Software

NLS Databases
The Version 6.0 Informix database servers provide Native Language Support
(NLS). This allows you to work in various European or Latin American lan-
guages other than English and conform to the customs of a specific locale
(various money and date formats, foreign characters in data or identifier
names, different collation order, and so on). You can activate the NLS func-
tionality by setting the DBNLS environment variable as well as other
NLS-specific environment variables.

Chapter 4 of the Informix Guide to SQL: Reference describes how to set the NLS
environment variables. For additional information on NLS databases, see
Chapter 1 in the Informix Guide to SQL: Reference.

The Database Software

You access your database through two layers of sophisticated software. The
top layer, or application, sends commands or queries to the database server.
The application calls on the bottom layer, or database server, and gets back
information. You command both layers using SQL.

The Applications
A database application, or simply application, is a program that uses the
database. It does so by calling on the database server. At its simplest, the
application sends SQL commands to the database server, and the database
server sends rows of data back to the application. Then the application
displays the rows to you, its user.
Alternatively, you command the application to add new data to the database.
It incorporates the new data as part of an SQL command to insert a row and
passes this command to the database server for execution.
Several types of applications exist. Some allow you to access the database
interactively using SQL; others present the stored data in different forms
related to its use.

The Database Server

The database server is the program that manages the contents of the database
as they are stored on disk. The database server knows how tables, rows, and
columns are actually organized in physical computer storage. The database
server also interprets and executes all SQL commands.

1-16 Introducing Database Features and SQL

 The Database Software

Interactive SQL
To carry out the examples in this book, and to experiment with SQL and
database design for yourself, you need a program that lets you execute SQL
statements interactively. DB-Access and INFORMIX-SQL are two such pro-
grams. They assist you in composing SQL statements; then they pass your
SQL to the database server for execution and display the results to you.

Alternatively, you may use a Wingz spreadsheet on a workstation or personal

computer. Wingz, with its optional DataLink extension, can talk to a database
server. Using the Wingz language, HyperScript, you can write SQL state-
ments and have rows of data returned to your spreadsheet for display.

Reports and Forms

After you perfect a query to return precisely the data in which you are inter-
ested, you need to format the data for display as a report or form on the ter-
minal screen. ACE is the report generator for INFORMIX-SQL. You provide to
ACE a SELECT statement that returns the rows of data, and a report specifica-
tion stating how the pages of the report are to be laid out. ACE compiles this
information into a program that you can run whenever you want to produce
that report.
PERFORM is the module of INFORMIX-SQL that generates interactive screen
forms. You prepare a form specification that relates display fields on the
screen to columns within tables in the database. PERFORM compiles this
specification into a program that you can run at any time. When run, the form
program interrogates the database to display a row or rows of data on the
screen in the format you specified. You can arrange the form so that the user
can type in sample values and have matching rows returned from the
database (the query-by-example feature).
For more information on these products, refer to the manuals that come with
them.

General Programming
You can write programs that incorporate SQL statements and exchange data
with the database server. That is, you can write a program to retrieve data
from the database and format it however you choose. You also can write pro-
grams that take data from any source in any format, prepare it, and insert it
into the database.

Introducing Database Features and SQL 1-17

Summary

The most convenient programming language for this kind of work is

INFORMIX-4GL, a nonprocedural language designed expressly for writing
database applications. However, you can also communicate with an Informix
database server from programs that contain embedded SQL written in C,
COBOL, Ada, and FORTRAN.

You also can write programs called stored procedures to work with database
data and objects. The stored procedures that you write are stored directly in
a database within tables. You can then execute a stored procedure from
DB-Access or an SQL API.

Chapter 5, “Programming with SQL,” and Chapter 6, “Modifying Data

Through SQL Programs,” present an overview of how SQL is used in
rograms.

Applications and Database Servers

Every program that uses data from a database operates the same way.
Regardless of whether it is a packaged program such as INFORMIX-SQL, a
report program compiled by ACE, or a program that you wrote using
INFORMIX-4GL or an SQL API, you find the same two layers in every case:

1. An application that interacts with the user, prepares and formats data,
and sets up SQL statements.
2. A database server that manages the database and interprets the SQL.
All the applications converge on the database server, and only the database
server manipulates the database files on disk.

Summary
A database contains a collection of related information but differs in a
fundamental way from other methods of storing data. The database contains
not only the data but also a data model that defines each data item and
specifies its meaning with respect to the other items and to the real world.
A database can be used and even modified by several computer users
working concurrently. Different users can be given different views of the con-
tents of a database, and their access to those contents can be restricted in
several ways.

1-18 Introducing Database Features and SQL

 Summary

A database can be crucially important to the success of an organization and

can require central administration and monitoring. The INFORMIX-OnLine
Dynamic Server database server caters to the needs of essential applications;
both OnLine and INFORMIX-SE support smaller databases for private or
group use.
You manipulate and query a database using SQL, which was pioneered by
IBM and standardized by ANSI. You can take advantage of the Informix
extensions to the ANSI-defined language, and your Informix tools also make
it possible to maintain strict compliance with ANSI standards.
Two layers of software mediate all your work with databases. The bottom
layer is always a database server that executes SQL statements and manages
the data on disk and in computer memory. The top layer is one of many appli-
cations, some from Informix and some written by you or written by other
vendors or your colleagues.

Introducing Database Features and SQL 1-19

Summary

1-20 Introducing Database Features and SQL

 Chapter

Composing Simple
SELECT Statements
2
Chapter Overview 3
Introducing the SELECT Statement 4
Some Basic Concepts 4
Privileges 5
Relational Operations 5
Selection and Projection 5
Joining 8
The Forms of SELECT 10
Special Data Types 11
Single-Table SELECT Statements 11
Selecting All Columns and Rows 12
Using the Asterisk 12
Reordering the Columns 13
Sorting the Rows 13
Selecting Specific Columns 18
ORDER BY and Native Language Support 25
Selecting Substrings 27
Using the WHERE Clause 28
Creating a Comparison Condition 29
Using Variable Text Searches 37
Using Exact Text Comparisons 37
Using a Single-Character Wildcard 38
MATCHES and Native Language Support 42
Comparing for Special Characters 43
Expressions and Derived Values 46
Arithmetic Expressions 46
Sorting on Derived Columns 52
 Using Functions in SELECT Statements 53
Aggregate Functions 53
Time Functions 56
Other Functions and Keywords 63
Using Stored Procedures in SELECT Statements 67
Multiple-Table SELECT Statements 69
Creating a Cartesian Product 69
Creating a Join 71
Equi-Join 71
Natural Join 75
Multiple-Table Join 77
Some Query Shortcuts 79
Using Aliases 79
The INTO TEMP Clause 83
Summary 84

2-2 Composing Simple SELECT Statements

Chapter Overview
SELECT is the most important and the most complex SQL statement. It is used,
along with the SQL statements INSERT, UPDATE, and DELETE, to manipulate
data. You can use the SELECT statement in the following ways:
• By itself to retrieve data from a database
• As part of an INSERT statement to produce new rows
• As part of an UPDATE statement to update information
The SELECT statement is the primary way to query on information in a
database. It is your key to retrieving data in a program, report, screen form,
or spreadsheet.
This chapter shows how you can use the SELECT statement to query on and
retrieve data in a variety of ways from a relational database. It discusses how
to tailor your statements to select columns or rows of information from one
or more tables, how to include expressions and functions in SELECT state-
ments, and how to create various join conditions between relational database
tables.
This chapter introduces the basic methods for retrieving data from a
relational database. More complex uses of SELECT statements, such as sub-
queries, outer joins, and unions, are discussed in Chapter 3 of this manual.
The syntax and usage for the SELECT statement are described in detail in
Chapter 1 of the Informix Guide to SQL: Syntax.
Most of the examples in this chapter come from the nine tables in the stores6
demonstration database, which is installed with the software for your Infor-
mix SQL API. In the interest of brevity, the examples may show only part of
the data that is retrieved for each SELECT statement. See Appendix A in the
Informix Guide to SQL: Reference for information on the structure and contents
of the stores6 database. For emphasis, keywords are shown in uppercase
letters in the examples, although SQL is not case sensitive.

Composing Simple SELECT Statements 2-3

Introducing the SELECT Statement

Introducing the SELECT Statement

The SELECT statement is constructed of clauses that let you look at data in a
relational database. These clauses let you select columns and rows from one
or more database tables or views, specify one or more conditions, order and
summarize the data, and put the selected data in a temporary table.
This chapter shows how to use five SELECT statement clauses. You must
include these clauses in a SELECT statement in the following order:
1. SELECT clause
2. FROM clause
3. WHERE clause
4. ORDER BY clause
5. INTO TEMP clause
Only the SELECT and FROM clauses are required. These two clauses form the
basis for every database query because they specify the tables and columns
to be retrieved.
• Add a WHERE clause to select specific rows or create a join condition.
• Add an ORDER BY clause to change the order in which data is produced.
• Add an INTO TEMP clause to save the results as a table for further queries.
Two additional SELECT statement clauses, GROUP BY and HAVING, enable
you to perform more complex data retrieval. They are introduced in
Chapter 3, “Composing Advanced SELECT Statements.” Another clause,
INTO, is used to specify the program or host variable to receive data from a
SELECT statement in the INFORMIX-4GL and SQL APIs. Complete syntax and
rules for using the SELECT statement are shown in Chapter 1 of the Informix
Guide to SQL: Syntax.

Some Basic Concepts

The SELECT statement, unlike INSERT, UPDATE, and DELETE statements,
does not modify the data in a database. It is used simply to query on the data.
Whereas only one user at a time can modify data, multiple users can query
on or select the data concurrently. The statements that modify data are dis-
cussed in Chapter 4, “Modifying Data.” The INSERT, UPDATE, and DELETE
statements appear in Chapter 1 of the Informix Guide to SQL: Syntax.

2-4 Composing Simple SELECT Statements

 Introducing the SELECT Statement

In a relational database, a column is a data element that contains a particular

type of information that occurs in every row in the table. A row is a group of
related items of information about a single entity across all columns in a
database table.
You can select columns and rows from a database table; from a system catalog
table, a file that contains information on the database; or from a view, which is
a virtual table created to contain a customized set of data. System catalog
tables are shown in Chapter 2 of the Informix Guide to SQL: Reference. Views
are discussed in Chapter 1 of the Informix Guide to SQL: Syntax.

Privileges
Before you can query on data, you must have the Connect privilege to the
database and the Select privilege to the tables in it. These privileges normally
are granted to all users as a matter of course. Database access privileges are
discussed in Chapter 11, “Granting and Limiting Access to Your Database,”
of this manual and in the GRANT and REVOKE statements in Chapter 1 of the
Informix Guide to SQL: Syntax.

Relational Operations
A relational operation involves manipulating one or more tables, or relations, to
result in another table. The three kinds of relational operation are selection,
projection, and join. This chapter includes examples of selection, projection,
and simple joining.

Selection and Projection

In relational terminology, selection is defined as taking the horizontal subset of
rows of a single table that satisfies a particular condition. This kind of SELECT
statement returns some of the rows and all of the columns in a table. Selection
is implemented through the WHERE clause of a SELECT statement, as shown
in Query 2-1:

SELECT * FROM customer

WHERE state = 'NJ'

Query 2-1

Composing Simple SELECT Statements 2-5

Introducing the SELECT Statement

Query Result 2-1 contains the same number of columns as the customer table,
but only a subset of its rows. Because the data in the selected columns does
not fit on one line of the DB-Access or INFORMIX-SQL Interactive Schema
Editor (ISED) screen, the data is displayed vertically instead of horizontally.

customer_num 119
fname Bob
lname Shorter
company The Triathletes Club
address1 2405 Kings Highway
address2
city Cherry Hill
state NJ
zipcode 08002
phone 609-663-6079

customer_num 122
fname Cathy
lname O’Brian
company The Sporting Life
address1 543 Nassau Street
address2
city Princeton
state NJ
zipcode 08540
phone 609-342-0054

Query Result 2-1

In relational terminology, projection is defined as taking a vertical subset from

the columns of a single table that retains the unique rows. This kind of
SELECT statement returns some of the columns and all of the rows in a table.

Projection is implemented through the select list in the SELECT clause of a

SELECT statement, as shown in Query 2-2:

SELECT UNIQUE city, state, zipcode

FROM customer

Query 2-2

2-6 Composing Simple SELECT Statements

 Introducing the SELECT Statement

Query Result 2-2 contains the same number of rows as the customer table,
but it projects only a subset of the table’s columns:

city state zipcode

Bartlesville OK 74006
Blue Island NY 60406
Brighton MA 02135
Cherry Hill NJ 08002
Denver CO 80219
Jacksonville FL 32256
Los Altos CA 94022
Menlo Park CA 94025
Mountain View CA 94040
Mountain View CA 94063
Oakland CA 94609
Palo Alto CA 94303
Palo Alto CA 94304
Phoenix AZ 85008
Phoenix AZ 85016
Princeton NJ 08540
Redwood City CA 94026
Redwood City CA 94062
Redwood City CA 94063
San Francisco CA 94117
Sunnyvale CA 94085
Sunnyvale CA 94086
Wilmington DE 19898

Query Result 2-2

The most common kind of SELECT statement uses both selection and
projection. A query of this kind, shown in Query 2-3, returns some of the
rows and some of the columns in a table:

SELECT UNIQUE city, state, zipcode

FROM customer
WHERE state = 'NJ'

Query 2-3

Composing Simple SELECT Statements 2-7

Introducing the SELECT Statement

Query Result 2-3 contains a subset of the rows and a subset of the columns in
the customer table:

city state zipcode

Cherry Hill NJ 08002

Princeton NJ 08540

Query Result 2-3

Joining
A join occurs when two or more tables are connected by one or more columns
in common, creating a new table of results. Query 2-4 uses a subset of the
items and stock tables to illustrate the concept of a join, as shown in Figure
2-1.

2-8 Composing Simple SELECT Statements

 Introducing the SELECT Statement

SELECT unique item_num, order_num, stock.stock_num, description

FROM items, stock
WHERE items.stock_num = stock.stock_num

items Table (example) stock Table (example)

item_num order_num stock_num stock_num manu_code description

1 1001 1 1 HRO baseball gloves

1 1002 4 1 HSK baseball gloves

2 1002 3 2 HRO baseball

3 1003 5 4 HSK football

1 1005 5 5 NRG tennis racquet

item_num order_num stock_num description

1 1001 1 baseball gloves

1 1002 4 football

3 1003 5 tennis racquet

1 1005 5 tennis racquet

Figure 2-1 Example of a join between two tables

Query 2-4 joins the customer and state tables:

SELECT UNIQUE city, state, zipcode, sname

FROM customer, state
WHERE customer.state = state.code

Query 2-4

Composing Simple SELECT Statements 2-9

Introducing the SELECT Statement

Query Result 2-4 is composed of specified rows and columns from both the
customer and state tables:

city state zipcode sname

Bartlesville OK 74006 Oklahoma

Blue Island NY 60406 New York
Brighton MA 02135 Massachusetts
Cherry Hill NJ 08002 New Jersey
Denver CO 80219 Colorado
Jacksonville FL 32256 Florida
Los Altos CA 94022 California
Menlo Park CA 94025 California
Mountain View CA 94040 California
Mountain View CA 94063 California
Oakland CA 94609 California
Palo Alto CA 94303 California
Palo Alto CA 94304 California
Phoenix AZ 85008 Arizona
Phoenix AZ 85016 Arizona
Princeton NJ 08540 New Jersey
Redwood City CA 94026 California
Redwood City CA 94062 California
Redwood City CA 94063 California
San Francisco CA 94117 California
Sunnyvale CA 94085 California
Sunnyvale CA 94086 California
Wilmington DE 19898 Delaware

Query Result 2-4

The Forms of SELECT

Although the syntax remains the same across all Informix products, the form
of a SELECT statement and the location and formatting of the resulting output
depends on the application. The examples in this chapter and in Chapter 3,
“Composing Advanced SELECT Statements,” display the SELECT statements
and their output as they appear when you use the interactive Query-
language option in DB-Access or INFORMIX-SQL.
You also can use SELECT statements to query on data noninteractively
through INFORMIX-SQL ACE reports, you can embed them in a language
such as INFORMIX-ESQL/C (where they are treated as executable code), or
you can incorporate them in INFORMIX-4GL as part of its fourth-generation
language.

2-10 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Special Data Types

The examples in this chapter use the INFORMIX-OnLine Dynamic Server
database server, which enables database applications to include the data
types VARCHAR, TEXT, and BYTE. These data types are not available to
applications that run on INFORMIX-SE.
With the DB-Access or INFORMIX-SQL Interactive Editor, when you issue a
SELECT statement that includes one of these three data types, the results of
the query are displayed differently.
• If you execute a query on a VARCHAR column, the entire VARCHAR value
is displayed, just as CHARACTER values are displayed.
• If you select a TEXT column, the contents of the TEXT column are
displayed and you can scroll through them.
• If you query on a BYTE column, the words <BYTE value> are displayed
instead of the actual value.
Differences specific to VARCHAR, TEXT, and BYTE are noted as appropriate
throughout this chapter.
When Native Language Support (NLS) is enabled, you can issue a SELECT
statement that queries on NCHAR or NVARCHAR columns instead of their
CHAR or VARCHAR counterparts. The NCHAR and NVARCHAR data types
allow foreign characters in character data.
For additional information on these and other data types, see Chapter 9,
“Implementing Your Data Model,” in this manual, and Chapter 3 in the
Informix Guide to SQL: Reference.

Single-Table SELECT Statements

You can query a single table in a database in many ways. You can tailor a
SELECT statement to perform the following actions:
• Retrieve all or specific columns
• Retrieve all or specific rows
• Perform computations or other functions on the retrieved data
• Order the data in various ways

Composing Simple SELECT Statements 2-11

Single-Table SELECT Statements

Selecting All Columns and Rows

The most basic SELECT statement contains only the two required clauses,
SELECT and FROM.

Using the Asterisk

Query 2-5a specifies all the columns in the manufact table in a select list. A
select list is a list of the column names or expressions that you want to project
from a table.

SELECT manu_code, manu_name, lead_time

FROM manufact

Query 2-5a

Query 2-5b uses the wildcard * (asterisk), which is shorthand for the select list.
The * represents the names of all the columns in the table. You can use the *
when you want all the columns, in their defined order.

SELECT * FROM manufact

Query 2-5b

Query 2-5a and Query 2-5b are equivalent and display the same results; that
is, a list of every column and row in the manufact table. Query Result 2-5
shows the results as they would appear on a DB-Access or INFORMIX-SQL
Interactive Schema Editor (ISED) screen.

manu_code manu_name lead_time

SMT Smith 3
ANZ Anza 5
NRG Norge 7
HSK Husky 5
HRO Hero 4
SHM Shimara 30
KAR Karsten 21
NKL Nikolus 8
PRC ProCycle 9

Query Result 2-5

2-12 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Reordering the Columns

Query 2-6 shows how you can change the order in which the columns are
listed by changing their order in your select list:

SELECT manu_name, manu_code, lead_time

FROM manufact

Query 2-6

Query Result 2-6 includes the same columns as the previous one, but because
the columns are specified in a different order, the display is also different:

manu_name manu_code lead_time

Smith SMT 3
Anza ANZ 5
Norge NRG 7
Husky HSK 5
Hero HRO 4
Shimara SHM 30
Karsten KAR 21
Nikolus NKL 8
ProCycle PRC 9

Query Result 2-6

Sorting the Rows

You can direct the system to sort the data in a specific order by adding an
ORDER BY clause to your SELECT statement. The columns you want to use in
the ORDER BY clause must be included in the select list either explicitly or
implicitly.
An explicit select list, shown in Query 2-7a, includes all the column names
that you want to retrieve:

SELECT manu_code, manu_name, lead_time

FROM manufact
ORDER BY lead_time

Query 2-7a

Composing Simple SELECT Statements 2-13

Single-Table SELECT Statements

An implicit select list uses the * symbol, as shown in Query 2-7b:

SELECT * FROM manufact

ORDER BY lead_time

Query 2-7b

Query 2-7a and Query 2-7b produce the same display. Query Result 2-7
shows a list of every column and row in the manufact table, in order of
lead_time:

manu_code manu_name lead_time

SMT Smith 3
HRO Hero 4
HSK Husky 5
ANZ Anza 5
NRG Norge 7
NKL Nikolus 8
PRC ProCycle 9
KAR Karsten 21
SHM Shimara 30

Query Result 2-7

Ascending Order
The retrieved data is sorted and displayed, by default, in ascending order.
Ascending order is uppercase A to lowercase z for CHARACTER data types,
and lowest to highest value for numeric data types. DATE and DATETIME
type data are sorted from earliest to latest, and INTERVAL data is ordered
from shortest to longest span of time.

Descending Order
Descending order is the opposite of ascending order, from lowercase z to
uppercase A for character types and highest to lowest for numeric data types.
DATE and DATETIME data are sorted from latest to earliest, and INTERVAL
data is ordered from longest to shortest span of time. Query 2-8 shows an
example of descending order:

SELECT * FROM manufact

ORDER BY lead_time DESC

Query 2-8

2-14 Composing Simple SELECT Statements

 Single-Table SELECT Statements

The keyword DESC following a column name causes the retrieved data to be
sorted in descending order, as shown in Query Result 2-8:

manu_code manu_name lead_time

SHM Shimara 30
KAR Karsten 21
PRC ProCycle 9
NKL Nikolus 8
NRG Norge 7
HSK Husky 5
ANZ Anza 5
HRO Hero 4
SMT Smith 3

Query Result 2-8

You can specify any column (except TEXT or BYTE) in the ORDER BY clause,
and the database server sorts the data based on the values in that column.

Sorting on Multiple Columns

You can also ORDER BY two or more columns, creating a nested sort. The
default is still ascending, and the column listed first in the ORDER BY clause
takes precedence.
The following two queries and corresponding query results show nested
sorts. The order in which selected data is displayed is modified by changing
the order of the two columns named in the ORDER BY clause.

SELECT * FROM stock

ORDER BY manu_code, unit_price

Query 2-9

Composing Simple SELECT Statements 2-15

Single-Table SELECT Statements

In Query Result 2-9, the manu_code column data appears in alphabetical

order and, within each set of rows with the same manu_code (for example,
ANZ, HRO), the unit_price is listed in ascending order:

stock_num manu_code description unit_price unit unit_descr

5 ANZ tennis racquet $19.80 each each

9 ANZ volleyball net $20.00 each each
6 ANZ tennis ball $48.00 case 24 cans/case
313 ANZ swim cap $60.00 box 12/box
201 ANZ golf shoes $75.00 each each
310 ANZ kick board $84.00 case 12/case
301 ANZ running shoes $95.00 each each
304 ANZ watch $170.00 box 10/box
110 ANZ helmet $244.00 case 4/case
205 ANZ 3 golf balls $312.00 case 24/case
8 ANZ volleyball $840.00 case 24/case
302 HRO ice pack $4.50 each each
309 HRO ear drops $40.00 case 20/case
.
.
.
113 SHM 18-spd, assmbld $685.90 each each
5 SMT tennis racquet $25.00 each each
6 SMT tennis ball $36.00 case 24 cans/case
1 SMT baseball gloves $450.00 case 10 gloves/case

Query Result 2-9

Query 2-10 shows the reversed order of the columns in the ORDER BY clause:

SELECT * FROM stock

ORDER BY unit_price, manu_code

Query 2-10

2-16 Composing Simple SELECT Statements

 Single-Table SELECT Statements

In Query Result 2-10, the data appears in ascending order of unit_price and,
where two or more rows have the same unit_price (for example, $20.00,
$48.00, $312.00), the manu_code is in alphabetical order:

stock_num manu_code description unit_price unit unit_descr

302 HRO ice pack $4.50 each each

302 KAR ice pack $5.00 each each
5 ANZ tennis racquet $19.80 each each
9 ANZ volleyball net $20.00 each each
103 PRC frnt derailleur $20.00 each each
106 PRC bicycle stem $23.00 each each
5 SMT tennis racquet $25.00 each each
.
.
.
301 HRO running shoes $42.50 each each
204 KAR putter $45.00 each each
108 SHM crankset $45.00 each each
6 ANZ tennis ball $48.00 case 24 cans/case
305 HRO first-aid kit $48.00 case 4/case
303 PRC socks $48.00 box 24 pairs/box
311 SHM water gloves $48.00 box 4 pairs/box
.
.
.
110 HSK helmet $308.00 case 4/case
205 ANZ 3 golf balls $312.00 case 24/case
205 HRO 3 golf balls $312.00 case 24/case
205 NKL 3 golf balls $312.00 case 24/case
1 SMT baseball gloves $450.00 case 10 gloves/case
4 HRO football $480.00 case 24/case
102 PRC bicycle brakes $480.00 case 4 sets/case
111 SHM 10-spd, assmbld $499.99 each each
112 SHM 12-spd, assmbld $549.00 each each
7 HRO basketball $600.00 case 24/case
203 NKL irons/wedge $670.00 case 2 sets/case
113 SHM 18-spd, assmbld $685.90 each each
1 HSK baseball gloves $800.00 case 10 gloves/case
8 ANZ volleyball $840.00 case 24/case
4 HSK football $960.00 case 24/case

Query Result 2-10

Composing Simple SELECT Statements 2-17

Single-Table SELECT Statements

The order of the columns in the ORDER BY clause is important, and so is the
position of the DESC keyword. Although the statements in Query 2-11 con-
tain the same components in the ORDER BY clause, each produces a different
result (not shown):

SELECT * FROM stock

ORDER BY manu_code, unit_price DESC

SELECT * FROM stock

ORDER BY unit_price, manu_code DESC

SELECT * FROM stock

ORDER BY manu_code DESC, unit_price

SELECT * FROM stock

ORDER BY unit_price DESC, manu_code

Query 2-11

Selecting Specific Columns

The previous section showed how to select and order all data from a table.
However, often all you want to see is the data in one or more specific col-
umns. Again, the formula is to use the SELECT and FROM clauses, specify the
columns and table, and perhaps order the data in ascending or descending
order with an ORDER BY clause.
If you want to find all the customer numbers in the orders table, you would
use a statement such as the one in Query 2-12:

SELECT customer_num FROM orders

Query 2-12

2-18 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Query Result 2-12 shows how the statement simply selects all data in the
customer_num column in the orders table and lists the customer numbers on
all the orders, including duplicates:

customer_num

101
104
104
104
104
106
106
110
110
111
112
115
116
117
117
119
120
121
122
123
124
126
127

Query Result 2-12

The output includes several duplicates because some customers have placed
more than one order. Sometimes you want to see duplicate rows in a projec-
tion. At other times, you want to see only the distinct values, not how often
each value appears.
You can cause duplicate rows to be suppressed by including the keyword
DISTINCT or its synonym UNIQUE at the start of the select list, as shown in
Query 2-13:

SELECT DISTINCT customer_num FROM orders

SELECT UNIQUE customer_num FROM orders

Query 2-13

Composing Simple SELECT Statements 2-19

Single-Table SELECT Statements

Query 2-13 limits the display to show each customer number in the orders
table only once, producing a more readable list, as shown in Query Result
2-13:

customer_num

101
104
106
110
111
112
115
116
117
119
120
121
122
123
124
126
127

Query Result 2-13

Suppose you are handling a customer call, and you want to locate purchase
order number DM354331. You decide to list all the purchase order numbers in
the orders table, using a statement such as the one in Query 2-14:

SELECT po_num FROM orders

Query 2-14

2-20 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Query Result 2-14 shows how the statement retrieves data in the po_num
column in the orders table:

po_num

B77836
9270
B77890
8006
2865
Q13557
278693
LZ230
4745
429Q
B77897
278701
B77930
8052
MA003
PC6782
DM354331
S22942
Z55709
W2286
C3288
W9925
KF2961

Query Result 2-14

Composing Simple SELECT Statements 2-21

Single-Table SELECT Statements

However, the list is not in a very useful order. You can add an ORDER BY
clause to sort the column data in ascending order and make it easier to find
that particular po_num, as shown in Query 2-15:

SELECT po_num FROM orders

ORDER BY po_num

Query 2-15

po_num

278693
278701
2865
429Q
4745
8006
8052
9270
B77836
B77890
B77897
B77930
C3288
DM354331
KF2961
LZ230
MA003
PC6782
Q13557
S22942
W2286
W9925
Z55709

Query Result 2-15

To select multiple columns from a table, list them in the select list in the
SELECT clause. Query 2-16 shows the order in which the columns are selected
is the order in which they are produced, from left to right:

SELECT paid_date, ship_date, order_date,

customer_num, order_num, po_num
FROM orders
ORDER BY paid_date, order_date, customer_num

Query 2-16

2-22 Composing Simple SELECT Statements

 Single-Table SELECT Statements

As shown in “Sorting on Multiple Columns” on page 2-15, you can use the
ORDER BY clause to sort the data in ascending or descending order and
perform nested sorts. Query Result 2-16 shows descending order:

paid_date ship_date order_date customer_num order_num po_num

05/30/1993 05/22/1993 106 1004 8006

05/30/1993 112 1006 Q13557
06/05/1993 05/31/1993 117 1007 278693
06/29/1993 06/18/1993 117 1012 278701
07/12/1993 06/29/1993 119 1016 PC6782
07/13/1993 07/09/1993 120 1017 DM354331
06/03/1993 05/26/1993 05/21/1993 101 1002 9270
06/14/1993 05/23/1993 05/22/1993 104 1003 B77890
06/21/1993 06/09/1993 05/24/1993 116 1005 2865
07/10/1993 07/03/1993 06/25/1993 106 1014 8052
07/21/1993 07/06/1993 06/07/1993 110 1008 LZ230
07/22/1993 06/01/1993 05/20/1993 104 1001 B77836
07/31/1993 07/10/1993 06/22/1993 104 1013 B77930
08/06/1993 07/13/1993 07/10/1993 121 1018 S22942
08/06/1993 07/16/1993 07/11/1993 122 1019 Z55709
08/21/1993 06/21/1993 06/14/1993 111 1009 4745
08/22/1993 06/29/1993 06/17/1993 115 1010 429Q
08/22/1993 07/25/1993 07/23/1993 124 1021 C3288
08/22/1993 07/30/1993 07/24/1993 127 1023 KF2961
08/29/1993 07/03/1993 06/18/1993 104 1011 B77897
08/31/1993 07/16/1993 06/27/1993 110 1015 MA003
09/02/1993 07/30/1993 07/24/1993 126 1022 W9925
09/20/1993 07/16/1993 07/11/1993 123 1020 W2286

Query Result 2-16

Composing Simple SELECT Statements 2-23

Single-Table SELECT Statements

When you use SELECT and ORDER BY on several columns in a table, you
might find it helpful to use integers to refer to the position of the columns in
the ORDER BY clause.The statements in Query 2-17 retrieve and display the
same data, shown in Query Result 2-17.

SELECT customer_num, order_num, po_num, order_date

FROM orders
ORDER BY 4, 1

SELECT customer_num, order_num, po_num, order_date

FROM orders
ORDER BY order_date, customer_num

Query 2-17

customer_num order_num po_num order_date

104 1001 B77836 05/20/1993

101 1002 9270 05/21/1993
104 1003 B77890 05/22/1993
106 1004 8006 05/22/1993
116 1005 2865 05/24/1993
112 1006 Q13557 05/30/1993
117 1007 278693 05/31/1993
110 1008 LZ230 06/07/1993
111 1009 4745 06/14/1993
115 1010 429Q 06/17/1993
104 1011 B77897 06/18/1993
117 1012 278701 06/18/1993
104 1013 B77930 06/22/1993
106 1014 8052 06/25/1993
110 1015 MA003 06/27/1993
119 1016 PC6782 06/29/1993
120 1017 DM354331 07/09/1993
121 1018 S22942 07/10/1993
122 1019 Z55709 07/11/1993
123 1020 W2286 07/11/1993
124 1021 C3288 07/23/1993
126 1022 W9925 07/24/1993
127 1023 KF2961 07/24/1993

Query Result 2-17

2-24 Composing Simple SELECT Statements

 Single-Table SELECT Statements

You can include the DESC keyword in the ORDER BY clause when you assign
integers to column names, as shown in Query 2-18:

SELECT customer_num, order_num, po_num, order_date

FROM orders
ORDER BY 4 DESC, 1

Query 2-18

In this case, data is first sorted in descending order by order_date and in

ascending order by customer_num.

ORDER BY and Native Language Support

You can set the DBNLS environment variable to activate Native Language
Support (NLS) functionality. You can then specify a language environment
(locale) by setting the LANG environment variable and can set the
LC_COLLATE X/Open category to further define the collation order.

Query 2-18 uses a SELECT statement with an ORDER BY clause on a column

containing data of type NCHAR (the NLS equivalent of CHAR). This NLS table,
abonnés, is provided separately from the demonstration database; this
SELECT statement example is provided as the sel_nls1.sql command file with
NLS-enabled DB-Access.

SELECT numéro,nom,prénom
FROM abonnés
ORDER BY nom;

Query 2-19

Composing Simple SELECT Statements 2-25

Single-Table SELECT Statements

The collation order for the results of this query may vary, depending on
whether NLS is activated and the settings for the LANG and LC_COLLATE
environment variables, as well as how uppercase versus lowercase letters are
handled by the particular system. Query Result 2-19a and Query Result 2-19b
show two sample sets of output:

numéro nom prénom

13612 Azevedo Edouardo Freire

13606 Dupré Michèle Françoise
13607 Hammer Gerhard
13602 Hämmerle Greta
13604 LaForêt Jean-Noël
13610 LeMaître Héloïse
13613 Llanero Gloria Dolores
13603 Montaña José Antonio
13611 Oatfield Emily
13609 Tiramisù Paolo Alfredo
13600 da Sousa João Lourenço Antunes
13615 di Girolamo Giuseppe
13601 Ålesund Sverre
13608 Étaix Émile
13605 Ötker Hans-Jürgen
13614 Øverst Per-Anders

Query Result 2-19a

2-26 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Query Result 2-19a follows the standard ASCII sorting order, which ranks
capital letters before lowercase letters and moves the names that start with a
capitalized foreign character (Ålesund, Étaix, Ötker, and Øverst) to the end
of the list.

numéro nom prénom

13601 Ålesund Sverre

13612 Azevedo Edouardo Freire
13600 da Sousa João Lourenço Antunes
13615 di Girolamo Giuseppe
13606 Dupré Michèle Françoise
13608 Étaix Émile
13607 Hammer Gerhard
13602 Hämmerle Greta
13604 LaForêt Jean-Noël
13610 LeMaître Héloïse
13613 Llanero Gloria Dolores
13603 Montaña José Antonio
13611 Oatfield Emily
13605 Ötker Hans-Jürgen
13614 Øverst Per-Anders
13609 Tiramisù Paolo Alfredo

Query Result 2-19b

Query Result 2-19b indicates that with the appropriate NLS environment
variable settings, names starting with foreign characters (Ålesund, Étaix,
Ötker, and Øverst) are collated differently than they are in the standard ASCII
order so they are integrated in the list. It does not distinguish between
uppercase and lowercase letters.

Selecting Substrings
You can select part of the value of a CHARACTER column by including a
substring in the select list. Suppose your marketing department is planning a
mailing to your customers and wants a rough idea of their geographical
distribution based on zip codes. You could write a query similar to Query
2-20:

SELECT zipcode[1,3], customer_num

FROM customer
ORDER BY zipcode

Query 2-20

Composing Simple SELECT Statements 2-27

Single-Table SELECT Statements

Query 2-20 uses a substring to select the first three characters of the zipcode
column (which identify the state) and the full customer_num, and lists them
in ascending order by zip code, as shown in Query Result 2-20:

zipcode customer_num

021 125
080 119
085 122
198 121
322 123
604 127
740 124
802 126
850 128
850 120
940 105
940 112
940 113
940 115
940 104
940 116
940 110
940 114
940 106
940 108
940 117
940 111
940 101
940 109
941 102
943 103
943 107
946 118

Query Result 2-20

Using the WHERE Clause

Perhaps you want to see only those orders placed by a particular customer or
the calls entered by a particular customer service representative. You can add
a WHERE clause to a SELECT statement to retrieve these specific rows from a
table.
You can use the WHERE clause to set up a comparison condition or a join
condition. This section demonstrates only the first use. Join conditions are
described in a later section and in the next chapter.

2-28 Composing Simple SELECT Statements

 Single-Table SELECT Statements

The set of rows returned by a SELECT statement is the active set for that
statement. A singleton SELECT statement returns a single row. In
INFORMIX-4GL or an SQL API, the retrieval of multiple rows requires the use
of a cursor. See Chapter 5, “Programming with SQL,” and Chapter 6,
“Modifying Data Through SQL Programs.”

Creating a Comparison Condition

The WHERE clause of a SELECT statement specifies the rows you want to see.
A comparison condition employs specific keywords and operators to define the
search criteria.
For example, you might use one of the keywords BETWEEN, IN, LIKE, or
MATCHES to test for equality, or the keywords IS NULL to test for null values.
You can combine the keyword NOT with any of these keywords to specify the
opposite condition.
Figure 2-2 lists the relational operators you can use in a WHERE clause in place
of a keyword to test for equality.

Operator Operation
= equals
!= or <> does not equal
> greater than
>= greater than or equal to
< less than
<= less than or equal to

Figure 2-2 Relational operators that test for equality

For CHAR expressions, “greater than” means after in ASCII collating order,
where lowercase letters are after uppercase letters, and both are after numer-
als. See the ASCII Character Set chart in Chapter 1 of the Informix Guide to
SQL: Syntax. For DATE and DATETIME expressions, “greater than” means
later in time, and for INTERVAL expressions, it means of longer duration. You
cannot use TEXT or BYTE columns in string expressions, except when you test
for null values.
You can use the preceding keywords and operators in a WHERE clause to
create comparison condition queries that perform the following actions:
• Include values
• Exclude values
• Find a range of values

Composing Simple SELECT Statements 2-29

Single-Table SELECT Statements

• Find a subset of values

• Identify null values
You can also use the preceding keywords and operators in a WHERE clause
to create comparison condition queries, which perform variable text searches
using the following criteria:
• Exact text comparison
• Single-character wildcards
• Restricted single-character wildcards
• Variable-length wildcards
• Subscripting
The following section contains examples that illustrate these types of queries.

Including Rows
Use the relational operator = to include rows in a WHERE clause, as shown in
Query 2-21:

SELECT customer_num, call_code, call_dtime, res_dtime

FROM cust_calls
WHERE user_id = 'maryj'

Query 2-21

Query 2-21 returns the set of rows shown in Query Result 2-21:

customer_num call_code call_dtime res_dtime

106 D 1993-06-12 08:20 1993-06-12 08:25

121 O 1993-07-10 14:05 1993-07-10 14:06
127 I 1993-07-31 14:30

Query Result 2-21

Excluding Rows
Use the relational operators != or <> to exclude rows in a WHERE clause.
Query 2-22 assumes that you are selecting from an ANSI-compliant database;
the statements specify the owner or login name of the creator of the customer
table. This qualifier is not required when the creator of the table is the current

2-30 Composing Simple SELECT Statements

 Single-Table SELECT Statements

user, or when the database is not ANSI-compliant, although it is not incorrect

to include it in either case. For a complete discussion of owner naming, see
Chapter 1 in the Informix Guide to SQL: Syntax.

SELECT customer_num, company, city, state

FROM odin.customer
WHERE state != 'CA'

SELECT customer_num, company, city, state

FROM odin.customer
WHERE state <> 'CA'

Query 2-22

Both statements in Query 2-22 exclude values by specifying that, in the

customer table owned by the user odin, the value in the state column should
not be equal to CA.

customer_num company city state

119 The Triathletes Club Cherry Hill NJ

120 Century Pro Shop Phoenix AZ
121 City Sports Wilmington DE
122 The Sporting Life Princeton NJ
123 Bay Sports Jacksonville FL
124 Putnum’s Putters Bartlesville OK
125 Total Fitness Sports Brighton MA
126 Neelie’s Discount Sp Denver CO
127 Big Blue Bike Shop Blue Island NY
128 Phoenix College Phoenix AZ

Query Result 2-22

Specifying Rows
Query 2-23 shows two ways to specify rows in a WHERE clause:

SELECT catalog_num, stock_num, manu_code, cat_advert

FROM catalog
WHERE catalog_num BETWEEN 10005 AND 10008

SELECT catalog_num, stock_num, manu_code, cat_advert

FROM catalog
WHERE catalog_num >= 10005 AND catalog_num <= 10008

Query 2-23

Composing Simple SELECT Statements 2-31

Single-Table SELECT Statements

Each statement in Query 2-23 specifies a range for catalog_num from 10005
through 10008, inclusive. The first statement uses keywords, and the second
uses relational operators to retrieve the rows shown in Query Result 2-23:

catalog_num 10005
stock_num 3
manu_code HSK
cat_advert High-Technology Design Expands the Sweet Spot

catalog_num 10006
stock_num 3
manu_code SHM
cat_advert Durable Aluminum for High School and Collegia
te Athletes

catalog_num 10007
stock_num 4
manu_code HSK
cat_advert Quality Pigskin with Joe Namath Signature

catalog_num 10008
stock_num 4
manu_code HRO
cat_advert Highest Quality Football for High School
and Collegiate Competitions

Query Result 2-23

Note that although the catalog table includes a column with the BYTE data
type, that column is not included in this SELECT statement because the out-
put would show only the words <BYTE value> by the column name. You
can display TEXT and BYTE values by using the PROGRAM attribute when
using forms in INFORMIX-SQL or INFORMIX-4GL or by writing a 4GL or an
SQL API to do so.

2-32 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Excluding a Range of Rows

In Query 2-24, by using the keywords NOT BETWEEN, the condition excludes
rows that have the character range 94000 through 94999 in the zipcode
column, as shown in Query Result 2-24:

SELECT fname, lname, company, city, state

FROM customer
WHERE zipcode NOT BETWEEN '94000' AND '94999'
ORDER BY state

Query 2-24

fname lname company city state

Fred Jewell Century* Pro Shop Phoenix AZ

Frank Lessor Phoenix University Phoenix AZ
Eileen Neelie Neelie’s Discount Sp Denver CO
Jason Wallack City Sports Wilmington DE
Marvin Hanlon Bay Sports Jacksonville FL
James Henry Total Fitness Sports Brighton MA
Bob Shorter The Triathletes Club Cherry Hill NJ
Cathy O’Brian The Sporting Life Princeton NJ
Kim Satifer Big Blue Bike Shop Blue Island NY
Chris Putnum Putnum’s Putters Bartlesville OK

Query Result 2-24

Using a WHERE Clause to Find a Subset of Values

As shown in “Excluding Rows” on page 2-30, Query 2-25 also assumes the
use of an ANSI-compliant database. The owner qualifier is in quotation marks
to preserve the case sensitivity of the literal string.

SELECT lname, city, state, phone

FROM 'Aleta'.customer
WHERE state = 'AZ' OR state = 'NJ'
ORDER BY lname

SELECT lname, city, state, phone

FROM 'Aleta'.customer
WHERE state IN ('AZ', 'NJ')
ORDER BY lname

Query 2-25

Composing Simple SELECT Statements 2-33

Single-Table SELECT Statements

Each statement in Query 2-25 retrieves rows that include the subset of AZ or
NJ in the state column of the Aleta.customer table, as shown in Query Result
2-25:

lname city state phone

Jewell Phoenix AZ 602-265-8754

Lessor Phoenix AZ 602-533-1817
O’Brian Princeton NJ 609-342-0054
Shorter Cherry Hill NJ 609-663-6079

Query Result 2-25

Note that you cannot test a TEXT or BYTE column with the IN keyword.
In Query 2-26, an example of a query on an ANSI-compliant database, no
quotation marks exist around the table owner name. Whereas the two state-
ments in Query 2-25 searched the Aleta.customer table, Query 2-26 searches
the table ALETA.customer, which is a different table, due to how ANSI-
compliant databases look at owner names.

SELECT lname, city, state, phone

FROM Aleta.customer
WHERE state NOT IN ('AZ', 'NJ')
ORDER BY state

Query 2-26

2-34 Composing Simple SELECT Statements

 Single-Table SELECT Statements

In Query 2-26, by adding the keyword NOT IN, the subset changes to exclude
the subsets AZ and NJ in the state column. Query Result 2-26 shows the
results in order of state:

lname city state phone

Pauli Sunnyvale CA 408-789-8075

Sadler San Francisco CA 415-822-1289
Currie Palo Alto CA 415-328-4543
Higgins Redwood City CA 415-368-1100
Vector Los Altos CA 415-776-3249
Watson Mountain View CA 415-389-8789
Ream Palo Alto CA 415-356-9876
Quinn Redwood City CA 415-544-8729
Miller Sunnyvale CA 408-723-8789
Jaeger Redwood City CA 415-743-3611
Keyes Sunnyvale CA 408-277-7245
Lawson Los Altos CA 415-887-7235
Beatty Menlo Park CA 415-356-9982
Albertson Redwood City CA 415-886-6677
Grant Menlo Park CA 415-356-1123
Parmelee Mountain View CA 415-534-8822
Sipes Redwood City CA 415-245-4578
Baxter Oakland CA 415-655-0011
Neelie Denver CO 303-936-7731
Wallack Wilmington DE 302-366-7511
Hanlon Jacksonville FL 904-823-4239
Henry Brighton MA 617-232-4159
Satifer Blue Island NY 312-944-5691
Putnum Bartlesville OK 918-355-2074

Query Result 2-26

Identifying Null Values

Use the IS NULL or IS NOT NULL option to check for null values. A null
represents either the absence of data or a value that is unknown or not
applicable. A null value is not the same as a zero or a blank.
Query 2-27 returns all rows that have a null paid_date:

SELECT order_num, customer_num, po_num, ship_date

FROM orders
WHERE paid_date IS NULL
ORDER BY customer_num

Query 2-27

Composing Simple SELECT Statements 2-35

Single-Table SELECT Statements

order_num customer_num po_num ship_date

1004 106 8006 05/30/1993

1006 112 Q13557
1007 117 278693 06/05/1993
1012 117 278701 06/29/1993
1016 119 PC6782 07/12/1993
1017 120 DM354331 07/13/1993
Query Result 2-27

Forming Compound Conditions

You can connect two or more comparison conditions, or Boolean expressions,
by using the logical operators AND, OR, and NOT. A Boolean expression eval-
uates as true or false or, if null values are involved, as unknown. You can
use TEXT or BYTE objects in a Boolean expression only when you test for a
null value.
In Query 2-28, the operator AND combines two comparison expressions in
the WHERE clause:

SELECT order_num, customer_num, po_num, ship_date

FROM orders
WHERE paid_date IS NULL
AND ship_date IS NOT NULL
ORDER BY customer_num

Query 2-28

Query 2-28 returns all rows that have a null paid_date and the ones that do
not also have a null ship_date, as shown in Query Result 2-28:

order_num customer_num po_num ship_date

1004 106 8006 05/30/1993

1007 117 278693 06/05/1993
1012 117 278701 06/29/1993
1016 119 PC6782 07/12/1993

Query Result 2-28

2-36 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Using Variable Text Searches

You can use the keywords LIKE and MATCHES for variable text queries based
on substring searches of CHARACTER fields. Include the keyword NOT to
indicate the opposite condition. The keyword LIKE is the ANSI standard,
whereas MATCHES is an Informix extension.
Variable text search strings can include the wildcards listed in Figure 2-3 with
LIKE or MATCHES.

Symbol Meaning
LIKE
% evaluates to zero or more characters
_ evaluates to a single character
\ escapes special significance of next character
MATCHES
* evaluates to zero or more characters
? evaluates to a single character (except null)
[] evaluates to single character or range of values
\ escapes special significance of next character

Figure 2-3 Wildcards used with LIKE and MATCHES

You cannot test a TEXT or BYTE column with LIKE or MATCHES.

Using Exact Text Comparisons

The following examples include a WHERE clause that searches for exact text
comparisons by using the keyword LIKE or MATCHES or the relational oper-
ator =. Unlike earlier examples, these examples illustrate how to query on an
external table in an ANSI-compliant database.
An external table is a table that is not in the current database. You can access
only external tables that are part of an ANSI-compliant database.
Whereas the database used previously in this chapter was the demonstration
database called stores6, the FROM clause in the following examples specifies
the manatee table, created by the owner bubba, which resides in an ANSI-
compliant database named syzygy. For more information on defining
external tables, see Chapter 1 in the Informix Guide to SQL: Syntax.

Composing Simple SELECT Statements 2-37

Single-Table SELECT Statements

Each statement in Query 2-29 retrieves all the rows that have the single word
helmet in the description column:

SELECT * FROM syzygy:bubba.manatee

WHERE description = 'helmet'
ORDER BY mfg_code

SELECT * FROM syzygy:bubba.manatee

WHERE description LIKE 'helmet'
ORDER BY mfg_code

SELECT * FROM syzygy:bubba.manatee

WHERE description MATCHES 'helmet'
ORDER BY mfg_code

Query 2-29

stock_no mfg_code description unit_price unit unit_type

991 ANT helmet $222.00 case 4/case

991 BKE helmet $269.00 case 4/case
991 JSK helmet $311.00 each 4/case
991 PRM helmet $234.00 case 4/case
991 SHR helmet $245.00 case 4/case

Query Result 2-29

Using a Single-Character Wildcard

The statements in Query 2-30 illustrate the use of a single-character wildcard
in a WHERE clause. Further, they demonstrate a query on an external table.
The stock table is in the external database sloth, which, beside being outside
the current stores6 database, is on a separate database server called meerkat.

2-38 Composing Simple SELECT Statements

 Single-Table SELECT Statements

For details on external tables, external databases, and networks, see

Chapter 12, “Understanding Informix Networking,” in this manual and
Chapter 1 in the Informix Guide to SQL: Syntax.

SELECT * FROM sloth@meerkat:stock

WHERE manu_code LIKE '_R_'
AND unit_price >= 100
ORDER BY description, unit_price

SELECT * FROM sloth@meerkat:stock

WHERE manu_code MATCHES '?R?'
AND unit_price >= 100
ORDER BY description, unit_price

Query 2-30

Each statement in Query 2-30 retrieves only those rows for which the middle
letter of the manu_code is R, as shown in Query Result 2-30:

stock_num manu_code description unit_price unit unit_descr

205 HRO 3 golf balls $312.00 case 24/case

2 HRO baseball $126.00 case 24/case
1 HRO baseball gloves $250.00 case 10 gloves/case
7 HRO basketball $600.00 case 24/case
102 PRC bicycle brakes $480.00 case 4 sets/case
114 PRC bicycle gloves $120.00 case 10 pairs/case
4 HRO football $480.00 case 24/case
110 PRC helmet $236.00 case 4/case
110 HRO helmet $260.00 case 4/case
307 PRC infant jogger $250.00 each each
306 PRC tandem adapter $160.00 each each
308 PRC twin jogger $280.00 each each
304 HRO watch $280.00 box 10/box

Query Result 2-30

The comparison '_R_' (for LIKE) or '?R?' (for MATCHES) specifies, from left
to right, the following items:
• Any single character
• The letter R
• Any single character

Composing Simple SELECT Statements 2-39

Single-Table SELECT Statements

WHERE Clause with Restricted Single-Character Wildcard

Query 2-31 selects only those rows where the manu_code begins with A
through H and returns the rows shown in Query Result 2-31. The class test
'[A-H]' specifies any single letter from A through H, inclusive. No equivalent
wildcard symbol exists for the LIKE keyword.

SELECT * FROM stock

WHERE manu_code MATCHES '[A-H]*'
ORDER BY description, manu_code, unit_price

Query 2-31

stock_num manu_code description unit_price unit unit_descr

205 ANZ 3 golf balls $312.00 case 24/case

205 HRO 3 golf balls $312.00 case 24/case
2 HRO baseball $126.00 case 24/case
3 HSK baseball bat $240.00 case 12/case
1 HRO baseball gloves $250.00 case 10 gloves/case
1 HSK baseball gloves $800.00 case 10 gloves/case
7 HRO basketball $600.00 case 24/case
.
.
.
110 ANZ helmet $244.00 case 4/case
110 HRO helmet $260.00 case 4/case
110 HSK helmet $308.00 case 4/case
.
.
.
301 ANZ running shoes $95.00 each each
301 HRO running shoes $42.50 each each
313 ANZ swim cap $60.00 box 12/box
6 ANZ tennis ball $48.00 case 24 cans/case
5 ANZ tennis racquet $19.80 each each
8 ANZ volleyball $840.00 case 24/case
9 ANZ volleyball net $20.00 each each
304 ANZ watch $170.00 box 10/box
304 HRO watch $280.00 box 10/box

Query Result 2-31

2-40 Composing Simple SELECT Statements

 Single-Table SELECT Statements

WHERE Clause with Variable-Length Wildcard

The statements in Query 2-32 use a wildcard at the end of a string to retrieve
all the rows where the description begins with the characters bicycle:

SELECT * FROM stock

WHERE description LIKE 'bicycle%'
ORDER BY description, manu_code

SELECT * FROM stock

WHERE description MATCHES 'bicycle*'
ORDER BY description, manu_code

Query 2-32

Either statement returns the rows found in Query Result 2-32:

stock_num manu_code description unit_price unit unit_descr

102 PRC bicycle brakes $480.00 case 4 sets/case

102 SHM bicycle brakes $220.00 case 4 sets/case
114 PRC bicycle gloves $120.00 case 10 pairs/case
107 PRC bicycle saddle $70.00 pair pair
106 PRC bicycle stem $23.00 each each
101 PRC bicycle tires $88.00 box 4/box
101 SHM bicycle tires $68.00 box 4/box
105 PRC bicycle wheels $53.00 pair pair
105 SHM bicycle wheels $80.00 pair pair

Query Result 2-32

The comparison 'bicycle%' or 'bicycle*' specifies the characters bicycle

followed by any sequence of zero or more characters. It matches bicycle
stem with stem matched by the wildcard. It matches to the characters
bicycle alone, if a row exists with that description.
Query 2-33 narrows the search by adding another comparison condition that
excludes a manu_code of PRC:

SELECT * FROM stock

WHERE description LIKE 'bicycle%'
AND manu_code NOT LIKE 'PRC'
ORDER BY description, manu_code

Query 2-33

Composing Simple SELECT Statements 2-41

Single-Table SELECT Statements

The statement retrieves only the rows found in Query Result 2-33:

stock_num manu_code description unit_price unit unit_descr

102 SHM bicycle brakes $220.00 case 4 sets/case

101 SHM bicycle tires $68.00 box 4/box
105 SHM bicycle wheels $80.00 pair pair

Query Result 2-33

When you select from a large table and use an initial wildcard in the
comparison string (such as '%cycle'), the query often takes longer to execute.
This is because indexes cannot be used, so every row is searched.

MATCHES and Native Language Support

As indicated on page 2-25, you can set the DBNLS, LANG, and LC_COLLATE
environment variables to influence NLS functionality regarding the collation
order of data returned by a query.
Query 2-34 uses an ORDER BY clause and a WHERE clause specifying a match-
ing range on a column containing data of type NCHAR. This NLS table, abon-
nés, is provided separately from the demonstration database. Query 2-34 is
provided as the sel_nls2.sql command file with NLS-enabled DB-Access.

SELECT numéro,nom,prénom
FROM abonnés
WHERE nom MATCHES '[E-P]*'
ORDER BY nom;

Query 2-34

The output in Query Result 2-34a and Query Result 2-34b show different
rows selected for the MATCHES range by the same query, based on different
NLS environment variable settings:

numéro nom prénom

13607 Hammer Gerhard

13602 Hämmerle Greta
13604 LaForêt Jean-Noël
13610 LeMaître Héloïse
13613 Llanero Gloria Dolores
13603 Montaña José Antonio
13611 Oatfield Emily

Query Result 2-34a

2-42 Composing Simple SELECT Statements

 Single-Table SELECT Statements

In Query Result 2-34a, the rows for Étaix, Ötker, and Øverst are not selected
and listed because, with standard ASCII ordering, the accented first letter of
each name is not in the E-P MATCHES range for the nom column.

numéro nom prénom

13608 Étaix Émile

13607 Hammer Gerhard
13602 Hämmerle Greta
13604 LaForêt Jean-Noël
13610 LeMaître Héloïse
13613 Llanero Gloria Dolores
13603 Montaña José Antonio
13611 Oatfield Emily
13605 Ötker Hans-Jürgen
13614 Øverst Per-Anders

Query Result 2-34b

In Query Result 2-34b, the rows for Étaix, Ötker, and Øverst are included in
the list because, with the appropriate NLS environment variable settings, the
accented first letter of each name is in the E-P MATCHES range for the nom
column.

Comparing for Special Characters

Query 2-35 uses the keyword ESCAPE with LIKE or MATCHES so you can
protect a special character from misinterpretation as a wildcard symbol:

SELECT * FROM cust_calls

WHERE res_descr LIKE '%!%%' ESCAPE '!'

Query 2-35

The ESCAPE keyword designates an escape character (it is an exclamation point

[!] in this example), which protects the next character so it is interpreted as
data and not as a wildcard. In the example, the escape character causes the
middle percent sign to be treated as data. By using the ESCAPE keyword, you

Composing Simple SELECT Statements 2-43

Single-Table SELECT Statements

can search for occurrences of a percent sign (%) in the res_descr column with
the LIKE wildcard %. The query retrieves the row shown in Query Result 2-35:

customer_num 116
call_dtime 1992-12-21 11:24
user_id mannyn
call_code I
call_descr Second complaint from this customer! Received
two cases right-handed outfielder gloves
(1 HRO) instead of one case lefties.
res_dtime 1992-12-27 08:19
res_descr Memo to shipping (Ava Brown) to send case of
left-handed gloves, pick up wrong case; memo
to billing requesting 5% discount to placate
customer due to second offense and lateness
of resolution because of holiday

Query Result 2-35

Using Subscripting in a WHERE Clause

You can use subscripting in the WHERE clause of a SELECT statement to specify
a range of characters or numbers in a column, as shown in Query 2-36:

SELECT catalog_num, stock_num, manu_code, cat_advert,

cat_descr
FROM catalog
WHERE cat_advert[1,4] = 'High'

Query 2-36

The subscript [1,4] causes Query 2-36 to retrieve all rows in which the first
four letters of the cat_advert column are High, as shown in Query Result
2-36:

2-44 Composing Simple SELECT Statements

 Single-Table SELECT Statements

catalog_num 10004
stock_num 2
manu_code HRO
cat_advert Highest Quality Ball Available, from
Hand-Stitching to the Robinson Signature
cat_descr
Jackie Robinson signature ball. Highest professional quality, used by
National League.

catalog_num 10005
stock_num 3
manu_code HSK
cat_advert High-Technology Design Expands the Sweet Spot
cat_descr
Pro-style wood. Available in sizes: 31, 32, 33, 34, 35.

catalog_num 10008
stock_num 4
manu_code HRO
cat_advert Highest Quality Football for High School and
Collegiate Competitions
cat_descr
NFL-style, pigskin.

catalog_num 10012
stock_num 6
manu_code SMT
cat_advert High-Visibility Tennis, Day or Night
cat_descr
Soft yellow color for easy visibility in sunlight or
artificial light.

catalog_num 10043
stock_num 202
manu_code KAR
cat_advert High-Quality Woods Appropriate for High School
Competitions or Serious Amateurs
cat_descr
Full set of woods designed for precision control and
power performance.

catalog_num 10045
stock_num 204
manu_code KAR
cat_advert High-Quality Beginning Set of Irons
Appropriate for High School Competitions
cat_descr
Ideally balanced for optimum control. Nylon covered shaft.

catalog_num 10068
stock_num 310
manu_code ANZ
cat_advert High-Quality Kickboard
cat_descr
White. Standard size.

Query Result 2-36

Composing Simple SELECT Statements 2-45

Single-Table SELECT Statements

Expressions and Derived Values

You are not limited to selecting columns by name. You can use the SELECT
clause of a SELECT statement to perform computations on column data and
to display information derived from the contents of one or more columns. You
do this by listing an expression in the select list.
An expression consists of a column name, a constant, a quoted string, a
keyword, or any combination of these items connected by operators. It also
can include host variables (program data) when the SELECT statement is
embedded in a program.
For more information, see the section “Expression,” in Chapter 1 of the
Informix Guide to SQL: Syntax.

Arithmetic Expressions
An arithmetic expression contains at least one of the arithmetic operators listed
in Figure 2-4 and produces a number. You cannot use TEXT or BYTE columns
in arithmetic expressions.

Operator Operation
+ addition
- subtraction
* multiplication
/ division

Figure 2-4 Arithmetic operators used in arithmetic expressions

Operations of this nature enable you to see the results of proposed

computations without actually altering the data in the database. You can add
an INTO TEMP clause to save the altered data in a temporary table for further
reference, computations, or impromptu reports. Query 2-37 calculates a 7
percent sales tax on the unit_price column when the unit_price is $400 or
more (but does not update it in the database).

SELECT stock_num, description, unit, unit_descr,

unit_price, unit_price * 1.07
FROM stock
WHERE unit_price >= 400

Query 2-37

2-46 Composing Simple SELECT Statements

 Single-Table SELECT Statements

If you are using the DB-Access or INFORMIX-SQL Interactive Schema Editor,

the result is displayed in a column labeled expression, as shown in Query
Result 2-37:

stock_num description unit unit_descr unit_price (expression)

1 baseball gloves case 10 gloves/case $800.00 $856.0000

1 baseball gloves case 10 gloves/case $450.00 $481.5000
4 football case 24/case $960.00 $1027.2000
4 football case 24/case $480.00 $513.6000
7 basketball case 24/case $600.00 $642.0000
8 volleyball case 24/case $840.00 $898.8000
102 bicycle brakes case 4 sets/case $480.00 $513.6000
111 10-spd, assmbld each each $499.99 $534.9893
112 12-spd, assmbld each each $549.00 $587.4300
113 18-spd, assmbld each each $685.90 $733.9130
203 irons/wedge case 2 sets/case $670.00 $716.9000

Query Result 2-37

Query 2-38 calculates a surcharge of $6.50 on orders when the quantity

ordered is less than 5:

SELECT item_num, order_num, quantity,

total_price, total_price + 6.50
FROM items
WHERE quantity < 5

Query 2-38

Composing Simple SELECT Statements 2-47

Single-Table SELECT Statements

If you are using DB-Access or INFORMIX-SQL, the result is displayed in a

column labeled expression, as shown in Query Result 2-38:

item_num order_num quantity total_price (expression)

1 1001 1 $250.00 $256.50

1 1002 1 $960.00 $966.50
2 1002 1 $240.00 $246.50
1 1003 1 $20.00 $26.50
2 1003 1 $840.00 $846.50
1 1004 1 $250.00 $256.50
2 1004 1 $126.00 $132.50
3 1004 1 $240.00 $246.50
4 1004 1 $800.00 $806.50
.
.
.
1 1021 2 $75.00 $81.50
2 1021 3 $225.00 $231.50
3 1021 3 $690.00 $696.50
4 1021 2 $624.00 $630.50
1 1022 1 $40.00 $46.50
2 1022 2 $96.00 $102.50
3 1022 2 $96.00 $102.50
1 1023 2 $40.00 $46.50
2 1023 2 $116.00 $122.50
3 1023 1 $80.00 $86.50
4 1023 1 $228.00 $234.50
5 1023 1 $170.00 $176.50
6 1023 1 $190.00 $196.50

Query Result 2-38

2-48 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Query 2-39 calculates and displays in an expression column (if you are using
DB-Access or INFORMIX-SQL) the interval between when the customer call
was received (call_dtime) and when the call was resolved (res_dtime), in
days, hours, and minutes.

SELECT customer_num, user_id, call_code,

call_dtime, res_dtime - call_dtime
FROM cust_calls
ORDER BY user_id

Query 2-39

customer_num user_id call_code call_dtime (expression)

116 mannyn I 1992-12-21 11:24 5 20:55

116 mannyn I 1992-11-28 13:34 0 03:13
106 maryj D 1993-06-12 08:20 0 00:05
121 maryj O 1993-07-10 14:05 0 00:01
127 maryj I 1993-07-31 14:30
110 richc L 1993-07-07 10:24 0 00:06
119 richc B 1993-07-01 15:00 0 17:21

Query Result 2-39

Using Display Labels

You can assign a display label to a computed or derived data column to replace
the default column header expression. In Query Result 2-37, Query Result
2-38, and Query Result 2-39, the derived data is shown in a column called
(expression). Query 2-40 also presents derived values, but the column
displaying the derived values now has the descriptive header, taxed.

SELECT stock_num, description, unit, unit_descr,

unit_price, unit_price * 1.07 taxed
FROM stock
WHERE unit_price >= 400

Query 2-40

Composing Simple SELECT Statements 2-49

Single-Table SELECT Statements

Query Result 2-40 shows the label taxed is assigned to the expression in the
select list that displays the results of the operation unit_price * 1.07:

stock_num description unit unit_descr unit_price taxed

1 baseball gloves case 10 gloves/case $800.00 $856.0000

1 baseball gloves case 10 gloves/case $450.00 $481.5000
4 football case 24/case $960.00 $1027.2000
4 football case 24/case $480.00 $513.6000
7 basketball case 24/case $600.00 $642.0000
8 volleyball case 24/case $840.00 $898.8000
102 bicycle brakes case 4 sets/case $480.00 $513.6000
111 10-spd, assmbld each each $499.99 $534.9893
112 12-spd, assmbld each each $549.00 $587.4300
113 18-spd, assmbld each each $685.90 $733.9130
203 irons/wedge case 2 sets/case $670.00 $716.9000

Query Result 2-40

In Query 2-41, the label surcharge is defined for the column that displays the
results of the operation total_price + 6.50:

SELECT item_num, order_num, quantity,

total_price, total_price + 6.50 surcharge
FROM items
WHERE quantity < 5

Query 2-41

2-50 Composing Simple SELECT Statements

 Single-Table SELECT Statements

The surcharge column is labeled in the output, as shown in Query Result

2-41:

item_num order_num quantity total_price surcharge

.
.
.
2 1013 1 $36.00 $42.50
3 1013 1 $48.00 $54.50
4 1013 2 $40.00 $46.50
1 1014 1 $960.00 $966.50
2 1014 1 $480.00 $486.50
1 1015 1 $450.00 $456.50
1 1016 2 $136.00 $142.50
2 1016 3 $90.00 $96.50
3 1016 1 $308.00 $314.50
4 1016 1 $120.00 $126.50
1 1017 4 $150.00 $156.50
2 1017 1 $230.00 $236.50
.
.
.

Query Result 2-41

Query 2-42 assigns the label span to the column that displays the results of
subtracting the DATETIME column call_dtime from the DATETIME column
res_dtime.

SELECT customer_num, user_id, call_code,

call_dtime, res_dtime - call_dtime span
FROM cust_calls
ORDER BY user_id

Query 2-42

The span column is labeled in the output, as shown in Query Result 2-42:

customer_num user_id call_code call_dtime span

116 mannyn I 1992-12-21 11:24 5 20:55

116 mannyn I 1992-11-28 13:34 0 03:13
106 maryj D 1993-06-12 08:20 0 00:05
121 maryj O 1993-07-10 14:05 0 00:01
127 maryj I 1993-07-31 14:30
110 richc L 1993-07-07 10:24 0 00:06
119 richc B 1993-07-01 15:00 0 17:21

Query Result 2-42

Composing Simple SELECT Statements 2-51

Single-Table SELECT Statements

Sorting on Derived Columns

When you want to use ORDER BY as an expression, you can use either the
display label assigned to the expression or an integer, as shown in Query
2-43:

SELECT customer_num, user_id, call_code,

call_dtime, res_dtime - call_dtime span
FROM cust_calls
ORDER BY span

Query 2-43

Query 2-43 retrieves the same data from the cust_calls table as Query 2-42. In
Query 2-43, the ORDER BY clause causes the data to be displayed in ascending
order of the derived values in the span column, as shown in Query Result
2-43:

customer_num user_id call_code call_dtime span

127 maryj I 1993-07-31 14:30

121 maryj O 1993-07-10 14:05 0 00:01
106 maryj D 1993-06-12 08:20 0 00:05
110 richc L 1993-07-07 10:24 0 00:06
116 mannyn I 1992-11-28 13:34 0 03:13
119 richc B 1993-07-01 15:00 0 17:21
116 mannyn I 1992-12-21 11:24 5 20:55

Query Result 2-43

Query 2-44 uses an integer to represent the result of the operation

res_dtime - call_dtime and retrieves the same rows that appear in
Query Result 2-43.

SELECT customer_num, user_id, call_code,

call_dtime, res_dtime - call_dtime span
FROM cust_calls
ORDER BY 5

Query 2-44

2-52 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Using Functions in SELECT Statements

In addition to column names and operators, an expression also can include
one or more functions.
Expressions supported include aggregate, function (which include
arithmetic functions), constant, and column expressions. These expressions
are described in Chapter 1 of the Informix Guide to SQL: Syntax.

Aggregate Functions
The aggregate functions are COUNT, AVG, MAX, MIN, and SUM. They take on
values that depend on all the rows selected and return information about
rows, not the rows themselves. You cannot use these functions with TEXT or
BYTE columns.

Aggregates often are used to summarize information about groups of rows

in a table. This use is discussed in Chapter 3, “Composing Advanced
SELECT Statements.” When you apply an aggregate function to an entire
table, the result contains a single row that summarizes all of the selected
rows.
Query 2-45 counts and displays the total number of rows in the stock table:

SELECT COUNT(*)
FROM stock

Query 2-45

(count(*))

73

Query Result 2-45

Composing Simple SELECT Statements 2-53

Single-Table SELECT Statements

Query 2-46 includes a WHERE clause to count specific rows in the stock table,
in this case, only those rows that have a manu_code of SHM:

SELECT COUNT (*)

FROM stock
WHERE manu_code = 'SHM'

Query 2-46

(count(*))

16

Query Result 2-46

By including the keyword DISTINCT (or its synonym UNIQUE) and a column
name in Query 2-47, you can tally the number of different manufacturer
codes in the stock table:

SELECT COUNT (DISTINCT manu_code)

FROM stock

Query 2-47

(count)

Query Result 2-47

Query 2-48 computes the average unit_price of all rows in the stock table:

SELECT AVG (unit_price)

FROM stock

Query 2-48

(avg)

Query Result 2-48

2-54 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Query 2-49 computes the average unit_price of just those rows in the stock
table that have a manu_code of SHM:

SELECT AVG (unit_price)

FROM stock
WHERE manu_code = 'SHM'

Query 2-49

(avg)

Query Result 2-49

You can combine aggregate functions in Query 2-50:

SELECT MAX (ship_charge), MIN (ship_charge)

FROM orders

Query 2-50

Query 2-50 finds and displays both the highest and lowest ship_charge in the
orders table, as shown in Query Result 2-50:

(max) (min)

Query Result 2-50

You can apply functions to expressions, and you can supply display labels for
their results, as shown in Query 2-51:

SELECT MAX (res_dtime - call_dtime) maximum,

MIN (res_dtime - call_dtime) minimum,
AVG (res_dtime - call_dtime) average
FROM cust_calls

Query 2-51

Composing Simple SELECT Statements 2-55

Single-Table SELECT Statements

Query 2-51 finds and displays the maximum, minimum, and average amount
of time (in days, hours, and minutes) between the reception and resolution of
a customer call and labels the derived values appropriately. These amounts
of time are shown in Query Result 2-51:

maximum minimum average

5 20:55 0 00:01 1 02:56

Query Result 2-51

Query 2-52 calculates the total ship_weight of orders shipped on July 13,
1993:

SELECT SUM (ship_weight)

FROM orders
WHERE ship_date = '07/13/1993'

Query 2-52

(sum)

Query Result 2-52

Time Functions
You can use the time functions DAY, MDY, MONTH, WEEKDAY, YEAR, and
DATE in either the SELECT clause or the WHERE clause of a query. These func-
tions return a value that corresponds to the expressions or arguments that
you use to call the function. You also can use the CURRENT function to return
a value with the current date and time, or the EXTEND function to adjust the
precision of a DATE or DATETIME value.

2-56 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Using DAY and CURRENT

Query 2-53 returns the day of the month for the call_dtime and res_dtime
columns in two expression columns, as shown in Query Result 2-53:

SELECT customer_num, DAY (call_dtime), DAY (res_dtime)

FROM cust_calls

Query 2-53

customer_num (expression) (expression)

106 12 12
110 7 7
119 1 2
121 10 10
127 31
116 28 28
116 21 27

Query Result 2-53

Query 2-54 uses the DAY and CURRENT functions to compare column values
to the current day of the month. It selects only those rows where the value is
earlier than the current day.

SELECT customer_num, DAY (call_dtime), DAY (res_dtime)

FROM cust_calls
WHERE DAY (call_dtime) < DAY (CURRENT)

Query 2-54

customer_num (expression) (expression)

106 12 12
110 7 7
119 1 2
121 10 10

Query Result 2-54

Composing Simple SELECT Statements 2-57

Single-Table SELECT Statements

Query 2-55 shows another use of the CURRENT function, selecting rows
where the day is earlier than the current one:

SELECT customer_num, call_code, call_descr

FROM cust_calls
WHERE call_dtime < CURRENT YEAR TO DAY

Query 2-55

customer_num 106
call_code D
call_descr Order was received, but two of the cans of ANZ tennis
balls within the case were empty

customer_num 116
call_code I
call_descr Received plain white swim caps (313 ANZ) instead of
navy with team logo (313 SHM)

customer_num 116
call_code I
call_descr Second complaint from this customer! Received
two cases right-handed outfielder gloves (1 HRO) instead of
one case lefties.

Query Result 2-55

2-58 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Using MONTH
Query 2-56 uses the MONTH function to extract and show what month the
customer call was received and resolved and uses display labels for the
resulting columns. However, it does not make a distinction between years.

SELECT customer_num,
MONTH (call_dtime) call_month,
MONTH (res_dtime) res_month
FROM cust_calls

Query 2-56

customer_num call_month res_month

106 6 6
110 7 7
119 7 7
121 7 7
127 7
116 11 11
116 12 12

Query Result 2-56

Query 2-57 uses the MONTH function plus DAY and CURRENT to show what
month the customer call was received and resolved if DAY is earlier than the
current day:

SELECT customer_num,
MONTH (call_dtime) called,
MONTH (res_dtime) resolved
FROM cust_calls
WHERE DAY (res_dtime) < DAY (CURRENT)

Query 2-57

customer_num called resolved

106 6 6
110 7 7
119 7 7
121 7 7

Query Result 2-57

Composing Simple SELECT Statements 2-59

Single-Table SELECT Statements

Using WEEKDAY
In Query 2-58, the WEEKDAY function is used to indicate which day of the
week calls are received and resolved (0 represents Sunday, 1 is Monday, and
so on), and the expression columns are labeled:

SELECT customer_num,
WEEKDAY (call_dtime) called,
WEEKDAY (res_dtime) resolved
FROM cust_calls
ORDER BY resolved

Query 2-58

customer_num called resolved

127 3
110 0 0
119 1 2
121 3 3
116 3 3
106 3 3
116 5 4

Query Result 2-58

Query 2-60 uses the COUNT and WEEKDAY functions to count how many
calls were received on a weekend. This kind of statement can give you an idea
of customer call patterns or indicate whether overtime pay might be
required.

SELECT COUNT(*)
FROM cust_calls
WHERE WEEKDAY (call_dtime) IN (0,6)

Query 2-59

(count(*))

Query Result 2-59

2-60 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Query 2-60 retrieves rows where the call_dtime is earlier than the beginning
of the current year.

SELECT customer_num, call_code,

YEAR (call_dtime) call_year,
YEAR (res_dtime) res_year
FROM cust_calls
WHERE YEAR (call_dtime) < YEAR (TODAY)

Query 2-60

customer_num call_code call_year res_year

116 I 1992 1992

116 I 1992 1992

Query Result 2-60

Formatting DATETIME Values

In Query 2-61, the EXTEND function restricts the two DATETIME values by
displaying only the specified subfields:

SELECT customer_num,
EXTEND (call_dtime, month to minute) call_time,
EXTEND (res_dtime, month to minute) res_time
FROM cust_calls
ORDER BY res_time

Query 2-61

Query Result 2-61 returns the month-to-minute range for the columns
labeled call_time and res_time and gives an indication of the workload:

customer_num call_time res_time

127 07-31 14:30

106 06-12 08:20 06-12 08:25
119 07-01 15:00 07-02 08:21
110 07-07 10:24 07-07 10:30
121 07-10 14:05 07-10 14:06
116 11-28 13:34 11-28 16:47
116 12-21 11:24 12-27 08:19

Query Result 2-61

Composing Simple SELECT Statements 2-61

Single-Table SELECT Statements

Using the DATE Function

Query 2-62 retrieves DATETIME values only when call_dtime is later than the
specified DATE:

SELECT customer_num, call_dtime, res_dtime

FROM cust_calls
WHERE call_dtime > DATE ('12/31/92')

Query 2-62

Query Result 2-62 returns the following rows:

customer_num call_dtime res_dtime

106 1993-06-12 08:20 1993-06-12 08:25

110 1993-07-07 10:24 1993-07-07 10:30
119 1993-07-01 15:00 1993-07-02 08:21
121 1993-07-10 14:05 1993-07-10 14:06
127 1993-07-31 14:30

Query Result 2-62

Query 2-63 converts DATETIME values to DATE format and displays the
values, with labels, only when call_dtime is greater than or equal to the
specified date.

SELECT customer_num,
DATE (call_dtime) called,
DATE (res_dtime) resolved
FROM cust_calls
WHERE call_dtime >= DATE ('1/1/93')

Query 2-63

customer_num called resolved

106 06/12/1993 06/12/1993

110 07/07/1993 07/07/1993
119 07/01/1993 07/02/1993
121 07/10/1993 07/10/1993
127 07/31/1993

Query Result 2-63

2-62 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Other Functions and Keywords

You also can use the LENGTH, USER, CURRENT, and TODAY functions
anywhere in an SQL expression that you would use a constant. In addition,
with INFORMIX-OnLine Dynamic Server, you can include the DBSERVER-
NAME keyword in a SELECT statement to display the name of the database
server where the current database resides.
You can use these functions and keywords to select an expression that
consists entirely of constant values or an expression that includes column
data. In the first instance, the result is the same for all rows of output.
In addition, you can use the HEX function to return the hexadecimal encoding
of an expression, the ROUND function to return the rounded value of an
expression, and the TRUNC function to return the truncated value of an
expression.
In Query 2-64, the LENGTH function calculates the number of bytes in the
combined fname and lname columns for each row where the length of
company is greater than 15:

SELECT customer_num,
LENGTH (fname) + LENGTH (lname) namelength
FROM customer
WHERE LENGTH (company) > 15

Query 2-64

customer_num namelength

101 11
105 13
107 11
112 14
115 11
118 10
119 10
120 10
122 12
124 11
125 10
126 12
127 10
128 11

Query Result 2-64

Composing Simple SELECT Statements 2-63

Single-Table SELECT Statements

Although it might not be useful when you work with the DB-Access or
INFORMIX-SQL Interactive Editor, the LENGTH function can be important to
determine the string length for programs and reports. LENGTH returns the
clipped length of a CHARACTER or VARCHAR string and the full number of
bytes in a TEXT or BYTE string.
The USER function can be handy when you want to define a restricted view
of a table that contains only your rows. For information on creating views, see
Chapter 11, “Granting and Limiting Access to Your Database,” in this manual
and the GRANT and CREATE VIEW statements in Chapter 1 of the Informix
Guide to SQL: Syntax.
Query 2-65a specifies the USER function and the cust_calls table:

SELECT USER from cust_calls

Query 2-65a

Query 2-65b returns the user name (login account name) of the user who
executes the query. It is repeated once for each row in the table.

SELECT * FROM cust_calls

WHERE user_id = USER

Query 2-65b

If the user name of the current user is richc, Query 2-65b retrieves only those
rows in the cust_calls table that are owned by that user.

customer_num 110
call_dtime 1993-07-07 10:24
user_id richc
call_code L
call_descr Order placed one month ago (6/7) not received.
res_dtime 1993-07-07 10:30
res_descr Checked with shipping (Ed Smith). Order sent
yesterday- we were waiting for goods from ANZ. Next
time will call with delay if necessary.

customer_num 119
call_dtime 1993-07-01 15:00
user_id richc
call_code B
call_descr Bill does not reflect credit from previous order
res_dtime 1993-07-02 08:21
res_descr Spoke with Jane Akant in Finance. She found the
error and is sending new bill to customer

Query Result 2-65

2-64 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Query 2-66, if issued when today’s system date is July 10, 1993, returns this
one row:

SELECT * FROM orders

WHERE order_date = TODAY

Query 2-66

order_num 1018
order_date 07/10/1993
customer_num 121
ship_instruct SW corner of Biltmore Mall
backlog n
po_num S22942
ship_date 07/13/1993
ship_weight 70.50
ship_charge $20.00
paid_date 08/06/1993

Query Result 2-66

You can include the keyword DBSERVERNAME (or its synonym, SITENAME)
in a SELECT statement on INFORMIX-OnLine Dynamic Server to find the
name of the database server. You can query on the DBSERVERNAME for any
table that has rows, including system catalog tables.
In Query 2-67, you assign the label server to the DBSERVERNAME expression
and also select the tabid column from the systables system catalog table. This
table describes database tables, and tabid is the serial interval table identifier.

SELECT DBSERVERNAME server, tabid FROM systables

WHERE tabid <= 4

Query 2-67

server tabid

montague 1
montague 2
montague 3
montague 4

Query Result 2-67

Composing Simple SELECT Statements 2-65

Single-Table SELECT Statements

Without the WHERE clause to restrict the values in the tabid, the database
server name would be repeated for each row of the systables table.
In Query 2-68, the HEX function returns the hexadecimal format of three
specified columns in the customer table:

SELECT HEX (customer_num) hexnum, HEX (zipcode) hexzip,

HEX (rowid) hexrow
FROM customer

Query 2-68

hexnum hexzip hexrow

0x00000065 0x00016F86 0x00000001

0x00000066 0x00016FA5 0x00000002
0x00000067 0x0001705F 0x00000003
0x00000068 0x00016F4A 0x00000004
0x00000069 0x00016F46 0x00000005
0x0000006A 0x00016F6F 0x00000006
0x0000006B 0x00017060 0x00000007
0x0000006C 0x00016F6F 0x00000008
0x0000006D 0x00016F86 0x00000009
0x0000006E 0x00016F6E 0x0000000A
0x0000006F 0x00016F85 0x0000000B
0x00000070 0x00016F46 0x0000000C
0x00000071 0x00016F49 0x0000000D
0x00000072 0x00016F6E 0x0000000E
0x00000073 0x00016F49 0x0000000F
0x00000074 0x00016F58 0x00000010
0x00000075 0x00016F6F 0x00000011
0x00000076 0x00017191 0x00000012
0x00000077 0x00001F42 0x00000013
0x00000078 0x00014C18 0x00000014
0x00000079 0x00004DBA 0x00000015
0x0000007A 0x0000215C 0x00000016
0x0000007B 0x00007E00 0x00000017
0x0000007C 0x00012116 0x00000018
0x0000007D 0x00000857 0x00000019
0x0000007E 0x0001395B 0x0000001A
0x0000007F 0x0000EBF6 0x0000001B
0x00000080 0x00014C10 0x0000001C

Query Result 2-68

2-66 Composing Simple SELECT Statements

 Single-Table SELECT Statements

Using Stored Procedures in SELECT Statements

We have seen examples of SELECT statement expressions consisting of
column names, operators, and functions. Another type of expression is one
that contains a stored procedure call.
Stored procedures are procedures that contain special Stored Procedure
Language (SPL) statements as well as SQL statements. For more information
on stored procedures, refer to Chapter 14, “Creating and Using Stored Proce-
dures.”
Stored procedures provide a way to extend the range of functions available;
you can perform a subquery on each row you select.
For example, suppose you want a listing of the customer number, the
customer’s last name, and the number of orders the customer has made. One
way to retrieve this information is shown in Query 2-69. The customer table
has customer_num and lname columns but no record of the number of
orders each customer has made. You could write a get_orders procedure
which queries the orders table for each customer_num and returns the
number of corresponding orders (labeled n_orders).

SELECT customer_num, lname, get_orders(customer_num) n_orders

FROM customer

Query 2-69

Composing Simple SELECT Statements 2-67

Single-Table SELECT Statements

The output from this stored-procedure query is shown in Query Result 2-69:

customer_num lname n_orders

101 Pauli 1
102 Sadler 0
103 Currie 0
104 Higgins 4
105 Vector 0
106 Watson 2
107 Ream 0
108 Quinn 0
109 Miller 0
110 Jaeger 2
111 Keyes 1
112 Lawson 1
113 Beatty 0
114 Albertson 0
115 Grant 1
116 Parmelee 1
117 Sipes 2
118 Baxter 0
119 Shorter 1
120 Jewell 1
121 Wallack 1
122 O’Brian 1
123 Hanlon 1
124 Putnum 1
125 Henry 0
126 Neelie 1
127 Satifer 1
128 Lessor 0

Query Result 2-69

You can use stored procedures to encapsulate operations that you frequently
perform in your queries. For example, the condition in Query 2-70 contains a
procedure, conv_price, that converts the unit price of a stock item to a
different currency and adds any import tariffs:

SELECT stock_num, manu_code, description FROM stock

WHERE conv_price(unit_price, ex_rate = 1.50, tarif = 50.00)
< 1000

Query 2-70

2-68 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

Multiple-Table SELECT Statements

You can select data from two or more tables by naming these tables in the
FROM clause. Add a WHERE clause to create a join condition between at least
one related column in each table. This creates a temporary composite table in
which each pair of rows that satisfies the join condition is linked to form a
single row.
A simple join combines information from two or more tables based on the
relationship between one column in each table. A composite join is a join
between two or more tables based on the relationship between two or more
columns in each table.
To create a join, you must specify a relationship, called a join condition,
between at least one column from each table. Because the columns are being
compared, they must have compatible data types. When you join large tables,
performance improves when you index the columns in the join condition.
Data types are described in Chapter 3 of the Informix Guide to SQL: Reference,
and indexing is discussed in Chapter 10, “Tuning Your Data Model,” and
Chapter 13, “Optimizing Your Queries.”

Creating a Cartesian Product

When you perform a multiple-table query that does not explicitly state a join
condition among the tables, you create a Cartesian product. A Cartesian prod-
uct consists of every possible combination of rows from the tables. This is
usually a very large, unwieldy result and the data is inaccurate.
Query 2-71 selects from two tables and produces a Cartesian product:

SELECT * FROM customer, state

Query 2-71

Composing Simple SELECT Statements 2-69

Multiple-Table SELECT Statements

Although only 52 rows exist in the state table and 28 rows in the customer
table, the effect of Query 2-71 is to multiply the rows of one table by the rows
of the other and retrieve an impractical 1456 rows, as shown in Query Result
2-71:

customer_num 101
fname Ludwig
lname Pauli
company All Sports Supplies
address1 213 Erstwild Court
address2
city Sunnyvale
state CA
zipcode 94086
phone 408-789-8075
code AK
sname Alaska

customer_num 101
fname Ludwig
lname Pauli
company All Sports Supplies
address1 213 Erstwild Court
address2
city Sunnyvale
state CA
zipcode 94086
phone 408-789-8075
code HI
sname Hawaii

customer_num 101
fname Ludwig
lname Pauli
company All Sports Supplies
address1 213 Erstwild Court
address2
city Sunnyvale
state CA
zipcode 94086
phone 408-789-8075
code CA
sname California
.
.
.

Query Result 2-71

2-70 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

Note that some of the data displayed in the concatenated rows is inaccurate.
For example, although the city and state from the customer table indicate an
address in California, the code and sname from the state table might be for a
different state.

Creating a Join
Conceptually, the first stage of any join is the creation of a Cartesian product.
To refine or constrain this Cartesian product and eliminate meaningless rows
of data, include a WHERE clause with a valid join condition in your SELECT
statement.
This section illustrates equi-joins, natural joins, and multiple-table joins.
Additional complex forms, such as self-joins and outer joins, are covered in
Chapter 3, “Composing Advanced SELECT Statements.”

Equi-Join
An equi-join is a join based on equality or matching values. This equality is
indicated with an equal sign (=) in the comparison operation in the WHERE
clause, as shown in Query 2-72:

SELECT * FROM manufact, stock

WHERE manufact.manu_code = stock.manu_code

Query 2-72

Composing Simple SELECT Statements 2-71

Multiple-Table SELECT Statements

Query 2-72 joins the manufact and stock tables on the manu_code column,
retrieving only those rows for which the values for the two columns are
equal, as shown in Query Result 2-72:

manu_code SMT
manu_name Smith
lead_time 3
stock_num 1
manu_code SMT
description baseball gloves
unit_price $450.00
unit case
unit_descr 10 gloves/case

manu_code SMT
manu_name Smith
lead_time 3
stock_num 5
manu_code SMT
description tennis racquet
unit_price $25.00
unit each
unit_descr each

manu_code SMT
manu_name Smith
lead_time 3
stock_num 6
manu_code SMT
description tennis ball
unit_price $36.00
unit case
unit_descr 24 cans/case

manu_code ANZ
manu_name Anza
lead_time 5
stock_num 5
manu_code ANZ
description tennis racquet
unit_price $19.80
unit each
unit_descr each
.
.
.
Query Result 2-72

2-72 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

Note that in this equi-join, Query Result 2-72 includes the manu_code
column from both the manufact and stock tables because the select list
requested every column.
You also can create an equi-join with additional constraints, one where the
comparison condition is based on the inequality of values in the joined col-
umns. These joins use a relational operator other than = in the comparison
condition specified in the WHERE clause.
When columns in the joined tables have the same name, the columns must be
preceded by the name of a specific table and a period, as shown in Query
2-73:

SELECT order_num, order_date, ship_date, cust_calls.*

FROM orders, cust_calls
WHERE call_dtime >= ship_date
AND cust_calls.customer_num = orders.customer_num
ORDER BY customer_num

Query 2-73

Composing Simple SELECT Statements 2-73

Multiple-Table SELECT Statements

Query 2-73 joins on the customer_num column and then selects only those
rows where the call_dtime in the cust_calls table is greater than or equal to
the ship_date in the orders table. It returns the rows shown in Query Result
2-73:

order_num 1004
order_date 05/22/1993
ship_date 05/30/1993
customer_num 106
call_dtime 1993-06-12 08:20
user_id maryj
call_code D
call_descr Order received okay, but two of the cans of
ANZ tennis balls within the case were empty
res_dtime 1993-06-12 08:25
res_descr Authorized credit for two cans to customer,
issued apology. Called ANZ buyer to report
the qa problem.

order_num 1008
order_date 06/07/1993
ship_date 07/06/1993
customer_num 110
call_dtime 1993-07-07 10:24
user_id richc
call_code L
call_descr Order placed one month ago (6/7) not received.
res_dtime 1993-07-07 10:30
res_descr Checked with shipping (Ed Smith). Order out
yesterday-was waiting for goods from ANZ.
Next time will call with delay if necessary.

order_num 1023
order_date 07/24/1993
ship_date 07/30/1993
customer_num 127
call_dtime 1993-07-31 14:30
user_id maryj
call_code I
call_descr Received Hero watches (item # 304) instead
of ANZ watches
res_dtime
res_descr Sent memo to shipping to send ANZ item 304
to customer and pickup HRO watches. Should
be done tomorrow, 8/1

Query Result 2-73

2-74 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

Natural Join
A natural join is structured so that the join column does not display data
redundantly, as shown in Query 2-74:

SELECT manu_name, lead_time, stock.*

FROM manufact, stock
WHERE manufact.manu_code = stock.manu_code

Query 2-74

Composing Simple SELECT Statements 2-75

Multiple-Table SELECT Statements

Like the example for equi-join, Query 2-74 joins the manufact and stock
tables on the manu_code column. Because the select list is more closely
defined, the manu_code is listed only once for each row retrieved, as shown
in Query Result 2-74:

manu_name Smith
lead_time 3
stock_num 1
manu_code SMT
description baseball gloves
unit_price $450.00
unit case
unit_descr 10 gloves/case

manu_name Smith
lead_time 3
stock_num 5
manu_code SMT
description tennis racquet
unit_price $25.00
unit each
unit_descr each

manu_name Smith
lead_time 3
stock_num 6
manu_code SMT
description tennis ball
unit_price $36.00
unit case
unit_descr 24 cans/case

manu_name Anza
lead_time 5
stock_num 5
manu_code ANZ
description tennis racquet
unit_price $19.80
unit each
unit_descr each
.
.
.
Query Result 2-74

All joins are associative, that is, the order of the joining terms in the WHERE
clause does not affect the meaning of the join.

2-76 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

Both of the statements in Query 2-75 create the same natural join:

SELECT catalog.*, description, unit_price, unit, unit_descr

FROM catalog, stock
WHERE catalog.stock_num = stock.stock_num
AND catalog.manu_code = stock.manu_code
AND catalog_num = 10017

SELECT catalog.*, description, unit_price, unit, unit_descr

FROM catalog, stock
WHERE catalog_num = 10017
AND catalog.manu_code = stock.manu_code
AND catalog.stock_num = stock.stock_num

Query 2-75

Each statement retrieves the row shown in Query Result 2-75:

catalog_num 10017
stock_num 101
manu_code PRC
cat_descr
Reinforced, hand-finished tubular. Polyurethane belted.
Effective against punctures. Mixed tread for super wear
and road grip.
cat_picture <BYTE value>

cat_advert Ultimate in Puncture Protection, Tires

Designed for In-City Riding
description bicycle tires
unit_price $88.00
unit box
unit_descr 4/box

Query Result 2-75

Note that Query Result 2-75 includes a TEXT column, cat_descr; a BYTE
column, cat_picture; and a VARCHAR column, cat_advert.

Multiple-Table Join
A multiple-table join connects more than two tables on one or more associ-
ated columns; it can be an equi-join or a natural join.

Composing Simple SELECT Statements 2-77

Multiple-Table SELECT Statements

Query 2-76 creates an equi-join on the catalog, stock, and manufact tables
and retrieves the following row:

SELECT * FROM catalog, stock, manufact

WHERE catalog.stock_num = stock.stock_num
AND stock.manu_code = manufact.manu_code
AND catalog_num = 10025

Query 2-76

Query Result 2-76 retrieves the rows shown in Query 2-76:

catalog_num 10025
stock_num 106
manu_code PRC
cat_descr
Hard anodized alloy with pearl finish; 6mm hex bolt hardware.
Available in lengths of 90-140mm in 10mm increments.
cat_picture <BYTE value>

cat_advert ProCycle Stem with Pearl Finish

stock_num 106
manu_code PRC
description bicycle stem
unit_price $23.00
unit each
unit_descr each
manu_code PRC
manu_name ProCycle
lead_time 9

Query Result 2-76

Note that the manu_code is repeated three times, once for each table, and
stock_num is repeated twice.
Because of the considerable duplication of a multiple-table query in Query
2-76, it is wise to more closely define the SELECT statement by including
specific columns in the select list, as shown in Query 2-77:

SELECT catalog.*, description, unit_price, unit,

unit_descr, manu_name, lead_time
FROM catalog, stock, manufact
WHERE catalog.stock_num = stock.stock_num
AND stock.manu_code = manufact.manu_code
AND catalog_num = 10025

Query 2-77

2-78 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

Query 2-77 uses a wildcard to select all columns from the table having the
most columns and then specifies columns from the other two tables. It pro-
duces the natural join, shown in Query Result 2-77, that displays the same
information as the previous example, but without duplication:

catalog_num 10025
stock_num 106
manu_code PRC
cat_descr
Hard anodized alloy with pearl finish. 6mm hex bolt hardware.
Available in lengths of 90-140mm in 10mm increments.
cat_picture <BYTE value>

cat_advert ProCycle Stem with Pearl Finish

description bicycle stem
unit_price $23.00
unit each
unit_descr each
manu_name ProCycle
lead_time 9

Query Result 2-77

Some Query Shortcuts

You can use aliases, the INTO TEMP clause, and display labels to speed your
way through joins and multiple-table queries and to produce output for other
uses.

Using Aliases
You can make multiple-table queries shorter and more readable by assigning
aliases to the tables in a SELECT statement. An alias is a word that immediately
follows the name of a table in the FROM clause. You can use it wherever the
table name would be used, for instance, as a prefix to the column names in
the other clauses.

SELECT s.stock_num, s.manu_code, s.description,

s.unit_price, s.unit, c.catalog_num,
c.cat_descr, c.cat_advert, m.lead_time
FROM stock s, catalog c, manufact m
WHERE s.stock_num = c.stock_num
AND s.manu_code = c.manu_code
AND s.manu_code = m.manu_code
AND s.manu_code IN ('HRO', 'HSK')
AND s.stock_num BETWEEN 100 AND 301
ORDER BY catalog_num

Query 2-78a

Composing Simple SELECT Statements 2-79

Multiple-Table SELECT Statements

The associative nature of the SELECT statement allows you to use an alias
before you define it. In Query 2-78a, the aliases s for the stock table, c for the
catalog table, and m for the manufact table are specified in the FROM clause
and used throughout the SELECT and WHERE clauses as column prefixes.
Compare the length of Query 2-78a with Query 2-78b, which does not use
aliases:

SELECT stock.stock_num, stock.manu_code, stock.description,

stock.unit_price, stock.unit, catalog.catalog_num,
catalog.cat_descr, catalog.cat_advert,
manufact.lead_time
FROM stock, catalog, manufact
WHERE stock.stock_num = catalog.stock_num
AND stock.manu_code = catalog.manu_code
AND stock.manu_code = manufact.manu_code
AND stock.manu_code IN ('HRO', 'HSK')
AND stock.stock_num BETWEEN 100 AND 301
ORDER BY catalog_num

Query 2-78b

2-80 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

Query 2-78a and Query 2-78b are equivalent and retrieve the data shown in
Query Result 2-78:

stock_num 110
manu_code HRO
description helmet
unit_price $260.00
unit case
catalog_num 10033
cat_descr
Newest ultralight helmet uses plastic shell. Largest ventilation
channels of any helmet on the market. 8.5 oz.
cat_advert Lightweight Plastic Slatted with Vents Assures Cool
Comfort Without Sacrificing Protection
lead_time 4

stock_num 110
manu_code HSK
description helmet
unit_price $308.00
unit each
catalog_num 10034
cat_descr
Aerodynamic (teardrop) helmet covered with anti-drag fabric.
Credited with shaving 2 seconds/mile from winner’s time in
Tour de France time-trial. 7.5 oz.
cat_advert Teardrop Design Endorsed by Yellow Jerseys,
You Can Time the Difference
lead_time 5

stock_num 205
manu_code HRO
description 3 golf balls
unit_price $312.00
unit each
catalog_num 10048
cat_descr
Combination fluorescent yellow and standard white.
cat_advert HiFlier Golf Balls: Case Includes Fluorescent
Yellow and Standard White
lead_time 4

stock_num 301
manu_code HRO
description running shoes
unit_price $42.50
unit each
catalog_num 10050
cat_descr
Engineered for serious training with exceptional stability.
Fabulous shock absorption. Great durability. Specify
mens/womens, size.
cat_advert Pronators and Supinators Take Heart: A Serious
Training Shoe For Runners Who Need Motion Control
lead_time 4

Query Result 2-78

Note that you cannot ORDER BY the TEXT column cat_descr or the BYTE
column cat_picture.

Composing Simple SELECT Statements 2-81

Multiple-Table SELECT Statements

You also can use aliases to shorten your queries on external tables residing in
external databases.
Query 2-79 joins columns from two tables that reside in different databases
and systems, neither of which is the current database or system:

SELECT order_num, lname, fname, phone

FROM masterdb@central:customer c, sales@western:orders o
WHERE c.customer_num = o.customer_num
AND order_num <= 1010

Query 2-79

By assigning the aliases c and o to the long database@system:table names,

masterdb@central:customer and sales@western:orders, respectively, you
can use the aliases to shorten the expression in the WHERE clause and retrieve
the data shown in Query Result 2-79:

order_num lname fname phone

1001 Higgins Anthony 415-368-1100

1002 Pauli Ludwig 408-789-8075
1003 Higgins Anthony 415-368-1100
1004 Watson George 415-389-8789
1005 Parmelee Jean 415-534-8822
1006 Lawson Margaret 415-887-7235
1007 Sipes Arnold 415-245-4578
1008 Jaeger Roy 415-743-3611
1009 Keyes Frances 408-277-7245
1010 Grant Alfred 415-356-1123
Query Result 2-79

For more information on external tables and external databases, see

Chapter 12, “Understanding Informix Networking,” in this manual and
Chapter 1 in the Informix Guide to SQL: Syntax.
You also can use synonyms as shorthand references to the long names of exter-
nal and current tables and views. For details on how to create and use syn-
onyms, see Chapter 12, “Understanding Informix Networking,” in this
manual and the CREATE SYNONYM statement in Chapter 1 of the Informix
Guide to SQL: Syntax.

2-82 Composing Simple SELECT Statements

 Multiple-Table SELECT Statements

The INTO TEMP Clause

By adding an INTO TEMP clause to your SELECT statement, you can
temporarily save the results of a multiple-table query in a separate table that
can be queried or manipulated without modifying the database. Temporary
tables are dropped when you end your SQL session or when your program or
report terminates.
Query 2-80 creates a temporary table called stockman and stores the results
of the query in it. Because all columns in a temp table must have names, the
alias adj_price is required.

SELECT DISTINCT stock_num, manu_name, description,

unit_price, unit_price * 1.05 adj_price
FROM stock, manufact
WHERE manufact.manu_code = stock.manu_code
INTO TEMP stockman

Query 2-80

stock_num manu_name description unit_price adj_price

1 Hero baseball gloves $250.00 $262.5000

1 Husky baseball gloves $800.00 $840.0000
1 Smith baseball gloves $450.00 $472.5000
2 Hero baseball $126.00 $132.3000
3 Husky baseball bat $240.00 $252.0000
4 Hero football $480.00 $504.0000
4 Husky football $960.00 $1008.0000
.
.
.
306 Shimara tandem adapter $190.00 $199.5000
307 ProCycle infant jogger $250.00 $262.5000
308 ProCycle twin jogger $280.00 $294.0000
309 Hero ear drops $40.00 $42.0000
309 Shimara ear drops $40.00 $42.0000
310 Anza kick board $84.00 $88.2000
310 Shimara kick board $80.00 $84.0000
311 Shimara water gloves $48.00 $50.4000
312 Hero racer goggles $72.00 $75.6000
312 Shimara racer goggles $96.00 $100.8000
313 Anza swim cap $60.00 $63.0000
313 Shimara swim cap $72.00 $75.6000

Query Result 2-80

You can query on this table and join it with other tables, which avoids a mul-
tiple sort and lets you move more quickly through the database. Temporary
tables are discussed at greater length in Chapter 13, “Optimizing Your Que-
ries.”

Composing Simple SELECT Statements 2-83

Summary

Summary
This chapter introduced sample syntax and results for basic kinds of SELECT
statements that are used to query on a relational database. Earlier sections of
the chapter showed how to perform the following activities:
• Select all columns and rows from a table with the SELECT and FROM
clauses.
• Select specific columns from a table with the SELECT and FROM clauses.
• Select specific rows from a table with the SELECT, FROM, and WHERE
clauses.
• Use the DISTINCT or UNIQUE keyword in the SELECT clause to eliminate
duplicate rows from query results.
• Sort retrieved data with the ORDER BY clause and the DESC keyword.
• Select and order data containing foreign characters.
• Use the BETWEEN, IN, MATCHES, and LIKE keywords and various rela-
tional operators in the WHERE clause to create a comparison condition.
• Create comparison conditions that include values, exclude values, find a
range of values (with keywords, relational operators, and subscripting),
and find a subset of values.
• Perform variable text searches using exact text comparisons, variable-
length wildcards, and restricted and unrestricted wildcards.
• Use the logical operators AND, OR, and NOT to connect search conditions
or Boolean expressions in a WHERE clause.
• Use the ESCAPE keyword to protect special characters in a query.
• Search for null values with the IS NULL and IS NOT NULL keywords in the
WHERE clause.
• Use arithmetic operators in the SELECT clause to perform computations
on number fields and display derived data.
• Use substrings and subscripting to tailor your queries.
• Assign display labels to computed columns as a formatting tool for
reports.
• Use the aggregate functions COUNT, AVG, MAX, MIN, and SUM in the
SELECT clause to calculate and retrieve specific data.
• Include the time functions DATE, DAY, MDY, MONTH, WEEKDAY, YEAR,
CURRENT, and EXTEND plus the TODAY, LENGTH, and USER functions in
your SELECT statements.
• Include stored procedures in your SELECT statements.

2-84 Composing Simple SELECT Statements

 Summary

This chapter also introduced simple join conditions that enable you to select
and display data from two or more tables. The section “Multiple-Table
SELECT Statements” described how to perform the following actions:
• Create a Cartesian product.
• Constrain a Cartesian product by including a WHERE clause with a valid
join condition in your query.
• Define and create a natural join and an equi-join.
• Join two or more tables on one or more columns.
• Use aliases as a shortcut in multiple-table queries.
• Retrieve selected data into a separate, temporary table with the INTO
TEMP clause to perform computations outside the database.

The next chapter explains more complex queries and subqueries; self-joins
and outer joins; the GROUP BY and HAVING clauses; and the UNION,
INTERSECTION, and DIFFERENCE set operations.

Composing Simple SELECT Statements 2-85

Summary

2-86 Composing Simple SELECT Statements

 Chapter

Composing
Advanced SELECT
Statements
Chapter Overview 3
3
Using the GROUP BY and HAVING Clauses 4
Using the GROUP BY Clause 4
Using the HAVING Clause 9
Creating Advanced Joins 11
Self-Joins 11
Outer Joins 21
Simple Join 22
Simple Outer Join on Two Tables 24
Outer Join for a Simple Join to a Third Table 26
Outer Join for an Outer Join to a Third Table 28
Outer Join of Two Tables to a Third Table 30
Subqueries in SELECT Statements 32
Using ALL 33
Using ANY 34
Single-Valued Subqueries 36
Correlated Subqueries 37
Using EXISTS 38
Set Operations 42
Union 43
Intersection 51
Difference 53
Summary 54
3-2 Composing Advanced SELECT Statements
Chapter Overview
Chapter 2, “Composing Simple SELECT Statements,” demonstrated some of
the basic ways to retrieve data from a relational database with the SELECT
statement. This chapter increases the scope of what you can do with this pow-
erful SQL statement and enables you to perform more complex database
queries and data manipulation.
Whereas the previous chapter focused on five of the clauses in SELECT
statement syntax, this chapter adds two more. You can use the GROUP BY
clause with aggregate functions to organize rows returned by the FROM
clause. You can include a HAVING clause to place conditions on the values
returned by the GROUP BY clause.
This chapter extends the earlier discussion of joins. It illustrates self-joins,
which enable you to join a table to itself, and four kinds of outer joins, where
you apply the keyword OUTER to treat two or more joined tables unequally.
It also introduces correlated and uncorrelated subqueries and their opera-
tional keywords, shows how to combine queries with the UNION operator,
and defines the set operations known as union, intersection, and difference.

Composing Advanced SELECT Statements 3-3

Using the GROUP BY and HAVING Clauses

Examples in this chapter show how to use some or all of the SELECT
statement clauses in your queries. The clauses must appear in the following
order:
1. SELECT
2. FROM
3. WHERE
4. GROUP BY
5. HAVING
6. ORDER BY
7. INTO TEMP
An additional SELECT statement clause, INTO, which you can use to specify
program and host variables in INFORMIX-4GL and SQL APIs is described in
Chapter 5, “Programming with SQL,” as well as in the manuals that come
with the product.

Using the GROUP BY and HAVING Clauses

The optional GROUP BY and HAVING clauses add functionality to your
SELECT statement. You can include one or both in a basic SELECT statement
to increase your ability to manipulate aggregates.
The GROUP BY clause combines similar rows, producing a single result row
for each group of rows that have the same values for each column listed in the
select list. The HAVING clause sets conditions on those groups after you form
them. You can use a GROUP BY clause without a HAVING clause, or a HAVING
clause without a GROUP BY clause.

Using the GROUP BY Clause

The GROUP BY clause divides a table into sets. It is most often combined with
aggregate functions that produce summary values for each of those sets.
Some of the examples in Chapter 2, “Composing Simple SELECT State-
ments” show the use of aggregate functions applied to a whole table. This
chapter illustrates aggregate functions applied to groups of rows.

3-4 Composing Advanced SELECT Statements

 Using the GROUP BY and HAVING Clauses

Using the GROUP BY clause without aggregates is much like using the
DISTINCT (or UNIQUE) keyword in the SELECT clause. Chapter 2, “Compos-
ing Simple SELECT Statements” included the statement found in Query 3-1a:

SELECT DISTINCT customer_num FROM orders

Query 3-1a

You also could write the statement as shown in Query 3-1b:

SELECT customer_num
FROM orders
GROUP BY customer_num

Query 3-1b

Query 3-1a and Query 3-1b return the rows shown in Query Result 3-1:

customer_num

101
104
106
110
111
112
115
116
117
119
120
121
122
123
124
126
127

Query Result 3-1

The GROUP BY clause collects the rows into sets so that each row in each set
has equal customer numbers. With no other columns selected, the result is a
list of the unique customer_num values.
The power of the GROUP BY clause is more apparent when you use it with
aggregate functions.

Composing Advanced SELECT Statements 3-5

Using the GROUP BY and HAVING Clauses

Query 3-2 retrieves the number of items and the total price of all items for
each order:

SELECT order_num, COUNT (*) number, SUM (total_price) price

FROM items
GROUP BY order_num

Query 3-2

The GROUP BY clause causes the rows of the items table to be collected into
groups, each group composed of rows that have identical order_num values
(that is, the items of each order are grouped together). After you form the
groups, the aggregate functions COUNT and SUM are applied within each
group.
Query 3-2 returns one row for each group. It uses labels to give names to the
results of the COUNT and SUM expressions, as shown in Query Result 3-2:

order_num number price

1001 1 $250.00
1002 2 $1200.00
1003 3 $959.00
1004 4 $1416.00
1005 4 $562.00
1006 5 $448.00
1007 5 $1696.00
1008 2 $940.00
.
.
.
1015 1 $450.00
1016 4 $654.00
1017 3 $584.00
1018 5 $1131.00
1019 1 $1499.97
1020 2 $438.00
1021 4 $1614.00
1022 3 $232.00
1023 6 $824.00

Query Result 3-2

Query Result 3-2 collects the rows of the items table into groups that have
identical order numbers and computes the COUNT of rows in each group and
the sum of the prices.

3-6 Composing Advanced SELECT Statements

 Using the GROUP BY and HAVING Clauses

Note that you cannot include a column having a TEXT or BYTE data type in a
GROUP BY clause. To group, you must be able to sort, and no natural sort order
exists for TEXT or BYTE data.
Unlike the ORDER BY clause, the GROUP BY clause does not order data.
Include an ORDER BY clause after your GROUP BY clause if you want to sort
data in a particular order or to sort on an aggregate in the select list.
Query 3-3 is the same as Query 3-2 but includes an ORDER BY clause to sort
the retrieved rows in ascending order of price, as shown in Query Result 3-3:

SELECT order_num, COUNT(*) number, SUM (total_price) price

FROM items
GROUP BY order_num
ORDER BY price

Query 3-3

order_num number price

1010 2 $84.00
1011 1 $99.00
1013 4 $143.80
1022 3 $232.00
1001 1 $250.00
1020 2 $438.00
1006 5 $448.00
1015 1 $450.00
1009 1 $450.00
.
.
.
1018 5 $1131.00
1002 2 $1200.00
1004 4 $1416.00
1014 2 $1440.00
1019 1 $1499.97
1021 4 $1614.00
1007 5 $1696.00

Query Result 3-3

As stated in Chapter 2, “Composing Simple SELECT Statements,” you can

use an integer in an ORDER BY clause to indicate the position of a column in
the select list. You also can use an integer in a GROUP BY clause to indicate the
position of column names or display labels in the group list.

Composing Advanced SELECT Statements 3-7

Using the GROUP BY and HAVING Clauses

Query 3-3 returns the same rows as Query 3-4, as shown in Query Result 3-3.

SELECT order_num, COUNT(*) number, SUM (total_price) price

FROM items
GROUP BY 1
ORDER BY 3

Query 3-4

When you build a query, remember that all nonaggregate columns that are in
the select list in the SELECT clause must also be included in the group list in
the GROUP BY clause. The reason for this is that a SELECT with GROUP BY
must return only one row per group. Columns that are listed after GROUP BY
are certain to reflect only one distinct value within a group, and that value
can be returned. However, a column not listed after GROUP BY might contain
different values in the rows that are contained in a group.
As shown in Query 3-5, you can use the GROUP BY clause in a SELECT
statement that joins tables:

SELECT o.order_num, SUM (i.total_price)

FROM orders o, items i
WHERE o.order_date > '01/01/92'
AND o.customer_num = 110
AND o.order_num = i.order_num
GROUP BY o.order_num

Query 3-5

Query 3-5 joins the orders and items tables, assigns table aliases to them, and
returns the rows shown in Query Result 3-5:

order_num (sum)

1008 $940.00
1015 $450.00

Query Result 3-5

3-8 Composing Advanced SELECT Statements

 Using the GROUP BY and HAVING Clauses

Using the HAVING Clause

The HAVING clause usually complements a GROUP BY clause by applying
one or more qualifying conditions to groups after they are formed, which is
similar to the way the WHERE clause qualifies individual rows. One
advantage to using a HAVING clause is that you can include aggregates in the
search condition, whereas you cannot include aggregates in the search
condition of a WHERE clause.
Each HAVING condition compares one column or aggregate expression of the
group with another aggregate expression of the group or with a constant. You
can use HAVING to place conditions on both column values and aggregate
values in the group list.
Query 3-6 returns the average total price per item on all orders that have
more than two items. The HAVING clause tests each group as it is formed and
selects those composed of two or more rows.

SELECT order_num, COUNT(*) number, AVG (total_price) average

FROM items
GROUP BY order_num
HAVING COUNT(*) > 2

Query 3-6

order_num number average

1003 3 $319.67
1004 4 $354.00
1005 4 $140.50
1006 5 $89.60
1007 5 $339.20
1013 4 $35.95
1016 4 $163.50
1017 3 $194.67
1018 5 $226.20
1021 4 $403.50
1022 3 $77.33
1023 6 $137.33

Query Result 3-6

If you use a HAVING clause without a GROUP BY clause, the HAVING

condition applies to all rows that satisfy the search condition. In other words,
all rows that satisfy the search condition make up a single group.

Composing Advanced SELECT Statements 3-9

Using the GROUP BY and HAVING Clauses

Query 3-7, a modified version of Query 3-6, returns just one row, the average
of all total_price values in the table:

SELECT AVG (total_price) average

FROM items
HAVING count(*) > 2

Query 3-7

average

$270.97

Query Result 3-7

If Query 3-7, like Query 3-6, had included the nonaggregate column
order_ num in the select list, you would have to include a GROUP BY clause
with that column in the group list. In addition, if the condition in the HAVING
clause was not satisfied, the output would show the column heading and a
message would indicate that no rows were found.
Query 3-8 contains all the SELECT statement clauses that you can use in the
Informix version of interactive SQL (the INTO clause naming program or host
variables is available only in an INFORMIX-4GL or an SQL API:

SELECT o.order_num, SUM (i.total_price) price,

paid_date - order_date span
FROM orders o, items i
WHERE o.order_date > '01/01/92'
AND o.customer_num > 110
AND o.order_num = i.order_num
GROUP BY 1, 3
HAVING COUNT (*) < 5
ORDER BY 3
INTO TEMP temptab1

Query 3-8

3-10 Composing Advanced SELECT Statements

 Creating Advanced Joins

Query 3-8 joins the orders and items tables; employs display labels, table
aliases, and integers used as column indicators; groups and orders the data;
and puts the following results in a temporary table, as shown in Query Result
3-8:

order_num price span

1017 $584.00
1016 $654.00
1012 $1040.00
1019 $1499.97 26
1005 $562.00 28
1021 $1614.00 30
1022 $232.00 40
1010 $84.00 66
1009 $450.00 68
1020 $438.00 71

Query Result 3-8

Creating Advanced Joins

Chapter 2, “Composing Simple SELECT Statements,” showed how to
include a WHERE clause in a SELECT statement to join two or more tables on
one or more columns. It illustrated natural joins and equi-joins.
This chapter discusses the uses of two more complex kinds of joins: self-joins
and outer joins. As described for simple joins, you can define aliases for tables
and assign display labels to expressions to shorten your multiple-table que-
ries. You can also sort data with an ORDER BY clause and SELECT query
results into a temporary table.

Self-Joins
A join does not always have to involve two different tables. You can join a
table to itself, creating a self-join. This can be useful when you want to
compare values in a column to other values in the same column.
To create a self-join, list a table twice in the FROM clause, assigning it a
different alias each time. Use the aliases to refer to the table in the SELECT and
WHERE clauses as if it were two separate tables. (Aliases in SELECT state-
ments are shown in Chapter 2, “Composing Simple SELECT Statements” of
this manual and discussed in Chapter 1 of the Informix Guide to SQL: Syntax.)

Composing Advanced SELECT Statements 3-11

Creating Advanced Joins

Just as in joins between tables, you can use arithmetic expressions in

self-joins. You can test for null values, and you can ORDER BY a specified
column in ascending or descending order.
Query 3-9 finds pairs of orders where the ship_weight differs by a factor of
five or more and the ship_date is not null, and orders the data by
ship_date:

SELECT x.order_num, x.ship_weight, x.ship_date,

y.order_num, y.ship_weight, y.ship_date
FROM orders x, orders y
WHERE x.ship_weight >= 5 * y.ship_weight
AND x.ship_date IS NOT NULL
AND y.ship_date IS NOT NULL
ORDER BY x.ship_date

Query 3-9

order_num ship_weight ship_date order_num ship_weight ship_date

1004 95.80 05/30/1993 1011 10.40 07/03/1993

1004 95.80 05/30/1993 1020 14.00 07/16/1993
1004 95.80 05/30/1993 1022 15.00 07/30/1993
1007 125.90 06/05/1993 1015 20.60 07/16/1993
1007 125.90 06/05/1993 1020 14.00 07/16/1993
1007 125.90 06/05/1993 1022 15.00 07/30/1993
1007 125.90 06/05/1993 1011 10.40 07/03/1993
1007 125.90 06/05/1993 1001 20.40 06/01/1993
1007 125.90 06/05/1993 1009 20.40 06/21/1993
1005 80.80 06/09/1993 1011 10.40 07/03/1993
1005 80.80 06/09/1993 1020 14.00 07/16/1993
1005 80.80 06/09/1993 1022 15.00 07/30/1993
1012 70.80 06/29/1993 1011 10.40 07/03/1993
1012 70.80 06/29/1993 1020 14.00 07/16/1993
1013 60.80 07/10/1993 1011 10.40 07/03/1993
1017 60.00 07/13/1993 1011 10.40 07/03/1993
1018 70.50 07/13/1993 1011 10.40 07/03/1993
.
.
.

Query Result 3-9

Suppose you want to select the results of a self-join into a temporary table.
You would, of course, append an INTO TEMP clause to the SELECT statement.
However, because you are, in effect, creating a new table, you also must
rename at least one set of column names by assigning them display labels.
Otherwise, you get an error message that indicates duplicate column names,
and the temporary table is not created.

3-12 Composing Advanced SELECT Statements

 Creating Advanced Joins

Query 3-10, which is similar to Query 3-9, labels all columns selected from the
orders table and puts them in a temporary table called shipping:

SELECT x.order_num orders1, x.po_num purch1,

x.ship_date ship1, y.order_num orders2,
y.po_num purch2, y.ship_date ship2
FROM orders x, orders y
WHERE x.ship_weight >= 5 * y.ship_weight
AND x.ship_date IS NOT NULL
AND y.ship_date IS NOT NULL
ORDER BY orders1, orders2
INTO TEMP shipping

Query 3-10

If you SELECT * from that table, you see the rows in Query Result 3-10:

orders1 purch1 ship1 orders2 purch2 ship2

1004 8006 05/30/1993 1011 B77897 07/03/1993

1004 8006 05/30/1993 1020 W2286 07/16/1993
1004 8006 05/30/1993 1022 W9925 07/30/1993
1005 2865 06/09/1993 1011 B77897 07/03/1993
1005 2865 06/09/1993 1020 W2286 07/16/1993
1005 2865 06/09/1993 1022 W9925 07/30/1993
1007 278693 06/05/1993 1001 B77836 06/01/1993
1007 278693 06/05/1993 1009 4745 06/21/1993
1007 278693 06/05/1993 1011 B77897 07/03/1993
1007 278693 06/05/1993 1015 MA003 07/16/1993
1007 278693 06/05/1993 1020 W2286 07/16/1993
1007 278693 06/05/1993 1022 W9925 07/30/1993
1012 278701 06/29/1993 1011 B77897 07/03/1993
1012 278701 06/29/1993 1020 W2286 07/16/1993
1013 B77930 07/10/1993 1011 B77897 07/03/1993
1017 DM354331 07/13/1993 1011 B77897 07/03/1993
1018 S22942 07/13/1993 1011 B77897 07/03/1993
1018 S22942 07/13/1993 1020 W2286 07/16/1993
1019 Z55709 07/16/1993 1011 B77897 07/03/1993
1019 Z55709 07/16/1993 1020 W2286 07/16/1993
1019 Z55709 07/16/1993 1022 W9925 07/30/1993
1023 KF2961 07/30/1993 1011 B77897 07/03/1993

Query Result 3-10

You can join a table to itself more than once. The maximum number of
self-joins depends on the resources available to you.

Composing Advanced SELECT Statements 3-13

Creating Advanced Joins

The self-join in Query 3-11 creates a list of those items in the stock table that
are supplied by three manufacturers. By including the last two conditions in
the WHERE clause, it eliminates duplicate manufacturer codes in rows
retrieved.

SELECT s1.manu_code, s2.manu_code, s3.manu_code,

s1.stock_num, s1.description
FROM stock s1, stock s2, stock s3
WHERE s1.stock_num = s2.stock_num
AND s2.stock_num = s3.stock_num
AND s1.manu_code < s2.manu_code
AND s2.manu_code < s3.manu_code
ORDER BY stock_num

Query 3-11

manu_code manu_code manu_code stock_num description

HRO HSK SMT 1 baseball gloves

ANZ NRG SMT 5 tennis racquet
ANZ HRO HSK 110 helmet
ANZ HRO PRC 110 helmet
ANZ HRO SHM 110 helmet
ANZ HSK PRC 110 helmet
ANZ HSK SHM 110 helmet
ANZ PRC SHM 110 helmet
HRO HSK PRC 110 helmet
HRO HSK SHM 110 helmet
HRO PRC SHM 110 helmet
HSK PRC SHM 110 helmet
ANZ KAR NKL 201 golf shoes
ANZ HRO NKL 205 3 golf balls
ANZ HRO KAR 301 running shoes
.
.
.
HRO PRC SHM 301 running shoes
KAR NKL PRC 301 running shoes
KAR NKL SHM 301 running shoes
KAR PRC SHM 301 running shoes
NKL PRC SHM 301 running shoes

Query Result 3-11

3-14 Composing Advanced SELECT Statements

 Creating Advanced Joins

Say you want to select rows from a payroll table to determine which
employees earn more than their manager. You can construct the self-join
shown in Query 3-12a:

SELECT emp.employee_num, emp.gross_pay, emp.level,

emp.dept_num, mgr.employee_num, mgr.gross_pay,
mgr.dept_num, mgr.level
FROM payroll emp, payroll mgr
WHERE emp.gross_pay > mgr.gross_pay
AND emp.level < mgr.level
AND emp.dept_num = mgr.dept_num
ORDER BY 4

Query 3-12a

Query 3-12b uses a correlated subquery to retrieve and list the 10 highest-priced
items ordered:

SELECT order_num, total_price

FROM items a
WHERE 10 >
(SELECT COUNT (*)
FROM items b
WHERE b.total_price < a.total_price)
ORDER BY total_price

Query 3-12b

Query 3-12b returns the 10 rows shown in Query Result 3-12:

order_num total_price

1018 $15.00
1013 $19.80
1003 $20.00
1005 $36.00
1006 $36.00
1013 $36.00
1010 $36.00
1013 $40.00
1022 $40.00
1023 $40.00

Query Result 3-12

You can create a similar query to find and list the 10 employees in the
company who have the most seniority.

Composing Advanced SELECT Statements 3-15

Creating Advanced Joins

Correlated and uncorrelated subqueries are described later in “Subqueries in

SELECT Statements” on page 3-32.

Using Rowid Values

You can use the hidden rowid column in a self-join to locate duplicate values
in a table. In the following example, the condition x.rowid != y.rowid is
equivalent to saying “row x is not the same row as row y.”
Query 3-13 selects data twice from the cust_calls table, assigning it the table
aliases x and y:

SELECT x.rowid, x.customer_num

FROM cust_calls x, cust_calls y
WHERE x.customer_num = y.customer_num
AND x.rowid != y.rowid

Query 3-13

Query 3-13 searches for duplicate values in the customer_num column, and
for their rowids, finding the pair shown in Query Result 3-13:

rowid customer_num

515 116
769 116

Query Result 3-13

You can write the last condition in Query 3-13 as shown in the following
example:

AND x.rowid != y.rowid

AND NOT x.rowid = y.rowid

3-16 Composing Advanced SELECT Statements

 Creating Advanced Joins

Another way to locate duplicate values is with a correlated subquery, as

shown in Query 3-14:

SELECT x.customer_num, x.call_dtime

FROM cust_calls x
WHERE 1 <
(SELECT COUNT (*) FROM cust_calls y
WHERE x.customer_num = y.customer_num)

Query 3-14

Query 3-14 locates the same two duplicate customer_num values as Query
3-13 and returns the rows shown in Query Result 3-14:

customer_num call_dtime

116 1992-11-28 13:34

116 1992-12-21 11:24

Query Result 3-14

You can use the rowid, shown earlier in a self-join, to locate the internal
record number associated with a row in a database table. The rowid is, in
effect, a hidden column in every table. The sequential values of rowid have
no special significance and may vary depending on the location of the phys-
ical data in the chunk. Your rowid may vary from the example shown. See
Chapter 13, “Optimizing Your Queries” in this manual for a discussion of
performance issues and the rowid value. The use of rowid is also discussed
in detail in the INFORMIX-OnLine Dynamic Server Administrator’s Guide.

Composing Advanced SELECT Statements 3-17

Creating Advanced Joins

Query 3-16 uses rowid and the wildcard * in the SELECT clause to retrieve
every row in the manufact table and their corresponding rowids:

SELECT rowid, * FROM manufact

Query 3-15

rowid manu_code manu_name lead_time

257 SMT Smith 3

258 ANZ Anza 5
259 NRG Norge 7
260 HSK Husky 5
261 HRO Hero 4
262 SHM Shimara 30
263 KAR Karsten 21
264 NKL Nikolus 8
265 PRC ProCycle 9

Query Result 3-15

You also can use rowid when you select a specific column, as shown in Query
3-16:

SELECT rowid, manu_code FROM manufact

Query 3-16

rowid manu_code

258 ANZ
261 HRO
260 HSK
263 KAR
264 NKL
259 NRG
265 PRC
262 SHM
257 SMT

Query Result 3-16

3-18 Composing Advanced SELECT Statements

 Creating Advanced Joins

You also can use the rowid in the WHERE clause to retrieve rows based on
their internal record number. This method is handy when no other unique
column exists in a table. Query 3-17 uses the rowid from Query 3-16:

SELECT * FROM manufact WHERE rowid = 263

Query 3-17

Query 3-17 returns the row shown in Query Result 3-17:

manu_code manu_name lead_time

KAR Karsten 21

Query Result 3-17

Using the USER Function

To obtain additional information about a table, you can combine the rowid
with the USER function.
Query 3-18 assigns the label username to the USER expression column and
returns this information about the cust_calls table:

SELECT USER username, rowid FROM cust_calls

Query 3-18

username rowid

zenda 257
zenda 258
zenda 259
zenda 513
zenda 514
zenda 515
zenda 769

Query Result 3-18

You also can use the USER function in a WHERE clause when you select the
rowid.

Composing Advanced SELECT Statements 3-19

Creating Advanced Joins

Query 3-19 returns the rowid for only those rows inserted or updated by the
user who performs the query:

SELECT rowid FROM cust_calls WHERE user_id = USER

Query 3-19

For example, if the user richc used Query 3-19, the output would be as shown
in Query Result 3-19:

rowid

258
259

Query Result 3-19

Using the DBSERVERNAME Function

With INFORMIX-OnLine Dynamic Server, you can add the DBSERVERNAME
keyword (or its synonym, SITENAME) to a query to find out where the current
database resides.
Query 3-20 finds the database server name and the user name as well as the
rowid and the tabid, which is the serial interval table identifier for system
catalog tables:

SELECT DBSERVERNAME server, tabid, rowid, USER username

FROM systables
WHERE tabid >= 105 OR rowid <= 260
ORDER BY rowid

Query 3-20

3-20 Composing Advanced SELECT Statements

 Creating Advanced Joins

Query 3-20 assigns display labels to the DBSERVERNAME and USER

expressions and returns the 10 rows from the systables system catalog table,
as shown in Query Result 3-20:

server tabid rowid username

manatee 1 257 zenda

manatee 2 258 zenda
manatee 3 259 zenda
manatee 4 260 zenda
manatee 105 274 zenda
manatee 106 1025 zenda
manatee 107 1026 zenda
manatee 108 1027 zenda
manatee 109 1028 zenda
manatee 110 1029 zenda

Query Result 3-20

Note that you should never store a rowid in a permanent table or attempt to
use it as a foreign key because the rowid can change. For example, if a table
is dropped and then reloaded from external data, all the rowids are different.
USER and DBSERVERNAME were discussed in Chapter 2, “Composing Sim-
ple SELECT Statements.”

Outer Joins
Chapter 2, “Composing Simple SELECT Statements,” showed how to create
and use some simple joins. Whereas a simple join treats two or more joined
tables equally, an outer join treats two or more joined tables unsymmetrically.
It makes one of the tables dominant (also called “preserved”) over the other
subservient tables.
Outer joins occur in four basic types:
• A simple outer join on two tables
• A simple outer join to a third table
• An outer join for a simple join to a third table
• An outer join for an outer join to a third table
This section discusses these types of outer joins. See the discussion of outer
joins in Chapter 1 of the Informix Guide to SQL: Syntax for full information on
their syntax, use, and logic.

Composing Advanced SELECT Statements 3-21

Creating Advanced Joins

In a simple join, the result contains only the combinations of rows from the
tables that satisfy the join conditions. Rows that do not satisfy the join conditions
are discarded.
In an outer join, the result contains the combinations of rows from the tables
that satisfy the join conditions. Rows from the dominant table that would other-
wise be discarded are preserved, even though no matching row was found in the sub-
servient table. The dominant-table rows that do not have a matching
subservient-table row receive a row of nulls before the selected columns are
projected.
An outer join applies conditions to the subservient table while sequentially
applying the join conditions to the rows of the dominant table. The condi-
tions are expressed in a WHERE clause.
An outer join must have a SELECT clause, a FROM clause, and a WHERE
clause. You transform a simple join into an outer join by inserting the key-
word OUTER directly before the name of the subservient tables in the FROM
clause. As shown later in this section, you can include the OUTER keyword
more than once in your query.
Before you use outer joins heavily, you should determine whether one or
more simple joins can work. You often can get by with a simple join when you
do not need supplemental information from other tables.
The examples in this section use table aliases for brevity. Table aliases are
discussed in the preceding chapter.

Simple Join
Query 3-21 is an example of the type of simple join on the customer and
cust_calls tables shown in Chapter 2, “Composing Simple SELECT State-
ments.”

SELECT c.customer_num, c.lname, c.company,

c.phone, u.call_dtime, u.call_descr
FROM customer c, cust_calls u
WHERE c.customer_num = u.customer_num

Query 3-21

3-22 Composing Advanced SELECT Statements

 Creating Advanced Joins

Query 3-21 returns only those rows where the customer has made a call to
customer service, as shown in Query Result 3-21:

customer_num 106
lname Watson
company Watson & Son
phone 415-389-8789
call_dtime 1993-06-12 08:20
call_descr Order was received, but two of the cans of
ANZ tennis balls within the case were empty

customer_num 110
lname Jaeger
company AA Athletics
phone 415-743-3611
call_dtime 1993-07-07 10:24
call_descr Order placed one month ago (6/7) not received.

customer_num 119
lname Shorter
company The Triathletes Club
phone 609-663-6079
call_dtime 1993-07-01 15:00
call_descr Bill does not reflect credit from previous order

customer_num 121
lname Wallack
company City Sports
phone 302-366-7511
call_dtime 1993-07-10 14:05
call_descr Customer likes our merchandise. Requests that we
stock more types of infant joggers. Will call back
to place order.

customer_num 127
lname Satifer
company Big Blue Bike Shop
phone 312-944-5691
call_dtime 1993-07-31 14:30
call_descr Received Hero watches (item # 304) instead of
ANZ watches

customer_num 116
lname Parmelee
company Olympic City
phone 415-534-8822
call_dtime 1992-11-28 13:34
call_descr Received plain white swim caps (313 ANZ) instead
of navy with team logo (313 SHM)

customer_num 116
lname Parmelee
company Olympic City
phone 415-534-8822
call_dtime 1992-12-21 11:24
call_descr Second complaint from this customer! Received
two cases right-handed outfielder gloves (1 HRO)
instead of one case lefties.

Query Result 3-21

Composing Advanced SELECT Statements 3-23

Creating Advanced Joins

Simple Outer Join on Two Tables

Query 3-22 uses the same select list, tables, and comparison condition as the
preceding example, but this time it creates a simple outer join:

SELECT c.customer_num, c.lname, c.company,

c.phone, u.call_dtime, u.call_descr
FROM customer c, OUTER cust_calls u
WHERE c.customer_num = u.customer_num

Query 3-22

The addition of the keyword OUTER in front of the cust_calls table makes it
the subservient table. An outer join causes the query to return information on
all customers, whether or not they have made calls to customer service. All
rows from the dominant customer table are retrieved, and null values are
assigned to corresponding rows from the subservient cust_calls table.

3-24 Composing Advanced SELECT Statements

 Creating Advanced Joins

customer_num 101
lname Pauli
company All Sports Supplies
phone 408-789-8075
call_dtime
call_descr

customer_num 102
lname Sadler
company Sports Spot
phone 415-822-1289
call_dtime
call_descr

customer_num 103
lname Currie
company Phil’s Sports
phone 415-328-4543
call_dtime
call_descr

customer_num 104
lname Higgins
company Play Ball!
phone 415-368-1100
call_dtime
call_descr

customer_num 105
lname Vector
company Los Altos Sports
phone 415-776-3249
call_dtime
call_descr

customer_num 106
lname Watson
company Watson & Son
phone 415-389-8789
call_dtime 1993-06-12 08:20
call_descr Order was received, but two of the cans of
ANZ tennis balls within the case were empty

customer_num 107
lname Ream
company Athletic Supplies
phone 415-356-9876
call_dtime
call_descr

customer_num 108
lname Quinn
company Quinn’s Sports
phone 415-544-8729
call_dtime
call_descr
.
.
.

Query Result 3-22

Composing Advanced SELECT Statements 3-25

Creating Advanced Joins

Outer Join for a Simple Join to a Third Table

Query 3-23 shows an outer join that is the result of a simple join to a third
table. This second type of outer join is known as a nested simple join.

SELECT c.customer_num, c.lname, o.order_num,

i.stock_num, i.manu_code, i.quantity
FROM customer c, OUTER (orders o, items i)
WHERE c.customer_num = o.customer_num
AND o.order_num = i.order_num
AND manu_code IN ('KAR', 'SHM')
ORDER BY lname

Query 3-23

3-26 Composing Advanced SELECT Statements

 Creating Advanced Joins

Query 3-23 first performs a simple join on the orders and items tables,
retrieving information on all orders for items with a manu_code of KAR or
SHM. It then performs an outer join to combine this information with data
from the dominant customer table. An optional ORDER BY clause reorganizes
the data into the form shown in Query Result 3-23:

customer_num lname order_num stock_num manu_code quantity

114 Albertson
118 Baxter
113 Beatty
103 Currie
115 Grant
123 Hanlon 1020 301 KAR 4
123 Hanlon 1020 204 KAR 2
125 Henry
104 Higgins
110 Jaeger
120 Jewell 1017 202 KAR 1
120 Jewell 1017 301 SHM 2
111 Keyes
112 Lawson
128 Lessor
109 Miller
126 Neelie
122 O’Brian 1019 111 SHM 3
116 Parmelee
101 Pauli
124 Putnum 1021 202 KAR 3
108 Quinn
107 Ream
102 Sadler
127 Satifer 1023 306 SHM 1
127 Satifer 1023 105 SHM 1
127 Satifer 1023 110 SHM 1
119 Shorter 1016 101 SHM 2
117 Sipes
105 Vector
121 Wallack 1018 302 KAR 3
106 Watson

Query Result 3-23

Composing Advanced SELECT Statements 3-27

Creating Advanced Joins

Outer Join for an Outer Join to a Third Table

Query 3-24 creates an outer join that is the result of an outer join to a third
table. This third type is known as a nested outer join.

SELECT c.customer_num, lname, o.order_num,

stock_num, manu_code, quantity
FROM customer c, OUTER (orders o, OUTER items i)
WHERE c.customer_num = o.customer_num
AND o.order_num = i.order_num
AND manu_code IN ('KAR', 'SHM')
ORDER BY lname

Query 3-24

Query 3-24 first performs an outer join on the orders and items tables,
retrieving information on all orders for items with a manu_code of KAR or
SHM. It then performs an outer join, which combines this information with
data from the dominant customer table. Query 3-24 preserves order numbers

3-28 Composing Advanced SELECT Statements

 Creating Advanced Joins

that the previous example eliminated, returning rows for orders that do not
contain items with either manufacturer code. An optional ORDER BY clause
reorganizes the data, as shown in Query Result 3-24:

customer_num lname order_num stock_num manu_code quantity

114 Albertson
118 Baxter
113 Beatty
103 Currie
115 Grant 1010
123 Hanlon 1020 204 KAR 2
123 Hanlon 1020 301 KAR 4
125 Henry
104 Higgins 1011
104 Higgins 1001
104 Higgins 1013
104 Higgins 1003
110 Jaeger 1008
110 Jaeger 1015
120 Jewell 1017 301 SHM 2
120 Jewell 1017 202 KAR 1
111 Keyes 1009
112 Lawson 1006
128 Lessor
109 Miller
126 Neelie 1022
122 O’Brian 1019 111 SHM 3
116 Parmelee 1005
101 Pauli 1002
124 Putnum 1021 202 KAR 3
108 Quinn
107 Ream
102 Sadler
127 Satifer 1023 110 SHM 1
127 Satifer 1023 105 SHM 1
127 Satifer 1023 306 SHM 1
119 Shorter 1016 101 SHM 2
117 Sipes 1012
117 Sipes 1007
105 Vector
121 Wallack 1018 302 KAR 3
106 Watson 1014
106 Watson 1004

Query Result 3-24

You can state the join conditions in two ways when you apply an outer join
to the result of an outer join to a third table. The two subservient tables are
joined, but you can join the dominant table to either subservient table with-
out affecting the results if the dominant table and the subservient table share
a common column.

Composing Advanced SELECT Statements 3-29

Creating Advanced Joins

Outer Join of Two Tables to a Third Table

Query 3-25 shows an outer join that is the result of an outer join of each of two
tables to a third table. In this fourth type of outer join, join relationships are
possible only between the dominant table and the subservient tables.

SELECT c.customer_num, lname, o.order_num,

order_date, call_dtime
FROM customer c, OUTER orders o, OUTER cust_calls x
WHERE c.customer_num = o.customer_num
AND c.customer_num = x.customer_num
ORDER BY lname
INTO TEMP service

Query 3-25

3-30 Composing Advanced SELECT Statements

 Creating Advanced Joins

Query 3-25 individually joins the subservient tables orders and cust_calls to
the dominant customer table; it does not join the two subservient tables. An
INTO TEMP clause selects the results into a temporary table for further
manipulation or queries, as shown in Query Result 3-25.

customer_num lname order_num order_date call_dtime

114 Albertson
118 Baxter
113 Beatty
103 Currie
115 Grant 1010 06/17/1993
123 Hanlon 1020 07/11/1993
125 Henry
104 Higgins 1003 05/22/1993
104 Higgins 1001 05/20/1993
104 Higgins 1013 06/22/1993
104 Higgins 1011 06/18/1993
110 Jaeger 1015 06/27/1993 1993-07-07 10:24
110 Jaeger 1008 06/07/1993 1993-07-07 10:24
120 Jewell 1017 07/09/1993
111 Keyes 1009 06/14/1993
112 Lawson 1006 05/30/1993
109 Miller
128 Moore
126 Neelie 1022 07/24/1993
122 O’Brian 1019 07/11/1993
116 Parmelee 1005 05/24/1993 1992-12-21 11:24
116 Parmelee 1005 05/24/1993 1992-11-28 13:34
101 Pauli 1002 05/21/1993
124 Putnum 1021 07/23/1993
108 Quinn
107 Ream
102 Sadler
127 Satifer 1023 07/24/1993 1993-07-31 14:30
119 Shorter 1016 06/29/1993 1993-07-01 15:00
117 Sipes 1007 05/31/1993
117 Sipes 1012 06/18/1993
105 Vector
121 Wallack 1018 07/10/1993 1993-07-10 14:05
106 Watson 1004 05/22/1993 1993-06-12 08:20
106 Watson 1014 06/25/1993 1993-06-12 08:20

Query Result 3-25

Note that if Query 3-25 had tried to create a join condition between the two
subservient tables o and x, as shown in Query 3-26, an error message would
have indicated the creation of a two-sided outer join:

WHERE o.customer_num = x.customer_num

Query 3-26

Composing Advanced SELECT Statements 3-31

Subqueries in SELECT Statements

Subqueries in SELECT Statements

A SELECT statement nested in the WHERE clause of another SELECT statement
(or in an INSERT, DELETE, or UPDATE statement) is called a subquery. Each
subquery must contain a SELECT clause and a FROM clause, and it must be
enclosed in parentheses, which tells the database server to perform that
operation first.
Subqueries can be correlated or uncorrelated. A subquery (or inner SELECT
statement) is correlated when the value it produces depends on a value pro-
duced by the outer SELECT statement that contains it. Any other kind of
subquery is considered uncorrelated.
The important feature of a correlated subquery is that, because it depends on
a value from the outer SELECT, it must be executed repeatedly, once for every
value produced by the outer SELECT. An uncorrelated subquery is executed
only once.
You can construct a SELECT statement with a subquery to replace two
separate SELECT statements.
Subqueries in SELECT statements allow you to perform the following actions:
• Compare an expression to the result of another SELECT statement.
• Determine whether an expression is included in the results of another
SELECT statement.
• Determine whether any rows are selected by another SELECT statement.
An optional WHERE clause in a subquery often is used to narrow the search
condition.
A subquery selects and returns values to the first or outer SELECT statement.
A subquery can return no value, a single value, or a set of values.
• If it returns no value, the query does not return any rows. Such a subquery
is equivalent to a null value.
• If it returns one value, the subquery returns either one aggregate
expression or else selects exactly one row and one column. Such a
subquery is equivalent to a single number or character value.
• If it returns a list or set of values, the subquery returns either one row or
one column.

3-32 Composing Advanced SELECT Statements

 Subqueries in SELECT Statements

The following keywords introduce a subquery in the WHERE clause of a

SELECT statement:

• ALL
• ANY
• IN
• EXISTS

You can use any of the relational operators with ALL and ANY to compare
something to every one of (ALL), or to any one of (ANY), the values that the
subquery produces. You can use the keyword SOME in place of ANY. The
operator IN is equivalent to =ANY. To create the opposite search condition,
use the keyword NOT or a different relational operator.
The EXISTS operator tests a subquery to see if it found any values; that is, it
asks if the result of the subquery is not null.
See Chapter 1 in the Informix Guide to SQL: Syntax for the complete syntax
used in creating a condition with a subquery. See also Chapter 13,
“Optimizing Your Queries,” of this manual for information on performance
implications for correlated and uncorrelated subqueries.

Using ALL
Use the keyword ALL preceding a subquery to determine whether a
comparison is true for every value returned. If the subquery returns no val-
ues, the search condition is true. (If it returns no values, the condition is true
of all of the zero values.)

Composing Advanced SELECT Statements 3-33

Subqueries in SELECT Statements

Query 3-27 lists the following information for all orders that contain an item
for which the total price is less than the total price on every item in order
number 1023:

SELECT order_num, stock_num, manu_code, total_price

FROM items
WHERE total_price < ALL
(SELECT total_price FROM items
WHERE order_num = 1023)

Query 3-27

order_num stock_num manu_code total_price

1003 9 ANZ $20.00

1005 6 SMT $36.00
1006 6 SMT $36.00
1010 6 SMT $36.00
1013 5 ANZ $19.80
1013 6 SMT $36.00
1018 302 KAR $15.00

Query Result 3-27

Using ANY
Use the keyword ANY (or its synonym SOME) preceding a subquery to
determine whether a comparison is true for at least one of the values
returned. If the subquery returns no values, the search condition is false.
(Because there were no values, the condition cannot be true for one of them.)

3-34 Composing Advanced SELECT Statements

 Subqueries in SELECT Statements

Query 3-28 finds the order number of all orders that contain an item for
which the total price is greater than the total price of any one of the items in
order number 1005.

SELECT DISTINCT order_num

FROM items
WHERE total_price > ANY
(SELECT total_price
FROM items
WHERE order_num = 1005)

Query 3-28

order_num

1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023

Query Result 3-28

Composing Advanced SELECT Statements 3-35

Subqueries in SELECT Statements

Single-Valued Subqueries
You do not need to include the keyword ALL or ANY if you know the
subquery can return exactly one value to the outer-level query. A subquery that
returns exactly one value can be treated like a function. This kind of subquery
often uses an aggregate function because aggregate functions always return
single values
Query 3-29 uses the aggregate function MAX in a subquery to find the
order_num for orders that include the maximum number of volleyball nets.

SELECT order_num FROM items

WHERE stock_num = 9
AND quantity =
(SELECT MAX (quantity)
FROM items
WHERE stock_num = 9)

Query 3-29

order_num

1012

Query Result 3-29

3-36 Composing Advanced SELECT Statements

 Subqueries in SELECT Statements

Query 3-30 uses the aggregate function MIN in the subquery to select items
for which the total price is higher than 10 times the minimum price:

SELECT order_num, stock_num, manu_code, total_price

FROM items x
WHERE total_price >
(SELECT 10 * MIN (total_price)
FROM items
WHERE order_num = x.order_num)

Query 3-30

order_num stock_num manu_code total_price

1003 8 ANZ $840.00

1018 307 PRC $500.00
1018 110 PRC $236.00
1018 304 HRO $280.00

Query Result 3-30

Correlated Subqueries
Query 3-31 is an example of a correlated subquery, which returns a list of the
10 earliest shipping dates in the orders table. It includes an ORDER BY clause
after the subquery to order the results because you cannot include ORDER BY
within a subquery.

SELECT po_num, ship_date FROM orders main

WHERE 10 >
(SELECT COUNT (DISTINCT ship_date)
FROM orders sub
WHERE sub.ship_date > main.ship_date)
AND ship_date IS NOT NULL
ORDER BY ship_date, po_num

Query 3-31

The subquery is correlated because the number it produces depends on

main.ship_date, a value produced by the outer SELECT. Thus, the subquery
must be executed anew for every row that the outer query considers.

Composing Advanced SELECT Statements 3-37

Subqueries in SELECT Statements

Query 3-31 uses the COUNT function to return a value to the main query. The
ORDER BY clause then orders the data. The query locates and returns the 13
rows that have the 10 latest shipping dates, as shown in Query Result 3-31:

po_num ship_date

4745 06/21/1993
278701 06/29/1993
429Q 06/29/1993
8052 07/03/1993
B77897 07/03/1993
LZ230 07/06/1993
B77930 07/10/1993
PC6782 07/12/1993
DM354331 07/13/1993
S22942 07/13/1993
MA003 07/16/1993
W2286 07/16/1993
Z55709 07/16/1993
C3288 07/25/1993
KF2961 07/30/1993
W9925 07/30/1993

Query Result 3-31

If you use a correlated subquery, such as Query 3-31, on a very large table,
you should index the ship_date column to improve performance. Otherwise,
this SELECT statement might be considered somewhat inefficient because it
executes the subquery once for every row of the table. Indexing and
performance issues are discussed in Chapter 10, “Tuning Your Data Model.”

Using EXISTS
The keyword EXISTS is known as an existential qualifier because the subquery
is true only if the outer SELECT, shown in Query 3-32a, finds at least one row:

SELECT UNIQUE manu_name, lead_time

FROM manufact
WHERE EXISTS
(SELECT * FROM stock
WHERE description MATCHES '*shoe*'
AND manufact.manu_code = stock.manu_code)

Query 3-32a

3-38 Composing Advanced SELECT Statements

 Subqueries in SELECT Statements

You often can construct a query with EXISTS that is equivalent to one that uses
IN. You also can substitute =ANY for IN, as shown in Query 3-32b:

SELECT UNIQUE manu_name, lead_time

FROM stock, manufact
WHERE manufact.manu_code IN
(SELECT manu_code FROM stock
WHERE description MATCHES '*shoe*')
AND stock.manu_code = manufact.manu_code

Query 3-32b

Query 3-32a and Query 3-32b return rows for the manufacturers that produce
a kind of shoe as well as the lead time for ordering the product. These results
are shown in Query Result 3-32:

manu_name lead_time

Anza 5
Hero 4
Karsten 21
Nikolus 8
ProCycle 9
Shimara 30

Query Result 3-32

Note that you cannot use the predicate IN for a subquery that contains a
column with a TEXT or BYTE data type.
Add the keyword NOT to IN or to EXISTS to create a search condition that is
the opposite of the one in the preceding queries. You also can substitute
!=ALL for NOT IN.
Query 3-33 shows two ways to do the same thing. One way might allow the
database server to do less work than the other, depending on the design of
the database and the size of the tables. To find out which query might be

Composing Advanced SELECT Statements 3-39

Subqueries in SELECT Statements

better, you can use the SET EXPLAIN command to get a listing of the query
plan. SET EXPLAIN is discussed in Chapter 13, “Optimizing Your Queries,” of
this manual and in Chapter 1 of the Informix Guide to SQL: Syntax.

SELECT customer_num, company FROM customer

WHERE customer_num NOT IN
(SELECT customer_num FROM orders
WHERE customer.customer_num = orders.customer_num)

SELECT customer_num, company FROM customer

WHERE NOT EXISTS
(SELECT * FROM orders
WHERE customer.customer_num = orders.customer_num)

Query 3-33

Each statement in Query 3-33 returns the rows shown in Query Result 3-33,
which identify customers who have not placed orders:

customer_num company

102 Sports Spot

103 Phil’s Sports
105 Los Altos Sports
107 Athletic Supplies
108 Quinn’s Sports
109 Sport Stuff
113 Sportstown
114 Sporting Place
118 Blue Ribbon Sports
125 Total Fitness Sports
128 Phoenix University

Query Result 3-33

Note that the keywords EXISTS and IN are used for the set operation known
as intersection, and the keywords NOT EXISTS and NOT IN are used for the set
operation known as difference. These concepts are discussed in “Set Opera-
tions” on page 3-42.

3-40 Composing Advanced SELECT Statements

 Subqueries in SELECT Statements

Query 3-34 identifies all the items in the stock table that have not been
ordered yet by performing a subquery on the items table:

SELECT stock.* FROM stock

WHERE NOT EXISTS
(SELECT * FROM items
WHERE stock.stock_num = items.stock_num
AND stock.manu_code = items.manu_code)

Query 3-34

It returns the rows shown in Query Result 3-34:

stock_num manu_code description unit_price unit unit_descr

101 PRC bicycle tires $88.00 box 4/box

102 SHM bicycle brakes $220.00 case 4 sets/case
102 PRC bicycle brakes $480.00 case 4 sets/case
105 PRC bicycle wheels $53.00 pair pair
106 PRC bicycle stem $23.00 each each
107 PRC bicycle saddle $70.00 pair pair
108 SHM crankset $45.00 each each
109 SHM pedal binding $200.00 case 4 pairs/case
110 ANZ helmet $244.00 case 4/case
110 HRO helmet $260.00 case 4/case
112 SHM 12-spd, assmbld $549.00 each each
113 SHM 18-spd, assmbld $685.90 each each
201 KAR golf shoes $90.00 each each
202 NKL metal woods $174.00 case 2 sets/case
203 NKL irons/wedge $670.00 case 2 sets/case
205 NKL 3 golf balls $312.00 case 24/case
205 HRO 3 golf balls $312.00 case 24/case
301 NKL running shoes $97.00 each each
301 HRO running shoes $42.50 each each
301 PRC running shoes $75.00 each each
301 ANZ running shoes $95.00 each each
302 HRO ice pack $4.50 each each
303 KAR socks $36.00 box 24 pairs/box
305 HRO first-aid kit $48.00 case 4/case
306 PRC tandem adapter $160.00 each each
308 PRC twin jogger $280.00 each each
309 SHM ear drops $40.00 case 20/case
310 SHM kick board $80.00 case 10/case
310 ANZ kick board $84.00 case 12/case
311 SHM water gloves $48.00 box 4 pairs/box
312 SHM racer goggles $96.00 box 12/box
312 HRO racer goggles $72.00 box 12/box
313 SHM swim cap $72.00 box 12/box
313 ANZ swim cap $60.00 box 12/box

Query Result 3-34

Composing Advanced SELECT Statements 3-41

Set Operations

Note that no logical limit exists to the number of subqueries a SELECT

statement can have, but the size of any statement is physically limited when
considered as a character string. However, this limit is probably larger than
any practical statement you are likely to compose.
Perhaps you want to check whether information has been entered correctly
in the database. One way to find errors in a database is to write a query that
returns output only when errors exist. A subquery of this type serves as a
kind of audit query, as shown in Query 3-35:

SELECT * FROM items

WHERE total_price != quantity *
(SELECT unit_price FROM stock
WHERE stock.stock_num = items.stock_num
AND stock.manu_code = items.manu_code)

Query 3-35

Query Result 3-35 returns only those rows for which the total price of an item
on an order is not equal to the stock unit price times the order quantity.
Assuming that no discount has been applied, such rows must have been
entered incorrectly in the database.

item_num order_num stock_num manu_code quantity total_price

1 1004 1 HRO 1 $960.00

2 1006 5 NRG 5 $190.00

Query Result 3-35

Query 3-35 returns rows only when errors occur. If information is correctly
inserted into the database, no rows are returned.

Set Operations
The standard set operations union, intersection, and difference let you
manipulate database information. These three operations enable you to use
SELECT statements to check the integrity of your database after you perform
an update, insert, or delete. They can be useful when you transfer data to a
history table, for example, and want to verify that the correct data is in the
history table before you delete it from the original table.

3-42 Composing Advanced SELECT Statements

 Set Operations

Union
The union operation uses the UNION keyword, or operator, to combine two
queries into a single compound query. You can use the UNION keyword
between two or more SELECT statements to unite them and produce a tempo-
rary table containing rows that exist in any or all of the original tables. (Note
that you cannot use a UNION operator inside a subquery or in the definition
of a view.) Figure 3-1 illustrates the union set operation.

SELECT DISTINCT stock_num, manu_code

FROM stock
WHERE unit_price < 25.00
quantity
UNION
greater less than
SELECT stock_num, manu_code
than 3 or equal unit_price < 25.00
FROM items
to 3

less than
25.00 qualifies qualifies
unit_price quantity > 3
greater than
or equal to qualifies
25.00

Figure 3-1 The union set operation

The UNION keyword selects all rows from the two queries, removes
duplicates, and returns what is left. Because the results of the queries are
combined into a single result, the select list in each query must have the same
number of columns. Also, the corresponding columns selected from each
table must be of the same data type (CHARACTER type columns must be the
same length), and these corresponding columns must either all allow or all
disallow nulls.

Composing Advanced SELECT Statements 3-43

Set Operations

Query 3-36 performs a union on the stock_num and manu_code columns in

the stock and items tables:

SELECT DISTINCT stock_num, manu_code

FROM stock
WHERE unit_price < 25.00

UNION

SELECT stock_num, manu_code

FROM items
WHERE quantity > 3

Query 3-36

Query 3-36 selects those items that have a unit price of less than $25.00 or that
have been ordered in quantities greater than three and lists their stock_num
and manu_code, as shown in Query Result 3-36:

stock_num manu_code

5 ANZ
5 NRG
5 SMT
9 ANZ
103 PRC
106 PRC
201 NKL
301 KAR
302 HRO
302 KAR

Query Result 3-36

3-44 Composing Advanced SELECT Statements

 Set Operations

If you include an ORDER BY clause, it must follow Query 3-36 and use an
integer, not an identifier, to refer to the ordering column. Ordering takes place
after the set operation is complete.

SELECT DISTINCT stock_num, manu_code

FROM stock
WHERE unit_price < 25.00

UNION

SELECT stock_num, manu_code

FROM items
WHERE quantity > 3
ORDER BY 2

Query 3-37

The compound query in Query 3-37 selects the same rows as Query 3-36 but
displays them in order of manufacturer code, as shown in Query Result 3-37:

stock_num manu_code

5 ANZ
9 ANZ
302 HRO
301 KAR
302 KAR
201 NKL
5 NRG
103 PRC
106 PRC
5 SMT

Query Result 3-37

Composing Advanced SELECT Statements 3-45

Set Operations

By default, the UNION keyword excludes duplicate rows. Add the optional
keyword ALL, as shown in Query 3-38, to retain the duplicate values:

SELECT stock_num, manu_code

FROM stock
WHERE unit_price < 25.00

UNION ALL

SELECT stock_num, manu_code

FROM items
WHERE quantity > 3
ORDER BY 2
INTO TEMP stockitem

Query 3-38

Query 3-38 uses the UNION ALL keywords to unite two SELECT statements
and puts the results into a temporary table by adding an INTO TEMP clause
after the final SELECT. It returns the same rows as Query 3-37 but also
includes duplicate values.

stock_num manu_code

9 ANZ
5 ANZ
9 ANZ
5 ANZ
9 ANZ
5 ANZ
5 ANZ
5 ANZ
302 HRO
302 KAR
301 KAR
201 NKL
5 NRG
5 NRG
103 PRC
106 PRC
5 SMT
5 SMT

Query Result 3-38

Corresponding columns in the select lists for the combined queries must
have identical data types, but the columns do not need to use the same
identifier.

3-46 Composing Advanced SELECT Statements

 Set Operations

Query 3-39 selects the state column from the customer table and the
corresponding code column from the state table:

SELECT DISTINCT state

FROM customer
WHERE customer_num BETWEEN 120 AND 125

UNION

SELECT DISTINCT code

FROM state
WHERE sname MATCHES '*a'

Query 3-39

Query Result 3-39 returns state code abbreviations for customer numbers 120
through 125, or for states whose sname ends in A or a:

state

AK
AL
AZ
CA
DE
FL
GA
IA
IN
LA
MA
MN
MT
NC
ND
NE
NJ
NV
OK
PA
SC
SD
VA
WV

Query Result 3-39

Composing Advanced SELECT Statements 3-47

Set Operations

In compound queries, the column names or display labels in the first SELECT
statement are the ones that appear in the results. Thus, in Query 3-40, the col-
umn name state from the first SELECT statement is used instead of the
column name code from the second.
Query 3-40 performs a union on three tables. The maximum number of
unions depends on the practicality of the application and any memory
limitations.

SELECT stock_num, manu_code

FROM stock
WHERE unit_price > 600.00

UNION ALL

SELECT stock_num, manu_code

FROM catalog
WHERE catalog_num = 10025

UNION ALL

SELECT stock_num, manu_code

FROM items
WHERE quantity = 10
ORDER BY 2

Query 3-40

Query 3-40 selects items where the unit_price in the stock table is greater
than $600, or the catalog_num in the catalog table is 10025, or the quantity
in the items table is 10, and orders it by manu_code. The results are shown
in Query Result 3-40:

stock_num manu_code

5 ANZ
9 ANZ
8 ANZ
4 HSK
1 HSK
203 NKL
5 NRG
106 PRC
113 SHM

Query Result 3-40

3-48 Composing Advanced SELECT Statements

 Set Operations

See Chapter 1 of the Informix Guide to SQL: Syntax for the complete syntax of
the SELECT statement and the UNION operator. See also Chapter 5, “Pro-
gramming with SQL,”and Chapter 6, “Modifying Data Through SQL
Programs,”as well as the product manuals for information specific to the
INFORMIX-4GL and INFORMIX-ESQL/C products and any limitations
involving the INTO clause and compound queries.
Query 3-41 uses a combined query to select data into a temporary table and
then adds a simple query to order and display it. You must separate the
combined and simple queries with a semicolon.
The combined query uses a literal in the select list to tag the output of part of
a union so it can be distinguished later. The tag is given the label sortkey. The
simple query uses that tag as a sort key for ordering the retrieved rows.

SELECT '1' sortkey, lname, fname, company,

city, state, phone
FROM customer x
WHERE state = 'CA'

UNION

SELECT '2' sortkey, lname, fname, company,

city, state, phone
FROM customer y
WHERE state <> 'CA'
INTO TEMP calcust;

SELECT * FROM calcust

ORDER BY 1

Query 3-41

Composing Advanced SELECT Statements 3-49

Set Operations

Query 3-41 creates a list where the California customers, the ones called most
frequently, appear first, as shown in Query Result 3-41:

sortkey 1
lname Baxter
fname Dick
company Blue Ribbon Sports
city Oakland
state CA
phone 415-655-0011

sortkey 1
lname Beatty
fname Lana
company Sportstown
city Menlo Park
state CA
phone 415-356-9982

sortkey 1
lname Currie
fname Philip
company Phil’s Sports
city Palo Alto
state CA
phone 415-328-4543

sortkey 1
lname Grant
fname Alfred
company Gold Medal Sports
city Menlo Park
state CA
phone 415-356-1123
.
.
.
sortkey 2
lname Satifer
fname Kim
company Big Blue Bike Shop
city Blue Island
state NY
phone 312-944-5691

sortkey 2
lname Shorter
fname Bob
company The Triathletes Club
city Cherry Hill
state NJ
phone 609-663-6079

sortkey 2
lname Wallack
fname Jason
company City Sports
city Wilmington
state DE
phone 302-366-7511

Query Result 3-41

3-50 Composing Advanced SELECT Statements

 Set Operations

Intersection
The intersection of two sets of rows produces a table containing rows that exist
in both of the original tables. Use the keyword EXISTS or IN to introduce sub-
queries that show the intersection of two sets. Figure 3-2 illustrates the
intersection set operation.

SELECT stock_num, manu_code, unit_price

FROM stock
WHERE stock_num IN
(SELECT stock_num FROM items) stock_num
ORDER BY stock_num

exists in not in stock table

items table items table

exists in
stock table qualifies
stock_num items table
not in stock
table

Figure 3-2 The intersection set operation

Query 3-42 is an example of a nested SELECT statement that shows the

intersection of the stock and items tables:

SELECT stock_num, manu_code, unit_price

FROM stock
WHERE stock_num IN
(SELECT stock_num FROM items)
ORDER BY stock_num

Query 3-42

Query Result 3-42 contains all the elements from both sets, returning the
following 57 rows:

Composing Advanced SELECT Statements 3-51

Set Operations

stock_num manu_code unit_price

1 HRO $250.00
1 HSK $800.00
1 SMT $450.00
2 HRO $126.00
3 HSK $240.00
3 SHM $280.00
4 HRO $480.00
4 HSK $960.00
5 ANZ $19.80
5 NRG $28.00
5 SMT $25.00
6 ANZ $48.00
6 SMT $36.00
7 HRO $600.00
8 ANZ $840.00
9 ANZ $20.00
101 PRC $88.00
101 SHM $68.00
103 PRC $20.00
104 PRC $58.00
105 PRC $53.00
105 SHM $80.00
109 PRC $30.00
109 SHM $200.00
110 ANZ $244.00
110 HRO $260.00
110 HSK $308.00
110 PRC $236.00
110 SHM $228.00
111 SHM $499.99
114 PRC $120.00
201 ANZ $75.00
201 KAR $90.00
201 NKL $37.50
202 KAR $230.00
202 NKL $174.00
204 KAR $45.00
205 ANZ $312.00
205 HRO $312.00
205 NKL $312.00
301 ANZ $95.00
301 HRO $42.50
301 KAR $87.00
301 NKL $97.00
301 PRC $75.00
301 SHM $102.00
302 HRO $4.50
302 KAR $5.00
303 KAR $36.00
303 PRC $48.00
304 ANZ $170.00
304 HRO $280.00
306 PRC $160.00
306 SHM $190.00
307 PRC $250.00
309 HRO $40.00
309 SHM $40.00

Query Result 3-42

3-52 Composing Advanced SELECT Statements

 Set Operations

Difference
The difference between two sets of rows produces a table containing rows in
the first set that are not also in the second set. Use the keywords NOT EXISTS
or NOT IN to introduce subqueries that show the difference between two sets.
Figure 3-3 illustrates the difference set operation.

SELECT stock_num, manu_code, unit_price

FROM stock
WHERE stock_num NOT IN
(SELECT stock_num FROM items) stock_num
ORDER BY stock_num

exists in not in stock table

items table items table

exists in
stock table qualifies
stock_num items table
not in stock
table

Figure 3-3 The difference set operation

Query 3-43 is an example of a nested SELECT statement that shows the

difference between the stock and items tables:

SELECT stock_num, manu_code, unit_price

FROM stock
WHERE stock_num NOT IN
(SELECT stock_num FROM items)
ORDER BY stock_num

Query 3-43

Composing Advanced SELECT Statements 3-53

Summary

Query Result 3-43 contains all the elements from only the first set, returning
the following 17 rows:

stock_num manu_code unit_price

102 PRC $480.00

102 SHM $220.00
106 PRC $23.00
107 PRC $70.00
108 SHM $45.00
112 SHM $549.00
113 SHM $685.90
203 NKL $670.00
305 HRO $48.00
308 PRC $280.00
310 ANZ $84.00
310 SHM $80.00
311 SHM $48.00
312 HRO $72.00
312 SHM $96.00
313 ANZ $60.00
313 SHM $72.00

Query Result 3-43

Summary
This chapter was built on concepts introduced in Chapter 2, “Composing
Simple SELECT Statements.” It provided sample syntax and results for more
advanced kinds of SELECT statements, which are used to perform a query on
a relational database. This chapter
• Introduced the GROUP BY and HAVING clauses, which can be used with
aggregates to return groups of rows and apply conditions to those
groups.
• Described how to use the rowid to retrieve internal record numbers from
tables and system catalog tables and discussed the serial internal table
identifier or tabid.
• Showed how to join a table to itself with a self-join to compare values in
a column with other values in the same column and to identify
duplicates.
• Introduced the keyword OUTER, explained how an outer join treats two
or more tables asymmetrically, and provided examples of the four kinds
of outer join.

3-54 Composing Advanced SELECT Statements

 Summary

• Described how to create correlated and uncorrelated subqueries by

nesting a SELECT statement in the WHERE clause of another SELECT
statement and showed the use of aggregate functions in subqueries.
• Demonstrated the use of the keywords ALL, ANY, EXISTS, IN, and SOME
in creating subqueries, and the effect of adding the keyword NOT or a
relational operator.
• Discussed the set operations union, intersection, and difference.
• Showed how to use the UNION and UNION ALL keywords to create
compound queries consisting of two or more SELECT statements.

Composing Advanced SELECT Statements 3-55

Summary

3-56 Composing Advanced SELECT Statements

 Chapter

Modifying Data
Chapter Overview 3
Statements That Modify Data 3
Deleting Rows 4
Deleting All Rows of a Table 4

4
Deleting a Known Number of Rows 4
Deleting an Unknown Number of Rows 5
Complicated Delete Conditions 6
Inserting Rows 6
Single Rows 7
Multiple Rows and Expressions 9
Restrictions on the Insert-Selection 10
Updating Rows 11
Selecting Rows to Update 12
Updating with Uniform Values 13
Impossible Updates 13
Updating with Selected Values 14
Database Privileges 15
Displaying Table Privileges 16
Data Integrity 17
Entity Integrity 18
Semantic Integrity 18
Referential Integrity 19
Using the ON DELETE CASCADE Option 21
Interrupted Modifications 22
The Transaction 23
Transaction Logging 24
Logging and Cascading Deletes 24
Specifying Transactions 24
 Archives and Logs 25
Archiving with INFORMIX-SE 26
Archiving with INFORMIX-OnLine Dynamic Server 26
Concurrency and Locks 27
Data Replication 28
INFORMIX-OnLine Dynamic Server Data Replication 29
Summary 29

4-2 Modifying Data

Chapter Overview
Modifying data is fundamentally different from querying data. Querying
data involves examining the contents of tables. Modifying data involves
changing the contents of tables.
Think about what happens if the system hardware or software fails during a
query. In this case, the effect on the application can be severe but the database
itself is unharmed. However, if the system fails while a modification is under
way, the state of the database is in doubt. Obviously, this can have far-reach-
ing implications. Before you delete, insert, or update rows in a database, ask
yourself the following questions:
• Is user access to the database and its tables secure, that is, are specific
users given limited database and table-level privileges?
• Does the modified data preserve the existing integrity of the database?
• Are systems in place that make the database relatively immune to
external events that might cause system or hardware failures?
If you are unable to answer yes to each of these questions, do not panic.
Solutions to all these problems are built in to the Informix database servers.
After an introduction to the statements that modify data, this chapter dis-
cusses these solutions. Chapters 8 through 11 talk about these topics in
greater detail.

Statements That Modify Data

The following three statements modify data:
• DELETE
• INSERT
• UPDATE
Although these SQL statements are relatively simple when compared with
the more advanced SELECT statements, use them carefully because they
change the contents of the database.
Modifying Data 4-3
Statements That Modify Data

Deleting Rows
The DELETE statement removes any row or combination of rows from a table.
You cannot recover a deleted row after the transaction is committed. (Trans-
actions are discussed under “Interrupted Modifications” on page 4-22. For
now, think of a transaction and a statement as the same thing.)
When deleting a row, you must also be careful to delete any rows of other
tables whose values depend on the deleted row. If, however, your database
enforces referential constraints, you can use the ON DELETE CASCADE option
of the CREATE TABLE or ALTER TABLE statements to allow deletes to cascade
from one table in a relationship to another. For more information on referen-
tial constraints and the ON DELETE CASCADE option, refer to the section
“Referential Integrity” on page 4-19.

Deleting All Rows of a Table

The DELETE statement specifies a table and usually contains a WHERE clause
that designates the row or rows that are to be removed from the table. If the
WHERE clause is left out, all rows are deleted. Do not execute the following
statement:

DELETE FROM customer

Because this DELETE statement does not contain a WHERE clause, all rows
from the customer table are deleted. If you attempt an unconditional delete
using the DB-Access or INFORMIX-SQL menu options, the program warns
you and asks for confirmation. However, an unconditional delete from
within a program can occur without warning.

Deleting a Known Number of Rows

The WHERE clause in a DELETE statement has the same form as the WHERE
clause in a SELECT statement. You can use it to designate exactly which row
or rows should be deleted. You can delete a customer with a specific customer
number, as shown in the following example:

DELETE FROM customer WHERE customer_num = 175

In this example, because the customer_num column has a unique constraint,

you are sure that no more than one row is deleted.

4-4 Modifying Data

 Statements That Modify Data

Deleting an Unknown Number of Rows

You can also choose rows based on nonindexed columns, as shown in the fol-
lowing example:

DELETE FROM customer WHERE company = 'Druid Cyclery'

Because the column tested does not have a unique constraint, this statement
might delete more than one row. (Druid Cyclery may have two stores, both
with the same name but different customer numbers.)
You can find out how many rows might be affected by a DELETE statement
by selecting the count of qualifying rows from the customer table for Druid
Cyclery.

SELECT COUNT(*) FROM customer WHERE company = 'Druid Cyclery'

You can also select the rows and display them, to be sure they are the ones
you want to delete.
Using a SELECT statement as a test is only an approximation, however, when
the database is available to multiple users concurrently. Between the time you
execute the SELECT statement and the subsequent DELETE statement, other
users could have modified the table and changed the result. In this example,
another user might perform the following actions:
• Insert a new row for another customer named Druid Cyclery
• Delete one or more of the Druid Cyclery rows before you do so
• Update a Druid Cyclery row to have a new company name, or update
some other customer to have the name Druid Cyclery
Although it is not likely that other users would do these things in that brief
interval, the possibility does exist. This same problem affects the UPDATE
statement. Ways of addressing this problem are discussed under “Concur-
rency and Locks” on page 4-27, and in greater detail in Chapter 7, “Program-
ming for a Multiuser Environment.”
Another problem you may encounter is a hardware or software failure before
the statement finishes. In this case, the database may have deleted no rows,
some rows, or all specified rows. The state of the database is unknown, which
is undesirable. You can prevent this situation by using transaction logging, as
discussed in “Interrupted Modifications” on page 4-22.

Modifying Data 4-5

Statements That Modify Data

Complicated Delete Conditions

The WHERE clause in a DELETE statement can be almost as complicated as the
one in a SELECT statement. It can contain multiple conditions connected by
AND and OR, and it may contain subqueries.

Suppose you discover that some rows of the stock table contain incorrect
manufacturer codes. Rather than update them, you want to delete them so
they can be reentered. You know that these rows, unlike the correct ones,
have no matching rows in the manufact table. This allows you to write a
DELETE statement such as the one shown in the following example:

DELETE FROM stock

WHERE 0 = (SELECT COUNT(*) FROM manufact
WHERE manufact.manu_code = stock.manu_code)

The subquery counts the number of rows of manufact that match; the count
is 1 for a correct row of stock and 0 for an incorrect one. The latter rows are
chosen for deletion.
One way to develop a DELETE statement with a complicated condition is
to first develop a SELECT statement that returns precisely the rows to be
deleted. Write it as SELECT *; when it returns the desired set of rows, change
SELECT * to read DELETE and execute it once more.
The WHERE clause of a DELETE statement cannot use a subquery that tests the
same table. That is, when you delete from stock, you cannot use a subquery
in the WHERE clause that also selects from stock.
The key to this rule is in the FROM clause. If a table is named in the FROM
clause of a DELETE statement, it cannot also appear in the FROM clause of a
subquery of the DELETE statement.

Inserting Rows
The INSERT statement adds a new row, or rows, to a table. The statement has
two basic functions: It can create a single new row using column values you
supply, or it can create a group of new rows using data selected from other
tables.

4-6 Modifying Data

 Statements That Modify Data

Single Rows
In its simplest form, the INSERT statement creates one new row from a list of
column values, and puts that row in the table. The following statement is an
example of adding a row to the stock table:

INSERT INTO stock

VALUES(115, 'PRC', 'tire pump', 108, 'box', '6/box')

The stock table has the following columns:

stock_num (a number identifying the type of merchandise)
manu_code (a foreign key to the manufact table)
description (a description of the merchandise)
unit_price (the unit price of the merchandise)
unit (of measure)
unit_descr (characterizing the unit of measure)
Notice that the values listed in the VALUES clause in the preceding example
have a one-to-one correspondence with the columns of this table. To write a
VALUES clause, you must know the columns of the tables as well as their
sequence from first to last.

Possible Column Values

The VALUES clause accepts only constant values, not expressions. You can
supply the following values:
• Literal numbers
• Literal datetime values
• Literal interval values
• Quoted strings of characters
• The word NULL for a null value
• The word TODAY for today’s date
• The word CURRENT for the current date and time
• The word USER for your user name
• The word DBSERVERNAME (or SITENAME) for the name of the computer
where the database server is running
Some columns of a table might not allow null values. If you attempt to insert
NULL in such a column, the statement is rejected. Or a column in the table
may not permit duplicate values. If you specify a value that is a duplicate of

Modifying Data 4-7

Statements That Modify Data

one already in such a column, the statement is rejected. Some columns may
even restrict the possible column values allowed. These restrictions are
placed on columns using data integrity constraints. For more information on
data restrictions, see the section “Database Privileges” on page 4-15.
Only one column in a table can have the SERIAL data type. The database
server generates values for a serial column. To make this happen, when you
insert values, specify the value zero for the serial column. The database
server generates the next actual value in sequence. Serial columns do not
allow null values.
You can specify a nonzero value for a serial column (as long as it does not
duplicate any existing value in that column) and the database server uses the
value. However, that nonzero value may set a new starting point for values
that the database server generates. The next value the database server gener-
ates for you is one greater than the maximum value in the column.
Do not specify the currency symbols for columns that contain money values.
Just specify the numeric value of the amount.
The database server can convert between numeric and character data types.
You can give a string of numeric characters (for example, '-0075.6') as the
value of a numeric column. The database server converts the numeric string
to a number. An error occurs only if the string does not represent a number.
You can specify a number or a date as the value for a character column. The
database server converts that value to a character string. For example, if you
specify TODAY as the value for a character column, a character string repre-
senting today’s date is used. (The format used is specified by the DBDATE
environment variable.)

Listing Specific Column Names

You do not have to specify values for every column. Instead, you can list the
column names after the table name and then supply values for only those col-
umns you named. The following example shows a statement that inserts a
new row into the stock table:

INSERT INTO stock (stock_num, description, unit_price, manu_code)

VALUES (115, 'tyre pump', 114, 'SHM')

4-8 Modifying Data

 Statements That Modify Data

Notice that only the data for the stock number, description, unit price, and
manufacturer code is provided. The database server supplies the following
values for the remaining columns:
• It generates a serial number for an unlisted serial column.
• It generates a default value for a column with a specific default associated
with it.
• It generates a null value for any column that allows nulls but does not
specify a default value for any column that specifies null as the default
value.
This means that you must list and supply values for all columns that do not
specify a default value or do not permit nulls. However, you can list the col-
umns in any order—as long as the values for those columns are listed in the
same order. For information about setting a default value for a column, see
Chapter 9, “Implementing Your Data Model.”
After the INSERT statement is executed, the following new row is inserted
into the stock table:

stock_num manu_code description unit_price unit unit_descr

115 SHM tyre pump 114

Both the unit and unit_descr are blank, indicating that null values are in
those two columns. Because the unit column permits nulls, one can only
guess the number of tire pumps that were purchased for $114. Of course, if a
default value of “box” was specified for this column, then “box” would be the
unit of measure. In any case, when inserting values into specific columns of
a table, pay attention to what data is needed for that row.

Multiple Rows and Expressions

The other major form of the INSERT statement replaces the VALUES clause
with a SELECT statement. This feature allows you to insert the following data:
• Multiple rows with only one statement (a row is inserted for each row
returned by the SELECT statement)
• Calculated values (the VALUES clause only permits constants) because
the select list can contain expressions

Modifying Data 4-9

Statements That Modify Data

For example, suppose a follow-up call is required for every order that has
been paid for but not shipped. The INSERT statement in the following exam-
ple finds those orders and inserts a row in cust_calls for each order:

INSERT INTO cust_calls (customer_num, call_descr)

SELECT customer_num, order_num FROM orders
WHERE paid_date IS NOT NULL
AND ship_date IS NULL

This SELECT statement returns two columns. The data from these columns (in
each selected row) is inserted into the named columns of the cust_calls table.
Then, an order number (from order_num, a serial column) is inserted into the
call description, which is a character column. Remember that the database
server allows you to insert integer values into a character column. It automat-
ically converts the serial number to a character string of decimal digits.

Restrictions on the Insert-Selection

The following list contains the restrictions on the SELECT statement for insert-
ing rows:
• It cannot contain an INTO clause.
• It cannot contain an INTO TEMP clause.
• It cannot contain an ORDER BY clause.
• It cannot refer to the table into which you are inserting rows.
The INTO, INTO TEMP, and ORDER BY clause restrictions are minor. The INTO
clause is not useful in this context. (It is discussed in Chapter 5, “Program-
ming with SQL.”) You can work around the INTO TEMP clause restriction by
first selecting the data you want to insert into a temporary table and then
inserting the data from the temporary table with the INSERT statement. Like-
wise, the lack of an ORDER BY clause is not important. If you need to ensure
that the new rows are physically ordered in the table, you can first select them
into a temporary table and order it, and then insert from the temporary table.
You can also apply a physical order to the table using a clustered index after
all insertions are done.
The last restriction is more serious because it prevents you from naming the
same table in both the INTO clause of the INSERT statement and the FROM
clause of the SELECT statement. (This saves the database server from getting
into an endless loop in which each inserted row is reselected and reinserted.)
In some cases, however, you may want to do this. For example, suppose that
you have learned that the Nikolus company supplies the same products as

4-10 Modifying Data

 Statements That Modify Data

the Anza company, but at half the price. You want to add rows to the stock
table to reflect this. Optimally, you want to select data from all the Anza stock
rows and reinsert it with the Nikolus manufacturer code. However, you
cannot select from the same table into which you are inserting.
You can get around this restriction by selecting the data you want to insert
into a temporary table. Then, select from that table in the INSERT statement.
The following statements are required to accomplish this:

SELECT stock_num, 'HSK' temp_manu, description, unit_price/2

half_price, unit, unit_descr FROM stock
WHERE manu_code = 'ANZ'
AND stock_num < 110
INTO TEMP anzrows;

INSERT INTO stock SELECT * FROM anzrows;

DROP TABLE anzrows;

This SELECT statement takes existing rows from stock and substitutes a literal
value for the manufacturer code and a computed value for the unit price.
These rows are then saved in a temporary table, anzrows, which is immedi-
ately inserted into the stock table.
When you insert multiple rows, a risk exists that one of the rows contains
invalid data that might cause the database server to report an error. When
this happens, the statement terminates early. Even if no error occurs, a very
small risk exists that there might be a hardware or software failure while the
statement is executing (for example, the disk might fill up).
In either event, you cannot easily tell how many new rows were inserted. If
you repeat the statement in its entirety, you might create duplicate rows—or
you might not. Because the database is in an unknown state, you cannot
know what to do. The answer lies in using transactions, as discussed in
“Interrupted Modifications” on page 4-22.

Updating Rows
You use the UPDATE statement to change the contents of one or more
columns in one or more existing rows of a table. This statement takes two
fundamentally different forms. One lets you assign specific values to col-
umns by name; the other lets you assign a list of values (that might be
returned by a SELECT statement) to a list of columns. In either case, if you are

Modifying Data 4-11

Statements That Modify Data

updating rows and some of the columns have data integrity constraints, the
data you change must be within the constraints placed on those columns. For
more information, refer to “Database Privileges” on page 4-15.

Selecting Rows to Update

Either form of the UPDATE statement can end with a WHERE clause that
determines which rows are modified. If you omit it, all rows are modified.
The WHERE clause can be quite complicated to select the precise set of rows
that need changing. The only restriction on it is that the table that you are
updating cannot be named in the FROM clause of a subquery.
The first form of an UPDATE statement uses a series of assignment clauses to
specify new column values, as shown in the following example:

UPDATE customer
SET fname = 'Barnaby', lname = 'Dorfler'
WHERE customer_num = 103

The WHERE clause selects the row to be updated. In the stores6 database, the
customer.customer_num column is the primary key for that table, so this
statement can update no more than one row.
You can also use subqueries in the WHERE clause. Suppose that the Anza
Corporation issues a safety recall of their tennis balls. As a result, any
unshipped orders that include stock number 6 from manufacturer ANZ must
be put on back order, as shown in the following example:

UPDATE orders
SET backlog = 'y'
WHERE ship_date IS NULL
AND order_num IN
(SELECT DISTINCT items.order_num FROM items
WHERE items.stock_num = 6
AND items.manu_code = 'ANZ')

This subquery returns a column of order numbers (zero or more). The

UPDATE operation then tests each row of orders against the list and performs
the update if that row matches.

4-12 Modifying Data

 Statements That Modify Data

Updating with Uniform Values

Each assignment after the keyword SET specifies a new value for a column.
That value is applied uniformly to every row that you update. In the exam-
ples in the previous section, the new values were constants, but you can
assign any expression, including one based on the column value itself. Sup-
pose the manufacturer code HRO has raised all prices by 5 percent, and you
must update the stock table to reflect this. You would use a statement such as
the one shown in the following example:

UPDATE stock
SET unit_price = unit_price * 1.05
WHERE manu_code = 'HRO'

You can also use a subquery as part of the assigned value. When a subquery
is used as an element of an expression, it must return exactly one value (one
column and one row). Suppose that you decide that for any stock number,
you must charge a higher price than any manufacturer of that product. You
need to update the prices of all unshipped orders. The SELECT statements in
the following example specify the criteria:

UPDATE items
SET total_price = quantity *
(SELECT MAX (unit_price) FROM stock
WHERE stock.stock_num = items.stock_num)
WHERE items.order_num IN
(SELECT order_num FROM orders
WHERE ship_date IS NULL)

The first SELECT statement returns a single value—the highest price in the
stock table for a particular product. This is a correlated subquery; because
when a value from items appears in its WHERE clause, you must execute it
for every row that you update.
The second SELECT statement produces a list of the order numbers of
unshipped orders. It is an uncorrelated subquery that is executed once.

Impossible Updates
Restrictions exist on the use of subqueries when you modify data. In
particular, you cannot query the table that is being modified. You can refer to
the present value of a column in an expression, as in the example in which the
unit_price column was incremented by 5 percent. You can refer to a value of

Modifying Data 4-13

Statements That Modify Data

a column in a WHERE clause in a subquery, as in the example that updated

the stock table, in which the items table is updated and items.stock_num is
used in a join expression.
The need to update and query a table at the same time does not occur often
in a well-designed database. (Database design is covered in Chapters 8
through 11 of this manual.) However, you may want to update and query at
the same time when a database is first being developed, before its design has
been carefully thought through. A typical problem arises when a table inad-
vertently and incorrectly contains a few rows with duplicate values in a col-
umn that should be unique. You might want to delete the duplicate rows or
update only the duplicate rows. Either way, a test for duplicate rows inevita-
bly requires a subquery, which is not allowed in an UPDATE statement or
DELETE statement. Chapter 6, “Modifying Data Through SQL Programs,”
discusses how to use an update cursor to perform this kind of modification.

Updating with Selected Values

The second form of UPDATE statement replaces the list of assignments with
a single bulk assignment, in which a list of columns is set equal to a list of val-
ues. When the values are simple constants, this form is nothing more than the
form of the previous example with its parts rearranged, as shown in the
following example:

UPDATE customer
SET (fname, lname) = ('Barnaby', 'Dorfler')
WHERE customer_num = 103

No advantage exists to writing the statement this way. In fact, it is harder to

read because it is not obvious which values are assigned to which columns.
However, when the values to be assigned come from a single SELECT
statement, this form makes sense. Suppose that changes of address are to be
applied to several customers. Instead of updating the customer table each
time a change is reported, the new addresses are collected in a single tempo-
rary table named newaddr. It contains columns for the customer number and
the address-related fields of the customer table. Now the time comes to apply
all the new addresses at once. (This idea of collecting routine changes in a

4-14 Modifying Data

 Database Privileges

separate table and applying them in a batch, outside of peak hours, is one of
the performance techniques covered in Chapter 10, “Tuning Your Data
Model.”)

UPDATE customer
SET (address1, address2, city, state, zipcode) =
((SELECT address1, address2, city, state, zipcode
FROM newaddr
WHERE newaddr.customer_num=customer.customer_num))
WHERE customer_num IN
(SELECT customer_num FROM newaddr)

Notice that the values for multiple columns are produced by a single SELECT
statement. If you rewrite this example in the other form, with an assignment
for each updated column, you must write five SELECT statements, one for
each column to be updated. Not only is such a statement harder to write, it
also takes much longer to execute.
Note: In INFORMIX-4GL and the SQL API programs, you can use record or host
variables to update values. For more information, refer to Chapter 5, “Programming
with SQL.”

Database Privileges
Two levels of privileges exist in a database: database-level privileges and
table-level privileges. When you create a database, you are the only one who
can access it until you, as the owner (or Database Administrator [DBA]) of the
database, grant database-level privileges to others. When you create a table
in a database that is not ANSI-compliant, all users have access privileges to
the table until you, as the owner of the table, revoke table-level privileges
from specific users.
The following list contains database-level privileges:
Connect privilege allows you to open a database, issue queries, and
create and place indexes on temporary tables.
Resource privilege allows you to create permanent tables.
DBA privilege allows you to perform several additional functions
as the DBA.

Seven table-level privileges exist. However, only the first four are covered
here:

Modifying Data 4-15

Database Privileges

Select privilege is granted on a table-by-table basis and allows you

to select rows from a table (This privilege can be
limited by specific columns in a table.)
Delete privilege allows you to delete rows.
Insert privilege allows you to insert rows.
Update privilege allows you to update existing rows (that is, to
change their content).
The people who create databases and tables often grant the Connect and
Select privileges to public so that all users have them. If you can query a table,
you have at least the Connect and Select privileges for that database and
table. For more information about public, see “The Users and the Public” on
page 11-6.
You need the other table-level privileges to modify data. The owners of tables
often withhold these privileges or grant them only to specific users. As a
result, you may not be able to modify some tables that you can query freely.
Because these privileges are granted on a table-by-table basis, you can have
only Insert privileges on one table and only Update privileges on another, for
example. The Update privileges can be restricted even further to specific
columns in a table.
Chapter 11, “Granting and Limiting Access to Your Database,” discusses
granting privileges from the standpoint of the DBA. A complete list of privi-
leges and a summary of the GRANT and REVOKE statements can be found in
Chapter 1 of the Informix Guide to SQL: Syntax.

Displaying Table Privileges

If you are the owner of a table (that is, if you created it), you have all
privileges on that table. Otherwise, you can determine the privileges you
have for a certain table by querying the system catalog. The system catalog
consists of system tables that describe the database structure. The privileges
granted on each table are recorded in the systabauth system table. To display
these privileges, you must also know the unique identifier number of the
table. This number is specified in the systables system table. So, to display
privileges granted on the orders table, you might enter the following SELECT
statement:

SELECT * FROM systabauth

WHERE tabid = (SELECT tabid FROM systables
WHERE tabname = 'orders')

4-16 Modifying Data

 Data Integrity

The output of the query resembles the following display:

grantor grantee tabid tabauth

tfecit mutator 101 su-i-x--

tfecit procrustes 101 s--idx--
tfecit public 101 s--i-x--

The grantor is the user who grants the privilege. The grantor is usually the
owner of the table but can be another user empowered by the grantor. The
grantee is the user to whom the privilege is granted, and the grantee public
means “any user with Connect privilege.” If your user name does not appear,
you have only those privileges granted to public.
The tabauth column specifies the privileges granted. The letters in each row
of this column are the initial letters of the privilege names except that i
means Insert and x means Index. In this example, public has Select, Insert,
and Index privileges. Only the user mutator has Update privileges, and
only the user procrustes has Delete privileges.
Before the database server performs any action for you (for example, execute
a DELETE statement), it performs a query similar to the preceding one. If you
are not the owner of the table, and if it cannot find the necessary privilege on
the table for your user name or for public, it refuses to perform the operation.

Data Integrity
The INSERT, UPDATE, and DELETE statements modify data in an existing
database. Whenever you modify existing data, the integrity of the data can be
affected. For example, an order for a nonexistent product could be entered
into the orders table. Or a customer with outstanding orders could be deleted
from the customer table. Or the order number could be updated in the orders
table and not in the items table. In each of these cases, the integrity of the
stored data is lost.
Data integrity is actually made up of three parts:
• Entity integrity
Each row of a table has a unique identifier.
• Semantic integrity
The data in the columns properly reflects the types of information the
column was designed to hold.

Modifying Data 4-17

Data Integrity

• Referential integrity
The relationships between tables are enforced.
Well-designed databases incorporate these principles so that when you
modify data, the database itself prevents you from doing anything that might
harm the data integrity.

Entity Integrity
An entity is any person, place, or thing to be recorded in a database. Each
entity represents a table, and each row of a table represents an instance of that
entity. For example, if order is an entity, the orders table represents the idea of
order and each row in the table represents a specific order.
To identify each row in a table, the table must have a primary key. The
primary key is a unique value that identifies each row. This requirement is
called the entity integrity constraint.
For example, the orders table primary key is order_num. The order_num
column holds a unique system-generated order number for each row in the
table. To access a row of data in the orders table you can use the following
SELECT statement:

SELECT * FROM orders WHERE order_num = 1001

Using the order number in the WHERE clause of this statement enables you
to access a row easily because the order number uniquely identifies that row.
If the table allowed duplicate order numbers, it would be almost impossible
to access one single row, because all other columns of this table allow
duplicate values.
Refer to Chapter 8, “Building Your Data Model,” for more information on
primary keys and entity integrity.

Semantic Integrity
Semantic integrity ensures that data entered into a row reflects an allowable
value for that row. This means that the value must be within the domain, or
allowable set of values, for that column. For example, the quantity column of
the items table permits only numbers. If a value outside the domain can be
entered into a column, the semantic integrity of the data is violated.

4-18 Modifying Data

 Data Integrity

Semantic integrity is enforced using the following constraints:

• Data type
The data type defines the types of values that you can store in a column.
For example, the data type SMALLINT allows you to enter values from
-32,767 to 32,767 into a column.
• Default value
The default value is the value inserted into the column when an explicit
value is not specified. For example, the user_id column of the cust_calls
table defaults to the login name of the user if no name is entered.
• Check constraint
The check constraint specifies conditions on data inserted into a column.
Each row inserted into a table must meet these conditions. For example,
the quantity column of the items table may check for quantities greater
than or equal to 1.
For more information on using semantic integrity constraints in database
design, refer to “Defining the Domains” on page 9-3.

Referential Integrity
Referential integrity refers to the relationship between tables. Because each
table in a database must have a primary key, it is possible that this primary
key appears in other tables because of its relationship to data within those
tables. When a primary key from one table appears in another table, it is
called a foreign key.
Foreign keys join tables and establish dependencies between tables. Tables
can form a hierarchy of dependencies in such a way that if you change or
delete a row in one table, you destroy the meaning of rows in other tables. For
example, in Figure 4-1, the customer_num column of the customer table is a
primary key for that table and a foreign key in the orders and cust_call tables.
Customer number 106, George Watson, is referenced in both the orders and
cust_calls tables. If customer 106 is deleted from the customer table, the link
between the three tables and this particular customer is destroyed.
When you delete a row containing a primary key or update it with a different
primary key, you destroy the meaning of any rows that contain that value as
a foreign key. Referential integrity is the logical dependency of a foreign key
on a primary key. The integrity of a row that contains a foreign key depends
on the integrity of the row that it references—the row that contains the
matching primary key.

Modifying Data 4-19

Data Integrity

By default, INFORMIX-OnLine Dynamic Server does not allow you to violate

referential integrity and gives you an error message if you attempt to delete
rows from the parent table before you delete rows from the child table. You
can, however, use the ON DELETE CASCADE option to cause deletes from a
parent table to trip deletes on child tables. See “Using the ON DELETE
CASCADE Option” on page 4-21.

customer Table
(detail)
customer_num fname lname
103 Philip Currie
106 George Watson

orders Table
(detail)
order_num order_date customer_num
1002 05/21/1993 101
1003 05/22/1993 104
1004 05/22/1993 106

cust_calls Table
(detail)
customer_num call_dtime user_id
106 1993-06-12 8:20 maryj
110 1993-07-07 10:24 richc
119 1993-07-01 15:00 richc

Figure 4-1 Referential integrity in the stores6 database

You can define primary and foreign keys, and the relationship between them,
using the CREATE TABLE and ALTER TABLE statements. For more informa-
tion on these statements, refer to Chapter 1 of the Informix Guide to SQL: Syn-
tax. For information on building data models using primary and foreign
keys, refer to Chapter 8, “Building Your Data Model.”

4-20 Modifying Data

 Data Integrity

Using the ON DELETE CASCADE Option

You can maintain referential integrity when deleting rows from a primary
key for a table by using the ON DELETE CASCADE option in the REFERENCES
clause of the CREATE TABLE and ALTER TABLE statements. This option
allows you to delete a row from a parent table and its corresponding rows in
matching child tables with a single delete command.

Locking During Cascading Deletes

During deletes, locks are held on all qualifying rows of the parent and child
tables. When you specify a delete, the delete requested from the parent table
occurs before any referential actions are performed.

What Happens to Multiple Children Tables

If you have a parent table with two child constraints, one child with
cascading deletes specified and one child without cascading deletes, and you
attempt to delete a row from the parent table that applies to both child tables,
then the DELETE statement fails and no rows are deleted from either the
parent or child tables.

Logging Must Be Turned On

You must turn logging on in your current database for cascading deletes to
work. Logging and cascading deletes are discussed in “Transaction Logging”
on page 4-24.

Example
Say you have two tables with referential integrity rules applied, a parent
table, accounts, and a child table, sub_accounts. The following CREATE
TABLE statements define the referential constraints:

CREATE TABLE accounts (

acc_num SERIAL primary key,
acc_type INT,
acc_descr CHAR(20));

CREATE TABLE sub_accounts (

sub_acc INTEGER primary key,
ref_num INTEGER REFERENCES references accounts (acc_num) ON DELETE CASCADE,
sub_descr CHAR(20));

Modifying Data 4-21

Interrupted Modifications

Notice that the primary key of the accounts table, the acc_num column, uses
a SERIAL data type, and the foreign key of the sub_accounts table, the
ref_num column, uses an INTEGER data type. Combining the SERIAL data
type on the primary key and the INTEGER data type on the foreign key is
allowed. This is the only condition where you can mix and match data types.
The SERIAL data type is an INTEGER and the values for the column are auto-
matically generated by the database. All other primary and foreign key com-
binations must match explicitly. For example, a primary key defined as CHAR
must match a foreign key defined as CHAR.
To delete a row from the accounts table that will cascade a delete to the
sub_accounts table, you must turn on logging. After logging is turned on,
you can delete the account number 2 from both tables, as shown in the
following example:

DELETE FROM accounts WHERE acc_num = 2

Restrictions on Cascading Deletes

You can use cascading deletes for most deletes, including deletes on self-
referencing and cyclic queries. The only exception is correlated subqueries. In
correlated subqueries, the subquery (or inner SELECT) is correlated when the
value it produces depends on a value produced by the outer SELECT state-
ment that contains it. If you have implemented cascading deletes, you cannot
write deletes that use a child table in the correlated subquery. You receive an
error when you attempt to delete from a correlated subquery.

Interrupted Modifications
Even if all the software is error-free and all the hardware is utterly reliable,
the world outside the computer can interfere. It is possible for lightning to
strike the building, interrupting the electrical supply and stopping the com-
puter in the middle of your UPDATE statement. It is more likely, however, that
a disk fills up or a user supplies incorrect data, causing your multirow insert
to stop early with an error. In any case, as you are modifying data, you must
assume that some unforeseen event can interrupt the modification.
When a modification is interrupted by an external cause, you cannot be sure
how much of the operation was completed. Even in a single-row operation,
you cannot know whether the data reached the disk or the indexes were
properly updated.

4-22 Modifying Data

 Interrupted Modifications

If multirow modifications are a problem, multistatement ones are worse. It

does not help that they are usually embedded in programs so that you do not
see the individual SQL statements being executed. For example, the job of
entering a new order in the stores6 database requires you to perform the
following steps:
• Insert a row in the orders table. (This generates an order number.)
• For each item ordered, insert a row in the items table.
Two ways to program an order-entry application exist. One way is to make it
completely interactive so that the program inserts the first row immediately,
and then inserts each item as the operator enters data. This is the wrong
approach because it exposes the operation to the possibility of many more
unforeseen events: the customer’s telephone disconnecting, the operator
pressing the wrong key, the operator’s terminal losing power, and so on.
The right way to build an order-entry application is described in the
following list:
• Accept all the data interactively.
• Validate the data and expand it (by looking up codes in stock and
manufact, for example).
• Display the information on the screen for inspection.
• Wait for the operator to make a final commitment.
• Perform the insertions quickly.
Even when this is done, sometimes an unforeseen circumstance can halt the
program after it inserts the order but before it finishes inserting the items. If
that happens, the database is in an unpredictable condition: Its data integrity
is compromised.

The Transaction
The solution to all these potential problems is called the transaction. A
transaction is a sequence of modifications that either must be accomplished
completely or not at all. The database server guarantees that operations per-
formed within the bounds of a transaction are either completely and perfectly
committed to disk, or the database is restored to the state it was in before the
transaction started.
The transaction is not merely protection against unforeseen failures; it also
offers a program a way to escape when the program detects a logical error.
(This is discussed further in Chapter 6, “Modifying Data Through SQL
Programs.”)

Modifying Data 4-23

Interrupted Modifications

Transaction Logging
The database server can keep a record of each change that it makes to the
database during a transaction. If something happens to cancel the transac-
tion, the database server automatically uses the records to reverse the
changes. Many things can make a transaction fail. The program that issues
the SQL statements can crash or be terminated, or there might be a hardware
or software failure in any other component of the system. As soon as the data-
base server discovers that the transaction failed, which might be only after
the computer and the database server are restarted, it returns the database to
the state it was in before the transaction began using the records from the
transaction.
The process of keeping records of transactions is called transaction logging, or
simply logging. The records of the transactions, called log records, are stored in
a portion of disk space separate from the database. In INFORMIX-OnLine
Dynamic Server, this space is called the logical log (because the log records
represent logical units of the transactions). In INFORMIX-SE, the space used
to store log records is called the transaction log file.
Databases do not generate transaction records automatically. The database
administrator (or the INFORMIX-OnLine Dynamic Server administrator, for
OnLine databases) decides whether to make a database use transaction
logging. Without transaction logging, you cannot roll back transactions.

Logging and Cascading Deletes

Logging must be turned on in your database for cascading deletes to work.
This is because when you specify a cascading delete, the delete is first per-
formed on the primary key of the parent table. If the system crashes after the
rows of the primary key of the parent table are performed but before the rows
of the foreign key of the child table are deleted, referential integrity is vio-
lated. If logging is turned off, even temporarily, deletes do not cascade. After
logging is turned back on, however, deletes can cascade again. Turn logging
on with the CREATE DATABASE statement for INFORMIX-OnLine Dynamic
Server database servers.

Specifying Transactions
The boundaries of transactions are specified with SQL statements. You can do
this in two ways. In the most common style, you specify the start of a multi-
statement transaction by executing the BEGIN WORK statement. In databases

4-24 Modifying Data

 Archives and Logs

that are created with the MODE ANSI option, no need exists to mark the
beginning of a transaction. One is always in effect; you indicate only the end
of each transaction.
In both styles, you specify the end of a successful transaction by executing the
COMMIT WORK statement. This statement tells the database server that you
reached the end of a series of statements that must succeed together. The
database server does whatever is necessary to make sure that all
modifications are properly completed and committed to disk.
It is also possible for a program to cancel a transaction deliberately. It does so
by executing the ROLLBACK WORK statement. This statement asks the
database server to cancel the current transaction and undo any changes.
For example, an order-entry application can use a transaction when creating
a new order in the following ways:
• Accept all data interactively
• Validate and expand it
• Wait for the operator to make a final commitment
• Execute BEGIN WORK
• Insert rows in the orders and items tables, checking the error code
returned by the database server
• If there were no errors, execute COMMIT WORK, otherwise execute
ROLLBACK WORK

If any external failure prevents the transaction from being completed, the
partial transaction rolls back when the system restarts. In all cases, the data-
base is in a predictable state: Either the new order is completely entered, or it
is not entered at all.

Archives and Logs

By using transactions, you can ensure that the database is always in a
consistent state and that your modifications are properly recorded on disk.
But the disk itself is not perfectly safe. It is vulnerable to mechanical failures
and to flood, fire, and earthquake. The only safeguard is to keep multiple
copies of the data. These redundant copies are called archive copies.
The transaction log (also called the logical log) complements the archive copy
of a database. Its contents are a history of all modifications that occurred since
the last time the database was archived. If it should ever be necessary to
restore the database from the archive copy, the transaction log can be used to
roll the database forward to its most recent state.
Modifying Data 4-25
Archives and Logs

Archiving with INFORMIX-SE

If a database is stored in operating system files (INFORMIX-SE), archive
copies are made using the normal methods for making backup copies in your
operating system. Only two special considerations exist for databases.
The first is a practical consideration: A database can grow to great size. It may
become the largest file or set of files in the system. It can also be awkward or
very time consuming to make a copy of it. You may need a special procedure
for copying the database, separate from the usual backup procedures, and
the job may not be done too frequently.
The second consideration is the special relationship between the database
and the transaction log file. An archive copy is an image of the database at
one instant. The log file contains the history of modifications that were made
during that instant. It is important that those two instants are identical; in
other words, it is important to start a new transaction log file immediately
upon making an archive copy of the database. Then, if you must restore the
database from the archive tape, the transaction log contains exactly the
history needed to bring it forward in time from that instant to the latest
update.
The statement that applies a log to a restored database is ROLLFORWARD
DATABASE. You start a new log file by using whatever operating system com-
mands are needed to delete the file and re-create it empty or simply to set the
length of the file to zero.
A transaction log file can grow to extreme size. If you update a row ten times,
just one row exists in the database—but ten update events are recorded in the
log file. If the size of the log file is a problem, you can start a fresh log. Choose
a time when the database is not being updated (so no transactions are active),
and copy the existing log to another medium. That copy represents all mod-
ifications for some period of time; preserve it carefully. Then start a new log
file. If you ever have to restore the database, you must apply all the log files
in their correct sequence.

Archiving with INFORMIX-OnLine Dynamic Server

The OnLine database server contains elaborate features to support archiving
and logging. They are described in the INFORMIX-OnLine Dynamic Server
Archive and Backup Guide.

4-26 Modifying Data

 Concurrency and Locks

Conceptually, the facilities of OnLine are similar to those already described

for INFORMIX-SE, but they are more elaborate for the following reasons:
• OnLine has very stringent requirements for performance and reliability
(for example, it supports making archive copies while databases are in
use).
• OnLine manages its own disk space that is devoted to logging.
• It performs logging concurrently for all databases using a limited set of
log files. The log files can be copied to another medium (backed up) while
transactions are active.
These facilities are usually managed from a central location by the OnLine
administrator, so the users of the databases never have to be concerned with
them.
If you want to make a personal archive copy of a single database or table that
is held by OnLine, you can do it with the onunload utility. This program cop-
ies a table or a database to tape. Its output consists of binary images of the
disk pages as they were stored in OnLine. As a result, the copy can be made
very quickly, and the corresponding onload program can restore the file very
quickly. However, the data format is not meaningful to any other programs.
If your OnLine administrator is using ON-Archive to create archives and back
up logical logs, you also may be able to create your own backup copies using
ON-Archive. See your INFORMIX-OnLine Dynamic Server Archive and Backup
Guide for more information.

Concurrency and Locks

If your database is contained in a single-user workstation, without a network
connecting it to other machines, concurrency is unimportant. In all other
cases, you must allow for the possibility that, while your program is modify-
ing data, another program is also reading or modifying the same data. This is
concurrency: two or more independent uses of the same data at the same time.
A high level of concurrency is crucial to good performance in a multiuser
database system. Unless controls exist on the use of data, however, concur-
rency can lead to a variety of negative effects. Programs could read obsolete
data; modifications could be lost even though it seems they were entered
successfully.

Modifying Data 4-27

Data Replication

The database server prevents errors of this kind by imposing a system of

locks. A lock is a claim, or reservation, that a program can place on a piece of
data. The database server guarantees that, as long as the data is locked, no
other program can modify it. When another program requests the data, the
database server either makes the program wait or turns it back with an error.
You use two statements to control the effect that locks have on your data
access: SET LOCK MODE and SET ISOLATION. You can understand the details
of these statements are best understood after a discussion on the use of cur-
sors from within programs. Cursors are covered in Chapter 5, “Programming
with SQL,” and Chapter 6, “Modifying Data Through SQL Programs.” Also
see Chapter 7, “Programming for a Multiuser Environment,” for more
information about locking.

Data Replication
Data replication, in the broadest sense of the term, is when database objects
have more than one representation at more than one distinct site. For exam-
ple, a database that is copied to a database server on a different machine—so
that reports can be run against the data without disturbing client applications
using the original database—is one way data can be replicated.
The following list describes the advantages of data replication:
• Clients accessing replicated data locally—as opposed to remote data that
is not replicated—experience improved performance because they do not
have to use network services.
• Clients at all sites experience improved availability with replicated data,
because if local replicated data is unavailable, a copy of the data is still
available, albeit remotely.
These advantages do not come without a cost. Data replication obviously
requires more storage for replicated data than for unreplicated data, and
updating replicated data can take more processing time than updating a
single object.
Data replication can actually be implemented in the logic of client
applications, by explicitly specifying where data should be found or
updated. However, this way of achieving data replication is costly, error-
prone, and difficult to maintain. Instead, the concept of data replication is
often coupled with replication transparency. Replication transparency is func-
tionality built into a database server (instead of client applications) to
automatically handle the details of locating and maintaining data replicas.

4-28 Modifying Data

 Summary

INFORMIX-OnLine Dynamic Server Data Replication

Within the broad framework of data replication, OnLine implements nearly
transparent data replication of entire database servers. All the data managed
by one OnLine database server is replicated and dynamically updated on
another OnLine database server, usually at a remote site. OnLine data repli-
cation is sometimes called hot site backup, because it provides a means of
maintaining a backup copy of the entire database server that can be used
quickly in the event of a catastrophic failure.
Because OnLine provides replication transparency, you generally do not
need to be concerned with or aware of data replication; the OnLine adminis-
trator takes care of it. However, if your organization decides to use data rep-
lication, you should be aware that special connectivity considerations exist
for client applications in a data replication environment. These
considerations are described in the INFORMIX-OnLine Dynamic Server
Administrator’s Guide.

Summary
Database access is regulated by the privileges that the database owner grants
to you. The privileges that let you query data are often granted automatically,
but the ability to modify data is regulated by specific Insert, Delete, and
Update privileges that are granted on a table-by-table basis.
If data integrity constraints are imposed on the database, your ability to
modify data is restricted by those constraints. Your database- and table-level
privileges, along with any data constraints, control how and when you can
modify data.
You can delete one or more rows from a table with the DELETE statement. Its
WHERE clause selects the rows; use a SELECT statement with the same clause
to preview the deletes.
Rows are added to a table with the INSERT statement. You can insert a single
row containing specified column values, or you can insert a block of rows
generated by a SELECT statement.
You use the UPDATE statement to modify the contents of existing rows. You
specify the new contents with expressions that can include subqueries, so
that you can use data based on other tables or the updated table itself. The
statement has two forms: in the first form you specify new values column by
column; you use the other form when the new values are generated as a set
from a SELECT statement or a record variable.

Modifying Data 4-29

Summary

You use the REFERENCES clause of the CREATE TABLE and ALTER TABLE
statements to create relationships between tables. The ON DELETE CASCADE
option of the REFERENCES clause allows you to delete rows from parent and
associated child tables with one DELETE statement.
You use transactions to prevent unforeseen interruptions in a modification
from leaving the database in an indeterminate state. When modifications are
performed within a transaction, they are rolled back following an error. The
transaction log also extends the periodically made archive copy of the data-
base, so that if the database must be restored, it can be brought back to its
most recent state.
Data replication, which is transparent to users, offers another type of
protection from catastrophic failures.

4-30 Modifying Data

 Chapter

Programming with
SQL
Chapter Overview 3
SQL in Programs 3
Static Embedding 5
Dynamic Statements 5
Program Variables and Host Variables 5
Calling the Database Server 8
The SQL Communications Area 8
The SQLCODE Field 11
End of Data 11
Negative Codes 11
The SQLERRD Array 12
The SQLAWARN Array 12
5
The SQLSTATE Value 12
Retrieving Single Rows 13
Data Type Conversion 14
Dealing with Null Data 15
Dealing with Errors 16
End of Data 17
Serious Errors 17
Using Default Values 17
End of Data with Databases that are not
ANSI-Compliant 18
Interpreting End of Data with Aggregate
Functions 18
Retrieving Multiple Rows 19
Declaring a Cursor 20
Opening a Cursor 20
 Fetching Rows 21
Detecting End of Data 21
Locating the INTO Clause 22
Cursor Input Modes 23
The Active Set of a Cursor 24
Creating the Active Set 24
The Active Set for a Sequential Cursor 25
The Active Set for a Scroll Cursor 25
The Active Set and Concurrency 25
Using a Cursor: A Parts Explosion 26
Dynamic SQL 28
Preparing a Statement 29
Executing Prepared SQL 31
Using Prepared SELECT Statements 31
Dynamic Host Variables 32
Freeing Prepared Statements 33
Quick Execution 34
Embedding Data Definition Statements 34
Embedding Grant and Revoke Privileges 34
Summary 37

5-2 Programming with SQL

Chapter Overview
In the examples in the previous chapters, SQL is treated as if it were an
interactive computer language; that is, as if you could type a SELECT
statement directly into the database server and see rows of data rolling back
to you.
Of course, that is not how things are. Many layers of software stand between
you and the database server. The database server retains data in a binary
form that must be formatted before it can be displayed. It does not return a
mass of data at once; it returns one row at a time, as a program requests it.
You can access information in your database in two ways: through interactive
access using DB-Access or INFORMIX-SQL or through application programs
written using an SQL API.
Almost any program can contain SQL statements, execute them, and retrieve
data from a database server. This chapter explains how these activities are
performed and indicates how you can write programs that perform them.
This chapter is only an introduction to the concepts that are common to SQL
programming in any language. Before you can write a successful program in
a particular programming language, you must first become fluent in that lan-
guage. Then, because the details of the process are slightly different in every
language, you must become familiar with the manual for the Informix SQL
API specific to that language.

SQL in Programs
You can write a program in any of several languages and mix SQL statements
in among the other statements of the program, just as if they were ordinary
statements of that programming language. The SQL statements are said to be
embedded in the program, and the program is said to contain embedded SQL,
often abbreviated by Informix as ESQL.

Programming with SQL 5-3

SQL in Programs

Informix produces SQL APIs for the following programming languages:

• C
• COBOL
• FORTRAN (pre-version 6.0)
• Ada (version 4.0 only)
All SQL APIs products work in a similar way, as shown in Figure 5-1. You
write a source program in which you treat SQL statements as executable code.
Your source program is processed by an embedded SQL preprocessor, a pro-
gram that locates the embedded SQL statements and converts them into a
series of procedure calls and special data structures.

ESQL source ESQL Source program Language Executable

program preprocessor with procedure calls compiler program

Figure 5-1 Overview of processing a program with embedded SQL statements

The converted source program then passes through the programming

language compiler. The compiler output becomes an executable program
after it is linked with a library of SQL API procedures. When the program
runs, the SQL API library procedures are called; they set up communications
with the database server to carry out the SQL operations.
Whereas SQL APIs products allow you to embed SQL in the host language,
some languages have SQL as a natural part of their statement set. INFORMIX-
4GL incorporates the SQL language as a natural part of the fourth-generation
language it supports. Informix Stored Procedure Language (SPL) also uses
SQL as a natural part of its statement set. You use INFORMIX-4GL or an SQL
API product to write application programs. You use SPL to write procedures
that are stored with a database and are called from an application program.

5-4 Programming with SQL

 SQL in Programs

Static Embedding
You can introduce SQL statements into a program in two ways. The simpler
and more common way is by static embedding, which means that the SQL
statements are written as part of the source program text. The statements are
static because they are a fixed part of the source text.

Dynamic Statements
Some applications require the ability to compose SQL statements in response
to user input. For example, a program might have to select different columns
or apply different criteria to rows, depending on what the user wants.
This can be done with dynamic SQL, in which the program composes an SQL
statement as a string of characters in memory and passes it to the database
server to be executed. Dynamic statements are not part of the program source
text; they are constructed in memory during execution.

Program Variables and Host Variables

Application programs can use program variables within SQL statements.
In INFORMIX-4GL and SPL, you put the program variable in the SQL
statement as syntax allows. For example, a DELETE statement can use a pro-
gram variable in its WHERE clause. Figure 5-2 shows a program variable in
INFORMIX-4GL.

MAIN
.
.
.
DEFINE drop_number INT
LET drop_number = 108
DELETE FROM items WHERE order_num = drop_number
.
.
.

Figure 5-2 Using a program variable in INFORMIX-4GL

Programming with SQL 5-5

SQL in Programs

Figure 5-3 shows a program variable in SPL.

CREATE PROCEDURE delete_item (drop_number INT)

.
.
.
DELETE FROM items WHERE order_num = drop_number
.
.
.

Figure 5-3 Using a program variable in SPL

In applications that use embedded SQL statements, the SQL statements can
refer to the contents of program variables. A program variable that is named
in an embedded SQL statement is called a host variable because the SQL
statement is thought of as being a “guest” in the program.
The following example is a DELETE statement as it might appear when
embedded in a COBOL source program:

EXEC SQL
DELETE FROM items
WHERE order_num = :o-num
END-EXEC.

The first and last lines mark off embedded SQL from the normal COBOL
statements. Between them you see an ordinary DELETE statement, as
described in Chapter 4, “Modifying Data.” When this part of the COBOL
program is executed, a row of the items table is deleted; multiple rows can
also be deleted.
The statement contains one new feature. It compares the order_num column
to an item written as :o-num, which is the name of a host variable.
Each SQL API product provides a means of delimiting the names of host
variables when they appear in the context of an SQL statement. In COBOL,
host variable names are designated with an initial colon. The example state-
ment asks the database server to delete rows in which the order number
equals the current contents of the host variable named :o-num. This is a
numeric variable that has been declared and assigned a value earlier in the
program.

5-6 Programming with SQL

 SQL in Programs

The same statement embedded in a FORTRAN program looks like the

following example:

EXEC SQL
DELETE FROM items
WHERE order_num = :onum

The same statement embedded in an Ada program looks like the following
example:

EXEC SQL
DELETE FROM items
WHERE order_num = $onum;

In INFORMIX-ESQL/Ada, a host variable is indicated by a leading dollar sign,

and statements end with a semicolon. The same is true of INFORMIX-ESQL/C,
where the corresponding DELETE statement is written as shown in the fol-
lowing example:

EXEC SQL delete FROM items

WHERE order_num = :onum;

In INFORMIX-ESQL/C, an SQL statement can be introduced with either a

leading dollar sign or the words EXEC SQL.
These differences of syntax are trivial; the essential points in all languages (an
SQL API, INFORMIX-4GL, or SPL) are described in the following list:
• You can embed SQL statements in a source program as if they were
executable statements of the host language.
• You can use program variables in SQL expressions the way literal values
are used.
If you have programming experience, you can immediately see the
possibilities. In the example, the order number to be deleted is passed in the
variable onum. That value comes from any source that a program can use: it
can be read from a file, the program can prompt a user to enter it, or it can be
read from the database. The DELETE statement itself can be part of a subrou-
tine (in which case onum can be a parameter of the subroutine); the
subroutine can be called once or repetitively.

Programming with SQL 5-7

Calling the Database Server

In short, when you embed SQL statements in a program, you can apply all the
power of the host language to them. You can hide the SQL under a multitude
of interfaces, and you can embellish its functions in a multitude of ways.

Calling the Database Server

Executing an SQL statement is essentially calling the database server as a
subroutine. Information must pass from the program to the database server
and information must be returned.
Some of this communication is done through host variables. You can think of
the host variables named in an SQL statement as the parameters of the proce-
dure call to the database server. In the preceding example, a host variable acts
as a parameter of the WHERE clause. Also, as described in “Retrieving Multi-
ple Rows” on page 5-19, host variables receive data that is returned by the
database server.

The SQL Communications Area

The database server always returns a result code, and possibly other informa-
tion about the effect of an operation, in a data structure known as the SQL
Communications Area (SQLCA). If the database server executes an SQL state-
ment in a stored procedure, the SQLCA of the calling application contains the
values triggered by the SQL statement in the procedure.
The principal fields of the SQLCA are listed in Figure 5-4 and Figure 5-5. The
syntax you use to describe a data structure such as the SQLCA, as well as the
syntax you use to refer to a field in it, differs among programming languages.
See your SQL API manual for details.
You can also use the SQLSTATE variable of the GET DIAGNOSTICS statement
to detect, handle, and diagnose errors. See “The SQLSTATE Value” on
page 5-12.

5-8 Programming with SQL

 Calling the Database Server

integer
SQLCODE

0 Success.
100 No more data/not found.
negative Error code.
array of 6
integers
SQLERRD

first Following successful prepare of a SELECT, UPDATE, INSERT, or DELETE statement or

after a select cursor is opened, this field contains the estimated number of affected rows.
It is not used in INFORMIX-ESQL/Ada.
second When SQLCODE contains an error code, this field contains either zero or an additional
error code, called the ISAM error code, that explains the cause of the main error.
After a failed CONNECT or DATABASE statement, this field contains the ISAM,
operating system, or network-specific protocol error code.
If an application connects to an INFORMIX-Gateway with DRDA, this field contains
the application server error code. In this case, sqlca.sqlcode is -29000.
Following a successful insert operation of a single row, this field contains the value of a
generated serial number for that row.
third Following a successful multirow insert, update, or delete operation, this field contains the
number of processed rows.
Following a multirow insert, update, or delete operation that ends with an error, this field
contains the number of rows successfully processed before the error was detected.
fourth Following successful prepare of a SELECT, UPDATE, INSERT, or DELETE statement or
after a select cursor has been opened, this field contains the estimated weighted sum of
disk accesses and total rows processed. It is not used in INFORMIX-ESQL/Ada.
fifth Following a syntax error in a PREPARE, EXECUTE IMMEDIATE, DECLARE, or static
SQL statement, this field contains the offset in the statement text where the error was
detected.
sixth Following a successful fetch of a selected row or a successful insert, update, or delete
operation, this field contains the rowid (physical address) of the last row processed.

character SQLERRP
(8)

Internal use only.

Figure 5-4 The uses of SQLCODE, SQLERRD, and SQLERRP

Programming with SQL 5-9

Calling the Database Server

array of 8
characters SQLAWARN

When Opening or Connecting to a Database:

first Set to W when any field is set to W. If this field is blank, the others need not be checked.
second Set to W when the database now open uses a transaction log.
third Set to W when the database now open is ANSI-compliant.
fourth Set to W when the database server is INFORMIX-OnLine Dynamic Server.
fifth Set to W when the database server stores the FLOAT data type in DECIMAL form (done
when the host system lacks support for FLOAT types).
sixth Not used.
seventh Set to W when the application is connected to an INFORMIX-OnLine Dynamic Server
database server, which is a secondary server in a data replication pair (that is, the database
server is available only for read operations.
eighth Not used.

All Other Operations:

first Set to W when any other field is set to W.
second Set to W when a column value is truncated as it is fetched into a host variable.
third Set to W when an aggregate function encounters a null value.
fourth On a SELECT or on opening a cursor, set to W when the number of items in the select list
is not the same as the number of host variables given in the INTO clause to receive them.
fifth Set to W when any part of a statement involves a host database server which stores the
FLOAT data type in DECIMAL form (done when the host system lacks support for
FLOAT types).

sixth Set to W following execution of a statement that does not use ANSI-standard SQL syntax
(provided that the DBANSIWARN environment variable is set).

seventh
Not used.
eighth
Not used.
character
(71) SQLERRM

Contains the variable, such as table name, that is placed in the error message. For some
networked applications, contains an error message generated by networking software.

Figure 5-5 The uses of SQLAWARN and SQLERRM

5-10 Programming with SQL

 Calling the Database Server

In particular, the subscript by which you name one element of the SQLERRD
and SQLAWARN arrays differs: array elements are numbered starting with
zero in INFORMIX-ESQL/C, but starting with one in the other languages. In
this discussion, the fields are named using specific words such as third, and
you must translate into the syntax of your programming language.

The SQLCODE Field

The SQLCODE field is the primary return code of the database server. After
every SQL statement, SQLCODE is set to an integer value as shown in Figure
5-4. When that value is zero, the statement is performed without error. In par-
ticular, when a statement is supposed to return data into a host variable, a
code of zero means that the data has been returned and can be used. Any
nonzero code means the opposite: no useful data was returned to host
variables.
In INFORMIX-4GL, SQLCODE is also accessible under the name STATUS.

End of Data
The database server sets SQLCODE to 100 when the statement is performed
correctly but no rows are found. Two situations exist where this can occur.
The first situation involves a query that uses a cursor. (Queries using cursors
are described under “Retrieving Multiple Rows” on page 5-19.) In these que-
ries the FETCH statement retrieves each value from the active set into mem-
ory. After the last row is retrieved, a subsequent FETCH statement cannot
return any data. When this occurs the database server sets SQLCODE to 100,
which says end of data-no rows found.
The second situation involves a query that does not use a cursor. In this case
the server sets SQLCODE to 100 when no rows satisfy the query condition. In
ANSI-compliant databases, SELECT, DELETE, UPDATE, and INSERT state-
ments all set SQLCODE to 100 if no rows are returned. In databases that are
not ANSI-compliant, only a SELECT statement that returns no rows causes
SQLCODE to be set to 100.

Negative Codes
When something unexpected goes wrong during a statement, the database
server returns a negative number in SQLCODE to explain the problem. The
meanings of these codes are documented in the Informix Error Messages
manual and in the on-line error message file.

Programming with SQL 5-11

Calling the Database Server

The SQLERRD Array

Some error codes that can be reported in SQLCODE reflect general problems,
and the database server can set a more detailed code in the second field of
SQLERRD. This second code (referred to as the ISAM error) reveals the low-
level error encountered by the database server I/O routines or by the
underlying operating system.
The integers in the array named SQLERRD are set to different values
following different statements. The first and fourth elements of the array are
used only in INFORMIX-4GL, INFORMIX-ESQL/C, INFORMIX-ESQL/COBOL,
and INFORMIX-ESQL/FORTRAN. The fields are used as shown in Figure 5-4
on page 5-9.
These additional details can be very useful. For example, you can use the
value in the third field to report how many rows were deleted or updated.
When your program prepares an SQL statement entered by the user and an
error is found, the value in the fifth field enables you to display the exact
point of error to the user. (DB-Access and INFORMIX-SQL use this feature to
position the cursor when you ask to modify a statement after an error.)

The SQLAWARN Array

The eight character fields in the SQLAWARN array are set to either a blank or
a W to indicate a variety of special conditions; only six are used. Their
meanings depend on the statement just executed.
A set of warning flags appear when a database opens, that is, following a
CONNECT, DATABASE or CREATE DATABASE statement. These flags tell you
some characteristics of the database as a whole.
A second set of flags appear following any other statement. These flags reflect
unusual events that occur during the statement, events that might not be
reflected by SQLCODE.
Both sets of SQLAWARN values are summarized in Figure 5-5.

The SQLSTATE Value

Certain Informix products, such as INFORMIX-ESQL/C and
INFORMIX-ESQL/COBOL, support the SQLSTATE value in compliance with
X/Open and ANSI SQL standards. The GET DIAGNOSTICS statement reads the
SQLSTATE value in order to diagnose errors after you run an SQL statement.
The database server returns a result code in a five-character string that is

5-12 Programming with SQL

 Retrieving Single Rows

stored in a variable called SQLSTATE. The SQLSTATE error code, or value, tells
you the following information about the most recently executed SQL
statement:
• If the statement was successful
• If the statement was successful but generated warnings
• If the statement was successful but generated no data
• If the statement failed
For more information on GET DIAGNOSTICS, the SQLSTATE variable, and the
meanings of the SQLSTATE return codes, see the section called “GET DIAG-
NOSTICS” in Chapter 1 of the Informix Guide to SQL: Syntax. If your Informix
product supports GET DIAGNOSTICS and SQLSTATE, Informix recommends
that you use them as the primary structure to detect, handle, and diagnose
errors.

Retrieving Single Rows

You can use embedded SELECT statements to retrieve single rows from the
database into host variables. When a SELECT statement returns more than
one row of data, however, a program must use a more complicated method
to fetch the rows one at a time. Multiple-row select operations are discussed
in “Retrieving Multiple Rows” on page 5-19.
To retrieve a single row of data, simply embed a SELECT statement in your
program. The following example shows how it can be written using
INFORMIX-ESQL/C:

exec sql select avg (total_price)

into :avg_price
from items
where order_num in
(select order_num from orders
where order_date < date('6/1/93'));

The INTO clause is the only detail that distinguishes this statement from any
example in Chapter 2, “Composing Simple SELECT Statements,” or
Chapter 3, “Composing Advanced SELECT Statements.” This clause
specifies the host variables that are to receive the data that is produced.

Programming with SQL 5-13

Retrieving Single Rows

When the program executes an embedded SELECT statement, the database

server performs the query. The example statement selects an aggregate value,
so that it produces exactly one row of data. The row has only a single column,
and its value is deposited in the host variable named avg_price. Subsequent
lines of the program can use that variable.
You can use statements of this kind to retrieve single rows of data into host
variables. The single row can have as many columns as desired. In the follow-
ing INFORMIX-4GL example, host variables are used in two ways, as
receivers of data and in the WHERE clause:

DEFINE cfname, clname, ccompany CHAR(20)

DEFINE cnumbr INTEGER
LET cnumbr = 104
SELECT fname, lname, company
INTO cfname, clname, ccompany
FROM customer
WHERE customer_num = cnumbr

Because the customer_num column has a unique index (implemented

through a constraint), this query returns only one row. If a query produces
more than one row of data, the database server cannot return any data. It
returns an error code instead.
You should list as many host variables in the INTO clause as there are items
in the select list. If, by accident, these lists are of different lengths, the data-
base server returns as many values as it can and sets the warning flag in the
fourth field of SQLAWARN.

Data Type Conversion

The following example retrieves the average of a DECIMAL column, which is
itself a DECIMAL value. However, the host variable into which it is placed is
not required to have that data type.

exec sql select avg (total_price) into :avg_price

from items;

5-14 Programming with SQL

 Retrieving Single Rows

The declaration of the receiving variable avg_price in the previous example

of ESQL/C code is not shown. It could be any one of the following definitions:

int avg_price;
double avg_price;
char avg_price[16];
dec_t avg_price; /* typedef of decimal number structure */

The data type of each host variable used in a statement is noted and passed
to the database server along with the statement. The database server does its
best to convert column data into the form used by the receiving variables.
Almost any conversion is allowed, although some conversions cause a loss of
precision. The results of the preceding example differ depending on the data
type of the receiving host variable, as described in the following list:
FLOAT The database server converts the decimal result to FLOAT,
possibly truncating some fractional digits.
If the magnitude of a decimal exceeds the maximum
magnitude of the FLOAT format, an error is returned.
INTEGER The database server converts the result to integer, truncating
fractional digits if necessary.
If the integer part of the converted number does not fit the
receiving variable, an error occurs.
CHARACTER The database server converts the decimal value to a
character string.
If the string is too long for the receiving variable, it is
truncated. The second field of SQLAWARN is set to W and the
value in the SQLSTATE variable is 01000.

Dealing with Null Data

What if the program retrieves a null value? Null values can be stored in the
database, but the data types supported by programming languages do not
recognize a null state. A program must have some way of recognizing a null
item to avoid processing it as data.
Indicator variables meet this need in the SQL APIs. An indicator variable is an
additional variable that is associated with a host variable that might receive
a null item. When the database server puts data in the main variable, it also

Programming with SQL 5-15

Retrieving Single Rows

puts a special value in the indicator variable to show whether the data is null.
In the following INFORMIX-ESQL/C example, a single row is selected and a
single value is retrieved into the host variable op_date:

exec sql select paid_date

into :op_date:op_d_ind
from orders
where order_num = $the_order;
if (op_d_ind < 0) /* data was null */
rstrdate ('01/01/1900', :op_date);

Because the value might be null, an indicator variable named op_d_ind is

associated with the host variable. (It must be declared as a short integer
elsewhere in the program.)
Following execution of the SELECT statement, the program tests the indicator
variable for a negative value. A negative number (usually -1) means that the
value retrieved into the main variable is null. If that is the case, this program
uses a C function to assign a default value to the host variable. (The function
rstrdate is part of the INFORMIX-ESQL/C product.)
The syntax you use to associate an indicator variable differs with the
language you are using, but the principle is the same in all. However, indica-
tor variables are not used explicitly in INFORMIX-4GL or in SPL because in
those languages, null values are supported for variables. In 4GL, the
preceding example is written as follows:

SELECT paid_date
INTO op_date
FROM orders
WHERE order_num = the_order
IF op_date IS NULL THEN
LET op_date = date ('01/01/1900')
END IF

Dealing with Errors

Although the database server handles conversion between data types
automatically, several things still can go wrong with a SELECT statement. In
SQL programming, as in any kind of programming, you must anticipate
errors and provide for them at every point.

5-16 Programming with SQL

 Retrieving Single Rows

End of Data
One common event is that no rows satisfy a query. This is signalled by an
SQLSTATE code of 02000 and by a code of 100 in SQLCODE (sqlca.sqlcode,
as it is known in ESQL/C) following a SELECT statement. This code indicates
an error or a normal event, depending entirely on your application. If you are
sure that there ought to be a row or rows—for example, if you are reading a
row using a key value that you just read from a row of another table—then
the end-of-data code represents a serious failure in the logic of the program.
On the other hand, if you select a row based on a key supplied by a user or
some other source that is less reliable than a program, a lack of data can be a
normal event.

Serious Errors
Errors that set SQLCODE to a negative value are usually serious. Programs
which you have developed and are in production should rarely report these
errors. Nevertheless, it is difficult to anticipate every problematic situation,
so your program must be able to deal with them.
For example, a query can return error -206, which means table name is
not in the database. This happens if someone dropped the table since
the program was written or if, through some error of logic or mistake in
input, the program opened the wrong database.

Using Default Values

You can handle these inevitable errors in many ways. In some applications,
more lines of code to handle errors are used than for normal situations. In the
examples in this section, however, one of the simplest methods, the default
value, should work.

avg_price = 0; /* set default for errors */

exec sql select avg (total_price)
into :avg_price:null_flag
from items;
if (null_flag < 0) /* probably no rows */
avg_price = 0; /* set default for 0 rows */

Programming with SQL 5-17

Retrieving Single Rows

End of Data with Databases that are not ANSI-Compliant

If your database is not ANSI-compliant, the end-of-data return code, 100, is
set in SQLCODE only following SELECT statements. In addition, the SQL-
STATE value is set to 02000. (Other statements, such as INSERT, UPDATE,
and DELETE, set the third element of SQLERRD to show how many rows
they affected; this topic is covered in Chapter 6, “Modifying Data Through
SQL Programs.”)

Interpreting End of Data with Aggregate Functions

A SELECT statement that selects an aggregate function such as SUM, MIN, or
AVG always succeeds in returning at least one row of data. This is true even
when no rows satisfy the WHERE clause; an aggregate value based on an
empty set of rows is null, but it exists nonetheless.
However, an aggregate value is also null if it is based on one or more rows
that all contain null values. If you must be able to detect the difference
between an aggregate value based on no rows and one based on some rows
that are all null, you must include a COUNT function in the statement and
an indicator variable on the aggregate value. You can then work out the
following three cases:

Count Value Indicator Case

0 -1 zero rows selected
>0 -1 some rows selected; all were null
>0 0 some non-null rows selected

The previous example deals with the following considerations:

• If the query selects some non-null rows, the correct value is returned and
used. This is the expected and most frequent result.
• If the query selects no rows, or in the much less likely event that it selects
only rows that have null values in the total_price column (a column that
should never be null), the indicator variable is set and the default value is
assigned.
• If any serious error occurs, the host variable is left unchanged; it contains
the default value initially set. At this point in the program, the
programmer sees no need to trap such errors and report them.

5-18 Programming with SQL

 Retrieving Multiple Rows

The following example is an expansion of an earlier INFORMIX-4GL example

that displays default values if it cannot find the company that the user
requests:

DEFINE cfname, clname, ccompany CHAR(20)

DEFINE cnumbr INTEGER
PROMPT 'Enter the customer number: ' FOR cnumbr
LET cfname = 'unknown'
LET clname = 'person'
LET ccompany = 'noplace'
SELECT fname, lname, company
INTO cfname, clname, ccompany
WHERE customer_num = cnumbr
DISPLAY cfname,' ', clname,' at ', ccompany

This query does not use aggregates, so if no row matches the user-specified
customer number, SQLCODE is set to 100 and the host variables remain
unchanged.

Retrieving Multiple Rows

When any chance exists that a query could return more than one row, the
program must execute the query differently. Multirow queries are handled in
two stages. First, the program starts the query. (No data is returned
immediately.) Then, the program requests the rows of data one at a time.
These operations are performed using a special data object called a cursor. A
cursor is a data structure that represents the current state of a query. The
following list is the general sequence of program operations:
1. The program declares the cursor and its associated SELECT statement. This
merely allocates storage to hold the cursor.
2. The program opens the cursor. This starts the execution of the associated
SELECT statement and detects any errors in it.
3. The program fetches a row of data into host variables and processes it.
4. The program closes the cursor after the last row is fetched.
These operations are performed with SQL statements named DECLARE,
OPEN, FETCH, and CLOSE.

Programming with SQL 5-19

Retrieving Multiple Rows

Declaring a Cursor
You declare a cursor using the DECLARE statement. This statement gives the
cursor a name, specifies its use, and associates it with a statement. The
following simple example is written in INFORMIX-4GL:

DECLARE the_item CURSOR FOR

SELECT order_num, item_num, stock_num
INTO o_num, i_num, s_num
FROM items

The declaration gives the cursor a name (the_item in this case) and associates
it with a SELECT statement. (Chapter 6, “Modifying Data Through SQL
Programs,” discusses how a cursor also can be associated with an INSERT
statement.)
The SELECT statement in this example contains an INTO clause. That is one of
two ways in which you can specify the variables that receive data. The other
way, using the FETCH statement, is discussed in “Locating the INTO Clause”
on page 5-22.
The DECLARE statement is not an active statement; it merely establishes the
features of the cursor and allocates storage for it. You can use the cursor
declared in the preceding example to read once through the items table. Cur-
sors can be declared to read both backward and forward (see “Cursor Input
Modes” on page 5-23). This cursor, because it lacks a FOR UPDATE clause,
probably is used only to read data, not to modify it. (The use of cursors to
modify data is covered in Chapter 6, “Modifying Data Through SQL
Programs.”)

Opening a Cursor
The program opens the cursor when it is ready to use it. The OPEN statement
activates the cursor. It passes the associated SELECT statement to the database
server, which begins the search for matching rows. The database server pro-
cesses the query to the point of locating or constructing the first row of out-
put. It does not actually return that row of data, but it does set a return code
in SQLSTATE for SQL APIs and SQLCODE for INFORMIX-4GL and SQL APIs.
The following example shows the OPEN statement in INFORMIX-4GL:

OPEN the_item

5-20 Programming with SQL

 Retrieving Multiple Rows

Because this is the first time that the database server has seen the query, it is
the time when many errors are detected. After opening the cursor, the
program should test SQLSTATE or SQLCODE. If the SQLSTATE value is greater
than 02000 or the SQLCODE contains a negative number, the cursor is not
usable. There may be an error in the SELECT statement, or some other
problem may be preventing the database server from executing the
statement.
If SQLSTATE is equal to 00000 or SQLCODE contains a zero, the SELECT
statement is syntactically valid, and the cursor is ready for use. At this point,
however, the program does not know if the cursor can produce any rows.

Fetching Rows
The program uses the FETCH statement to retrieve each row of output. This
statement names a cursor, and also can name the host variables to receive the
data. Here is the INFORMIX-4GL example completed:

DECLARE the_item CURSOR FOR

SELECT order_num, item_num, stock_num
INTO o_num, i_num, s_num
FROM items
OPEN the_item
WHILE sqlcode = 0
FETCH the_item
IF sqlcode = 0 THEN
DISPLAY o_num, i_num, s_num
END IF
END WHILE

Detecting End of Data

In the previous example, the WHILE condition prevents execution of the loop
in case the OPEN statement returns an error. The same condition terminates
the loop when SQLCODE is set to 100 to signal the end of data. However, the
loop contains a test of SQLCODE. This test is necessary because, if the SELECT

Programming with SQL 5-21

Retrieving Multiple Rows

statement is valid yet finds no matching rows, the OPEN statement returns a
zero but the first fetch returns 100, end of data, and no data. The following
example shows another way to write the same loop:

DECLARE the_item CURSOR FOR

SELECT order_num, item_num, stock_num
INTO o_num, i_num, s_num
FROM items
OPEN the_item
IF sqlcode = 0 THEN
FETCH the_item -- fetch first row
END IF
WHILE sqlcode = 0
DISPLAY o_num, i_num, s_num
FETCH the_item
END WHILE

In this version, the case of zero returned rows is handled early, so no second
test of sqlcode exists within the loop. These versions have no measurable dif-
ference in performance because the time cost of a test of sqlcode is a tiny
fraction of the cost of a fetch.

Locating the INTO Clause

The INTO clause names the host variables that are to receive the data returned
by the database server. It must appear in either the SELECT or the FETCH
statement but not both. The following example is reworked to specify host
variables in the FETCH statement:

DECLARE the_item CURSOR FOR

SELECT order_num, item_num, stock_num
FROM items
OPEN the_item
WHILE status = 0
FETCH the_item INTO o_num, i_num, s_num
IF status = 0 THEN
DISPLAY o_num, i_num, s_num
END IF
END WHILE

The second form has the advantage that different rows can be fetched into
different variables. For example, you can use this form to fetch successive
rows into successive elements of an array.

5-22 Programming with SQL

 Retrieving Multiple Rows

Cursor Input Modes

For purposes of input, a cursor operates in one of two modes: sequential or
scrolling. A sequential cursor can fetch only the next row in sequence so it can
read through a table only once each time it is opened. A scroll cursor can fetch
the next row or any prior row, so it can read rows multiple times. The
following example show a sequential cursor declared in INFORMIX-ESQL/C:

EXEC SQL DECLARE pcurs CURSOR FOR

SELECT customer_num, lname, city
FROM customer;

After it is opened, the cursor can be used only with a sequential fetch that
retrieves the next row of data, as shown in the following example:

EXEC SQL FETCH p_curs INTO :cnum, :clname, :ccity;

Each sequential fetch returns a new row.

A scroll cursor is declared with the keyword SCROLL, as shown in the
following example from INFORMIX-ESQL/FORTRAN:

EXEC SQL DECLARE s_curs SCROLL CURSOR FOR

+ SELECT order_num, order_date FROM orders
+ WHERE customer_num > 104

A scroll cursor can be used with a variety of fetch options. The ABSOLUTE
option specifies the rank number of the row to fetch.

EXEC SQL FETCH ABSOLUTE numrow s_curs

+ INTO :nordr, :nodat

This statement fetches the row whose position is given in the host variable
numrow. It is also possible to fetch the current row again or to fetch the first
row and then scan through the entire list again. However, these features are
obtained at a price, as described in the next section.

Programming with SQL 5-23

Retrieving Multiple Rows

The Active Set of a Cursor

Once a cursor is opened, it stands for some selection of rows. The set of all
rows that the query produces is called the active set of the cursor. It is easy to
think of the active set as a well-defined collection of rows and to think of the
cursor as pointing to one row of the collection. This situation is true as long
as no other programs are modifying the same data concurrently.

Creating the Active Set

When a cursor is opened, the database server does whatever is necessary to
locate the first row of selected data. Depending on how the query is phrased,
this can be very easy to do, or it can require a great deal of work and time.

DECLARE easy CURSOR FOR

SELECT fname, lname FROM customer
WHERE state = 'NJ'

Because this cursor queries only a single table in a simple way, the database
server can very quickly discover whether any rows satisfy the query and find
the first one. The first row is the only row it finds at this time. The rest of the
rows in the active set remain unknown.

DECLARE hard CURSOR FOR

SELECT C.customer_num, O.order_num, sum (items.total_price)
FROM customer C, orders O, items I
WHERE C.customer_num = O.customer_num
AND O.order_num = I.order_num
AND O.paid_date is null
GROUP BY C.customer_num, O.order_num

The active set of this cursor is generated by joining three tables and grouping
the output rows. The optimizer might be able to use indexes to produce the
rows in the correct order, but in general the use of ORDER BY or GROUP BY
clauses requires the database server to generate all the rows, copy them to a
temporary table, and sort the table, before it can know which row to present
first. This is discussed in Chapter 13, “Optimizing Your Queries.”
In cases where the active set is entirely generated and saved in a temporary
table, the database server can take quite some time to open the cursor. After-
ward, it can tell the program exactly how many rows the active set contains.
This information is not made available, however. One reason is that you can

5-24 Programming with SQL

 Retrieving Multiple Rows

never be sure which method the optimizer uses. If it can avoid sorts and
temporary tables, it does; but very small changes in the query, in the sizes of
the tables, or in the available indexes can change its methods.

The Active Set for a Sequential Cursor

The database server wants to tie up as few resources as possible in
maintaining the active set of a cursor. If it can do so, the database server never
retains more than the single row that is fetched next. It can do this for most
sequential cursors. On each fetch, it returns the contents of the current row
and locates the next one.

The Active Set for a Scroll Cursor

All the rows in the active set for a scroll cursor must be retained until the
cursor closes because the database server cannot be sure which row the
program will ask for next.
Most frequently, the database server implements the active set of a scroll
cursor as a temporary table. The database server might not fill this table
immediately, however (unless it created a temporary table to process the
query). Usually it creates the temporary table when the cursor is opened.
Then, the first time a row is fetched, the database server copies it into the tem-
porary table and returns it to the program. When a row is fetched for a second
time, it can be taken from the temporary table. This scheme uses the fewest
resources in the event that the program abandons the query before it fetches
all the rows. Rows that are never fetched are not created or saved.

The Active Set and Concurrency

When only one program is using a database, the members of the active set
cannot change. This is the situation in most desktop workstations, and it is
the easiest situation to think about. But some programs must be designed for
use in a multiprogramming system, where two, three, or dozens of different
programs can work on the same tables simultaneously.
When other programs can update the tables while your cursor is open, the
idea of the active set becomes less useful. Your program can see only one row
of data at a time, but all other rows in the table can be changing.
In the case of a simple query, when the database server holds only one row of
the active set, any other row can change. The instant after your program
fetches a row, another program can delete the same row or update it so that
if it is examined again, it is no longer part of the active set.

Programming with SQL 5-25

Retrieving Multiple Rows

When the active set, or part of it, is saved in a temporary table, stale data can
present a problem. That is, the rows in the actual tables, from which the
active-set rows are derived, can change. If they do, some of the active-set
rows no longer reflect the current table contents.
These ideas may seem unsettling at first, but as long as your program only
reads the data, there is no such thing as stale data, or rather, all data is equally
stale. No matter when it is taken, the active set is a snapshot of the data as it
is at one moment in time. A row is different the next day; it does not matter
if it is also different in the next millisecond. To put it another way, no practical
difference exists between changes that occur while the program is running
and changes that are saved and applied the instant the program terminates.
The only time that stale data can cause a problem is when the program
intends to use the input data to modify the same database; for example, when
a banking application must read an account balance, change it, and write it
back. Chapter 6, “Modifying Data Through SQL Programs,” discusses
programs that modify data.

Using a Cursor: A Parts Explosion

When you use a cursor, supplemented by program logic, you can solve prob-
lems that plain SQL cannot solve. One of these is the parts-explosion problem,
sometimes called Bill of Materials processing. At the heart of this problem is
a recursive relationship among objects; one object contains other objects,
which contain yet others.
The problem is usually stated in terms of a manufacturing inventory. A
company makes a variety of parts, for example. Some parts are discrete, but
some are assemblages of other parts.
These relationships are documented in a single table, which might be called
contains.The column contains.parent holds the part numbers of parts that
are assemblages. The column contains.child has the part number of a part
that is a component of the parent. If part #123400 is an assembly of nine parts,
nine rows exist with 123400 in the first column and other part numbers in the
second.

5-26 Programming with SQL

 Retrieving Multiple Rows

CONTAINS

PARENT CHILD
FK NN FK NN
123400 432100
432100 765899

The parts-explosion problem is this: given a part number, produce a list of all
parts that are components of that part. A sketch of one solution, as
implemented in INFORMIX-4GL, is shown in Figure 5-6.

DEFINE part_list ARRAY[200] OF INTEGER

FUNCTION boom (top_part)
DEFINE this_part, child_part INTEGER
DEFINE next_to_do, next_free SMALLINT
DECLARE part_scan CURSOR FOR
SELECT child INTO child_part FROM contains
WHERE parent = this_part

LET next_to_do = 1
LET part_list[next_to_do] = top_part
LET next_free = 2

WHILE next_to_do < next_free

this_part = part_list[next_to_do]
FOREACH part_scan
LET part_list[next_free] = child_part
LET next_free = next_free + 1
END FOREACH
LET next_to_do = next_to_do + 1
END WHILE
RETURN next_free - 1
END FUNCTION

Figure 5-6 A breadth-first algorithm to generate a parts explosion

Technically speaking, each row of the contains table is the head node of a
directed acyclic graph, or tree. The function in Figure 5-6 performs a breadth-
first search of the tree whose root is the part number passed as its parameter.
The function uses a cursor named part_scan to return all the rows with a par-
ticular value in the parent column. This is very easy to implement using the
INFORMIX-4GL statement FOREACH, which opens a cursor, iterates once for
each row in the selection set, and closes the cursor.

Programming with SQL 5-27

Dynamic SQL

This is the heart of the parts-explosion problem, but it is not a complete

solution. For example, the program in Figure 5-6 does not allow for compo-
nents that appear at more than one level in the tree. Furthermore, a practical
contains table would also have a column count, giving the count of child
parts used in each parent. A program that returns a total count of each
component part is much more complicated.
The iterative approach described earlier is not the only way to approach the
parts-explosion problem. If the number of generations has a fixed limit, you
can solve the problem with a single SELECT statement using nested, outer
self-joins.
If there can be up to four generations of parts contained within one top-level
part, the following SELECT statement returns all of them:

SELECT a.parent, a.child, b.child, c.child, d.child

FROM contains a
OUTER (contains b,
OUTER (contains c, outer contains d))
WHERE a.parent = top_part_number
AND a.child = b.parent
AND b.child = c.parent
AND c.child = d.parent

This SELECT statement returns one row for each line of descent rooted in the
part given as top_part_number. Null values are returned for levels that do
not exist. (Use indicator variables to detect them.) You can extend this solu-
tion to more levels by selecting additional nested outer joins of the contains
table.You also can revise this solution to return counts of the number of parts
at each level.

Dynamic SQL
Although static SQL is extremely useful, it requires that you know the exact
content of every SQL statement at the time you write the program. For exam-
ple, you must state exactly which columns are tested in any WHERE clause
and exactly which columns are named in any select list.
This is no problem when you write a program to perform a well-defined task.
But the database tasks of some programs cannot be perfectly defined in
advance. In particular, a program that must respond to an interactive user
might need the ability to compose SQL statements in response to what the
user enters.

5-28 Programming with SQL

 Dynamic SQL

Dynamic SQL allows a program to form an SQL statement during execution,

so that the contents of the statement can be determined by user input. This is
done in the following steps:
1. The program assembles the text of an SQL statement as a character string,
which is stored in a program variable.
2. It executes a PREPARE statement, which asks the database server to
examine the statement text and prepare it for execution.
3. It uses the EXECUTE statement to execute the prepared statement.
In this way, a program can construct and then use any SQL statement, based
on user input of any kind. For example, it can read a file of SQL statements
and prepare and execute each.
DB-Access, the utility that you use to explore SQL interactively, is an
INFORMIX-ESQL/C program that constructs, prepares, and executes SQL
statements dynamically. For example, it enables users to specify the columns
of a table using simple, interactive menus. When the user is finished,
DB-Access builds the necessary CREATE TABLE or ALTER TABLE statement
dynamically and prepares and executes it.

Preparing a Statement
In form, a dynamic SQL statement is like any other SQL statement that is
written into a program, except that it cannot contain the names of any host
variables.
This leads to two restrictions. First, if it is a SELECT statement, it cannot
include the INTO clause. The INTO clause names host variables into which
column data is placed, and host variables are not allowed in a dynamic state-
ment. Second, wherever the name of a host variable normally appears in an
expression, a question mark is written as a placeholder.
You can prepare a statement in this form for execution with the PREPARE
statement. The following example is in INFORMIX-ESQL/C:

exec sql prepare query_2 from

'select * from orders
where customer_num = ? and
order_date > ?';

Notice the two question marks in this example; they indicate that when the
statement is executed, the values of host variables are used at those two
points.

Programming with SQL 5-29

Dynamic SQL

You can prepare almost any SQL statement dynamically. The only ones that
cannot be prepared are the ones directly concerned with dynamic SQL and
cursor management, such as the PREPARE and OPEN statements. After you
prepare an UPDATE or DELETE statement, it is a good idea to test the fifth
field of SQLAWARN to see if you used a WHERE clause (see
“The SQLAWARN Array” on page 5-12).
The result of preparing a statement is a data structure that represents the
statement. This data structure is not the same as the string of characters that
produced it. In the PREPARE statement, you give a name to the data structure;
it is query_2 in the preceding example. This name is used to execute the
prepared SQL statement.
The PREPARE statement does not limit the character string to one statement.
It may contain multiple SQL statements, separated by semicolons. Figure 5-7
shows a fairly complex example in INFORMIX-ESQL/COBOL:

MOVE 'BEGIN WORK;

UPDATE account
SET balance = balance + ?
WHERE acct_number = ?;
UPDATE teller
SET balance = balance + ?
WHERE teller_number = ?;
UPDATE branch
SET balance = balance + ?
WHERE branch_number = ?;
INSERT INTO history VALUES(timestamp, values);

TO BIG-QUERY.

EXEC SQL
PREPARE BIG-Q FROM :BIG-QUERY
END-EXEC.

Figure 5-7 Preparing a string containing five SQL statements

When this list of statements is executed, host variables must provide values
for six place-holding question marks. Although it is more complicated to set
up a multistatement list, the performance is often better because fewer
exchanges take place between the program and the database server.

5-30 Programming with SQL

 Dynamic SQL

Executing Prepared SQL

Once a statement is prepared, it can be executed multiple times. Statements
other than SELECT statements, and SELECT statements that return only a
single row, are executed with the EXECUTE statement.
Figure 5-8 shows how INFORMIX-ESQL/C prepares and executes a
multistatement update of a bank account:

exec sql begin declare section;

char bigquery[270] = 'begin work;';
exec sql end declare section;
stcat ('update account set balance = balance + ? where ', bigquery);
stcat ('acct_number = ?;', bigquery);
stcat ('update teller set balance = balance + ? where ', bigquery);
stcat ('teller_number = ?;', bigquery);
stcat ('update branch set balance = balance + ? where ', bigquery);
stcat ('branch_number = ?;', bigquery);
stcat ('insert into history values(timestamp, values);', bigquery);

exec sql prepare bigq from :bigquery;

exec sql execute bigq using :delta, :acct_number, :delta,

:teller_number, :delta, :branch_number;

exec sql commit work;

Figure 5-8 Preparing and executing a multistatement operation in ESQL/C

The USING clause of the EXECUTE statement supplies a list of host variables
whose values are to take the place of the question marks in the prepared
statement.

Programming with SQL 5-31

Dynamic SQL

Using Prepared SELECT Statements

A dynamically prepared SELECT statement cannot simply be executed; it
might produce more than one row of data, and the database server, not
knowing which row to return, produces an error code.
Instead, a dynamic SELECT statement is attached to a cursor. Then, the cursor
is opened and used in the usual way. The cursor to be used with a prepared
statement is declared for that statement name. The following example is in
INFORMIX-4GL:

LET select_2 = 'select order_num, order_date from orders ',

'where customer_num = ? and order_date > ?'

PREPARE q_orders FROM select_2

DECLARE cu_orders CURSOR FOR q_orders

OPEN cu_orders USING q_c_number, q_o_date

FETCH cu_orders INTO f_o_num, f_o_date

The following list identifies the stages of processing in this example:

1. A character string expressing a SELECT statement is placed in a program
variable. It employs two place-holding question marks.
2. The PREPARE statement converts the string into a data structure that can
be executed. The data structure is associated with a name, q_orders.
3. A cursor named cu_orders is declared and associated with the name of
the prepared statement.
4. When the cursor is opened, the prepared statement is executed. The
USING clause in the OPEN statement provides the names of two host vari-
ables whose contents are substituted for the question marks in the
original statement.
5. The first row of data is fetched from the open cursor. The INTO clause of
the FETCH statement specifies the host variables that are to receive the
fetched column values.
Later, the cursor can be closed and reopened. While the cursor is closed, a dif-
ferent SELECT statement can be prepared under the name q_orders. In this
way, a single cursor can be used to fetch from different SELECT statements.

5-32 Programming with SQL

 Dynamic SQL

Dynamic Host Variables

In SQL APIs, which support dynamically allocated data objects, it is possible
to take dynamic statements one step further. It is possible to dynamically
allocate the host variables that receive column data.
This makes it possible to take an arbitrary SELECT statement from program
input, determine how many values it produces and their data types, and
allocate the host variables of the appropriate types to hold them.
The key to this ability is the DESCRIBE statement. It takes the name of a
prepared SQL statement and returns information about the statement and its
contents. It sets SQLCODE to specify the type of statement, that is, the verb
with which it begins. If the prepared statement is a SELECT, the DESCRIBE
statement also returns information about the selected output data. If the pre-
pared statement is an INSERT, the DESCRIBE statement returns information
about the input parameters. The data structure is a predefined data structure
allocated for this purpose and is known as a system descriptor area. If you are
using INFORMIX-ESQL/C, you can use a system descriptor area or, as an
alternative, an sqlda structure.
The data structure that a DESCRIBE statement returns or references for a
SELECT statement includes an array of structures. Each structure describes
the data that is returned for one item in the select list. The program can exam-
ine the array and discover that a row of data includes a decimal value, a
character value of a certain length, and an integer.
Using this information, the program can allocate memory to hold the
retrieved values and put the necessary pointers in the data structure for the
database server to use.

Freeing Prepared Statements

A prepared SQL statement occupies space in memory. With some database
servers, it can consume space owned by the database server as well as space
belonging to the program. This space is released when the program
terminates, but in general, you should free this space when you finish with it.
You can use the FREE statement to release this space. It takes either the name
of a statement or the name of a cursor that was declared FOR a statement
name, and releases the space allocated to the prepared statement. If more
than one cursor is defined on the statement, freeing the statement does not
free the cursor.

Programming with SQL 5-33

Embedding Data Definition Statements

Quick Execution
For simple statements that do not require a cursor or host variables, you can
combine the actions of the PREPARE, EXECUTE, and FREE statements into a
single operation. The EXECUTE IMMEDIATE statement, shown in the follow-
ing example, takes a character string and in one operation prepares it,
executes it, and frees the storage.

exec sql execute immediate 'drop index my_temp_index';

This makes it easy to write simple SQL operations. However, because no

USING clause is allowed, the EXECUTE IMMEDIATE statement cannot be used
for SELECT statements.

Embedding Data Definition Statements

Data definition statements, the SQL statements that create databases and
modify the definitions of tables, are not usually put into programs. The rea-
son is that they are rarely performed—a database is created once, but is
queried and updated many times.
The creation of a database and its tables is generally done interactively, using
DB-Access or INFORMIX-SQL. These tools can also be driven from a file of
statements, so that the creation of a database can be done with one operating
system command. (This is discussed in Chapter 10, “Tuning Your Data
Model.”)

Embedding Grant and Revoke Privileges

One task related to data definition is done repeatedly: the granting and
revoking of privileges. The reasons for this are discussed in Chapter 11,
“Granting and Limiting Access to Your Database.” Because privileges must
be granted and revoked frequently, and because this might be done by per-
sons who are not skilled in SQL, it can be useful to package the GRANT and
REVOKE statements in programs to give them a simpler, more convenient
user interface.

5-34 Programming with SQL

 Embedding Data Definition Statements

The GRANT and REVOKE statements are especially good candidates for
dynamic SQL. Each statement takes the following parameters:
• A list of one or more privileges
• A table name
• The name of a user
You probably need to supply at least some of these values based on program
input (from the user, command-line parameters, or a file) but none can be
supplied in the form of a host variable. The syntax of these statements does
not allow host variables at any point.
The only alternative is to assemble the parts of a statement into a character
string, and prepare and execute the assembled statement. Program input can
be incorporated into the prepared statement as characters.
Figure 5-9 shows a function in INFORMIX-4GL that assembles a GRANT
statement from the function parameters, and then prepares and executes it.

FUNCTION table_grant (priv_to_grant, table_name, user_id)

DEFINE priv_to_grant char(100),
table_name char(20),
user_id char(20),
grant_stmt char(200)
LET grant_stmt =' GRANT ', priv_to_grant,
' ON ', table_name,
' TO ', user_id
WHENEVER ERROR CONTINUE
PREPARE the_grant FROM grant_stmt
IF status = 0 THEN
EXECUTE the_grant
END IF
IF status <> 0 THEN
DISPLAY 'Sorry, got error #', status, 'attempting:'
DISPLAY ' ',grant_stmt
END IF
FREE the_grant
WHENEVER ERROR STOP
END FUNCTION

Figure 5-9 A 4GL function that builds, prepares, and executes a GRANT statement

The opening statement, shown in the following example, defines the name of
the function and the names of its three parameters:

FUNCTION table_grant (priv_to_grant, table_name, user_id)

Programming with SQL 5-35

Embedding Data Definition Statements

The DEFINE statement, shown in the following example, defines the

parameters and one additional variable local to the function. All four are
character strings of various lengths.

DEFINE priv_to_grant char(100),

table_name char(20),
user_id char(20),
grant_stmt char(200)

The variable grant_stmt holds the assembled GRANT statement, which is

created by concatenating the parameters and some constants. This is shown
in the following example:

LET grant_stmt ='GRANT ', priv_to_grant,

' ON ', table_name,
' TO ', user_id

In INFORMIX-4GL, the comma is used to concatenate strings. This

assignment statement concatenates the following six character strings:
• 'GRANT'
• The parameter specifying the privileges to be granted
• 'ON'
• The parameter specifying the table name
• 'TO'
• The parameter specifying the user.
The result is a complete GRANT statement composed partly of program
input. The same feat can be accomplished in other host languages using
different syntax, as shown in the following figure:

WHENEVER ERROR CONTINUE

PREPARE the_grant FROM grant_stmt

If the database server returns an error code in SQLCODE, the default action of
an INFORMIX-4GL program is to terminate. However, errors are quite likely
when you prepare an SQL statement composed of user-supplied parts, and
program termination is a poor way to diagnose the error. In the preceding
code, the WHENEVER statement prevents termination. Then the PREPARE
statement passes the assembled statement text to the database server for
parsing.

5-36 Programming with SQL

 Summary

If the database server approves the form of the statement, it sets a zero return
code. This does not guarantee that the statement is executed properly; it
means only that the statement has correct syntax. It might refer to a nonexist-
ent table or contain many other kinds of errors that can only be detected
during execution.

IF status = 0 THEN
EXECUTE the_grant
END IF

If the preparation is successful, the next step is to execute the prepared

statement. The function in Figure 5-9 displays an error message if anything
goes wrong. As written, it makes no distinction between an error from the
PREPARE operation and one from the EXECUTE operation. It does not attempt
to interpret the numeric error code, leaving it to the user to interpret.

Summary
SQL statements can be written into programs as if they were normal
statements of the programming language. Program variables can be used in
WHERE clauses, and data from the database can be fetched into them. A
preprocessor translates the SQL code into procedure calls and data structures.
Statements that do not return data, or queries that return only one row of
data, are written like ordinary imperative statements of the language. Que-
ries that can return more than one row are associated with a cursor that rep-
resents the current row of data. Through the cursor, the program can fetch
each row of data as it is needed.
Static SQL statements are written into the text of the program. However, the
program can form new SQL statements dynamically, as it runs, and execute
them also. In the most advanced cases, the program can obtain information
about the number and types of columns that a query returns and dynamically
allocate the memory space to hold them.

Programming with SQL 5-37

Summary

5-38 Programming with SQL

 Chapter

Modifying Data
Through SQL
Programs
Chapter Overview 3
Using DELETE 3
Direct Deletions 4
Errors During Direct Deletions 4
Using Transaction Logging 5
Coordinated Deletions 6
Deleting with a Cursor 7
Using INSERT 8
Using an Insert Cursor 8
Declaring an Insert Cursor 9
Inserting with a Cursor 10
Status Codes After PUT and FLUSH 11
Rows of Constants 11
6
An Insert Example 12
Using UPDATE 14
Using an Update Cursor 15
The Purpose of the Keyword UPDATE 15
Updating Specific Columns 16
UPDATE Keyword Not Always Needed 16
Cleaning up a Table 16
Summary 17
6-2 Modifying Data Through SQL Programs
Chapter Overview
The preceding chapter introduced the idea of putting SQL statements,
especially the SELECT statement, into programs written in other languages.
This enables a program to retrieve rows of data from a database.
This chapter covers the issues that arise when a program needs to modify the
database by deleting, inserting, or updating rows. As in Chapter 5, “Pro-
gramming with SQL,” this chapter aims to prepare you for reading the
manual for the Informix ESQL or 4GL product you are using.
The general use of the INSERT, UPDATE, and DELETE statements is covered in
Chapter 4, “Modifying Data.” This chapter examines their use from within a
program. It is quite easy to put the statements in a program, but it can be
quite difficult to handle errors and to deal with concurrent modifications
from multiple programs.

Using DELETE
A program deletes rows from a table by executing a DELETE statement. The
DELETE statement can specify rows in the usual way with a WHERE clause,
or it can refer to a single row, the last one fetched through a specified cursor.
Whenever you delete rows, you must consider whether rows in other tables
depend on the deleted rows. This problem of coordinated deletions is cov-
ered in Chapter 4, “Modifying Data.” The problem is the same when
deletions are made from within a program.

Modifying Data Through SQL Programs 6-3

Using DELETE

Direct Deletions
You can embed a DELETE statement in a program. The following example
uses INFORMIX-ESQL/C:

exec sql delete from items

where order_num = :onum;

You can also prepare and execute a statement of the same form dynamically.
In either case, the statement works directly on the database to affect one or
more rows.
The WHERE clause in the example uses the value of a host variable named
onum. Following the operation, results are posted in the SQLSTATE and in
SQLCA, as usual. The third element of the SQLERRD array contains the count
of rows deleted even if an error occurs. The value in SQLCODE shows the
overall success of the operation. If it is not negative, no errors occurred and
the third element of SQLERRD is the count of all rows that satisfied the
WHERE clause and were deleted.

Errors During Direct Deletions

When an error occurs, the statement ends prematurely. The values in
SQLSTATE and in SQLCODE and the second element of SQLERRD explain its
cause, and the count of rows reveals how many rows were deleted. For many
errors, that count is zero because they prevented the database server from
beginning the operation at all. For example, if the named table does not exist,
or if a column tested in the WHERE clause is renamed, no deletions are
attempted.
However, certain errors can be discovered after the operation begins and
some rows are processed. The most common of these errors is a lock conflict.
The database server must obtain an exclusive lock on a row before it can
delete that row. Other programs might be using the rows from the table, pre-
venting the database server from locking a row. Because the issue of locking
affects all types of modifications, it is discussed Chapter 7, “Programming for
a Multiuser Environment.”
Other, rarer types of error can strike after deletions begin, for example,
hardware errors that occur while the database is being updated.

6-4 Modifying Data Through SQL Programs

 Using DELETE

Using Transaction Logging

The best way to prepare for any kind of error during a modification is to use
transaction logging. In the event of an error of any kind, you can tell the data-
base server to put the database back the way it was. The following example
is based on the preceding INFORMIX-ESQL/C example, which is extended to
use transactions:

exec sql begin work; /* start the transaction*/

exec sql delete from items
where order_num = :onum;
del_result = sqlca.sqlcode; /* save two error */
del_isamno = sqlca.sqlerrd[1]; /* ...code numbers */
del_rowcnt = sqlca.sqlerrd[2]; /* ...and count of rows */
if (del_result < 0) /* some problem, */
exec sql rollback work; /* ...put everything back */
else /* everything worked OK, */
exec sql commit work; /* ...finish transaction */

An important point in this example is that the program saves the important
return values in the SQLCA before it ends the transaction. The reason is that
both the ROLLBACK WORK and COMMIT WORK statements, as with all SQL
statements, set return codes in the SQLCA. Executing a ROLLBACK WORK
statement after an error wipes out the error code; unless it was saved, it
cannot be reported to the user.
The advantage of using transactions is that no matter what goes wrong, the
database is left in a known, predictable state. No question remains about how
much of the modification is completed; either all of it is or none of it is.

Modifying Data Through SQL Programs 6-5

Using DELETE

Coordinated Deletions
The usefulness of transaction logging is particularly clear when you must
modify more than one table. For example, consider the problem of deleting
an order from the demonstration database. In the simplest form of the prob-
lem, you must delete rows from two tables, orders and items, as shown in the
INFORMIX-4GL example in Figure 6-1.

WHENEVER ERROR CONTINUE{do not terminate on error}

BEGIN WORK {start transaction}
DELETE FROM items
WHERE order_num = o_num
IF (status >= 0) THEN{no error on first delete}
DELETE FROM orders
WHERE order_num = o_num
END IF
IF (status >= 0) THEN{no error on either delete}
COMMIT WORK
ELSE {problem on some delete}
DISPLAY 'Error ', status,' deleting.'
ROLLBACK WORK
END IF

Figure 6-1 A fragment of 4GL that deletes from two tables

The logic of this program is much the same regardless if transactions are
used. If they are not used, the person who sees the error message has a much
more difficult set of decisions to make. Depending on when the error
occurred one of the following situations applies:
• No deletions were performed; all rows with this order number remain in
the database.
• Some, but not all, item rows were deleted; an order record with only some
items remains.
• All item rows were deleted, but the order row remains.
• All rows were deleted.
In the second and third cases, the database is corrupted to some extent; it
contains partial information that can cause some queries to produce wrong
answers. You must take careful action to restore consistency to the
information. When transactions are used, all these uncertainties are
prevented.

6-6 Modifying Data Through SQL Programs

 Using DELETE

Deleting with a Cursor

You can also write a DELETE statement through a cursor to delete the row that
was last fetched. This lets you program deletions based on conditions that
cannot be tested in a WHERE clause, as shown in Figure 6-2.

int delDupOrder()
{
int ord_num;
int dup_cnt, ret_code;

exec sql declare scan_ord cursor for

select order_num, order_date
into :ord_num, :ord_date
from orders for update;
exec sql open scan_ord;
if (sqlca.sqlcode != 0)
return (sqlca.sqlcode);
exec sql begin work;
for(;;)
{
exec sql fetch next scan_ord;
if (sqlca.sqlcode != 0) break;
dup_cnt = 0; /* default in case of error */
exec sql select count(*) into dup_cnt from orders
where order_num = :ord_num;
if (dup_cnt > 1)
{
exec sql delete where current of scan_ord;
if (sqlca.sqlcode != 0)
break;
}
}
ret_code = sqlca.sqlcode;
if (ret_code == 100)/* merely end of data */
exec sql commit work;
else /* error on fetch or on delete */
exec sql rollback work;
return (ret_code);
}

Figure 6-2 An unsafe ESQL/C function that deletes through a cursor (see Warning)

Warning: The design of the ESQL/C function in Figure 6-2 is unsafe. It depends for
! correct operation on the current “isolation level,” which is discussed later in the
chapter. Even when it works as intended, its effects depend on the physical order of
rows in the table, which is not generally a good idea.

Modifying Data Through SQL Programs 6-7

Using INSERT

The purpose of the function is to delete rows containing duplicate order

numbers. In fact, in the demonstration database, the orders.order_num col-
umn has a unique index, so duplicate rows cannot occur in it. However, a
similar function can be written for another database; this one uses familiar
column names.
The function declares scan_ord, a cursor to scan all rows in the orders table.
It is declared with the FOR UPDATE clause, which states that the cursor can
modify data. If the cursor opens properly, the function begins a transaction
and then loops over rows of the table. For each row, it uses an embedded
SELECT statement to determine how many rows of the table have the order
number of the current row. (This step fails without the correct isolation level,
as described in Chapter 7, “Programming for a Multiuser Environment.”)
In the demonstration database, with its unique index on this table, the count
returned to dup_cnt is always 1. However, if it is greater, the function deletes
the current row of the table, reducing the count of duplicates by one.
Clean-up functions of this sort are sometimes needed, but they generally
need more sophisticated design. This one deletes all duplicate rows except
the last one delivered by the database server. That ordering has nothing to do
with the contents of the rows or their meanings. You can improve the func-
tion in Figure 6-2 by adding, perhaps, an ORDER BY clause to the cursor dec-
laration. However, you cannot use ORDER BY and FOR UPDATE together. A
better approach is sketched later in “An Insert Example” on page 6-12.

Using INSERT
You can embed the INSERT statement in programs. Its form and use in a
program are the same as described in Chapter 4, “Modifying Data,” with the
additional feature that you can use host variables in expressions, both in the
VALUES and WHERE clauses. Moreover, a program has the additional ability
to insert rows using a cursor.

Using an Insert Cursor

The DECLARE CURSOR statement has many variations. Most are used to
create cursors for different kinds of scans over data, but one variation creates
a special kind of cursor called an insert cursor. You use an insert cursor with
the PUT and FLUSH statements to efficiently insert rows into a table in bulk.

6-8 Modifying Data Through SQL Programs

 Using INSERT

Declaring an Insert Cursor

You create an insert cursor by declaring a cursor to be FOR an INSERT
statement instead of a SELECT statement. You cannot use such a cursor to
fetch rows of data; you can use it only to insert them. An example of the
declaration of an insert cursor is shown in Figure 6-3.

DEFINE the_company LIKE customer.company,

the_fname LIKE customer.fname,
the_lname LIKE customer.lname
DECLARE new_custs CURSOR FOR
INSERT INTO customer (company, fname, lname)
VALUES (the_company, the_fname, the_lname)

Figure 6-3 A 4GL fragment that declares an insert cursor

When you open an insert cursor, a buffer is created in memory to hold a block
of rows. The buffer receives rows of data as the program produces them; then
they are passed to the database server in a block when the buffer is full. This
reduces the amount of communication between the program and the data-
base server, and it lets the database server insert the rows with less difficulty.
As a result, the insertions go faster.
The minimum size of the insert buffer is set for any implementation of
embedded SQL; you have no control over it (it is typically 1 or 2 kilobytes).
The buffer is always made large enough to hold at least two rows of inserted
values. It is large enough to hold more than two rows when the rows are
shorter than the minimum buffer size.

Modifying Data Through SQL Programs 6-9

Using INSERT

Inserting with a Cursor

The code in Figure 6-3 prepares an insert cursor for use. The continuation,
shown in Figure 6-4, shows how the cursor can be used. For simplicity, this
example assumes that a function named next_cust returns either information
about a new customer or null data to signal the end of input.

WHENEVER ERROR CONTINUE {do not terminate on error}

BEGIN WORK
OPEN new_custs
WHILE status = 0
CALL next_cust() RETURNING the_company, the_fname, the_lname
IF the_company IS NULL THEN
EXIT WHILE
END IF
PUT new_custs
END WHILE
IF status = 0 THEN {no problem in a PUT}
FLUSH new_custs {write any last rows}
END IF
IF status = 0 THEN {no problem writing}
COMMIT WORK {..make it permanent}
ELSE
ROLLBACK WORK {retract any changes}
END IF

Figure 6-4 Continuing Figure 6-3, the 4GL code that uses the insert cursor

The code in Figure 6-4 calls next_cust repeatedly. When it returns non-null
data, the PUT statement sends the returned data to the row buffer. When the
buffer fills, the rows it contains are automatically sent to the database server.
The loop normally ends when next_cust has no more data to return. Then the
FLUSH statement writes any rows that remain in the buffer, after which the
transaction terminates.
Examine the INSERT statement in Figure 6-3 once more. The statement by
itself, not part of a cursor definition, inserts a single row into the customer
table. In fact, the whole apparatus of the insert cursor can be dropped from
the example code, and the INSERT statement can be written into Figure 6-4
where the PUT statement now stands. The difference is that an insert cursor
causes a program to run somewhat faster.

6-10 Modifying Data Through SQL Programs

 Using INSERT

Status Codes After PUT and FLUSH

When a program executes a PUT statement, the program should test whether
the row is placed in the buffer successfully. If the new row fits in the buffer,
the only action of PUT is to copy the row to the buffer. No errors can occur in
this case. However, if the row does not fit, the entire buffer load is passed to
the database server for insertion, and an error can occur.
The values returned into the SQL Communications Area give the program the
information it needs to sort out each case. SQLCODE is set after every PUT
statement—to zero if no error occurs and to a negative error code if an error
occurs.
The third element of SQLERRD is set to the number of rows actually inserted
into the table: It is set to zero if the new row is merely moved to the buffer; to
the count of rows that are in the buffer, if the buffer load is inserted without
error; or to the count of rows inserted before an error occurs, if one does
occur.
Read the code (Figure 6-4) once again to see how SQLCODE is used. First, if
the OPEN statement yields an error, the loop is not executed (because the
WHILE condition fails), the FLUSH operation is not performed, and the
transaction rolls back.
Second, if the PUT statement returns an error, the loop ends (because of the
WHILE condition), the FLUSH operation is not performed, and the transaction
rolls back. This can occur only if the loop generates enough rows to fill the
buffer at least once; otherwise, the PUT statement cannot generate an error.
The program might end the loop with rows still in the buffer, possibly
without inserting any rows. At this point, the SQL status is zero, and the
FLUSH operation occurs. If the FLUSH operation produces an error code, the
transaction rolls back. Only when all inserts are successfully performed is the
transaction committed.

Rows of Constants
The insert cursor mechanism supports one special case where high
performance is easy to obtain. This is the case in which all of the values listed
in the INSERT statement are constants—no expressions and no host variables,
just literal numbers and strings of characters. No matter how many times
such an INSERT operation occurs, the rows it produces are identical. In that
case, there is no point in copying, buffering, and transmitting each identical
row.

Modifying Data Through SQL Programs 6-11

Using INSERT

Instead, for this kind of INSERT operation, the PUT statement does nothing
except to increment a counter. When a FLUSH operation is finally performed,
a single copy of the row, and the count of inserts, is passed to the database
server. The database server creates and inserts that many rows in one opera-
tion.
It is not common to insert a quantity of identical rows. You can do it when
you first establish a database, to populate a large table with null data.

An Insert Example
“Deleting with a Cursor” on page 6-7, contains an example of the DELETE
statement whose purpose is to look for and delete duplicate rows of a table.
A better way to do the same thing is to select the desired rows instead of
deleting the undesired ones. The code in Figure 6-5 shows one way to do this.
The example is written in INFORMIX-4GL to take advantage of some features
that make SQL programming easy.

6-12 Modifying Data Through SQL Programs

 Using INSERT

BEGIN WORK
INSERT INTO new_orders
SELECT * FROM ORDERS main
WHERE 1 = (SELECT COUNT(*) FROM ORDERS minor
WHERE main.order_num = minor.order_num)
COMMIT WORK

DEFINE ord_row RECORD LIKE orders,

last_ord LIKE orders.order_num
DECLARE dup_row CURSOR FOR
SELECT * FROM ORDERS main INTO ord_row.*
WHERE 1 < (SELECT COUNT(*) FROM ORDERS minor
WHERE main.order_num = minor.order_num)
ORDER BY order_date
DECLARE ins_row CURSOR FOR
INSERT INTO new_orders VALUES (ord_row.*)

BEGIN WORK
OPEN ins_row
LET last_ord = -1
FOREACH dup_row
IF ord_row.order_num <> last_ord THEN
PUT ins_row
LET last_ord = ord_row.order_num
END IF
END FOREACH
CLOSE ins_row
COMMIT WORK

Figure 6-5 A 4GL program that re-creates a table without duplicates

This example begins with an ordinary INSERT statement that finds all the
nonduplicated rows of the table and inserts them into another table, presum-
ably created before the program started. That leaves only the duplicate rows.
(Remember, in the demonstration database the orders table has a unique
index and cannot have duplicate rows. This example deals with some other
database.)
In INFORMIX-4GL, you can define a data structure like a table; the structure is
automatically given one element for each column in the table. The ord_row
structure is a buffer to hold one row of the table.
The code in Figure 6-5 then declares two cursors. The first, called dup_row,
returns the duplicate rows in the table. Because it is for input only, it can use
the ORDER BY clause to impose some order on the duplicates other than the
physical record order used in Figure 6-2 on page 6-7. In this example, the
duplicate rows are ordered by their dates (the oldest one remains), but you
can use any other order based on the data.

Modifying Data Through SQL Programs 6-13

Using UPDATE

The second cursor is an insert cursor. It is written to take advantage of the

asterisk notation of INFORMIX-4GL; you can supply values for all columns
simply by naming a record with an asterisk to indicate all fields.
The remainder of the code examines the rows returned through dup_row. It
inserts the first one from each group of duplicates into the new table and
disregards the rest.
This example uses the simplest kind of error handling. Unless told otherwise,
an INFORMIX-4GL program automatically terminates when an error code is
set in SQLCODE. In this event, the active transaction rolls back. This program
relies on that behavior; it assumes that if it reaches the end, no errors exist and
the transaction can be committed. This kind of error handling is acceptable
when errors are unlikely, and the program is used by people who do not need
to know why the program terminates.

Using UPDATE
You can embed the UPDATE statement in a program in any of the forms
described in Chapter 4, “Modifying Data,” with the additional feature that
you can name host variables in expressions, both in the SET and WHERE
clauses. Moreover, a program can update the row addressed by a cursor.

How Many Rows Were Affected? SQLCODE and SQLERRD

When your program uses a cursor to select rows, it can test SQLCODE for
100, the end-of-data return code. This is set to indicate that no rows, or no
more rows, satisfy the query conditions. The end-of-data return code is set
in SQLCODE only following SELECT statements; it is not used following
DELETE, INSERT, or UPDATE statements.

A query that finds no data is not a success. However, an UPDATE or DELETE

statement that happens to update or delete no rows is still considered a
success: it updated or deleted the set of rows that its WHERE clause said it
should; however, the set was empty.
In the same way, the INSERT statement does not set the end-of-data return
code even when the source of the inserted rows is a SELECT statement and
it selected no rows. The INSERT statement is a success because it inserted as
many rows as it was asked to do (that is, zero).
To find out how many rows are inserted, updated, or deleted, a program can
test the third element of SQLERRD. The count of rows is there, regardless of
the value (zero or negative) in SQLCODE.

6-14 Modifying Data Through SQL Programs

 Using UPDATE

Using an Update Cursor

An update cursor permits you to delete or update the current row; that is, the
most recently fetched row. The following example (in
INFORMIX-ESQL/COBOL) show the declaration of an update cursor:

EXEC SQL
DECLARE names CURSOR FOR
SELECT fname, lname, company
FROM customer
FOR UPDATE
END-EXEC

The program that uses this cursor can fetch rows in the usual way.

EXEC SQL
FETCH names INTO :FNAME, :LNAME, :COMPANY
END-EXEC.

If the program then decides that the row needs to be changed, it can do so.

IF COMPANY IS EQUAL TO 'SONY'

EXEC SQL
UPDATE customer
SET fname = 'Midori', lname = 'Tokugawa'
WHERE CURRENT OF names
END-EXEC.

The words CURRENT OF names take the place of the usual test expressions
in the WHERE clause. In other respects, the UPDATE statement is the same as
usual—even including the specification of the table name, which is implicit
in the cursor name but still required.

The Purpose of the Keyword UPDATE

The purpose of the keyword UPDATE in a cursor is to let the database server
know that the program can update (or delete) any row that it fetches. The
database server places a more demanding lock on rows that are fetched
through an update cursor and a less demanding lock when fetching a row for
a cursor that is not declared with that keyword. This results in better perfor-
mance for ordinary cursors and a higher level of concurrent use in a
multiprocessing system. (Levels of locks and concurrent use are discussed in
Chapter 7, “Programming for a Multiuser Environment.”)

Modifying Data Through SQL Programs 6-15

Using UPDATE

Updating Specific Columns

You can declare a cursor to update specific columns. Here is a revision of the
preceding ESQL/COBOL example has been modified to use this feature and is
shown in the following example:

EXEC SQL
DECLARE names CURSOR FOR
SELECT fname, lname, company, phone
INTO :FNAME,:LNAME,:COMPANY,:PHONE FROM customer
FOR UPDATE OF fname, lname
END-EXEC.

Only the fname and lname columns can be updated through this cursor. A
statement such as the following one is rejected as an error:

EXEC SQL
UPDATE customer
SET company = 'Siemens'
WHERE CURRENT OF names
END-EXEC.

If the program attempts such an update, an error code is returned and no

update takes place. An attempt to delete using WHERE CURRENT OF is also
rejected, because deletion affects all columns.

UPDATE Keyword Not Always Needed

The ANSI standard for SQL does not provide for the FOR UPDATE clause in a
cursor definition. When a program uses an ANSI-compliant database, it can
update or delete using any cursor.

Cleaning up a Table
A final, hypothetical example of using an update cursor presents a problem
that should never arise with an established database but could arise in the
initial design phases of an application.
A large table, named target, for example, is created and populated. A
character column, datcol, inadvertently acquires some null values. These
rows should be deleted. Furthermore, a new column, serials, is added to the

6-16 Modifying Data Through SQL Programs

 Summary

table with the ALTER TABLE statement. This column is to have unique integer
values installed. Figure 6-6 shows the INFORMIX-ESQL/C code needed to
accomplish these things.

exec sql begin declare section;

char dcol[80];
short dcolint;
int sequence;
exec sql end declare section;

exec sql declare target_row cursor for

select datcol
into :dcol:dcolint
from target
for update of serials;
exec sql begin work;
exec sql open target_row;
if (sqlca.sqlcode == 0) exec sql fetch next target_row;
for(sequence = 1; sqlca.sqlcode == 0; ++sequence)
{
if (dcolint < 0) /* null datcol */
exec sql delete where current of target_row;
else
exec sql update target set serials = :sequence
where current of target_row;
}
if (sqlca.sqlcode >= 0)exec sql commit work;
else exec sql rollback work;

Figure 6-6 Cleaning up a table using an update cursor

Summary
A program can execute the INSERT, DELETE, and UPDATE statements as
described in Chapter 4, “Modifying Data.” A program also can scan through
a table with a cursor, updating or deleting selected rows. It also can use a cur-
sor to insert rows, with the benefit that the rows are buffered and sent to the
database server in blocks.
In all these activities, the program must take pains to detect errors and return
the database to a known state when one occurs. The most important tool for
doing this is the transaction. Without transaction logging, the program has a
more difficult time recovering from errors.

Modifying Data Through SQL Programs 6-17

Summary

6-18 Modifying Data Through SQL Programs

 Chapter

Programming for a
Multiuser
Environment
Chapter Overview 3
Concurrency and Performance 3
Locking and Integrity 3
Locking and Performance 3
Concurrency Issues 4
How Locks Work 5
Kinds of Locks 6
Lock Scope 6
Database Locks 6
Table Locks 7
Page, Row, and Key Locks 8
The Duration of a Lock 8
Locks While Modifying 9
Setting the Isolation Level 9
Dirty Read Isolation 10
Committed Read Isolation 10
7
Cursor Stability Isolation 11
Repeatable Read Isolation 12
 Setting the Lock Mode 13
Waiting for Locks 13
Not Waiting for Locks 13
Waiting a Limited Time 14
Handling a Deadlock 14
Handling External Deadlock 14
Simple Concurrency 15
Locking with Other Database Servers 15
Isolation While Reading 16
Locking Updated Rows 17
Hold Cursors 17
Summary 19

7-2 Programming for a Multiuser Environment

Chapter Overview
If your database is contained in a single-user workstation and is not
connected on a network to other machines, your programs can modify data
freely. But in all other cases, you must allow for the possibility that, while
your program is modifying data, another program is reading or modifying
the same data. This is concurrency: two or more independent uses of the same
data at the same time. This chapter addresses concurrency, locking, and
isolation levels.

Concurrency and Performance

Concurrency is crucial to good performance in a multiprogramming system.
When access to the data is serialized so that only one program at a time can
use it, processing slows dramatically.

Locking and Integrity

Unless controls are placed on the use of data, concurrency can lead to a
variety of negative effects. Programs can read obsolete data or modifications
can be lost even though they were apparently completed.
The database server prevents errors of this kind by imposing a system of
locks. A lock is a claim, or reservation, that a program can place on a piece of
data. The database server guarantees that, as long as the data is locked, no
other program can modify it. When another program requests the data, the
database server either makes the program wait or turns it back with an error.

Locking and Performance

Because a lock serializes access to one piece of data, it reduces concurrency;
any other programs that want access to that data must wait. The database
server can place a lock on a single row, a disk page (which holds multiple

Programming for a Multiuser Environment 7-3

Concurrency Issues

rows), a whole table, or an entire database. The more locks it places and the
larger the objects it locks, the more concurrency is reduced. The fewer the
locks and the smaller the objects, the greater concurrency and performance
can be.
This section discusses how a program can achieve two goals:
• To place all the locks needed to ensure data integrity
• To lock the fewest, smallest pieces of data possible consistent with the
preceding goal

Concurrency Issues
To understand the hazards of concurrency, you must think in terms of
multiple programs, each executing at its own speed. The following example
suppose that your program is fetching rows through the following cursor:

DECLARE sto_curse CURSOR FOR

SELECT * FROM stock
WHERE manu_code = 'ANZ'

The transfer of each row from the database server to the program takes time.
During and between transfers, other programs can perform other database
operations. Say that at about the same time your program fetches the rows
produced by that query, another user’s program executes the following
update:

UPDATE stock
SET unit_price = 1.15 * unit_price
WHERE manu_code = 'ANZ'

In other words, both programs are reading through the same table, one
fetching certain rows and the other changing the same rows. The following
possibilities concerning what happens next:
1. The other program finishes its update before your program fetches its first
row.
Your program shows you only updated rows.
2. Your program fetches every row before the other program has a chance to
update it.
Your program shows you only original rows.

7-4 Programming for a Multiuser Environment

 How Locks Work

3. After your program fetches some original rows, the other program
catches up and goes on to update some rows that your program has yet
to read; then it executes the COMMIT WORK statement.
Your program might return a mixture of original rows and updated rows.
4. Same as number 3, except that after updating the table, the other program
issues a ROLLBACK WORK statement.
Your program can show you a mixture of original rows and updated rows
that no longer exist in the database.
The first two possibilities are harmless. In number 1, the update is complete
before your query begins. It makes no difference whether it finished a
microsecond or a week ago.
In number 2, your query is, in effect, complete before the update begins. The
other program might have been working just one row behind yours, or it
might not start until tomorrow night; it does not matter.
The second pair of chances, however, can be very important to the design of
some applications. In number 3, the query returns a mix of updated and orig-
inal data. That can be a negative thing in some applications. In others, for
instance one that is taking an average of all prices, it might not matter at all.
In number 4, it can be disastrous if a program returns some rows of data that,
because their transaction was cancelled, can no longer be found in the table.
Another concern arises when your program uses a cursor to update or delete
the last-fetched row. Erroneous results occur with the following sequence of
events:
• Your program fetches the row.
• Another program updates or deletes the row.
• Your program updates or deletes WHERE CURRENT OF names.
You control concurrent events such as these using the locking and isolation
level features of the database server.

How Locks Work

The INFORMIX-OnLine Dynamic Server database server supports a
complex, flexible set of locking features that is described in this section. Some
of these locking features work differently on an INFORMIX-SE database
server. (See Chapter 1 of the Informix Guide to SQL: Reference for a summary
of the differences between locking in SE and OnLine database servers.)

Programming for a Multiuser Environment 7-5

How Locks Work

Kinds of Locks
INFORMIX-OnLine Dynamic Server supports the following three kinds of
locks, which it uses in different situations:
shared A shared lock reserves its object for reading only. It prevents
the object from changing while the lock remains. More than
one program can place a shared lock on the same object.
exclusive An exclusive lock reserves its object for the use of a single
program. It is used when the program intends to change the
object.
An exclusive lock cannot be placed where any other kind of
lock exists. Once one has been placed, no other lock can be
placed on the same object.
promotable A promotable lock establishes the intent to update. It can
only be placed where no other promotable or exclusive lock
exists. Promotable locks can exist with shared locks. They
are “promoted” to an exclusive lock.

Lock Scope
You can apply locks to entire databases, entire tables, disk pages, single rows,
or index-key values. The size of the object being locked is referred to as the
scope of the lock (also called the lock granularity). In general, the larger the
scope of a lock, the more concurrency is reduced but the simpler
programming becomes.

Database Locks
You can lock an entire database. The act of opening a database places a shared
lock on the name of the database. A database is opened with the CONNECT,
DATABASE, or CREATE DATABASE statements. As long as a program has a
database open, the shared lock on the name prevents any other program from
dropping the database or putting an exclusive lock on it.
You can lock an entire database exclusively with the following statement:

DATABASE database name EXCLUSIVE

This statement succeeds if no other program has opened that database. Once
the lock is placed, no other program can open the database, even for reading
because its attempt to place a shared lock on the database name fails.

7-6 Programming for a Multiuser Environment

 How Locks Work

A database lock is released only when the database closes. That can be done
explicitly with the DISCONNECT or CLOSE DATABASE statements or
implicitly by executing another DATABASE statement.
Because locking a database reduces concurrency in that database to zero, it
makes programming very simple—concurrent effects cannot happen. How-
ever, you should lock a database only when no other programs need access.
Database locking is often used before applying massive changes to data
during off-peak hours.

Table Locks
You can lock entire tables. In some cases, this is done automatically.
INFORMIX-OnLine Dynamic Server always locks an entire table while it
performs any of the following statements:
• ALTER INDEX
• ALTER TABLE
• CREATE INDEX
• DROP INDEX
• RENAME COLUMN
• RENAME TABLE
The completion of the statement (or end of the transaction) releases the lock.
An entire table also can be locked automatically during certain queries.
You can use the LOCK TABLE statement to lock an entire table explicitly. This
statement allows you to place either a shared lock or an exclusive lock on an
entire table.
A shared table lock prevents any concurrent updating of that table while
your program is reading from it. INFORMIX-OnLine Dynamic Server
achieves the same degree of protection by setting the isolation level, as
described in the next section; this allows greater concurrency. However, all
Informix database servers support the LOCK TABLE statement.
An exclusive table lock prevents any concurrent use of the table. This has a
serious effect on performance if many other programs are contending for the
use of the table. As with an exclusive database lock, an exclusive table lock is
often used when massive updates are applied during off-peak hours. For
example, some applications do not update tables during the hours of peak
use. Instead, they write updates to an update journal. During off-peak hours,
that journal is read and all updates are applied in a batch.

Programming for a Multiuser Environment 7-7

How Locks Work

Page, Row, and Key Locks

One row of a table is the smallest object that can be locked. A program can
lock one row or a selection of rows while other programs continue to work
on other rows of the same table.
INFORMIX-OnLine Dynamic Server stores data in units called disk pages. (Its
disk-storage methods are described in detail in Chapter 10, “Tuning Your
Data Model.”) A disk page contains one or more rows. In some cases, it is
better to lock a disk page than to lock individual rows on it.
You choose between locking by rows or locking by pages when you create the
table. INFORMIX-OnLine Dynamic Server supports a clause, LOCK MODE, to
specify either page or row locking. You can specify lock mode in the CREATE
TABLE statement and later change it with the ALTER TABLE statement. (Other
Informix database servers do not offer the choice; they lock by row or by
page, whichever makes the better implementation.)
Page and row locking are used identically. Whenever INFORMIX-OnLine
Dynamic Server needs to lock a row, it locks either the row itself or the page
it is on, depending on the lock mode established for the table.
In certain cases, the database server has to lock a row that does not exist—in
effect, it locks the place in the table where the row would be if it did exist. The
database server does this by placing a lock on an index-key value. Key locks
are used identically to row locks. When the table uses row locking, key locks
are implemented as locks on imaginary rows. When it uses page locking, a
key lock is placed on the index page that contains the key or that would
contain the key if the it existed.

The Duration of a Lock

The program controls the duration of a database lock. A database lock is
released when the database closes.
The duration of a table lock depends on whether the database uses
transactions. If it does not (that is, if no transaction log exists and the COMMIT
WORK statement is not used), a table lock remains until it is removed by the
execution of the UNLOCK TABLE statement.
The duration of table, row, and index locks depends on what SQL statements
are used and on whether transactions are in use.
When transactions are used, the end of a transaction releases all table, row,
page, and index locks. This is an important point: When a transaction ends,
all locks are released.

7-8 Programming for a Multiuser Environment

 Setting the Isolation Level

Locks While Modifying

When the database server fetches a row through an update cursor, it places a
promotable lock on the fetched row. If this succeeds, the database server
knows that no other program can alter that row. Because a promotable lock is
not exclusive, other programs can continue to read the row. This helps per-
formance because the program that fetched the row can take some time
before it issues the UPDATE or DELETE statement, or it can simply fetch the
next row.
When it is time to modify a row, the database server obtains an exclusive lock
on the row. If it already had a promotable lock, it changes it to exclusive
status.
The duration of an exclusive row lock depends on whether transactions are
in use. If they are not in use, the lock is released as soon as the modified row
is written to disk. When transactions are in use, all such locks are held until
the end of the transaction. This prevents other programs from using rows that
might be rolled back to their original state.
When transactions are in use, a key lock is used whenever a row is deleted.
This prevents the following error from occurring:
• Program A deletes a row.
• Program B inserts a row having the same key.
• Program A rolls back its transaction, forcing the database server to restore
its deleted row. What is to be done with the row inserted by program B?
By locking the index, the database server prevents a second program from
inserting a row until the first program commits its transaction.
The locks placed while reading a row are controlled by the current isolation
level, which is discussed in the next section.

Setting the Isolation Level

The isolation level is the degree to which your program is isolated from the
concurrent actions of other programs. INFORMIX-OnLine Dynamic Server
offers a choice of four isolation levels. It implements them by setting different
rules for how a program uses locks when it is reading. (This does not apply
to reads performed on update cursors.)
You set the isolation level using the command SET ISOLATION LEVEL. All
INFORMIX-OnLine Dynamic Server database servers support this statement.

Programming for a Multiuser Environment 7-9

Setting the Isolation Level

Dirty Read Isolation

The simplest isolation level, Dirty Read, amounts to virtually no isolation.
When a program fetches a row, it places no locks and it respects none; it sim-
ply copies rows from the database without regard for what other programs
are doing.
A program always receives complete rows of data; even under Dirty Read
isolation, a program never sees a row in which some columns have been
updated and some have not. However, a program using Dirty Read isolation
sometimes reads updated rows before the updating program ends its trans-
action. If the updating program later rolls back its transaction, the reading
program processed data that never really existed (number 4 in the list of
concurrency issues on page 7-5).
Dirty Read is the most efficient level. The reading program never waits and
never makes another program wait. It is the preferred level in any of the
following cases:
• All tables are static; that is, concurrent programs only read and never
modify data.
• The database is held in an exclusive lock.
• It is certain that only one program is using the database.

Committed Read Isolation

When a program requests the Committed Read isolation level,
INFORMIX-OnLine Dynamic Server guarantees that it never returns a row
that is not committed to the database. This prevents the situation in number
4 in the list of concurrency issues on page 7-5 (reading data that is not
committed and that is subsequently rolled back).
Committed Read is implemented very simply. Before fetching a row, the
database server tests to determine whether an updating process placed a lock
on the row. If not, it returns the row. Because rows that are updated but not
committed have locks on them, this test ensures that the program does not
read uncommitted data.
Committed Read does not actually place a lock on the fetched row; hence, it
is almost as efficient as Dirty Read. It is appropriate for use when each row of
data is processed as an independent unit, without reference to other rows in
the same or other tables.

7-10 Programming for a Multiuser Environment

 Setting the Isolation Level

Cursor Stability Isolation

The next level is called Cursor Stability. When it is in effect, the database
server places a lock on the latest row fetched. It places a shared lock for an
ordinary cursor or a promotable lock for an update cursor. Only one row is
locked at a time; that is, each time a row is fetched, the lock on the previous
row is released (unless that row is updated, in which case the lock holds until
the end of the transaction).
Cursor Stability ensures that a row does not change while the program
examines it. This is important when the program updates some other table
based on the data it reads from this row. Because of Cursor Stability, the pro-
gram is assured that the update is based on current information. It prevents
the use of stale data.
Here is an example to illustrate this point. In terms of the demonstration
database, Program A wants to insert a new stock item for manufacturer Hero.
Concurrently, program B wants to delete manufacturer Hero and all stock
associated with it. The following sequence of events can occur:
1. Program A, operating under Cursor Stability, fetches the Hero row from
the manufact table to learn the manufacturer code: This places a shared
lock on the row.
2. Program B issues a DELETE statement for that row. Because of the lock, the
database server makes the program wait.
3. Program A inserts a new row in the stock table using the manufacturer
code it obtained from the manufact table.
4. Program A closes its cursor on the manufact table or reads a different row
of it, releasing its lock.
5. Program B, released from its wait, completes the deletion of the row and
goes on to delete the rows of stock that use manufacturer code HRO,
including the row just inserted by Program A.
If Program A used a lesser level of isolation, the following sequence could
have happened:
1. Program A reads the Hero row of the manufact table to learn the
manufacturer code. No lock is placed.
2. Program B issues a DELETE statement for that row. It succeeds.
3. Program B deletes all rows of stock that use manufacturer code HRO.

Programming for a Multiuser Environment 7-11

Setting the Isolation Level

4. Program B ends.
5. Program A, not aware that its copy of the Hero row is now invalid, inserts
a new row of stock using the manufacturer code HRO.
6. Program A ends.
At the end, a row occurs in stock that has no matching manufacturer code in
manufact. Furthermore, Program B apparently has a bug; it did not delete the
rows it was supposed to delete. The use of the Cursor Stability isolation level
prevents these effects.
(It is possible to rearrange the preceding scenario so that it fails even with
Cursor Stability. All that is required is for Program B to operate on tables in
the reverse sequence to Program A. If Program B deletes from stock before it
removes the row of manufact, no degree of isolation can prevent an error.
Whenever this kind of error is possible, it is essential that all programs
involved use the same sequence of access.)
Because Cursor Stability locks only one row at a time, it restricts concurrency
less than does a table lock or database lock.

Repeatable Read Isolation

The Repeatable Read isolation level asks the database server to put a lock on
every row the program examines and fetches. The locks that are placed are
shareable for an ordinary cursor and promotable for an update cursor. The
locks are placed individually as each row is examined. They are not released
until the cursor closes or a transaction ends.
Repeatable Read allows a program that uses a scroll cursor to read selected
rows more than once and to be sure that they are not modified or deleted
between readings. (Scroll cursors are described in Chapter 5, “Programming
with SQL.”) No lower isolation level guarantees that rows still exist and are
unchanged the second time they are read.
Repeatable Read isolation places the largest number of locks and holds them
the longest. Therefore, it is the level that reduces concurrency the most. If
your program uses this level of isolation, you must think carefully about how
many locks it places, how long they are held, and what the effect can be on
other programs.
In addition to the effect on concurrency, the large number of locks can be a
problem. The database server records only a fixed number of locks for all the
programs it serves. (This number is adjustable.) If that table fills up, the

7-12 Programming for a Multiuser Environment

 Setting the Lock Mode

database server cannot place a lock and returns an error code. The person
who administers an INFORMIX-OnLine Dynamic Server system can monitor
the lock table and tell you when it is heavily used.
Repeatable Read is automatically the isolation level in an ANSI-compliant
database. Repeatable Read is required to ensure operations behave in
accordance with the ANSI standard for SQL.

Setting the Lock Mode

The lock mode determines what happens when your program encounters
locked data. One of the following things occurs when a program attempts to
fetch or modify a locked row:
• The program receives an immediate return from the database server with
an error code in SQLCODE.
• The program is suspended until the lock is removed by the program that
placed it.
• The program is suspended for a time and then, if the lock is not removed,
it receives an error-return code from the database server.
You choose among these results with the command SET LOCK MODE.

Waiting for Locks

If you prefer to wait (and this is the best choice for many applications),
execute the following command:

SET LOCK MODE TO WAIT

When this lock mode is set, your program usually ignores the existence of
other concurrent programs. When it needs to access a row that another pro-
gram has locked, your program waits until the lock is removed, then
proceeds. The delays are usually imperceptible.

Programming for a Multiuser Environment 7-13

Setting the Lock Mode

Not Waiting for Locks

The disadvantage of waiting for locks is that the wait might become very
long (although properly designed applications should hold their locks very
briefly). When the possibility of a long delay is not acceptable, a program can
execute the following command:

SET LOCK MODE TO NOT WAIT

When the program requests a locked row, it immediately receives an error

code (for example, error -107, Record is locked), and the current SQL
statement terminates. It is up to the program to roll back its current
transaction and try again.
The initial setting is not waiting when a program starts up. If you are using
SQL interactively and see an error related to locking, set the lock mode to
wait. If you are writing a program, consider making that one of the first
embedded SQL commands that the program executes.

Waiting a Limited Time

When you use INFORMIX-OnLine Dynamic Server, you have an additional
choice: You can ask the database server to set an upper limit on a wait. You
can issue the following command:

SET LOCK MODE TO WAIT 17

This places an upper limit of 17 seconds on the length of any wait. If a lock is
not removed in that time, the error code is returned.

Handling a Deadlock
A deadlock is a situation in which a pair of programs block the progress of each
other. Each program has a lock on some object that the other program wants
to access. A deadlock arises only when all programs concerned set their lock
modes to wait for locks.
INFORMIX-OnLine Dynamic Server detects deadlocks immediately when
they involve only data at a single network server. It prevents the deadlock
from occurring by returning an error code (error -143 ISAM error: dead-
lock detected) to the second program to request a lock. The error code is

7-14 Programming for a Multiuser Environment

 Simple Concurrency

the one the program receives if it sets its lock mode to not wait for locks. Thus,
if your program receives an error code related to locks even after it sets lock
mode to wait, you know the cause is an impending deadlock.

Handling External Deadlock

A deadlock can also occur between programs on different database servers.
In this case, INFORMIX-OnLine Dynamic Server cannot instantly detect the
deadlock. (Perfect deadlock detection requires excessive communications
traffic among all database servers in a network.) Instead, each database
server sets an upper limit on the amount of time that a program can wait to
obtain a lock on data at a different database server. If the time expires, the
database server assumes that a deadlock was the cause and returns a lock-
related error code.
In other words, when external databases are involved, every program runs
with a maximum lock-waiting time. The maximum is set for the database
server and can be changed by the database administrator.

Simple Concurrency
If you are not sure which choice to make concerning locking and concurrency,
and if your application is straightforward, have your program execute the
following commands when it starts up (immediately after the first
DATABASE statement):

SET LOCK MODE TO WAIT

SET ISOLATION TO REPEATABLE READ

Ignore the return codes from both statements. Proceed as if no other

programs exist. If no performance problems arise, you do not need to read
this section again.

Programming for a Multiuser Environment 7-15

Locking with Other Database Servers

Locking with Other Database Servers

INFORMIX-OnLine Dynamic Server manages its own locking so that it can
provide the different kinds of locks and levels of isolation described in the
preceding topics. Other Informix database servers implement locks using the
facilities of the host operating system and cannot provide the same
conveniences.
Some host operating systems provide locking functions as operating system
services. In these systems, database servers support the SET LOCK MODE
statement.
Some host operating systems do not provide kernel-locking facilities. In these
systems, the database server performs its own locking based on small files
that it creates in the database directory. These files have the suffix .lok.
You can tell the kind of system in which your database server is running by
executing the SET LOCK MODE statement and testing the error code, as
shown in the following fragment of INFORMIX-ESQL/C code:

#define LOCK_ONLINE 1
#define LOCK_KERNEL 2
#define LOCK_FILES 3
int which_locks()
{
int locktype;

locktype = LOCK_FILES;
exec sql set lock mode to wait 30;
if (sqlca.sqlcode == 0)
locktype = LOCK_ONLINE;
else
{
exec sql set lock mode to wait;
if (sqlca.sqlcode == 0)
locktype = LOCK_KERNEL;
}
/* restore default condition */
exec sql set lock mode to not wait;
return(locktype);
}

If the database server does not support the SET LOCK MODE statement, your
program is effectively always in NOT WAIT mode; that is, whenever it tries to
lock a row that is locked by another program, it receives an error code
immediately.

7-16 Programming for a Multiuser Environment

 Locking with Other Database Servers

Isolation While Reading

Informix database servers other than INFORMIX-OnLine Dynamic Server do
not normally place locks when fetching rows. Nothing comparable exists to
the shared locks used by OnLine to implement the Cursor Stability isolation
level.
If your program fetches a row with a singleton SELECT statement or through
a cursor that is not declared FOR UPDATE, the row is fetched immediately,
regardless of whether it is locked or modified by an unfinished transaction.
This design produces the best performance, especially when locks are
implemented by writing notes in disk files, but you must be aware that the
program can read rows that are modified by uncommitted transactions.
You can obtain the effect of Cursor Stability isolation by declaring a cursor
FOR UPDATE, and then using it for input. Whenever the database server
fetches a row through an update cursor, it places a lock on the fetched row. (If
the row is already locked, the program waits or receives an error, depending
on the lock mode.) When the program fetches another row without updating
the current one, the lock on the current row is released and the new row is
locked.
You can be sure by fetching through an update cursor, that the fetched row is
locked as long as you use it. (The row cannot become stale.) You are also
assured of fetching only committed data, because locks on rows that are
updated are held until the end of the transaction. Depending on the host
operating system and the database server, you might experience a
performance penalty for using an update cursor this way.

Locking Updated Rows

When a cursor is declared FOR UPDATE, locks are handled as follows: Before
a row is fetched, it is locked; or if it cannot be locked, the program waits or
returns an error.
The next time a fetch is requested, the database server notes whether the
current row is modified (using either the UPDATE or DELETE statement with
WHERE CURRENT OF) and whether a transaction is in progress. If both these
things are true, the lock on the row is retained. Otherwise, the lock is released.
So if you perform updates within a transaction, all updated rows remain
locked until the transaction ends. Rows that are not updated are locked only
while they are current. Rows updated outside a transaction, or in a database
that does not use transaction logging, are also unlocked as soon as another
row is fetched.

Programming for a Multiuser Environment 7-17

Hold Cursors

Hold Cursors
When transaction logging is used, the database server guarantees that
anything done within a transaction can be rolled back at the end of it. To do
this reliably, the database server normally applies the following rules:
• All cursors are closed by ending a transaction.
• All locks are released by ending a transaction.
These rules are normal with all database systems that support transactions,
and for most applications they do not cause any trouble. However, circum-
stances exist where using standard transactions with cursors are not possible.
For example, a situation is sketched in Figure 7-1 which works fine without
transactions. However, when transactions are added, closing the cursor
conflicts with using two cursors simultaneously.

DECLARE master CURSOR FOR ...

DECLARE detail CURSOR FOR ... FOR UPDATE
OPEN master
LOOP:
FETCH master INTO ...
IF (the fetched data is appropriate) THEN
BEGIN WORK
OPEN detail USING data read from master
FETCH detail ...
UPDATE ... WHERE CURRENT OF detail
COMMIT WORK
END IF
END LOOP
CLOSE MASTER

Figure 7-1 A pseudocode sketch of one common form of database application

In this design, one cursor is used to scan a table. Selected records are used as
the basis for updating a different table. The problem is that when each update
is treated as a separate transaction (as shown in the pseudocode in Figure
7-1), the COMMIT WORK statement following the UPDATE closes all cursors—
including the master cursor.
The simplest alternative is to move the COMMIT WORK and BEGIN WORK
statements to be the last and first ones, respectively, so that the entire scan
over the master table is one large transaction. This is sometimes possible, but
it may become impractical if many rows need to be updated. The number of
locks can be too large, and they are held for the duration of the program.

7-18 Programming for a Multiuser Environment

 Summary

A solution supported by Informix database servers is to add the keywords

WITH HOLD to the declaration of the master cursor. Such a cursor is referred
to as a hold cursor and is not closed at the end of a transaction. The database
server still closes all other cursors, and it still releases all locks, but the hold
cursor remains open until it is explicitly closed.
Before you attempt to use a hold cursor, you must be sure that you
understand the locking mechanism described here, and you must also under-
stand the programs that are running concurrently. The reason is that when-
ever COMMIT WORK is executed, all locks are released, including any locks
placed on rows fetched through the hold cursor.
This has little importance if the cursor is used as intended, for a single
forward scan over a table. However, you are allowed to specify WITH HOLD
for any cursor, including update cursors and scroll cursors. Before you do
this, you must understand the implications of the fact that all locks (including
locks on entire tables) are released at the end of a transaction.

Summary
Whenever multiple programs have access to a database concurrently (and
when at least one of them can modify data), all programs must allow for the
possibility that another program can change the data even as they read it. The
database server provides a mechanism of locks and isolation levels that
usually allow programs to run as if they were alone with the data.

Programming for a Multiuser Environment 7-19

Summary

7-20 Programming for a Multiuser Environment

Section II: Designing and
Managing Databases
 Chapter

Building Your Data

Model
Chapter Overview 3
Why Build a Data Model 3
Entity-Relationship Data Model Overview 3
Identifying and Defining Your Principal Data Objects 4
Discovering Entities 4
Choosing Possible Entities 5
Pruning Your List of Entities 5
The Telephone-Directory Example 6
Diagramming Your Entities 8
Defining the Relationships 8
Connectivity 8
Existence Dependency 9
Cardinality 9
Discover the Relationships 10
Diagramming Your Relationships 14
Identifying Attributes 14
How to Identify an Attribute 15
Selecting Attributes for Your Entities 15
Listing Your Attributes 16

8
About Entity Occurrences 16
Diagramming Your Data Objects 16
How You Read Entity-Relationship Diagrams 18
The Telephone-Directory Example 18
What’s Next 19
 Translating E-R Data Objects Into Relational Constructs 20
Rules for Defining Tables, Rows, and Columns 20
Placing Constraints on Columns 21
Determining Keys for Tables 22
Primary Keys 22
Foreign Keys (Join Columns) 24
Adding Keys to the Telephone-Directory Diagram 25
Resolving Your Relationships 26
Resolving m:n Relationships 26
Resolving Other Special Relationships 28
Normalizing Your Data Model 29
First Normal Form 29
Second Normal Form 31
Third Normal Form 32
Summary of Normalization Rules 32
Summary 33

8-2 Building Your Data Model

Chapter Overview
The first step in creating a database is to construct a data model—a precise,
complete definition of the data to be stored. This chapter contains a cursory
overview of one method of doing this. The following chapters describe how
to implement a data model once you design it.
To understand the material in this chapter, you should have a basic
understanding of SQL and relational database theory.

Why Build a Data Model

You may already have an idea of a database. If you use some type of formal
notation to build your data model, you can help your design in two ways:
• It makes you think through the data model completely.
A mental model often contains unexamined assumptions; formalizing the
design reveals these points.
• It is easier to communicate your design to other people.
A formal statement makes the model explicit, so that others can return
comments and suggestions in the same form.

Entity-Relationship Data Model Overview

Different books present different formal methods of modeling data. Most
methods force you to be thorough and precise. If you have already learned
some method, by all means use it.

Building Your Data Model 8-3

Identifying and Defining Your Principal Data Objects

This chapter presents a summary of the entity-relationship (E-R) data model,

a modeling method taught in training courses presented by Informix and
other companies and based on the original data model by E. F. Codd. This
modeling method is carried out in the following steps:
1. Identify and define the principal data objects (entities, relationships, and
attributes).
2. Diagram the data objects using the entity-relationship approach.
3. Translate your entity-relationship data objects into relational constructs.
4. Resolve the logical data model.
5. Normalize the logical data model.
Steps 1-5 are discussed in this chapter. Chapter 9, “Implementing Your Data
Model,” discusses a final step—converting your logical data model to a
physical schema.
The end product of data modeling is a fully defined database design encoded
in a diagram like that in Figure 8-16 on page 8-31, which shows the final set
of tables for a personal telephone directory. The personal telephone directory
is an example developed in this chapter. It is used rather than the stores6
database used in the rest of this book because it is small enough to be
developed completely in one chapter but large enough to show the entire
method.

Identifying and Defining Your Principal Data Objects

The first step in building an entity-relationship data model is to identify and
define your principal data objects. The principal data objects are entities, rela-
tionships, and attributes.

Discovering Entities
An entity is a principal data object that is of significant interest to the user. It
is usually a person, place, thing, or event to be recorded in the database. If the
data model were a language, entities would be its nouns. The stores6
database contains the following entities: customer, orders, items, stock, catalog,
cust_calls, call_type, manufact, and state.
The first step in modeling is to choose the entities to record. Most of the
entities that you choose will become tables in the model.

8-4 Building Your Data Model

 Identifying and Defining Your Principal Data Objects

Choosing Possible Entities

If you have an idea for your database, you can probably list several entities
immediately. However, if other people use the database, you should poll
them for their understanding of what fundamental things the database
should contain. Make a preliminary list of all the entities you can identify.
Interview the potential users of the database for their opinions about what
must be recorded in the database. Determine basic characteristics for each
entity, such as “there must be at least one address associated with a name.”
All of the decisions you make in determining your entities become your busi-
ness rules. The section, “The Telephone-Directory Example” on page 8-6,
provides some of the business rules for the example shown in this chapter.
Later, when you normalize your data model, some of the entities may expand
or become other data objects. See “Normalizing Your Data Model” on
page 8-29 for additional information.

Pruning Your List of Entities

When the list of entities seems complete, prune it by making sure that each
one has the following qualities:
• It is significant.
List only entities that are important to the users of the database and worth
the trouble and expense of computer tabulation.
• It is generic.
List only types of things, not individual instances. For instance, symphony
might be an entity, but Beethoven’s Fifth would be an entity instance or
entity occurrence.
• It is fundamental.
List only entities that exist independently, without needing something
else to explain them. Anything you could call a trait, a feature, or a
description is not an entity. For example, a part number is a feature of the
fundamental entity called part. Also, do not list things that you can derive
from other entities; for example, avoid any sum, average, or other
quantity that you can calculate in a SELECT expression.
• It is unitary.
Be sure that each entity you name represents a single class—that it cannot
be broken down into subcategories, each with its own features. In plan-
ning the telephone-directory model (see “The Telephone-Directory
Example” on page 8-6) an apparently simple entity, the telephone num-
ber, turns out to consist of three categories, each with different features.

Building Your Data Model 8-5

Identifying and Defining Your Principal Data Objects

These choices are neither simple nor automatic. To discover the best choice of
entities, you must think deeply about the nature of the data you want to store.
Of course, that is exactly the point of making a formal data model. The
following section describes the example shown in this chapter in further
detail.

The Telephone-Directory Example

Suppose that you create a database that computerizes a personal telephone
directory. The database model must record the names, addresses, and
telephone numbers of people and organizations that its user deals with for
business and pleasure.
The first step is to define the entities, and the first thing you might do is look
carefully at a page from a telephone directory to see what entities are there.

NAME PHONE NAME PHONE

Thomas Morrison 503-776-3428 Catherine Morgan 206-789-5396
ADDRESS ADDRESS
866 Gage Rd. 429 Bridge Way
Klamath Falls Seattle, WA 98103
N
OR 97601
NAME PHONE NAME PHONE O
Thomas Morrison Norman Dearborn (206)598-8189 P
ADDRESS ADDRESS Q
866 Gage Rd. Morganthaler Industries R
Klamath Falls 12558 E. 10th Ave. Seattle, WA S
OR 97601 98102 FAX: 206 598-6872 T
NAME PHONE NAME PHONE U
Thomas Morrison Thomas Morrison 503-256-6031 V

ADDRESS ADDRESS W
866 Gage Rd. 866 Gage Rd. X
Klamath Falls Klamath Falls, OR 97601 Y
OR 97601 Z

Figure 8-1 Part of a page from a telephone directory

8-6 Building Your Data Model

 Identifying and Defining Your Principal Data Objects

The physical form of the existing data can be misleading. Do not let the layout
of pages and entries in the telephone directory mislead you into trying to
specify an entity that represents one entry in the book—some kind of alpha-
betized record with fields for name, number, and address. Remember it is not
the medium you want to model, it is the data.
At first glance, the entities recorded in a telephone directory include the
following items:
• Names (of persons and organizations)
• Addresses
• Telephone numbers
Do these entities meet the earlier criteria? They are clearly significant to the
model and are generic.
Are they fundamental? A good test is to ask if an entity can vary in number
independently of any other entity. After thinking about it, you realize that a
telephone directory sometimes lists people who have no number or current
address (people who move or change jobs). A telephone directory also can
list both addresses and numbers that are used by more than one person. All
three of these entities can vary in number independently; that strongly
suggests they are fundamental, not dependent.
Are they unitary? Names can be split into personal names and corporate
names. After thinking about it, you decide that all names should have the
same features in this model; that is, you do not plan to record different infor-
mation about a company than you would about a person. Likewise, you
decide only one kind of address exists; no need exists to treat home addresses
differently from business ones.
However, you also realize that more than one kind of telephone number
exists, three. Voice numbers are answered by a person, fax numbers connect
to a fax machine, and modem numbers connect to a computer. You decide that
you want to record different information about each kind of number, so these
three are different entities.
For the personal telephone-directory example, you decide that you want to
keep track of the following entities:
• Name
• Address (mailing)
• Telephone number (voice)
• Telephone number (fax)
• Telephone number (modem)

Building Your Data Model 8-7

Identifying and Defining Your Principal Data Objects

Diagramming Your Entities

A section in this chapter will teach you how to use the entity-relationship
diagrams. For now, create a separate, rectangular box for each entity in the
telephone-directory example. You will learn how to put the entities together
with relationships in “Diagramming Your Data Objects” on page 8-16.

name address voice fax modem

Defining the Relationships

After you choose your entities, you need to consider the relationships
between them. Relationships are not always obvious, but all of the ones
worth recording must be found. The only way to ensure that all the relation-
ships are found is to exhaustively list all possible relationships. You must
consider every pair of entities A and B and ask, “What is the relationship
between an A and a B?”
A relationship is an association between two entities. Usually, a verb or
preposition connecting two entities will imply a relationship. A relationship
between entities is described in terms of connectivity, existence dependency, and
cardinality.

Connectivity
Connectivity refers to the number of entity instances. An entity instance is a
particular occurrence of an entity. The three types of connectivity are one-to-
one (written 1:1), one-to-many (written 1:n), and many-to-many (written
m:n) shown graphically in Figure 8-2.

8-8 Building Your Data Model

 Identifying and Defining Your Principal Data Objects

one-to-one one-to-many many-to-many

Figure 8-2 Connectivity in relationships

For example, in the telephone-directory example, an address can be

associated with more than one name. The connectivity for the relationship
between the name and address entities is one-to-many (1:n).

Existence Dependency
Existence dependency describes whether an entity in a relationship is
optional or mandatory. Analyze your business rules to identify whether an
entity must exist in a relationship. For example, your business rules may dic-
tate that an address must be associated with a name. This would make the
existence dependency for the relationship between the name and address
entities mandatory. An optional existence dependency could be indicated by
a business rule that said a person may or may not have children.

Cardinality
Cardinality places a constraint on the number of times an entity can appear
in a relationship. The cardinality of a 1:1 relationship is always one. But the
cardinality of a 1:n relationship is open; n could be any number. If you need
to place an upper limit on n, you do it by specifying a cardinality for the rela-
tionship. For example, in a store sale example, you could limit the number of
sale items that a customer can purchase at one time. You usually place cardi-
nality constraints through your application program or through stored
procedures.
For additional information about cardinality, see any entity-relationship data
modeling text. References to two data modeling books are included in the
Summary of this chapter.

Building Your Data Model 8-9

Identifying and Defining Your Principal Data Objects

Discover the Relationships

A compact way to discover the relationships is to prepare a matrix that names
all the entities on the rows and again on the columns. Figure 8-3 is a matrix
that reflects the entities for the personal telephone directory.

name address number number number

(voice) (fax) (modem)

name

address

number
(voice)

number
(fax)

number
(modem)

Figure 8-3 A matrix that reflects the entities for a personal telephone directory

You can ignore the lower triangle of the matrix, as indicated by the shaded
area. You must consider the diagonal cells; that is, you must ask the question
“What is the relationship between an A and another A?” In this model, the
answer is always none. No relationship exists between a name and a name,
or an address and another address, at least none worth recording in this
model. When a relationship exists between an A and another A, you have
found a recursive relationship. (See “Resolving Other Special Relationships”
on page 8-28.)
For all cells for which the answer is clearly none, write none in the matrix.
Now the matrix looks like Figure 8-4.

8-10 Building Your Data Model

 Identifying and Defining Your Principal Data Objects

name address number number number

(voice) (fax) (modem)

name none

address none

number none
(voice)

number none
(fax)

number none
(modem)

Figure 8-4 A matrix with initial relationships included: No entities relate to themselves

Although in this model no entities relate to themselves, in other models this

is not always true. A typical example is an employee who is the manager of
another employee. Another example occurs in manufacturing, when a part
entity is a component of another part.
In the remaining cells, you write the connectivity relationship that exists
between the entity on the row and the entity on the column. To repeat, the fol-
lowing kinds of relationships are possible:
• One-to-one (written 1:1), in which never more than one entity A exists for
one entity B and never more than one B for one A.
• One-to-many (written 1:n), in which never more than one entity A exists,
but there can be several entities B related to it (or vice versa).
• Many-to-many (written m:n), in which several entities A can be related to
one B and several entities B can be related to one A.
One-to-many relationships are the most common. Examples of one-to-many
and many-to-many relationships are shown in the telephone-directory
model.
As you can see in Figure 8-4, the first unfilled cell represents the relationship
between names and addresses. What connectivity lies between these entities?
You can ask yourself, “how many names can be associated with an address?”

Building Your Data Model 8-11

Identifying and Defining Your Principal Data Objects

A name (you decide) can have zero or one address but no more than one. You
write 0-1 opposite name and below address, as shown in the following
figure:

name address

name none
0-1

Ask yourself how many addresses can be associated with a name. You decide
that an address can be associated with more than one name. For example, you
can know several people at one company or more than two people who live
at the same address.
Can an address be associated with zero names? That is, should it be possible
for an address to exist when no names use it? You decide that yes, it can.
Below address and opposite name, you write 0-n, as shown in the following
figure:

name address

name none 0-n

0-1

If you decide that an address could not exist without being associated with at
least one name, you write 1-n instead of 0-n.
When the cardinality of a relationship is limited on either side to 1, it is a 1:n
relationship. In this case, the relationship between names and addresses is a
1:n relationship.
Now consider the next cell, the relationship between a name and a voice
number. How many voice numbers can a name be associated with, one or
more than one? Glancing at your telephone directory, you see that you may
often have noted more than one telephone number for a person—for some
busy salesman you have a home number, an office number, a paging number,

8-12 Building Your Data Model

 Identifying and Defining Your Principal Data Objects

and a car phone number. But there also can be names without associated
numbers. You write 0-n opposite name and below number (voice), as
shown in the following figure:

name address number

(voice)

name none 0-n

0-1 0-n

What is the other side of this relationship? How many names can be
associated with a voice number? You decide that only one name can be asso-
ciated with a voice number. Can a number be associated with zero names?
No, you decide; no point exists for recording a number unless it is used by
someone. You write 1 under number (voice) and opposite name.

name address number

(voice)

name none 0-n 1

0-1 0-n

Fill out the rest of the matrix in the same fashion, using the following
decisions:
• A name can be associated with more than one fax number; for example, a
company can have several fax machines. Going the other way, a fax num-
ber can be associated with more than one name; for example, several
people can use the same fax number.
• A modem number must be associated with exactly one name. (This is an
arbitrary decree to complicate the example; pretend it is a requirement of
the design.) However, a name can have more than one associated modem
number; for example, a company computer can have several dial-up
lines.
• Although some relationship exists between a voice number and an
address, a modem number and an address, and a fax number and an

Building Your Data Model 8-13

Identifying and Defining Your Principal Data Objects

address in the real world, none needs to be recorded in this model. There
already is an indirect relationship through name.

name address number number number

(voice) (fax) (modem)

name none 0-n 1 1-n 1

0-1 0-n 0-n 0-n

address none none none none

number none none none

(voice)

number none none

(fax)

number none
(modem)

Figure 8-5 A completed matrix for an telephone directory

Other decisions reflected in the matrix is that no relationship exists between

a fax number and a modem number, between a voice number and a fax, or
between a voice number and a modem number.
You might disagree with some of these decisions (for example, why a
relationship between voice numbers and modem numbers is not supported).
For the sake of our example, these are our business rules.

Diagramming Your Relationships

For now, save the matrix that you created in this section. You will learn how
to create an entity-relationship diagram in “Diagramming Your Data
Objects” on page 8-16.

Identifying Attributes
Entities contain attributes, which are characteristics or modifiers, qualities,
amounts, or features. An attribute is a fact or nondecomposable piece of
information about an entity. Later, when you represent an entity as a table, its
attributes are added to the model as new columns.

8-14 Building Your Data Model

 Identifying and Defining Your Principal Data Objects

How to Identify an Attribute

Before you can identify your attributes, you must identify your entities. After
you determine your entities, ask yourself, “what characteristics do you need
to know about each entity?” For example, in an address entity, you probably
need information about street, city, and zipcode. Each of these characteristics of
the address entity become attributes.

Selecting Attributes for Your Entities

In selecting attributes, choose ones that have the following qualities:
• They are significant.
Include only attributes that are useful to the users of the database.
• They are direct, not derived.
An attribute that can be derived from existing attributes—for instance,
through an expression in a SELECT statement—should not be made part
of the model. The presence of derived data greatly complicates the
maintenance of a database.
At a later stage of the design (discussed in Chapter 10, “Tuning Your Data
Model”), you can consider adding derived attributes to improve
performance, but at this stage you should exclude them.
• They are nondecomposable.
An attribute can contain only single values, never lists or repeating
groups. Composite values must be broken into separate attributes.
• They contain data of the same type.
For example, you would want to enter only date values in a birthday
attribute, not names or telephone numbers.
The rules for defining attributes are the same as those for defining columns.
For more information about defining columns, see “Placing Constraints on
Columns” on page 8-21.
The following attributes are added to the telephone-directory example to
produce the diagram shown in Figure 8-10 on page 8-19:
• Street, city, state, and zip code are added to the address entity.
• Birth date is added to the name entity. Also added to the name entity are
e-mail address, anniversary date, and children’s first names.
• Type is added to the voice entity to distinguish car phones, home phones,
and office phones. A voice number can be associated with only one voice
type.

Building Your Data Model 8-15

Diagramming Your Data Objects

• The hours a fax machine is attended is added to the fax entity.

• Whether a modem supports 300-, 1,200-, or 2,400-baud rates is added to
the modem entity.

Listing Your Attributes

For now, simply list the attributes for the telephone-directory example with
the entities with which you think they belong. Your list should look
something like Figure 8-6:

name address voice fax modem

fname street vce_num fax_num mdm_num
lname city vce_type operfrom b300
bdate state opertill b1200
anniv zipcode b2400
email
child1
child2
child3

Figure 8-6 Attributes for the telephone-directory example

About Entity Occurrences

An additional data object that you need to know about is the entity
occurrence. Each row in a table represents a specific, single occurrence of the
entity. For example, if customer is an entity, a customer table represents the
idea of customer; in it, each row represents one specific customer, such as Sue
Smith. Keep in mind that entities will become tables, attributes will become
columns, and rows will become entity occurrences.

Diagramming Your Data Objects

At this point, you have already discovered and understood the entities and
relationships in your database. That is the most important part of the rela-
tional database design process. Once you have determined the entities and
relationships, you might find it helpful to have a method for displaying your
thought process during database design.

8-16 Building Your Data Model

 Diagramming Your Data Objects

Most data modeling methods provide some form of graphically displaying

your entities and relationships. Informix uses the entity-relationship diagram
approach originally developed by C. R. Bachman. Entity-relationship
diagrams serve the following purposes:
• They model the information needs of an organization.
• They identify entities and their relationships.
• They provide a starting point for data definition (data flow diagrams).
• They provide an excellent source of documentation for application
developers as well as database and system administrators.
• They create a logical design of the database which can be translated into
a physical schema.
Several different styles of documenting entity-relationship diagrams exist. If
you already have a style that you prefer, use it. A sample entity-relationship
diagram is represented by the following symbols:

name address

entity entity

relationship

Figure 8-7 Symbols of an entity-relationship diagram

Entities are represented by a box. Relationships are represented by a line

connecting the entities. In addition, you use several graphical items to dis-
play the following features of relationships as shown in Figure 8-8:
• A circle across a relationship link indicates optionality in the relationship
(zero instances).
• A small bar across a relationship link indicates that exactly one instance of
the entity is associated with another entity. (Consider the bar as a “1”.)
• The “crow’s feet” represents many in your relationship.

Building Your Data Model 8-17

Diagramming Your Data Objects

optionality optionality
ames Addres

many exactly
one

Figure 8-8 The parts of a relationship in an entity-relationship diagram

How You Read Entity-Relationship Diagrams

You read the diagrams first from left to right and then from right to left. In the
case of the name-address relationship diagrammed in Figure 8-9, you read the
relationships as follows: names can be associated with zero or exactly one
address; addresses can be associated with zero, one, or many names.

can have zero or

exactly 1
name address

can have zero

or many

Figure 8-9 Reading an entity-relationship diagram

The Telephone-Directory Example

Figure 8-10 shows the telephone-directory example and includes the entities,
relationships, and attributes. This diagram includes the relationships that
were established with the matrix. After studying the diagram symbols, com-
pare the entity-relationship diagram in Figure 8-10 with the matrix in Figure
8-5 on page 8-14. Verify for yourself that the relationships are the same in
both figures.

8-18 Building Your Data Model

 Diagramming Your Data Objects

A matrix such as Figure 8-5 is a useful tool when you are first designing your
model because, in filling it out, you are forced to think of every possible rela-
tionship. However, the same relationships appear in a diagram such as
Figure 8-10, and this type of diagram may be easier to read when you are
reviewing an existing model.

name
lname
fname
bdate address
anniv street
email city
child1 state
child2 zipcode
child3

voice fax modem

vce_num fax_num mdm_num
vce_type operfrom b300
opertill b1200
b2400

Figure 8-10 Preliminary entity-relationship diagram of the Telephone-Directory Example

What’s Next
Subsequent sections describe the following tasks:
• How to translate the entities, relationships, and attributes into relational
constructs
• How to resolve the entity-relationship data model
• How to normalize the entity-relationship data model
Chapter 9, “Implementing Your Data Model,” shows you how to create a
database from the entity-relationship data model.

Building Your Data Model 8-19

Translating E-R Data Objects Into Relational Constructs

Translating E-R Data Objects Into Relational Constructs

All the data objects you have learned about so far—entities, relationships,
attributes, and entity occurrences—will be translated into SQL tables, joins
between tables, columns, and rows. The tables, columns, and rows of your
database must fit the rules found in “Rules for Defining Tables, Rows, and
Columns.”
Your data objects should fit these rules before you normalize your data
objects. Normalizing your data objects is accomplished by analyzing the
dependencies between your entities, relationships, and attributes.
Normalizing is discussed in “Normalizing Your Data Model” on page 8-29.
After you normalize the data model, you can use SQL statements to create a
database based on your data model. Chapter 9, “Implementing Your Data
Model,” describes how to create your database and provides the database
schema for the telephone-directory example.
Each entity you choose is represented as a table in the model. The table stands
for the entity as an abstract concept, and each row represents a specific, indi-
vidual occurrence of the entity. In addition, each attribute of an entity is
represented by a column in the table.
The following ideas are fundamental to most relational data model methods,
including the entity-relationship data model. By following these rules while
designing your data model, you will save time and effort when you
normalize your model.

Rules for Defining Tables, Rows, and Columns

You are already familiar with the idea of a table composed of rows and
columns. But you must respect the following rules while you define the tables
of a formal data model:
• Rows must stand alone.
Each row of a table is independent and does not depend on any other row
of the same table. As a consequence, the order of the rows in a table is not
significant in the model. The model should still be correct even if all the
rows of a table are shuffled into random order.
After the database is implemented, you can tell the database server to
store rows in a certain order for the sake of efficiency, but that does not
affect the model.

8-20 Building Your Data Model

 Translating E-R Data Objects Into Relational Constructs

• Rows must be unique.

In every row, some column must contain a unique value. If no single
column has this property, the values of some group of columns taken as a
whole must be different in every row.
• Columns must stand alone.
The order of columns within a table has no meaning in the model. The
model should still be correct even if the columns are rearranged.
After the database is implemented, programs and stored queries that use
an asterisk to mean all columns are dependent on the final order of
columns, but that does not affect the model.
• Column values must be unitary.
A column can contain only single values, never lists or repeating groups.
Composite values must be broken into separate columns. For example, if
you decide to treat a person’s first and last names as separate values, as
shown in the examples in this chapter, they must be in separate columns,
not in a single name column.
• Each column must have a unique name.
Two columns within the same table cannot share the same name.
However, you may have columns that contain similar information. For
example, the name table in the telephone-directory example contains col-
umns for children’s names. You can name each column, child1, child2,
and so on.
• Each column must contain data of a single type.
A column must contain information of the same data type. For example,
a column identified as an integer must contain only numeric information,
not characters from a name.
If your previous experience is only with data organized as arrays or
sequential files, these rules might seem unnatural. However, relational data-
base theory shows that you can represent all types of data using only tables,
rows, and columns that follow these rules. With a little practice, these rules
become automatic.

Placing Constraints on Columns

When you define your table and columns with the CREATE TABLE statement,
you constrain each column. These constraints specify whether you want the
column to contain characters or numbers, the form that you want dates to

Building Your Data Model 8-21

Translating E-R Data Objects Into Relational Constructs

use, and other constraints. A domain describes the constraints on and

identifies the set of valid values attributes may assume. The domain
characteristics of a column can consist of the following items:
• Data type (INTEGER, CHAR, DATE, and so on)
• Format (for example, yy/mm/dd)
• Range (for example, 1,000-5,400)
• Meaning (for example, personnel number)
• Allowable values (for example, only grades S or U)
• Uniqueness
• Null support
• Default value
• Referential constraints
You define the domain characteristics when you create your tables. Defining
domains and creating your tables and database are discussed in Chapter 9,
“Implementing Your Data Model.”

Determining Keys for Tables

The columns of a table are either key columns or descriptor columns. A key
column is one that uniquely identifies a particular row in the table. For exam-
ple, a social-security number is unique for each employee. A descriptor col-
umn specifies the nonunique characteristics of a particular row in the table.
For example, two employees may have the same first name, Sue. The first
name Sue is a nonunique characteristic of an employee. The main types of
keys in a table are: primary keys and foreign keys.
You designate primary and foreign keys when you create your tables.
Primary and foreign keys are used to relate tables together physically. Your
next task is to specify a primary key for each table. That is, you must identify
some quantifiable characteristic of the table that distinguishes each row from
every other.

Primary Keys
The primary key of a table is the column whose values are different in every
row. Because they are different, they make each row unique. If no one such
column exists, the primary key is a composite of two or more columns whose
values, taken together, are different in every row.

8-22 Building Your Data Model

 Translating E-R Data Objects Into Relational Constructs

Every table in the model must have a primary key. This follows automatically
from the rule that all rows must be unique. If necessary, the primary key is
composed of all the columns taken together.
The primary key should be a numeric data type (INT or SMALLINT), SERIAL
data type, or a short character string (as used for codes). It is recommended
that you avoid using long character strings as primary keys.
Null values are never allowed in a primary-key column. Null values are not
comparable; that is, they cannot be said to be alike or different. Hence, they
cannot make a row unique from other rows. If a column permits null values,
it cannot be part of a primary key.

What is a Null Value?

Columns in a table may be designated as containing null values. This means
that the value for the column may be unknown or not applicable. For exam-
ple, in the telephone-directory example, the anniv column of the name table
may contain null values; if you do not know the person’s anniversary, you
do not specify it. Be careful not to confuse null value with zero or blank
value.

Some entities have ready-made primary keys such as catalog codes or

identity numbers, which are defined outside the model. These are
user-assigned keys.
Sometimes more than one column or group of columns can be used as the
primary key. All columns or groups that qualify to be primary keys are called
candidate keys. All candidate keys are worth noting because their property of
uniqueness makes them predictable in a SELECT operation. When you select
the columns of a candidate key, you know the result can contain no duplicate
rows. Sometimes that allows you to predict the uniqueness of the rows
returned. It always means that the result of a SELECT operation can be a table
in its own right, with the selected candidate key as its primary key.

Composite Keys
Some entities lack features that are reliably unique. Different people can have
identical names; different books can have identical titles. You can usually find
a composite of attributes that work as a primary key. For example, it is rare
for people to have identical names and identical addresses or for different
books to have identical titles, authors, and publication dates.

Building Your Data Model 8-23

Translating E-R Data Objects Into Relational Constructs

System-Assigned Keys
A system-assigned primary key is usually preferable to a composite key. A
system-assigned key is a number or code that is attached to each instance of
an entity when it is first entered into the database. The easiest system-
assigned keys to implement are serial numbers because the database server
can generate them automatically. Informix offers the SERIAL data type for
serial numbers. However, the people who use the database might not like a
plain numeric code. Other codes can be based on actual data; for example, an
employee identification code could be based on the person’s initials com-
bined with the digits of the date they were hired. In the telephone-directory
example, a system-assigned primary key is used for the name table.

Foreign Keys (Join Columns)

A foreign key is simply a column or group of columns in one table that
contains values that match the primary key in another table. Foreign keys are
used to join tables; in fact, most of the join columns referred to earlier in this
book are foreign-key columns. Figure 8-11 shows the primary and foreign
keys of the customer and order tables from the stores6 database.

customer orders
customer_num order_num customer_num

primary key foreign key

Figure 8-11 Primary-foreign keys in the customer-order relationships

Foreign keys are noted wherever they appear in the model because their
presence can restrict your ability to delete rows from tables. Before you can
delete a row safely, either you must delete all rows that refer to it through
foreign keys or you must define the relationship using special syntax that
allows you to delete rows from primary-key and foreign-key columns with a
single delete command. The database server disallows deletes that violate
referential integrity.
You can always preserve referential integrity by deleting all foreign-key rows
before you delete the primary key to which they refer. If you are imposing ref-
erential constraints on your database, the database server does not permit
you to delete primary keys with matching foreign keys. It also does not

8-24 Building Your Data Model

 Translating E-R Data Objects Into Relational Constructs

permit you to add a foreign-key value that does not reference an existing
primary-key value. Referential integrity is discussed in Chapter 4,
“Modifying Data.”

Adding Keys to the Telephone-Directory Diagram

The initial choices of primary and foreign keys is shown in Figure 8-12. This
diagram reflects some important decisions.
For the name table, the primary key rec_num is chosen. Note that the data
type for rec_num is SERIAL. The values for rec_num are system-generated. If
you look at the other columns (or attributes) in the name table, you see that
the data types associated with the columns are mostly character- based. None
of these columns alone are good candidates for a primary key. If you combine
elements of the table into a composite key, you create an exceedingly cumber-
some key. By using a SERIAL data type, you have a key that you can also use
to join the other tables to the name table.
For the voice, fax, and modem tables, the telephone numbers are shown as
primary keys. Note that these tables are joined to the name table through the
rec_num key.
The address table also uses a system-generated primary key, id_num. The
address table must have a primary key because the business rules state that an
address can exist when no names use it. If the business rules prevent an
address from existing unless a name is associated with it, then the address
table could be joined to the name table with the foreign key rec_num only.

Building Your Data Model 8-25

Resolving Your Relationships

name
rec_num PK
lname
fname address
bdate id_num PK
anniv rec_num FK
email street
child1 city
child2 state
child3 zipcode

voice fax modem

vce_num PK fax_num PK mdm_num PK
rec_num FK rec_num FK rec_num FK
vce_type operfrom b300
opertill b1200
b2400

Figure 8-12 Telephone-directory diagram with primary and foreign keys added

Resolving Your Relationships

The aim of a good data model is to create a structure that provides the
database server with quick access. The telephone-directory data model can
be further refined by resolving the relationships and normalizing the data
model. This section addresses the hows and whys of resolving your
relationships. Normalizing your data model is discussed in “Normalizing
Your Data Model” on page 8-29.

Resolving m:n Relationships

Many-to-many relationships add complexity and confusion to your model
and to the application development process. The key to resolving many-to-
many relationships is to separate the two entities and create two one-to-many
relationships between them with a third intersect entity. The intersect entity
usually contains attributes from both connecting entities.

8-26 Building Your Data Model

 Resolving Your Relationships

To resolve a many-to-many relationship, analyze your business rules again.

Have you accurately diagrammed the relationship? In the telephone-
directory example, we have a many-to-many relationship between the name
and fax entities as shown in Figure 8-12 on page 8-26. To resolve the relation-
ship between name and fax, we carefully reviewed the business rules. The
business rules say: “One person can have zero, one, or many fax numbers; a
fax number can be for several people.” Based on what we selected earlier as our
primary key for the voice entity, a many-to-many relationship exists.
A problem exists in the fax entity because the telephone number, which is
designated as the primary key, can appear more than one time in the fax
entity; this violates the qualification of a primary key. Remember, the primary
key must be unique.
This many-to-many relationship is resolved by adding an intersect entity
between name and fax entities. The new intersect entity, faxname, contains two
attributes, fax_num and rec_num. The primary key for the entity is a com-
posite of both attributes. Individually, each attribute is a foreign key that ref-
erences the table from which it came. The relationship between the name and
faxname tables is one-to-many because one name can be associated with many
fax numbers; in the other direction, each faxname combination can be associ-
ated with one rec_num. The relationship between the fax and faxname tables
is one-to-many because each number can be associated with many faxname
combinations.

Building Your Data Model 8-27

Resolving Your Relationships

name
rec_num PK rec_num PK
lname lname
Intersect
fname fname
Entity
bdate bdate
anniv anniv
email faxname email
child1 fax_num PK FK child1
child2 rec_num PK FK child2
child3 child3

fax
fax_num PK fax
rec_num FK
oper_from fax_num PK
oper_till oper_from
oper_till

BEFORE AFTER

Figure 8-13 Resolving a many-to-many relationship

Resolving Other Special Relationships

You may encounter other special relationships that can hamper a smooth
running database. These relationships are shown in the following list:
• Complex relationships
• Recursive relationships
• Redundant relationships
A complex relationship is an association among three or more entities. All the
entities must be present for the relationship to exist. To reduce this complex-
ity, reclassify all complex relationships as an entity, related through binary
relationships to each of the original entities.
A recursive relationship is an association between occurrences of the same
entity type. These types of relationships do not occur often. Examples of
recursive relationships are bill-of-materials (parts are composed of subparts)
and organizational structures (employee manages employees). See
Chapter 5, “Programming with SQL,” for an extended example of a recursive
relationship. You may choose not to resolve recursive relationships.

8-28 Building Your Data Model

 Normalizing Your Data Model

A redundant relationship is when two or more relationships are used to

represent the same concept. Redundant relationships add complexity to the
data model and may lead a developer to incorrectly place attributes in the
model. Redundant relationships may appear as duplicated entries in your
entity-relationship diagram. For example, you may have two entities that
contain the same attributes. To resolve a redundant relationship, review your
data model. Do you have more than one entity that contains the same
attributes? You may need to add an entity to the model to resolve the
redundancy. Chapter 10, “Tuning Your Data Model,” discusses additional
topics related to redundancy in a data model

Normalizing Your Data Model

The telephone-directory example shown in this chapter appears to be a good
model. It could be implemented at this point into a database, but it might
present problems later on with application development and data-manipula-
tion operations. Normalization is a formal approach to applying a set of rules
used in associating attributes with entities.
Normalizing your data model can do the following things:
• Produce greater flexibility in your design
• Ensure that attributes are placed in the proper tables
• Reduce data redundancy
• Increase programmer effectiveness
• Lower application maintenance costs
• Maximize stability of the data structure
Normalization consists of several steps to reduce the entities to more
desirable physical properties. These steps are called normalization rules, also
referred to as normal forms. Several normal forms exist; this chapter discusses
the first three normal forms. Each normal form constrains the data to be more
organized than the last form. Because of this, you must achieve first normal
form before you can achieve second normal form, and you must achieve
second normal form before you can achieve third normal form.

First Normal Form

An entity is in first normal form if it contains no repeating groups. In
relational terms, this means that a table is in first normal form if it contains
no repeating columns. Repeating columns make your data less flexible, waste

Building Your Data Model 8-29

Normalizing Your Data Model

disk space, and make it more difficult to search for data. In the telephone-
directory example, it appears that the name table contains repeating columns,
child1, child2, and child3, as shown in Figure 8-14:

name
rec_num lname fname bdate anniv email child1 child2 child3

repeating columns

Figure 8-14 Unnormalized name entity

Some problems can be noted in the current table. Space is always reserved on
the disk for three child records, whether the person has children or not. The
maximum number of children that you could record is three; but some of
your acquaintances might have four or more children. If you look for a
particular child, you would have to search all three columns in every row.
You can eliminate the repeating columns and bring the table to first normal
form by separating the table into two tables as shown in Figure 8-15. Put the
repeating columns into one of the tables. The association between the two
tables is established with a primary and foreign key combination. Because a
child cannot exist without an association in the name table, we can reference
the name table with a foreign key, rec_num.

name
rec_num lname fname bdate anniv email
Primary Key

child
rec_num child_name

Foreign Key

Figure 8-15 First normal form reached for name entity

In addition, the modem-name relationship is not at first normal form because

the columns b300, b1200, and b2400 are considered repeating columns. This
relationship is normalized in a similar manner as the name-child relationship,

8-30 Building Your Data Model

 Normalizing Your Data Model

as shown in Figure 8-16. A b_type attribute is added to the entity which can
contain occurrences of b300, b1200, and b2400. Figure 8-16 shows the data
model normalized through first normal form.

name child
rec_num PK rec_num FK
address
lname child_name
fname id_num PK
bdate rec_num FK
anniv street
email city
state
zipcode
voice faxname
vce_num PK fax_num PK FK
rec_num rec_num PK FK
vce_type modem
mdm_num PK
fax rec_num FK
fax_num PK b_type
operfrom
opertill

Figure 8-16 The data model of a personal telephone directory

Second Normal Form

An entity is in second normal form if it is in first normal form and all its
attributes depend on the whole (primary) key. In relational terms, this means
that every column in a table must be functionally dependent on the whole pri-
mary key of that table. Functional dependency indicates that a link exists
between the values in two different columns.
To say that the value of an attribute depends on a column means that, if the
value in the column changes, the value of the attribute must also change. The
attribute is a function of the column. The following explanations make this
more specific:
• If the table has a one-column primary key, the attribute must depend on
that key.
• If the table has a composite primary key, the attribute must depend on the
values in all of its columns taken as a whole, not on just one or some of
them.

Building Your Data Model 8-31

Normalizing Your Data Model

• If the attribute also depends on other columns, these must be columns of

a candidate key, that is, columns that are unique in every row.
If you do not convert your model to second normal form, you risk data
redundancy and difficulty in changing data. Convert first normal form tables
to second normal form tables by removing columns that are not dependent
on the whole primary key.

Third Normal Form

An entity is in third normal form if it is in second normal form and all its
attributes are nontransitively dependent on the primary key. Transitive depen-
dence means that descriptor key attributes depend not only on the whole
primary key but also on other descriptor key attributes that, in turn, depend
on the whole primary key. In SQL terms, third normal form means that no
column within a table is dependent on a descriptor column that, in turn,
depends on the primary key.
To convert to third normal form, remove attributes that depend on other
descriptor key attributes.

Summary of Normalization Rules

The following normal forms are discussed in this section:
• First normal form: A table is in first normal form if it contains no
repeating columns.
• Second normal form: A table is in second normal form if it is in first
normal form and contains only columns that are dependent on the whole
primary key.
• Third normal form: A table is in third normal form if it is in second
normal form and contains only columns that are nontransitively
dependent on the primary key.
When you follow these rules, the tables of the model are in what E.F. Codd,
the inventor of relational databases, calls third-normal form. When tables are
not in third-normal form, either redundant data exists in the model or
problems exist when you attempt to update the tables.
If you cannot find a place for an attribute that observes these rules, then you
have probably made one of the following errors:
• The attribute is not well-defined.
• The attribute is derived, not direct.
• The attribute is really an entity or a relationship.

8-32 Building Your Data Model

 Summary

• Some entity or relationship is missing from the model.

Summary
This chapter summarized and illustrated the following steps of entity-
relationship data modeling:
1. Identify and define your principal data objects, including the following
options:
• Entities
• Relationships
• Attributes
2. Diagram your data objects using the entity-relationship diagram
approach.
3. Translate your entity-relationship data objects into relational constructs.
• Determine the primary and foreign keys for each entity.
4. Resolve your relationships, particularly the following relationships:
• 1:1 relationships
• m:n relationships
• Other special relationships
5. Normalize your data model in one of the following forms:
• First normal form
• Second normal form
• Third normal form
When the process is done right, you must examine every aspect of the data
not once, but several times.
If you are interested in learning more about database design, you can attend
the Informix course, Relational Database Design. This thorough course
presents how to create an entity-relationship data model.
If you are interested in pursuing more about database design on your own,
the following excellent books are recommended:
• Database Modeling and Design, The Entity-Relationship Approach, by Toby J.
Teorey, Morgan Kauffman Publishers, Inc., 1990
• Handbook of Relational Database Design, by Candace C. Fleming and
Barbara von Halle, Addison-Wesley Publishing Company, 1989

Building Your Data Model 8-33

Summary

8-34 Building Your Data Model

 Chapter

Implementing Your
Data Model
Chapter Overview 3
Defining the Domains 3
Data Types 4
Choosing a Data Type 4
Numeric Types 7
Chronological Types 12
Character Types 15
Changing the Data Type 19
Default Values 19
Check Constraints 20
Creating the Database 20
Using CREATE DATABASE 21
Using CREATE DATABASE with INFORMIX-
OnLine Dynamic Server 21
Using CREATE DATABASE with Other Informix
Database Servers 23
Using CREATE TABLE 24
Using Command Scripts 26
Capturing the Schema 26
Executing the File 26
An Example 26
Populating the Tables 27
Summary 29

9
9-2 Implementing Your Data Model
Chapter Overview
Once a data model is prepared, it must be implemented as a database and
tables. This chapter covers the decisions that you must make to implement
the model.
The first step in implementation is to complete the data model by defining a
domain, or set of data values, for every column. The second step is to
implement the model using SQL statements.
The first section of this chapter covers defining domains in detail. The second
section shows how you create the database (using the CREATE DATABASE
and CREATE TABLE statements) and populate it with data.

Defining the Domains

To complete the data model described in Chapter 8, “Building Your Data
Model,” you must define a domain for each column. The domain of a column
describes the constraints on and identifies the set of valid values attributes (or
columns) may assume.
The purpose of a domain is to guard the semantic integrity of the data in the
model; that is, to ensure that it reflects reality in a sensible way. If a name can
be entered where a telephone number was planned or a fractional number
where an integer should be, the integrity of the data model is at risk.
You define a domain by first defining the constraints that a data value must
satisfy before it can be part of the domain. Column domains are specified
using the following constraints:
• Data types
• Default values
• Check constraints
In addition, referential constraints can be placed on columns by identifying
the primary and foreign keys in each table. How to identify these keys was
discussed in Chapter 8, “Building Your Data Model.”

Implementing Your Data Model 9-3

Defining the Domains

Data Types
The first constraint on any column is the one that is implicit in the data type
for the column. When you choose a data type, you constrain the column so
that it contains only values that can be represented by that type.
Each data type represents certain kinds of information and not others. The
correct data type for a column is the one that represents all the data values
that are proper for that column but as few as possible of the values that are
not proper for it.

Choosing a Data Type

Every column in a table must have a data type chosen from the types that the
database server supports. The choice of data type is important for the
following reasons:
• It establishes the basic domain of the column; that is, the set of valid data
items that the column can store.
• It determines the kinds of operations you can perform on the data. For
example, you cannot apply aggregate functions, such as SUM, to columns
with a character data type.
• It determines how much space each data item occupies on disk. This is
not important for small tables, but if a table has tens or hundreds of thou-
sands of rows, the difference between a 4-byte and an 8-byte type can be
crucial.

Using Data Types in Referential Constraints

Almost all data type combinations must match. For example, if a primary
key is defined as CHAR, the foreign key must also be defined as CHAR.
However, when you specify a SERIAL data type on a primary key in one
table, you specify an INTEGER on the foreign key of the relationship. The
SERIAL and INTEGER construction is the only data type combination that
you can mix in a relationship.
The decision tree shown in Figure 9-1 summarizes the choices among data
types. They are explained in the following sections.

9-4 Implementing Your Data Model

 Defining the Domains

Data items purely yes

numeric?
no
Numbers all integral? yes

no

All numbers between yes

-215 and 215-1?
no SMALLINT

All numbers between yes

-231 and 231-1?
no
INTEGER

DECIMAL(p,0)

Number of fractional yes

digits is fixed?
no
DECIMAL(p,s)

At most 8 significant yes

digits?
no
SMALLFLOAT

At most 16 significant yes

digits?
no
FLOAT

DECIMAL(p)

Figure 9-1 A diagram of the decisions to be made in choosing a data type (1 of 2)

Implementing Your Data Model 9-5

Defining the Domains

Data is chronological? yes

no Span of time or specific span

point in time?
point INTERVAL

Precise only to nearest yes

day?
no
DATE
DATETIME

Data contains yes

foreign characters?
no
No or little variance in yes
item lengths?
no
NCHAR(n)
NVARCHAR(m,r)

Data is ASCII yes

characters?
no
No or little variance in yes
item lengths?
no
Lengths under 32,511 yes
bytes?
no CHAR(n)
Length exceeds 255 yes
bytes?
no
TEXT
BYTE VARCHAR(m,r)

Figure 9-1 A diagram of the decisions to be made in choosing a data type (2 of 2)

9-6 Implementing Your Data Model

 Defining the Domains

Numeric Types
Informix database servers support eight numeric data types. Some are best
suited for counters and codes, some for engineering quantities, and some for
money.

Counters and Codes: INTEGER and SMALLINT

The INTEGER and SMALLINT data types hold small whole numbers. They are
suited for columns that contain counts, sequence numbers, numeric identity
codes, or any range of whole numbers when you know in advance the
maximum and minimum values to be stored.
Both types are stored as signed binary integers. INTEGER values have 32 bits
and can represent whole numbers from −231 through 231–1; that is, from
–2,147,483,647 through 2,147,483,647. (The maximum negative number,
–2,147,483,248, is reserved and cannot be used.)
SMALLINT values have only 16 bits. They can represent whole numbers from
–32,767 through 32,767. (The maximum negative number, -32,768, is reserved
and cannot be used.)
These data types have the following advantages:
• They take up little space (2 bytes per value for SMALLINT and 4 bytes per
value for INTEGER).
• Arithmetic expressions such as SUM and MAX as well as sort comparisons
can be done very efficiently on them.
The disadvantage to using INTEGER and SMALLINT is the limited range of
values they can store. The database server does not store a value that exceeds
the capacity of an integer. Of course, this is not a problem when you know the
maximum and minimum values to be stored.

Automatic Sequences: SERIAL

The SERIAL data type is simply INTEGER with a special feature. Whenever a
new row is inserted into a table, the database server automatically generates
a new value for a SERIAL column. A table can have only one SERIAL column.
Because the database server generates them, the serial values in new rows are
always different even when multiple users are adding rows at the same time.
This is a useful service, because it is quite difficult for an ordinary program
to coin unique numeric codes under those conditions.

Implementing Your Data Model 9-7

Defining the Domains

How Many Serialized Rows?

After inserting 231 rows in a table, the database server uses up all the
positive serial numbers. Should you be concerned about this? Probably not
because to make this happen you need to insert a row every second for 68
years. However, if it did occur, the database server would continue generat-
ing new numbers. It would treat the next-serial quantity as a signed integer.
Because it uses only positive values, it would simply wrap around and start
generating integer values beginning with 1.

The sequence of generated numbers always increases. When rows are deleted
from the table their serial numbers are not reused. This means that rows
sorted on a SERIAL column are returned in the order in which they were
created. That cannot be said of any other data type.
You can specify the initial value in a SERIAL column in the CREATE TABLE
statement. This makes it possible to generate different subsequences of
system-assigned keys in different tables. The stores6 database uses this tech-
nique. In stores6, the customer numbers begin at 101, and the order numbers
start at 1001. As long as this small business does not register more than 899
customers, all customer numbers have three digits and order numbers have
four.
A SERIAL column is not automatically a unique column. If you want to be
perfectly sure no duplicate serial numbers occur, you must apply a unique
constraint (see “Using CREATE TABLE” on page 9-24). However, if you
define the table using the interactive schema editor in DB-Access or
INFORMIX-SQL, it automatically applies a unique constraint to any SERIAL
column.
The SERIAL data type has the following advantages:
• It provides a convenient way to generate system-assigned keys.
• It produces unique numeric codes even when multiple users are updating
the table.
• Different tables can use different ranges of numbers.
It has the following disadvantages:
• Only one SERIAL column is permitted in a table.
• It can produce only arbitrary numbers (then again, arbitrary numeric
codes might not be acceptable to the database users).

9-8 Implementing Your Data Model

 Defining the Domains

Altering the Next SERIAL Number

The starting value for a SERIAL column is set when the column is created
(see “Using CREATE TABLE” on page 9-24). You can use the ALTER TABLE
statement later to reset the next value, the value that is used for the
next-inserted row.
You cannot set the next value below the current maximum value in the
column because that can lead the database server to generate duplicate
numbers. However, you can set the next value to any value higher than the
current maximum, thus creating gaps in the sequence.

Approximate Numbers: FLOAT and SMALLFLOAT

In scientific, engineering, and statistical applications, numbers are often
known to only a few digits of accuracy, and the magnitude of a number is as
important as its exact digits.
The floating-point data types are designed for these applications. They can
represent any numerical quantity, fractional or whole, over a wide range of
magnitudes from the cosmic to the microscopic. For example, they can easily
represent both the average distance from the Earth to the Sun (1.5 × 109
meters) or Planck’s constant (6.625 × 10-27). Their only restriction is their lim-
ited precision. Floating-point numbers retain only the most significant digits
of their value. If a value has no more digits than a floating-point number can
store, the value is stored exactly. If it has more digits, it is stored in
approximate form, with its least-significant digits treated as zeros.
This lack of exactitude is fine for many uses, but you should never use a
floating-point data type to record money or any other quantity for which it is
an error to change the least significant digits to zero.
Two sizes of floating-point data types exist. The FLOAT type is a
double- precision, binary floating-point number as implemented in the C lan-
guage on your computer. One usually takes up 8 bytes. The SMALLFLOAT
(also known as REAL) type is a single-precision, binary floating-point number
that usually takes up 4 bytes. The main difference between the two data types
is their precision. A FLOAT column retains about 16 digits of its values; a
SMALLFLOAT column retains only about 8 digits.
Floating-point numbers have the following advantages:
• They store very large and very small numbers, including fractional ones.
• They represent numbers compactly in 4 or 8 bytes.

Implementing Your Data Model 9-9

Defining the Domains

• Arithmetic functions such as AVG, MIN, and sort comparisons, are

efficient on these data types.
The main disadvantage of floating-point numbers is that digits outside their
range of precision are treated as zeros.

Adjustable-Precision Floating Point: DECIMAL(p)

The DECIMAL(p) data type is a floating-point type similar to FLOAT and
SMALLFLOAT. The important difference is that you specify how many signif-
icant digits it retains. The precision you write as p may range from 1 to 32,
from fewer than SMALLFLOAT up to twice the precision of FLOAT.
The magnitude of a DECIMAL(p) number ranges from 10-130 to 10124.
It is easy to be confused about decimal data types. The one under discussion
is DECIMAL(p); that is, DECIMAL with only a precision specified. The size of
DECIMAL(p) numbers depends on their precision; they occupy 1+p/2 bytes
(rounded up to a whole number, if necessary).
DECIMAL(p) has the following advantages over FLOAT:

• Precision can be set to suit the application, from highly approximate to

highly precise.
• Numbers with as many as 32 digits can be represented exactly.
• Storage is used in proportion to the precision of the number.
• Every Informix database server supports the same precision and range of
magnitudes, regardless of the host operating system.
The DECIMAL(p) data type has the following disadvantages:
• Arithmetic and sorting are somewhat slower than on FLOAT numbers.
• Many programming languages do not support the DECIMAL(p) data
format the way they support FLOAT and INTEGER. When a program
extracts a DECIMAL(p) value from the database, it may have to convert the
value to another format for processing. (However, INFORMIX-4GL
programs can use DECIMAL(p) values directly.)

Fixed-Point Numbers: DECIMAL and MONEY

Most commercial applications need to store numbers that have fixed
numbers of digits on the right and left of the decimal point. Amounts of
money are the most common examples. Amounts in U.S. and other curren-
cies are written with two digits to the right of the decimal point. Normally,

9-10 Implementing Your Data Model

 Defining the Domains

you also know the number of digits needed on the left, depending on whose
transactions are recorded—perhaps 5 digits for a personal budget, 7 for a
small business, and 12 or 13 for a national budget.
These numbers are fixed-point numbers because the decimal point is fixed at
a specific place, regardless of the value of the number. The DECIMAL(p,s) data
type is designed to hold them. When you specify a column of this type, you
write its precision (p) as the total number of digits it can store, from 1 to 32. You
write its scale (s) as the number of those digits that fall to the right of the dec-
imal point. (The relation between precision and scale is diagrammed in Fig-
ure 9-2.) Scale can be zero, meaning it stores only whole numbers. When this
is done, DECIMAL(p,s) provides a way of storing integers of up to 32 digits.

precision: 8 digits

DECIMAL(8,3) 31964.535
scale: 3 digits

Figure 9-2 The relation between precision and scale in a fixed-point number

As with the DECIMAL(p) data type, DECIMAL(p,s) takes up space in

proportion to its precision. One value occupies 1+p/2 bytes, rounded up to a
whole number of bytes.
The MONEY type is identical to DECIMAL(p,s), but with one extra feature.
Whenever the database server converts a MONEY value to characters for
display, it automatically includes a currency symbol.
The advantages of DECIMAL(p,s) over INTEGER and FLOAT are that much
greater precision is available (up to 32 digits as compared to 10 for INTEGER
and 16 for FLOAT), and both the precision and the amount of storage required
can be adjusted to suit the application.
The disadvantages are that arithmetic is less efficient and that many
programming languages do not support numbers in this form. Therefore,
when a program extracts a number, it usually must convert it to another
numeric form for processing. (However, INFORMIX-4GL programs can use
DECIMAL(p,s) and MONEY values directly.)

Implementing Your Data Model 9-11

Defining the Domains

Whose Money? Choosing a Currency Format

Each nation has its own way of punctuating amounts of money. When an
Informix database server displays a MONEY value, it refers to a currency
format defined to the operating system, usually in a variable named
DBMONEY. A currency symbol can precede or follow the amount, and the
decimal delimiter can be set to a period, a comma, or another character.
When in NLS mode, you can set the X/Open category LC_MONETARY to
specify a foreign currency symbol and format. For details, see Chapter 4 of
the Informix Guide to SQL: Reference.

Chronological Types
Informix database servers support three data types for recording time. The
DATE data type stores a calendar date. DATETIME records a point in time to
any degree of precision from a year to a fraction of a second. The INTERVAL
data type stores a span of time; that is, a duration.

Calendar Dates: DATE

The DATE data type stores a calendar date. A DATE value is actually a signed
integer whose contents are interpreted as a count of full days since midnight
on December 31, 1899. Most often it holds a positive count of days into the
current century.
The DATE format has ample precision to carry dates into the far future (58,000
centuries). Negative DATE values are interpreted as counts of days prior to
the epoch date; that is, a DATE of -1 represents the day December 30, 1899.
Because DATE values are integers, Informix database servers permit them to
be used in arithmetic expressions. For example, you can take the average of
a DATE column, or you can add 7 or 365 to a DATE column. In addition, a rich
set of functions exist specifically for manipulating DATE values. (See Chapter
1 of the Informix Guide to SQL: Syntax.)
The DATE data type is compact at 4 bytes per item. Arithmetic functions and
comparisons execute quickly on a DATE column.

Exact Points in Time: DATETIME

The DATETIME data type stores any moment in time in the era beginning
1 A.D. In fact, DATETIME is really a family of 28 data types, each with a dif-
ferent precision. When you define a DATETIME column, you specify its preci-
sion. It may contain any sequence from the list year, month, day, hour, minute,
second, and fraction. Thus, you can define a DATETIME column that stores only

9-12 Implementing Your Data Model

 Defining the Domains

M/D/Y? D-M-Y? Choosing a Date Format

You can punctuate and order the components of a date in many ways. When
an Informix database server displays a DATE value, it refers to a format
defined to the operating system, usually in a variable named DBDATE. The
numbers for day, month, and year can be shown in any order, separated by
a chosen delimiter. For more information, see Chapter 4 of the Informix
Guide to SQL: Reference.

a year, only a month and day, a date and time that is exact to the hour, or even
to the millisecond. The size of a DATETIME value ranges from 2 to 11 bytes
depending on its precision, as shown in Figure 9-3.
The advantage of DATETIME is that it can store dates more precisely than to
the nearest day, and it can store time values. Its sole disadvantage is an inflex-
ible display format, but this can be circumvented (see “Forcing the Format of
a DATETIME or INTERVAL Value” on page 9-16).

Precision Size* Precision Size*

year to year 3 day to hour 3
year to month 4 day to minute 4
year to day 5 day to second 5
year to hour 6 day to fraction(f) 5+f/2
year to minute 7 hour to hour 2
year to second 8 hour to minute 3
year to fraction (f) 8+f/2 hour to second 4
month to month 2 hour to fraction(f) 4+f/2
month to day 3 minute to minute 2
month to hour 4 minute to second 3
month to minute 5 minute to fraction(f) 3+f/2
month to second 6 second to second 2
month to fraction(f) 6+f/2 second to fraction(f) 2+f/2
day to day 2 fraction to fraction(f) 1+f/2

* When f is odd, round the size to the next full byte.

Figure 9-3 All possible precisions of the DATETIME data type, with their sizes in bytes

Implementing Your Data Model 9-13

Defining the Domains

Durations: INTERVAL
The INTERVAL data type stores a duration, that is, a length of time. The
difference between two DATETIME values is an INTERVAL, which represents
the span of time that separates them. The following examples might help to
clarify the differences:
• An employee began working on January 21, 1993 (either a DATE or a
DATETIME).
• She has worked for 254 days (an INTERVAL, the difference between the
TODAY function and the starting DATE or DATETIME).
• She begins work each day at 0900 hours (a DATETIME).
• She works 8 hours (an INTERVAL) with 45 minutes for lunch (another
INTERVAL).
• Her quitting time is 1745 hours (the sum of the DATETIME when she
begins work and the two INTERVALs).
Like DATETIME, INTERVAL is a family of types with different precisions. An
INTERVAL can represent a count of years and months; or it can represent a
count of days, hours, minutes, seconds, or fractions of seconds; 18 possible
precisions in all. The size of an INTERVAL value ranges from 2 to 12 bytes,
depending on the formulas shown in Figure 9-4.

Precision Size* Precision Size*

year(p) to year 1+p/2 hour(p) to minute 2+p/2
year(p) to month 2+p/2 hour(p) to second 3+p/2
month(p) to month 1+p/2 hour(p) to fraction(f) 4+(p+f)/2
day(p) to day 1+p/2 minute(p) to minute 1+p/2
day(p) to hour 2+p/2 minute(p) to second 2+p/2
day(p) to minute 3+p/2 minute(p) to fraction(f) 3+(p+f)/2
day(p) to second 4+p/2 second(p) to second 1+p/2
day(p) to fraction(f) 5+(p+f)/2 second(p) to fraction(f) 2+(p+f)/2
hour(p) to hour 1+p/2 fraction to fraction(f) 1+f/2

* Round a fractional size to the next full byte.

Figure 9-4 All possible precisions of the INTERVAL data type, with their sizes in bytes

INTERVAL values can be negative as well as positive. You can add or subtract
them, and you can scale them by multiplying or dividing by a number. This
is not true of either DATE or DATETIME. It is reasonable to ask, “What is one-
half the number of days until April 23?” but it is not reasonable to ask, “What
is one-half of April 23?”

9-14 Implementing Your Data Model

 Defining the Domains

Character Types
All Informix database servers support the CHAR(n) data type. When NLS has
been enabled, the Informix database servers support the NCHAR(n) data
type. INFORMIX-OnLine Dynamic Server also supports other special use
character data types.

Character Data: CHAR(n) and NCHAR(n)

The CHAR(n) data type contains a sequence of n ASCII characters. The length
n ranges from 1 to 32,767. (If you are using the INFORMIX-SE database server,
the maximum length is 32,511.) In NLS mode, the NCHAR(n) data type con-
tains a sequence of n native characters with the same length range as CHAR.
Whenever a CHAR(n) or NCHAR(n) value is retrieved or stored, exactly n
bytes are transferred. If an inserted value is short, it is extended with spaces
to make up n bytes. A value longer than n is truncated. No provision exists
for variable-length data in this format.
A CHAR(n) or NCHAR(n) value can include tabs and spaces but normally
contains no other nonprinting characters. When rows are inserted using
INSERT or UPDATE, or when rows are loaded with a utility program, no
means exists for entering nonprintable characters. However, when rows are
created by a program using embedded SQL, the program can insert any char-
acter except the null (binary zero) character. It is not a good idea to store non-
printable characters in a character column because standard programs and
utilities do not expect them.
The advantage of the CHAR(n) or NCHAR(n) data type is that it is available
on all database servers. The only disadvantage of CHAR(n) or NCHAR(n) is
that its length is fixed. When the length of data values varies widely from row
to row, space is wasted. NCHAR(n) is available on all database servers when
NLS has been enabled.

Variable-Length Strings: VARCHAR(m,r) and NVARCHAR (m,r)

Often the items in a character column have different lengths; that is, many
have an average length and only a few have the maximum length. The
VARCHAR(m,r) data type (or NVARCHAR(m, r) when NLS has been enabled)
is designed to save disk space when storing such data. A column defined as
VARCHAR(m,r) or NVARCHAR(m, r) is used just like one defined as CHAR(n).
When you define a VARCHAR(m,r) or NVARCHAR(m, r) column, you specify
m as the maximum length of a data item. Only the actual contents of each item
are stored on disk, with a 1-byte length field. The limit on m is 255; that is, a
VARCHAR(m,r) or NVARCHAR(m, r) value can store as many as 255
characters. If the column is indexed, the limit on m is 254.
Implementing Your Data Model 9-15
Defining the Domains

Forcing the Format of a DATETIME or INTERVAL Value

The database server always displays the components of an INTERVAL or
DATETIME value in the order year-month-day hour:minute:second.fraction. It
does not refer to the date format defined to the operating system, as it does
when formatting a DATE.
You can write a SELECT statement that displays the date part of a DATETIME
value in the system-defined format. The trick is to isolate the component
fields using the EXTEND function and pass them through the MDY()
function, which converts them to a DATE. The following code is a partial
example:

SELECT ... MDY (

EXTEND (DATE_RECEIVED, MONTH TO
MONTH),
EXTEND (DATE_RECEIVED, DAY TO DAY),
EXTEND (DATE_RECEIVED, YEAR TO YEAR))
FROM RECEIPTS ...

When you are designing a report using INFORMIX-4GL or INFORMIX-SQL,

you have the greater flexibility of the PRINT statement. Select each compo-
nent of a DATETIME or INTERVAL value as an expression using EXTEND.
Give each expression an alias for convenience.

SELECT ...
EXTEND (START_TIME, HOUR TO HOUR) H,
EXTEND (START_TIME, MINUTE TO MINUTE) M, ...

Then, in the report, combine the components in a PRINT expression with the
desired punctuation.:

PRINT 'Start work at ', H USING '&&', M USING '&&', 'hours.'

Start work at 0800 hours.

The second parameter, r, is an optional reserve length that sets a lower limit
on the length of an item as stored on disk. When an item is shorter than r,
r bytes are nevertheless allocated to hold it. The purpose is to save time when
rows are updated. (See “Variable-Length Execution Time” on page 9-17.)

9-16 Implementing Your Data Model

 Defining the Domains

The advantages of the VARCHAR(m,r) or NVARCHAR(m, r) data type over the

CHAR(n) or NCHAR(m, r) type are found in the following list:

• It conserves disk space when the lengths of data items vary widely or
when only a few items are longer than average.
• Queries on the more compact tables can be faster.
The following list describes its disadvantages:
• It does not allow lengths that exceed 255 characters.
• Updates of a table can be slower in some circumstances.
• It is not available with all Informix database servers.

Variable-Length Execution Time

When the VARCHAR(m,r) or NVARCHAR(m, r) data type is used, the rows of
a table have varying lengths instead of fixed lengths. This has mixed effects
on the speed of database operations.
Because more rows fit in a disk page, the database server can search the
table using fewer disk operations than if the rows were of fixed length. As
a result, queries can execute more quickly. Insert and delete operations can
be a little quicker for the same reason.
When you update a row, the amount of work the database server must do
depends on the length of the new row as compared with the old one. If the
new row is the same size or shorter, execution time is not significantly dif-
ferent than it is with fixed-length rows. But if the new row is longer than the
old one, the database server might have to perform several times as many
disk operations. Thus, updates of a table that uses VARCHAR(m,r) or NVAR-
CHAR(m, r) data can sometimes be slower than updates of a fixed-length
field.
You can mitigate this effect by specifying r as a size that covers a high
proportion of the data items. Then, most rows use the reserve length; only
a little space is wasted in padding, and updates are only slow when a
normal value is replaced with one of the rare, longer ones.

Large Character Objects: TEXT

The TEXT data type stores a block of text. It is designed to store self-contained
documents: business forms, program source or data files, or memos.
Although you can store any data in a TEXT item, Informix tools expect that a
TEXT item is printable, so this data type should be restricted to printable
ASCII text.

Implementing Your Data Model 9-17

Defining the Domains

How Blobs Are Used

Collectively, columns of type TEXT and BYTE are called binary large objects,
or blobs. The database server simply stores and retrieves them. Normally,
blob values are fetched and stored by programs written using
INFORMIX-4GL or a language that supports embedded SQL, such as
INFORMIX-ESQL/C. In such a program you can fetch, insert, or update a
blob value in a manner similar to the way you read or write a sequential file.
In any SQL statement, interactive or programmed, a blob column cannot be
used in the following ways:
• In arithmetic or Boolean expressions
• In a GROUP BY or ORDER BY clause
• In a UNIQUE test
• For indexing, either by itself or as part of a composite index
In a SELECT statement entered interactively, or in a form or report, a blob
can only perform the following functions:
• Be selected by name, optionally with a subscript to extract part of it
• Have its length returned by selecting LENGTH(column)
• Be tested with the IS [NOT] NULL predicate
In an interactive INSERT statement, you can use the VALUES clause to insert
a blob value, but the only value you can give that column is null. However,
you can use the SELECT form of the INSERT statement to copy a blob value
from another table.
In an interactive UPDATE statement, you can update a blob column to null
or to a subquery returning a blob column.

TEXT values are not stored with the rows of which they are a part. They are
allocated in whole disk pages, usually areas away from rows. (See “Locating
Blob Data” on page 10-18.)
The advantage of the TEXT data type over CHAR(n) and VARCHAR(m,r) is that
the size of a TEXT data item has no limit except the capacity of disk storage to
hold it. The disadvantages of the TEXT data type are found in the following
list:
• It is allocated in whole disk pages; a short item wastes space.
• Restrictions apply on how you can use a TEXT column in an SQL
statement. (See “How Blobs Are Used” on page 9-18.)

9-18 Implementing Your Data Model

 Defining the Domains

• It is not available with all Informix database servers.

You can display TEXT values in reports generated with INFORMIX-4GL
programs or the ACE report writer. You can display TEXT values on a screen
and edit them using screen forms generated with INFORMIX-4GL programs
or with the PERFORM screen-form processor.

Binary Objects: BYTE

The BYTE data type is designed to hold any data a program can generate:
graphic images, program object files, and documents saved by any word pro-
cessor or spreadsheet. The database server permits any kind of data of any
length in a BYTE column.
As with TEXT, BYTE data items are stored in whole disk pages in separate disk
areas from normal row data.
The advantage of the BYTE data type, as opposed to TEXT or CHAR(n), is that
it accepts any data. Its disadvantages are the same as those of the TEXT data
type.

Changing the Data Type

You can use the ALTER TABLE statement to change the data type assigned to
a column after the table is built. Although this is sometimes necessary, you
should avoid doing it for the following reasons:
• To change a data type, the database server must copy and rebuild the
table. For large tables, that can take a lot of time and disk space.
• Some data-type changes can cause a loss of information. For example,
when you change a column from a longer to a shorter character type, long
values are truncated; when you change to a less-precise numeric type,
low-order digits are truncated.
• Existing programs, forms, reports, and stored queries also might have to
be changed.

Default Values
A default value is the value inserted into a column when an explicit value is
not specified in an INSERT statement. A default value can be a literal charac-
ter string defined by you or one of the following SQL null, constant
expressions:
• USER
• CURRENT

Implementing Your Data Model 9-19

Creating the Database

• TODAY
• DBSERVERNAME
Not all columns need default values, but as you work with your data model
you may discover instances where the use of a default value saves data-entry
time or prevents data-entry error. For example, the telephone-directory
model has a State column. While looking at the data for this column, you dis-
cover that more than 50 percent of the addresses list California as the state. To
save time, you specify the string “CA” as the default value for the State
column.

Check Constraints
Check constraints specify a condition or requirement on a data value before
data can be assigned to a column during an INSERT or UPDATE statement. If
a row evaluates to false for any of the check constraints defined on a table dur-
ing an insert or update, the database server returns an error. You define a con-
straint using the CREATE TABLE or ALTER TABLE statements. For example,
the following requirement constrains the values of an integer domain to a
certain range:

Customer_Number >= 50000 AND Customer_Number <= 99999

You can express constraints on character-based domains this way using the
MATCHES predicate and the regular-expression syntax it supports. For exam-
ple, the following constraint restricts a Telephone domain to the form of a U.S.
local telephone number:

vce_num MATCHES '[2-9][2-9][0-9]-[0-9][0-9][0-9][0-9]'

For additional information about check constraints, see the CREATE TABLE
and ALTER TABLE statements in the Informix Guide to SQL: Syntax.

Creating the Database

Now you are ready to create the data model as tables in a database. You do
this with the CREATE DATABASE, CREATE TABLE, and CREATE INDEX state-
ments. The syntax of these statements is shown in detail in the Informix Guide
to SQL: Syntax. This section discusses the use of CREATE DATABASE and CRE-
ATE TABLE in implementing a data model. The use of CREATE INDEX is
covered in the next chapter.
9-20 Implementing Your Data Model
 Creating the Database

Keep in mind that the telephone-directory data model is used for illustrative
purposes only. For the sake of the example, we translate it into SQL
statements.
You might have to create the same database model more than once. However,
the statements that create the model can be stored and executed automati-
cally. See “Using Command Scripts” on page 9-26 for more information
about using this topic.
When the tables exist, you must populate them with rows of data. You can do
this manually, with a utility program, or with custom programming.

Using CREATE DATABASE

A database is a container that holds all the parts that go into a data model.
These parts include not only the tables but also views, indexes, synonyms,
and other objects associated with the database. You must create a database
before you can create anything else. Besides creating the database, the CRE-
ATE DATABASE statement establishes the kind of transaction logging to be
used. Transaction logging is discussed in Chapter 4, “Modifying Data.”
The INFORMIX-OnLine Dynamic Server database server differs from other
database servers in the way that it creates databases and tables. OnLine is
discussed first.

Using CREATE DATABASE with INFORMIX-OnLine Dynamic Server

When the OnLine database server creates a database, it sets up records that
show the existence of the database and its mode of logging. It manages disk
space directly, so these records are not visible to operating system commands.

Avoiding Name Conflicts

Normally, only one copy of OnLine is running on a machine, and it manages
the databases that belong to all users of that machine. It keeps only one list of
database names. The name of your database must be different from that of
any other database managed by that database server. (It is possible to run
more than one copy of the database server. This is sometimes done, for exam-
ple, to create a safe environment for testing apart from the operational data.
In that case, you must be sure you are using the correct database server when
you create the database, and again when you access it later.)

Implementing Your Data Model 9-21

Creating the Database

Selecting a Dbspace
OnLine offers you the option of creating the database in a particular dbspace.
A dbspace is a named area of disk storage. Ask your OnLine administrator
whether you should use a particular dbspace. The administrator can put a
database in a dbspace to isolate it from other databases or to locate it on a par-
ticular disk device. (Chapter 10, “Tuning Your Data Model” discusses
dbspaces and their relationship to disk devices.)
Some dbspaces are mirrored (duplicated on two different disk devices for high
reliability); your database can be put in a mirrored dbspace if its contents are
of exceptional importance.

Choosing the Type of Logging

OnLine offers the following four choices for transaction logging:
• No logging at all. This choice is not recommended; if the database is lost
to a hardware failure, all data alterations since the last archive are lost.

CREATE DATABASE db_with_no_log

When you do not choose logging, BEGIN WORK and other SQL statements
related to transaction processing are not permitted in the database. This
affects the logic of programs that use the database.
• Regular (unbuffered) logging. This is the best choice for most databases.
In the event of a failure, only uncommitted transactions are lost.

CREATE DATABASE a_logged_db WITH LOG

• Buffered logging. If the database is lost, a few or possibly none of the most
recent alterations are lost. In return for this small risk, performance
during alterations is improves slightly.

CREATE DATABASE buf_log_db WITH BUFFERED LOG

Buffered logging is best for databases that are updated frequently (so that
speed of updating is important), but the updates can be re-created from
other data in the event of a crash. You can use the SET LOG statement to
alternate between buffered and regular logging.

9-22 Implementing Your Data Model

 Creating the Database

• ANSI-compliant logging. This is the same as regular logging, but the ANSI
rules for transaction processing are also enforced. (See the discussion of
ANSI SQL in Chapter 1 of the Informix Guide to SQL: Reference.)

CREATE DATABASE std_rules_db WITH LOG MODE ANSI

The design of ANSI SQL prohibits the use of buffered logging.

The OnLine administrator can turn transaction logging on and off later. For
example, it can be turned off before inserting a large number of new rows.

Using CREATE DATABASE with Other Informix Database Servers

Other Informix database servers create a database as a set of one or more files
managed by the operating system. For example, under the UNIX operating
system, a database is a small group of files in a directory whose name is the
database name. (See the manual for your database server for details on how
it uses files.) This means that the rules for database names are the same as the
rules the operating system has for file names.

Choosing the Type of Logging

Other database servers offer the following three choices of logging:
• No logging at all. This choice is not recommended; if the database is lost
to a hardware failure, all data alterations since the last archive are lost.

CREATE DATABASE not_logged_db

When you do not choose logging, BEGIN WORK and other SQL statements
related to transaction processing are not permitted in the database. This
affects the logic of programs that use the database.
• Regular logging. This is the best choice for most databases. If the database
is lost, only the alteration in progress at the time of failure is lost.

CREATE DATABASE a_logged_db WITH LOG IN 'a-log-file'

You must specify a file to contain the transaction log. (The form of the file
name depends on the rules of your operating system.) This file grows
whenever the database is altered. Whenever the database files are
archived, you should set the log file back to an empty condition so it
reflects only transactions following the latest archive.

Implementing Your Data Model 9-23

Creating the Database

• ANSI-compliant logging. This is the same as regular logging, but the ANSI
rules for transaction processing are also enabled. (See the discussion of
ANSI-compliant databases in Chapter 1 of the Informix Guide to SQL: Ref-
erence.)

CREATE DATABASE std_rules_db WITH LOG IN 'a-log-file' MODE

ANSI

You can add a transaction log to a nonlogged database later using the START
DATABASE statement.

Using CREATE TABLE

Use the CREATE TABLE statement to create each table you designed in the
data model. This statement has a complicated form, but it is basically a list of
the columns of the table. For each column, you supply the following
information:
• The name of the column
• The data type (from the domain list you made)
• If the column (or columns) is a primary key, the constraint PRIMARY KEY
• If the column (or columns) is a foreign key, the constraint FOREIGN KEY
• If the column is not a primary key and should not allow nulls, the
constraint NOT NULL
• If the column is not a primary key and should not allow duplicates, the
constraint UNIQUE
• If the column has a default value, the constraint DEFAULT
• If the column has a check constraint, the constraint CHECK
In short, the CREATE TABLE statement is an image in words of the table as
you drew it in the data model diagram. Figure 9-5 shows the statements for
the telephone-directory model whose diagram is shown in Figure 8-16 on
page 8-31.

9-24 Implementing Your Data Model

 Creating the Database

CREATE TABLE name

(
rec_num SERIAL PRIMARY KEY,
lname CHAR(20),
fname CHAR(20),
bdate DATE,
anniv DATE,
email VARCHAR(25)
);

CREATE TABLE child

(
child CHAR(20),
rec_num INT,
FOREIGN KEY (rec_num) REFERENCES NAME (rec_num)
);

CREATE TABLE address

(
id_num SERIAL PRIMARY KEY,
rec_num INT,
street VARCHAR (50,20),
city VARCHAR (40,10),
state CHAR(5) DEFAULT ’CA’,
zipcode CHAR(10),
FOREIGN KEY (rec_num) REFERENCES name (rec_num)
);

CREATE TABLE voice

(
vce_num CHAR(13) PRIMARY KEY,
vce_type CHAR(10),
rec_num INT,
FOREIGN KEY (rec_num) REFERENCES name (rec_num)
);

CREATE TABLE fax

(
fax_num CHAR(13),
operfrom DATETIME HOUR TO MINUTE,
opertill DATETIME HOUR TO MINUTE,
PRIMARY KEY (fax_num)
);

CREATE TABLE faxname

(
fax_num CHAR(13),
rec_num INT,
PRIMARY KEY (fax_num, rec_num),
FOREIGN KEY (fax_num) REFERENCES fax (fax_num),
FOREIGN KEY (rec_num) REFERENCES name (rec_num)
);

CREATE TABLE modem

(
mdm_num CHAR(13) PRIMARY KEY,
rec_num INT,
b_type CHAR(5),
FOREIGN KEY (rec_num) REFERENCES name (rec_num)
);

Figure 9-5 The CREATE TABLE statements for the telephone-directory data model

Implementing Your Data Model 9-25

Creating the Database

Using Command Scripts

You can create the database and tables by entering the statements
interactively. But what if you have to do it again or several more times?
You might have to do it again to make a production version after a test
version is satisfactory. You might have to implement the same data model on
several machines. To save time and reduce the chance of errors, you can put
all the commands to create a database in a file and execute them
automatically.

Capturing the Schema

You can write the statements to implement your model into a file yourself.
However, you can also have a program do it for you. See Chapter 5 of the
Informix Guide to SQL: Reference. It documents the dbschema utility, a pro-
gram that examines the contents of a database and generates all the SQL state-
ments required to re-create it. You can build the first version of your database
interactively, making changes until it is exactly as you want it. Then you can
use dbschema to generate the SQL statements necessary to duplicate it.

Executing the File

DB-Access or INFORMIX-SQL, the programs you use to enter SQL statements
interactively, can be driven from a file of commands. The use of these prod-
ucts is covered in the DB-Access User Manual or the INFORMIX-SQL User
Guide. You can start DB-Access or INFORMIX-SQL and have them read and
execute a file of commands prepared by you or by dbschema.

An Example
Most Informix database server products come with a demonstration
database called stores6 (the one used for most of the examples in this book).
The stores6 database is delivered as an operating system command script
that calls Informix products to build the database. You can copy this
command script and use it as the basis for automating your own data model.

9-26 Implementing Your Data Model

 Creating the Database

Populating the Tables

For your initial tests, it is easiest to populate the tables interactively by typing
INSERT statements in DB-Access or INFORMIX-SQL. As shown in the follow-
ing example, to insert a row into the manufact table of the stores6 database
in DB-Access, you enter the following command:

INSERT INTO manufact VALUES ('MKL', 'Martin', 15)

Or, if you are preparing an application program in INFORMIX-4GL or another

language, you can use the program to enter rows.
Often, the initial rows of a large table can be derived from data held in tables
in another database or in operating system files. You can move the data into
your new database in a bulk operation. If the data is in another Informix
database, you can retrieve it in several ways.
If you are using INFORMIX-OnLine Dynamic Server, you can simply select
the data you want from the other database on another database server as part
of an INSERT statement in your database. As shown in the following example,
you could select information from the items table in the stores6 database to
insert into a new table:

INSERT INTO newtable

SELECT item_num, order_num, quantity, stock_num,
manu_code, total_price
FROM stores6@otherserver:items

Implementing Your Data Model 9-27

Creating the Database

If you are using INFORMIX-SE, you can select the data you want from one
database and insert it into another database, as long as the databases are on
the same database server. As shown in the following example, you could
select information from the catalog table in the stores6 database to insert into
a new table by using a temporary table:

CONNECT TO 'sharky/db1';

SELECT catalog_num, stock_num, manu_code

FROM catalog
INTO temptable;

DISCONNECT;

CONNECT TO 'sharky/db2';

SELECT * from temptable

INTO newsetable;

If you want to select data from another database server in INFORMIX-SE, you
must export the data to a file. You can use the UNLOAD statement in
DB-Access, INFORMIX-SQL, or INFORMIX-4GL, or you can write a report in
ACE or INFORMIX-4GL and direct the output to a file.

When the source is another kind of file or database, you must find a way to
convert it into a flat ASCII file; that is, a file of printable data in which each line
represents the contents of one table row.
After you have the data in a file, you can load it into a table using the dbload
utility. Read about dbload in Chapter 5 of the Informix Guide to SQL: Reference.
The LOAD statement in DB-Access, INFORMIX-SQL, or INFORMIX-4GL can
also load rows from a flat ASCII file. See Chapter 1 of the Informix Guide to
SQL: Syntax for information about the LOAD and UNLOAD statements.
Inserting hundreds or thousands of rows goes much faster if you turn off
transaction logging. No point exists in logging these insertions because in the
event of a failure you can easily re-create the lost work. The following list
contains the steps of a large bulk-load operation:
• If any chance exists that other users are using the database, exclude them
with the DATABASE EXCLUSIVE statement.
• If you are using INFORMIX-OnLine Dynamic Server, ask the
administrator to turn off logging for the database.
The existing logs can be used to recover the database to its present state,
while the bulk insertion can be run again to recover those rows if they are
lost soon after.
9-28 Implementing Your Data Model
 Summary

• Perform the statements or run the utilities that load the tables with data.
• Archive the newly loaded database.
If you are using INFORMIX-OnLine Dynamic Server, either ask the
administrator to perform a full or incremental archive, or use the
onunload utility to make a binary copy of your database only.
If you are using other database servers, use operating system commands
to back up the files that represent the database.
• Restore transaction logging and release the exclusive lock on the
database.
You can enclose the steps of populating a database in a script of operating
system commands. You can automate the INFORMIX-OnLine Dynamic
Server administrator commands by invoking the command-line equivalents
to ON-Monitor.

Summary
This chapter covered the following work, which you must do to implement
a data model:
• Specify the domains, or constraints, that are used in the model, and
complete the model diagram by assigning constraints to each column.
• Use interactive SQL to create the database and the tables in it.
• If you must create the database again, write the SQL statements to do so
into a script of commands for the operating system.
• Populate the tables of the model, first using interactive SQL and then by
bulk operations.
• Possibly write the bulk-load operation into a command script so you can
repeat it easily.
You can now use and test your data model. If it contains very large tables, or
if you must protect parts of it from certain classes of users, more work
remains to be done. That is the subject of Chapter 10, “Tuning Your Data
Model.”

Implementing Your Data Model 9-29

Summary

9-30 Implementing Your Data Model

 Chapter

10
Tuning Your Data
Model
Chapter Overview 3
INFORMIX-OnLine Dynamic Server Disk Storage 3
Chunks and Pages 4
Dbspaces and Blobspaces 4
Disk Mirroring 5
Databases 5
Tables and Spaces 6
Exploiting Mirroring 6
Sharing Temporary Space 6
Assigning Dedicated Hardware 6
Reducing Contention Between Programs Using the
Same Table 7
Tblspaces 7
Extents 8
Choosing Extent Sizes 8
Upper Limit on Extents 10
Reorganizing dbspaces and Tables 10
Calculating Table Sizes 12
Estimating Fixed-Length Rows 13
Estimating Variable-Length Rows 14
Estimating Index Pages 16
Estimating Blobpages 17
Locating Blob Data 18
Managing Indexes 19
Space Costs of Indexes 20
Time Costs of Indexes 20
 Choosing Indexes 21
Join Columns 21
Selective Filter Columns in Large Tables 22
Order-By and Group-By Columns 22
Duplicate Keys Slow Index Modifications 22
Dropping Indexes 24
Clustered Indexes 25
Denormalizing 26
Shorter Rows for Faster Queries 26
Expelling Long Strings 26
Using VARCHAR Strings 26
Changing Long Strings to TEXT 27
Building a Symbol Table of Repeated Strings 27
Moving Strings to a Companion Table 28
Splitting Wide Tables 28
Division by Bulk 28
Division by Frequency of Use 28
Division by Frequency of Update 29
Costs of Companion Tables 29
Splitting Tall Tables 29
Redundant and Derived Data 30
Adding Derived Data 30
Adding Redundant Data 31
Maximizing Concurrency 32
Easing Contention 32
Rescheduling Modifications 33
Using an Update Journal 33
Isolating and Dispersing Updates 35
Splitting Tables to Isolate Volatile Columns 35
Dispersing Bottleneck Tables 35
Summary 36

10-2 Tuning Your Data Model

Chapter Overview
The preceding chapters summarized the steps of creating and implementing
a data model that is theoretically sound. Normally, the resulting database
also yields adequate performance. If it does, the job is done.
However, some applications have more stringent performance requirements.
Some databases contain extremely large tables, others must be usable by
many programs concurrently or by programs that operate under tight
requirements for response time, and others must be extremely reliable.
You can tune your database design to meet such requirements. In some cases,
you might have to sacrifice theoretical correctness in the data model to get the
performance you need.
Before you make any changes, however, make sure that the queries are as
efficient as possible. Study Chapter 13, “Optimizing Your Queries,” to see
what you can do to improve performance without changing the data model.
The remainder of this chapter contains discussions on the following topics:
• The disk storage methods used by INFORMIX-OnLine Dynamic Server
and their implications for performance and reliability
• The methods for calculating the sizes of tables and indexes, and using
those numbers to estimate query-execution time
• The benefits and costs of adding indexes
• The ways you can “denormalize” the data model, that is, spoil its
theoretical purity to improve performance
• Suggestions for increasing concurrency among multiple programs

INFORMIX-OnLine Dynamic Server Disk Storage

OnLine manages disk storage directly. To the greatest extent possible, it
bypasses the host operating system and works with the raw disk space. (The
extent to which this is possible depends on the operating system.)

Tuning Your Data Model 10-3

INFORMIX-OnLine Dynamic Server Disk Storage

device 0x7f
device 0x7d blobspace rasters
host file system

device 0x7c device 0x7e

dbspace large

dbspace root dbspace temps

Figure 10-1 The relationship between chunks and dbspaces

Its methods of disk management are not complicated. However, several

terms are used for different amounts of space in varying contexts: the chunk,
the dbspace (or blobspace), the page, the extent, and the tblspace.

Chunks and Pages

The basic unit of physical storage is the chunk. A chunk is a unit of disk
storage that is dedicated to OnLine. Usually it is a physical device, that is, a
disk drive. However, a chunk can be part of one disk or simply a file. In any
case, a chunk is a unit of space that can be identified to the operating system
as being in the exclusive control of OnLine.
The page is the basic unit of disk input/output (I/O). All space in every chunk
is divided into pages. All I/O is done in units of whole pages. The size of a
page is the same in all chunks used for tables. It is set when an OnLine system
is installed.

Dbspaces and Blobspaces

A space always comprises one or more complete chunks. Often a space
contains exactly one chunk, so that the space and the chunk are identical.
However, when a space comprises more than one chunk, OnLine makes the
multiple chunks appear to be a single sequence of pages.
A space is either a dbspace or a blobspace, depending on its use. If a space
contains databases and tables, it is a dbspace. If it is dedicated to holding
blobs (the TEXT and BYTE data types), it is a blobspace. Chunks are assigned
to a space when it is created. Chunks also can be added to a space later.

10-4 Tuning Your Data Model

 INFORMIX-OnLine Dynamic Server Disk Storage

One dbspace, the root dbspace, is created first and always exists. It is the most
important space because it holds the control information that describes all
other spaces and chunks.
Parts of a single database can appear in two or more dbspaces. However, a
single table is always completely contained in one dbspace.

Disk Mirroring
Individual spaces can be mirrored. The chunks of a mirrored space are paired
with other chunks of equal size. Whenever a page is written to one of the
chunks, it is also written to the mirror chunk. When one chunk suffers a hard-
ware failure, the database server uses its mirror to continue processing with-
out interruption. When the broken chunk is restored to service (or when a
substitute chunk is assigned), OnLine automatically brings it back to equality
with the working chunk and continues operations.
When any dbspaces are mirrored, the root dbspace should also be mirrored.
Otherwise, if it is lost to hardware failure, all OnLine data is unusable
regardless of mirrors.
If your database has extreme requirements for reliability in the face of
hardware failure, you should place it in a mirrored dbspace. As indicated in
“Tables and Spaces” on page 10-6, it is possible to locate individual tables in
particular dbspaces. Hence, you can place some tables in mirrored spaces and
other tables in normal spaces.

Databases
A database resides initially in one dbspace. It is placed there by a parameter
of the CREATE DATABASE statement. The following example creates a
database in the dbspace named dev0x2d:

CREATE DATABASE reliable IN dev0x2d WITH BUFFERED LOG

When no dbspace is specified, a database is placed in the root dbspace. When

a database is in a dbspace it means only that the following conditions have
been met:
• Its system catalog tables are stored in that dbspace
• That dbspace is the default location for tables that are not explicitly
created in other dbspaces

Tuning Your Data Model 10-5

INFORMIX-OnLine Dynamic Server Disk Storage

Tables and Spaces

A table resides completely in one dbspace. (Its blob values can reside in
separate blobspaces or on an optical storage subsystem.) If no dbspace is
specified, a table resides in the dbspace where its database resides. The
following partial example creates a table in the dbspace named misctabs:

CREATE TABLE taxrates (…column specifications…) IN misctabs

You can achieve many aims by placing a table in a specific dbspace. Some of
these aims are explored in the following paragraphs.

Exploiting Mirroring
Place all the tables used by a critically important application in a mirrored
dbspace. Alternatively, create the database in the mirrored dbspace and let
the important tables reside there by default. Tables that are part of the data-
base, but that are not used by the critical programs, can be located in
nonmirrored spaces.

Sharing Temporary Space

When databases are large and disk space is limited, the normal dbspaces
might not have enough room for those large temporary tables that can be
helpful in improving query performance. However, a temporary table exists
only for the life of the program that creates it—even less time if the program
drops the table promptly. Set up a single dbspace for temporary tables, so it
can be used repeatedly by many programs.

Assigning Dedicated Hardware

A dbspace can equal a chunk, and a chunk can equal a device. Thus, you can
place a table on a disk device that is dedicated to its use. When disk drives
have different performance levels, you can put the tables with the highest
frequency of use on the fastest drives.
By giving a high-use table a dedicated access arm, you can reduce contention
with applications that use other tables. This does not reduce contention
between programs using the same table unless the table itself is spread across
multiple devices, as suggested in the next section.

10-6 Tuning Your Data Model

 INFORMIX-OnLine Dynamic Server Disk Storage

Reducing Contention Between Programs Using the Same Table

A dbspace can comprise multiple chunks, and each chunk can represent a
different disk. This permits you to assign multiple disk-access arms to one
table. Figure 10-2 shows a dbspace distributed over multiple disks.

device 0x21 device 0x27 device 0x22

dbspace three_arms

Figure 10-2 A dbspace distributed over three disks

The dbspace named three_arms comprises chunks on three different disks.

Each chunk is approximately one-third the size of an important, heavily used
table. (The third chunk is larger to allow for growth.) When multiple pro-
grams query the table, their I/O operations are distributed over the three
disks, reducing contention for the use of the hardware.
In an arrangement like this, the multiple disks can act as mirrors for each
other. For example, device 0x21 can be mirrored on 0x27, 0x27 on 0x22, and
0x22 on 0x21. As long as any two devices are operational, the dbspace is
usable. (However, if a single piece of hardware, such as a disk controller, is
common to all three disks, this type of mirror arrangement is still vulnerable
to the failure of that component.)

Tblspaces
The total of all disk space allocated to a table is the tblspace of the table. The
tblspace includes pages allocated to data (rows of that table) and pages allo-
cated to indexes. It also includes pages used by blob columns that are located
in the tblspace but not pages used by blob data in a separate blobspace or
blob data stored on an optical storage subsystem. (This choice is discussed
under “Locating Blob Data” on page 10-18.)
The tblspace is an accounting entity only; it does not correspond to any
particular part of a dbspace. The indexes and data extents that make up a
table can be scattered throughout the dbspace.

Tuning Your Data Model 10-7

INFORMIX-OnLine Dynamic Server Disk Storage

The oncheck utility (with the -pt option) returns information on the status of
tblspaces, including the counts of pages allocated and in use for each one.
You can use this information to monitor the growth of a table over time.

Extents
As you add rows to a table, OnLine allocates disk space to it in units called
extents. Each extent is a block of physically contiguous pages from the
dbspace. Even when the dbspace comprises more than one chunk, extents are
always allocated entirely within a single chunk to remain contiguous.
Contiguity is important to performance. When the pages of data are
contiguous, disk-arm motion is minimized when the database server reads
the rows sequentially. The mechanism of extents is a compromise between
the following competing requirements:
• Most dbspaces are shared among several tables.
• The size of some tables is not known in advance.
• Tables can grow at different times and different rates.
• All the pages of a table should be adjacent for best performance.
Because table sizes are not known, table space cannot be preallocated.
Therefore, extents are added only as they are needed, but all the pages in any
one extent are contiguous for better performance.

Choosing Extent Sizes

When you create a table, you can specify the size of the first extent as well as
the size of the extents to be added as the table grows. You can change the size
of added extents later with the ALTER TABLE statement. The following partial
example creates a table with a half-megabyte initial extent and 100-kilobyte
added extents:

CREATE TABLE big_one (…column specifications…)

IN big_space
EXTENT SIZE 512
NEXT SIZE 100

The following example changes the next-extent size of the table to 50

kilobytes. This has no effect on extents that already exist.

ALTER TABLE big_one MODIFY NEXT SIZE 50

10-8 Tuning Your Data Model

 INFORMIX-OnLine Dynamic Server Disk Storage

The next-extent sizes of the following kinds of tables are not very important
to performance:
• A small table has only one extent (otherwise it would not be small). If it is
heavily used, large parts of it are buffered in memory anyway.
• An infrequently used table is not important to performance no matter
what size it is.
• A table that resides in a dedicated dbspace always receives new extents
that are adjacent to its old extents. The size of these extents is not
important because, being adjacent, they perform like one large extent.
When you assign an extent size to these kinds of tables, the only
consideration is to avoid creating large numbers of extents. A large number
of extents causes the database server to spend a small amount of extra time
on bookkeeping. In addition, an upper limit exists on the number of extents
allowed. (This is covered in the section “Upper Limit on Extents” on
page 10-10.)
Next-extent sizes become important when two or more large and growing
tables share a dbspace. Because the extents added to the different tables are
interleaved, each new extent represents another long seek that must be made
when reading sequentially. It also extends the total distance over which the
disk arm must range when reading nonsequentially.
No upper limit exists on extent sizes except the size of the dbspace. When you
know the final size of a table (or can confidently predict it within 25 percent),
allocate all its space in the initial extent. When tables grow steadily to
unknown size, assign them next-extent sizes that let them share the dbspace
with a small number of extents each. The following list contains one possible
approach:
• Decide on the ratio in which the dbspace is shared among the tables. For
example, you might divide the dbspace among three tables in the ratio
0.4 : 0.2 : 0.3 (reserving 10 percent for small tables and overhead).
• Give each table one-fourth of its share of the dbspace as its initial extent.
• Assign each table one-eighth of its share as its next-extent size.
Monitor the growth of the tables regularly using oncheck.
What happens if, as the dbspace fills up, you do not have enough contiguous
space to create an extent of the size you specified? In that case, OnLine
allocates the largest contiguous extent that it can.

Tuning Your Data Model 10-9

INFORMIX-OnLine Dynamic Server Disk Storage

Upper Limit on Extents

No table should be allowed to acquire a large number of extents. But it can
happen, and an upper limit exists on the number of extents allowed. Trying
to add an extent after the limit is reached causes ISAM error -136 (No more
extents) to follow an INSERT request.
The upper limit depends on the page size and the table definition. To learn
the upper limit on extents for a particular table, calculate a series of values as
shown in the following list:
pagesize = the size of a page (reported by onstat -c under the head
BUFFSIZE)
colspace = 4 × the number of columns in the table
ixspace = 12 × the number of indexes on the table
ixparts = 4 × the number of columns named in each index
extspace = pagesize - (colspace + ixspace + ixparts + 84)
The table can have no more than extspace/8 extents. When the page size is
2,048, the customer table from the demonstration database can have no more
than 234 extents, as calculated in Figure 10-3.
To help ensure that the limit is not exceeded, OnLine checks the number of
extents each time it creates a new one. If the extent being created is the 16th
or a multiple of 16, the next-extent size for the table is automatically doubled.

Variable Value Basis

pagesize 2,048 output of onstat
colspace 40 ten columns
ixspace 36 three indexes
ixparts 12 one column in each index
extspace 1,876 2,048 - (40 + 36 + 12 + 84)
limit 234 extspace/8

Figure 10-3 Values used to determine upper limit on extents

Reorganizing dbspaces and Tables

Query performance can suffer after many extents are added to a table. This
can only be the case when two or more large, growing tables share a dbspace.
When tables grow at the same time, their new extents and index pages are
interleaved, creating large gaps between the extents of any one table. This
situation is shown in Figure 10-4.

10-10 Tuning Your Data Model

 INFORMIX-OnLine Dynamic Server Disk Storage

table 1

table 2
table 3

Figure 10-4 A dbspace with interleaved tables

Figure 10-4 also shows gaps of unused space, perhaps created by tables that
existed when extents were allocated but that have since been dropped.
Disorganization such as that depicted in Figure 10-4 hurts performance in
two ways. During sequential access to any table, several long seeks must be
made. For nonsequential access to any table, the disk might have to seek
across the entire width of the dbspace. It is possible to rebuild a dbspace so
that tables are compact once again, as shown in Figure 10-5. The relative
order of the reorganized tables within the dbspace is not important; all that
matters is that the pages of each reorganized table are together. When reading
a table sequentially, no long seeks occur. When reading a table
nonsequentially, the disk arm ranges only over the space occupied by that
table.

table 1

table 2
table 3

Figure 10-5 A dbspace reorganized to make tables compact

Use the following steps to reorganize a dbspace:

1. Copy the tables in the dbspace to tape individually, using the onunload
utility.
2. Drop all the tables in the dbspace.
3. Re-create the tables using the onload utility.
The onload utility re-creates the tables with the identical properties they had
before, including the same extent sizes. When a new extent is created
adjacent to the previous extent, the two are treated as a single extent.

Tuning Your Data Model 10-11

Calculating Table Sizes

You can also unload a table using the UNLOAD statement in DB-Access,
INFORMIX-SQL, or INFORMIX-4GL, and reload the table using the compan-
ion LOAD statement or the dbload utility. However, these operations convert
the table into character form, while onload and onunload work with binary
copies of disk pages.
You can reorganize a single table in two more ways. If you use the ALTER
TABLE statement to add or drop a column or to change the data type of a col-
umn, the table is copied and reconstructed. If you create a clustered index or
alter an index to cluster, the table is sorted and rewritten. (See “Clustered
Indexes” on page 10-25.) In both cases, the table is written on other areas of
the dbspace. However, if other tables are in the dbspace, no guarantee exists
that all the new extents are adjacent.

Calculating Table Sizes

This section discusses methods for calculating the approximate sizes of tables
and indexes in disk pages. As in the previous sections, this information
applies only to INFORMIX-OnLine Dynamic Server. Estimates of this kind
are useful when you are planning the layout of a database on disk.
These calculations aim to estimate the number of disk pages that are used to
hold different parts of a table. If the table already exists, or if you can build a
demonstration table of realistic size using simulated data, you do not have to
make estimates. You can run oncheck and obtain exact numbers.
You can calculate a series of values to produce the estimates. You need the
following values in all estimates. Let the values be as follows:
estrows = the number of rows you estimate the table has when it reaches
its normal size.
pagesize = the size of a page (reported by onstat -c under the head
BUFFSIZE).
pageuse = pagesize - 28 (the space available for data on a disk page).
In the calculations in this chapter, floor(x) means the largest integer smaller
than x, and ceiling(x) means the smallest integer larger than x.

10-12 Tuning Your Data Model

 Calculating Table Sizes

Estimating Fixed-Length Rows

When a table contains no VARCHAR columns, its rows all have the same size.
Calculate the following additional variables. Let the values be as shown in
the following list:
rowsize = the sum of the sizes of all columns in the table, plus 4 bytes.
Sizes of different data types are discussed in Chapter 9,
“Implementing Your Data Model.” Treat TEXT and BYTE
columns as having size 56.
homerow = if (rowsize ≤ pageuse) then rowsize
else 4 + remainder (rowsize/pageuse)
If a row is larger than a page, as many full pages as possible are
removed from the end of the row and stored in separate expan-
sion pages. Let homerow be the size of the leading part, which is
kept on the home page. The value homerow is simply rowsize if
a row fits a page.
overpage = if (rowsize ≤ pageuse) then 0
else floor (rowsize/pageuse)
Let overpage be the number of expansion pages needed for each
row; it is zero when a row fits on a page.
datrows = min (255,floor [pageuse/homerow])
Let datrows be the number of rows (or leading parts of rows)
that fits on one page. An upper limit exists of 255 rows per
page, even when rows are very short.
datpages = ceiling (estrows/datrows)
expages = estrows × overpage
The table requires approximately datpages + expages disk pages for rows.
(Additional pages for TEXT and BYTE columns are estimated in “Estimating
Blobpages” on page 10-17.) An estimate for the customer table is shown in
Figure 10-6.

Tuning Your Data Model 10-13

Calculating Table Sizes

Variable Estimate Basis of Estimate

estrows 1,500 company business plan
pagesize 2,048 output of onstat
pageuse 2,020 pagesize - 28
rowsize 138 sum of 4+ column data type size
customer_num SERIAL 4
fname CHAR(15) 15
lname CHAR(15) 15
company CHAR(20) 20
address1 CHAR(20) 20
address2 CHAR(20) 20
city CHAR(15) 15
state CHAR(2) 2
zipcode CHAR(5) 5
phone CHAR(18) 18
homerow 134 rowsize ≤ pageuse
overpage 0 rowsize ≤ pageuse
datrows 14 floor (pageuse/homerow)
datpages 108 ceiling (estrows/datrows)
expages 0 estrows × overpage

Figure 10-6 Estimating the size of the customer table

Estimating Variable-Length Rows

When a table contains one or more VARCHAR columns (or NVARCHAR
columns, when NLS has been enabled), its rows have varying lengths. This
introduces uncertainty. You must form an estimate of the typical size of each
VARCHAR column, based on your understanding of the data.

When INFORMIX-OnLine Dynamic Server allocates space to rows of varying

size, it considers a page to be full when no room exists for an additional row
of the maximum size. This can limit the allocation of rows to pages. Calculate
the following variables:
maxrow = the sum of the maximum sizes of all columns in the table, plus
4 bytes. Treat TEXT and BYTE columns as having size 56.
typrow = the sum of the estimated typical sizes of all columns in the
table, plus 4 bytes.
homerow = if (typrow ≤ pageuse) then typrow
else 4 + remainder (typrow/pageuse)
Let homerow be the amount of a typical row that fits on the
home page; it is typrow if one page is sufficient.

10-14 Tuning Your Data Model

 Calculating Table Sizes

overpage = if (typrow ≤ pageuse) then 0

else floor (typrow/pageuse)
Let overpage be the number of expansion pages needed for
the typical row; it is zero when a typical row fits on a page.
homemax = if (maxrow ≤ pageuse) then maxrow
else 4 + remainder (maxrow/pageuse)
Let homemax be the reserve requirement of the database
server.
datrows = floor ((pageuse − homemax)/typrow)
Let datrows be the number of typical rows that fits on a page
before the database server decides the page is full.
datpages = ceiling (estrows/datrows)
expages = estrows × overpage
The table requires approximately datpages + expages disk pages for rows.
(Additional pages for TEXT and BYTE columns are estimated in “Estimating
Blobpages” on page 10-17.) An estimate for the catalog table is shown in
Figure 10-6.

Variable Estimate Basis of Estimate

estrows 5,000 company business plan.
pagesize 2,048 output of onstat
pageuse 2,020 pagesize -28
maxrow 381 sum of 4+:
column data type max size typical
catalog_num SERIAL 4 4
stock_num SMALLINT 2 2
manu_code CHAR(3) 3 3
cat_descr TEXT 56 56
cat_picture BYTE 56 56
cat_adv VARCHAR(255,65) 256 66
typrow 191
homerow 191 typrow ≤ pageuse
overpage 0 typrow ≤ pageuse
homemax 381 maxrow ≤ pageuse
datrows 8 floor ((pageuse - homemax)/typrow)
datpages 625 ceiling (estrows/datrows)
expages 0 estrows × overpage

Figure 10-7 Estimating the size of the catalog table

Tuning Your Data Model 10-15

Calculating Table Sizes

Estimating Index Pages

The tblspace includes index pages as well as pages of data, and index pages
can be a significant fraction of the total.
An index entry consists of a key and a pointer. The key is a copy of the indexed
column or columns from one row of data. The pointer is a 4-byte value used
to locate the row containing that key value.
A unique index contains one such entry for every row in the table. It also
contains some additional pages of pointers that create the B+ tree structure.
(Index structure is described in “The Cost of Network Access” on
page 13-20.) When an index allows duplicates, it contains fewer keys than
pointers; that is, one key can be followed by a list of pointers.
If this were the whole story on indexes, it would be simple to estimate their
space requirements. However, the index leaf pages use a technique called key
compression, which allows the database server to put more keys on a page
than would otherwise fit. The space saved by key compression varies with
the content of the keys, from page to page and index to index.
To estimate the size of a compact index (disregarding space saved by key
compression), let the values be as shown in the following list:
keysize = the total width of the indexed column or columns.
pctuniq = the number of unique entries you expect in this index, divided
by estrows.
For a unique index, or one with only occasional duplicate
values, use 1.0. When duplicates are present in significant
numbers, let pctuniq be a fraction less than 1.0.
entsize = keysize + (5/pctuniq)
pagents = floor (pageuse/entsize)
Approximately pagents entries occur on each index page.
leaves = ceiling (estrows/pagents)
Approximately leaves leaf pages occur in the index.
branches = ceiling (leaves/pagents)
There are approximately branches nonleaf pages.
The index contains approximately leaves + branches pages when it is compact.
Estimates for two indexes in the customer table are carried out in Figure 10-8
on page 10-17.

10-16 Tuning Your Data Model

 Calculating Table Sizes

As rows are deleted and new ones inserted, the index can become sparse; that
is, the leaf pages might no longer be full of entries. On the other hand, if key
compression is effective, the index might be smaller. The method given here
should yield a conservative estimate for most indexes. If index space is
important, build a large test index using real data and check its size with the
oncheck utility.

Variable Estimate Basis for Estimate

estrows 1,500 company business plan
pagesize 2,048 output of onstat
pageuse 2,016 pagesize

keysize 4 customer_num is 4 bytes

pctuniq 1.0 unique index
entsize 12 4 + (5/1)
pagents 168 floor (2,016/12)
leaves 9 ceiling (1,500/168)
branches 1 ceiling (9/168)

keysize 2 state is CHAR(2)

pctuniq 0.033 50 states among 1,500 rows
entsize 8.066 2 + (5/0.033)
pagents 249 floor (2,016/8.066)
leaves 7 ceiling (1,500/249)
branches 1 ceiling (7/168)

Figure 10-8 Estimating the size of two indexes from the customer table

Estimating Blobpages
BYTE and TEXT data items that are stored on magnetic disk are stored in
separate pages, either interspersed among the row and index pages in the
tblspace or in a separate blobspace. (This discussion does not apply to blob
data stored on optical media.) Each blob data item occupies a whole number
of pages. For each blob column, let the values be as shown in the following
list:
typage = the number of whole pages required to store an item of
typical size.
nnpart = the fraction of rows in which this column has a non-null
value.
totpage= typage × estrows × nnpart.

Tuning Your Data Model 10-17

Calculating Table Sizes

The values for the column occupy approximately totpage pages. In the catalog
table (see Figure 10-6 on page 10-14) two blobpages occur. Their estimates
can be imagined as described in the following paragraphs:
For the cat_descr column, the text of a description is, at most, a
double-spaced page, which is 250 words or approximately 1,500 characters,
so typage = 1. A description exists for every entry, so nnpart = 1.0.
Thus, totpage = 1 × 5,000 × 1.0 = 5,000 pages of data.
The cat_picture column contains line art in a computer-readable form.
Examining several of these pictures reveals that they vary widely in size but
a typical file contains about 75,000 bytes of data. Thus typage = 38. The
marketing department estimates that they want to store a picture with one
entry in four: nnpart = 0.25. Therefore totpage = 38 × 5,000 × 0.25 = 47,500 pages
of data.
After a table is built and loaded, you can check the use of blobpages with the
oncheck utility -ptT option.

Locating Blob Data

When you create a BYTE or TEXT column on magnetic disk, you have the
option of locating the data of the column in the tblspace or in a blobspace.
(You can also store blob data on optical media, although this discussion does
not apply to blobs stored in this way.) In the following example, a TEXT value
is located in the tblspace and a BYTE value is located in a blobspace named
rasters:

CREATE TABLE examptab

(
pic_id SERIAL,
pic_desc TEXT IN TABLE,
pic_raster BYTE IN rasters
)

A TEXT or BYTE value is always stored apart from the rows of the table. Only
a 56-byte descriptor is stored with the row. However, the value itself occupies
at least one disk page.
When blob values are stored in the tblspace, the pages of their data are
interspersed among the pages containing rows. The result is to inflate the size
of the table. The blobpages separate the pages containing rows and spread
them out on the disk. When the database server reads only the rows and not
the blob data, the disk arm must move farther than it would if the blobpages

10-18 Tuning Your Data Model

 Managing Indexes

were stored apart. The database server scans only the row pages on any
SELECT operation that retrieves no blob column and whenever it tests rows
using a filter expression.
Another consideration is that disk I/O to and from a dbspace is buffered.
Pages are held in storage in case they are needed again soon; and, when
pages are written, the requesting program is allowed to continue before the
actual disk write takes place.
However, disk I/O to and from blobspaces is not buffered. Blobspace pages
are not retained in buffers to be read again, and the requesting program is not
allowed to proceed until all output to them is complete. The reason is that
blobspace I/O is expected to be voluminous. If these pages are passed
through the normal buffering mechanisms, they could monopolize the
buffers, driving out index pages and other pages that are useful to good
performance.
For best performance, then, you should locate a TEXT or BYTE column in a
blobspace in either of the following circumstances:
• When single data items are larger than one or two pages each; if kept in
the dbspace, their transfer dilutes the effectiveness of the page buffers.
• When the number of pages of blob data is more than half the number of
pages of row data; if kept in the dbspace, the table is inflated and queries
against it are slowed.
For a table that is both relatively small and nonvolatile, you can achieve the
effect of a dedicated blobspace by the following means: Load the entire table
with rows in which the blob columns are null. Create all indexes. The row
pages and the index pages are now contiguous. Update all the rows to install
the blob data. The blob pages follow the pages of row and index data in the
tblspace.

Managing Indexes
An index is necessary on any column (or composition of columns) that must
be unique. However, as discussed in Chapter 13, “Optimizing Your Queries,”
the presence of an index can also allow the query optimizer to speed up a
query. The optimizer can use an index in the following ways:
• To replace repeated sequential scans of a table with nonsequential access
• To avoid reading row data at all when processing expressions that name
only indexed columns

Tuning Your Data Model 10-19

Managing Indexes

• To avoid a sort (including building a temporary table) when executing the

GROUP BY and ORDER BY clauses

As a result, an index on the right column can save thousands, tens of

thousands, or in extreme cases, even millions of disk operations during a
query. However, costs are associated with indexes.

Space Costs of Indexes

The first cost of an index is disk space. An estimating method is given in
“Estimating Index Pages” on page 10-16. Loosely, an index contains a copy of
every unique data value in the indexed columns, plus a 4-byte pointer for
every row in the table. This information can add many pages to the table
space requirements; it is easy to have as many index pages as row pages.

Time Costs of Indexes

The second cost of an index is time while the table is modified. The
descriptions that follow assume that approximately two pages must be read
to locate an index entry. That is the case when the index consists of a root
page, one intermediate level, and leaf pages, and the root page is already in
a buffer. The index for a very large table has two intermediate levels, so
roughly three pages are read when an entry is looked up.
Presumably, one index has been used to locate the row being altered. Its index
pages are found in the page buffers. However, the pages for any other
indexes that need altering must be read from disk.
Under these assumptions, index maintenance adds time to different kinds of
modifications as shown in the following list:
• When you delete a row from a table, its entries must be deleted from all
indexes.
The entry for the deleted row must be looked up (two or three pages in)
and the leaf page must be rewritten (one page out), for a total of three or
four page accesses per index.
• When you insert a row, its entries must be inserted in all indexes.
The place for entry of the inserted row must be found (two or three pages
in) and rewritten (one page out), for a total of three or four page accesses
per index.
• When you update a row, its entries must be looked up in each index that
applies to a column that was altered (two or three pages in). The leaf page
must be rewritten to eliminate the old entry (one page out); then the new

10-20 Tuning Your Data Model

 Managing Indexes

column value must be located in the same index (two or three more pages
in) and the row entered (one more page out).
Insertions and deletions change the number of entries on a leaf page. In
virtually every pagents operation, some additional work must be done
because a leaf page has either filled up or been emptied. However, because
pagents is usually greater than 100, this occurs less than 1 percent of the time
and can be disregarded for estimating.
In short, when a row is inserted or deleted at random, allow three to four
added page I/O operations per index. When a row is updated, allow six to
eight page I/O operations for each index that applies to an altered column.
Bear in mind also that if a transaction is rolled back, all this work must be
undone. For this reason, rolling back a transaction can take a long time.
Because the alteration of the row itself requires only two page I/O operations,
it is clear that index maintenance is the most time-consuming part of data
modification. One way to reduce this cost is discussed under “Dropping
Indexes” on page 10-24.

Choosing Indexes
Indexes are required on columns that must be unique and are not specified as
primary keys. In addition, you should add an index in the following cases:
• Columns used in joins that are not specified as foreign keys
• Columns frequently used in filter expressions
• Columns frequently used for ordering or grouping

Join Columns
As discussed in Chapter 13, “Optimizing Your Queries,” at least one column
named in any join expression should have an index. If no index exists, the
database server may build a temporary index or perform a sort-merge before
the join and discard it afterward; that is almost always faster than performing
a join by repeated sequential scans over a table.
When both columns in a join expression have indexes, the optimizer has
more options when it constructs the query plan. As a general rule, you should
put an index on any column that is not identified as a primary or foreign key
and is used in a join expression more than occasionally.

Tuning Your Data Model 10-21

Managing Indexes

Selective Filter Columns in Large Tables

If a column is often used to filter the rows of a large table, consider placing an
index on it. The optimizer can use the index to select the desired columns,
avoiding a sequential scan of the entire table. One example might be a table
that contains a large mailing list. If you find that a postal-code column is often
used to filter a subset of rows, you should consider putting an index on it
even though it is not used in joins.
This strategy yields a net savings of time only when the selectivity of the
column is high; that is, when only a small fraction of rows holds any one
indexed value. Nonsequential access through an index takes several more
disk I/O operations than sequential access does, so if a filter expression on the
column passes more than a fourth of the rows, the database server might as
well read the table sequentially. As a rule, indexing a filter column saves time
in the following cases:
• The column is used in filter expressions in many queries or in slow
queries.
• The column contains at least 100 unique values.
• Most column values appear in fewer than 10 percent of the rows.

Order-By and Group-By Columns

When a large quantity of rows must be ordered or grouped, the database
server must put the rows in order. One way the database server does this is
to select all the rows into a temporary table and sort the table. But (as dis-
cussed in Chapter 13, “Optimizing Your Queries”) if the ordering columns
are indexed, the optimizer sometimes plans to read the rows in sorted order
through the index, thus avoiding a final sort.
Because the keys in an index are in sorted sequence, the index really
represents the result of sorting the table. By placing an index on the ordering
column or columns, you can replace many sorts during queries with a single
sort when the index is created.

Duplicate Keys Slow Index Modifications

When duplicate keys are permitted in an index, the entries that have any
single value are grouped in a list. When the selectivity of the column is high,
these lists are generally short. But when only a few unique values occur, the
lists become quite long, and in fact, can cross multiple leaf pages.

10-22 Tuning Your Data Model

 Managing Indexes

For example, in an index on a column whose only values are M for married
and S for single, all the index entries are contained in just two lists of dupli-
cates. Such an index is not of much use, but at least it works for querying; the
database server can read out the list of rows that have one value or the other.
When an entry must be deleted from a list of duplicates, the database server
must read the whole list and rewrite some part of it. When it adds an entry,
the database server puts the new row at the end of its list. Neither operation
is a problem when the list is short, which is normal. But when a list fills many
pages, the database server must read all the rows to find the end. When it
deletes an entry, it typically must update and rewrite half the pages in the list.
Thus, an index on a column that has a small number of distinct values, in a
table that has a large number of rows, can drastically reduce the speed of
updating. An example is a column whose values are the names or abbrevia-
tions of states or provinces. If 50 unique values occur in a mailing list of
100,000 rows, an average of 2,000 duplicates exists per value. But real data is
never so well distributed; in such a table, the more common values likely
have 10,000 or more duplicates, and their lists might approach 50 pages in
length.
When the database server inserts or deletes an entry in such a list, it is busy
for a long time. Worse still, it has to lock all the affected index pages while it
does the work, greatly reducing concurrent access to the table.
You can avoid this problem fairly simply at some cost in disk space. The trick
is to know that the database server uses the leading column of a composite
index in the same way as it uses an index on that column alone. So instead of
creating an index on a column with few unique values, create a composite
index on that column followed by one other column that has a wide
distribution of values.
For example, change the index on the column whose only values are M and S
into a composite index on that column and a birthdate column. You can use
any second column to disperse the key values as long as its value does not
change or changes at the same time as the real key. The shorter the second
column the better because its values are copied into the index and expand its
size.

Tuning Your Data Model 10-23

Managing Indexes

Dropping Indexes
In some applications, the majority of table updates can be confined to a single
time period. Perhaps all updates are applied overnight or on specified dates.
When this is the case, consider dropping all non-unique indexes while you
perform updates and then create new indexes afterward. This can have two
good effects.
First, because fewer indexes occur to update, the updating program can run
faster. Often, the total time to drop the indexes, update without them, and re-
create them afterward is less than the time to update with the indexes in
place. (The time cost of updating indexes is discussed under “Time Costs of
Indexes” on page 10-20.)
Second, newly made indexes are the most efficient ones. Frequent updates
tend to dilute the index structure, causing it to contain many partly full leaf
pages. This reduces the effectiveness of an index and wastes disk space.
As another timesaving measure, make sure that a batch-updating program
calls for rows in the sequence defined by the primary-key index. That
sequence causes the pages of the primary-key index to be read in order and
only one time each.
The presence of indexes also slows down the population of tables when you
use the LOAD statement or the dbload utility. Loading a table that has no
indexes is a very quick process (little more than a disk-to-disk sequential
copy), but updating indexes adds a great deal of overhead.
The fastest way to load a table is as described in the following list:
1. Drop the table (if it exists).
2. Create the table without specifying any unique constraints.
3. Load all rows into the table.
4. Alter the table to apply the unique constraints.
5. Create the non-unique indexes.
If you cannot guarantee that the loaded data satisfies all unique constraints,
you must create the unique indexes before you load the rows. It saves time if
the rows are presented in the correct sequence for at least one of the indexes
(if you have a choice, make it the one with the largest key). This minimizes
the number of leaf pages that must be read and written.

10-24 Tuning Your Data Model

 Managing Indexes

Clustered Indexes
The term clustered index is a misnomer. The index is nothing special; it is the
table that is modified so that its rows are physically ordered to agree with the
sequence of entries in the index. (Do not confuse the clustered index with an
optical cluster, which is a method for storing logically related blobs together
on an optical volume. For information on optical clusters, see the
INFORMIX-OnLine/Optical User Manual.)
When you know that a table is ordered by a certain index, you can take
advantage of the knowledge to avoid sorting. You can also be sure that when
the table is searched on that column, it is read (effectively) in sequential order
instead of nonsequentially. These points are covered in Chapter 13,
“Optimizing Your Queries.”
In the stores6 database, the orders table has an index, zip_ix, on the zip code
column. The following command causes the database server to put the rows
of the customer table into descending order by zip code:

ALTER INDEX zip_ix TO CLUSTER

To cluster a table on a nonindexed column, you must create an index. The

following command reorders the orders table by order date:

CREATE CLUSTERED INDEX o_date_ix ON orders (order_date ASC)

To reorder a table, the database server must copy the table. In the preceding
example, the database server reads all rows of the table and constructs an
index. Then it reads the index entries in sequence. For each entry, it reads the
matching row of the table and copies it to a new table. The rows of the new
table are in the desired sequence. This new table replaces the old table.
Clustering is not preserved when you alter a table. When you insert new
rows, they are stored physically at the end of the table regardless of their con-
tents. When you update rows and change the value of the clustering column,
the rows are written back into their original location in the table.
When clustering is disturbed by updates, it can be restored. The following
command reorders the table to restore the physical sequence:

ALTER INDEX o_date_ix TO CLUSTER

Tuning Your Data Model 10-25

Denormalizing

Reclustering is usually quicker than the original clustering because reading

out the rows of a nearly clustered table is almost a sequential scan.
Clustering and reclustering take a lot of space and time. You can avoid some
clustering by building the table in the desired order in the first place. The
physical order of rows is their insertion order, so if the table is initially loaded
with ordered data, no clustering is needed.

Denormalizing
The entity-relationship data model described in Chapter 8, “Building Your
Data Model,” produces tables that contain no redundant or derived data;
tables that are well-structured by the tenets of relational theory.
Sometimes, to meet extraordinary demands for high performance, you might
have to modify the data model in ways that are undesirable from a theoretical
standpoint. This section describes some modifications and their associated
costs.

Shorter Rows for Faster Queries

As a general principle, tables with shorter rows yield better performance
than ones with longer rows. This is because disk I/O is performed in pages,
not in rows. The shorter the rows of a table, the more rows occur on a page.
The more rows per page, the fewer I/O operations it takes to read the table
sequentially, and the more likely it is that a nonsequential access can be
satisfied from a buffer.
The entity-relationship data model put all the attributes of one entity into a
single table for that entity. For some entities, this can produce rows of awk-
ward length. You can shorten the rows. As the rows get shorter, query
performance should improve.

Expelling Long Strings

The most bulky attributes are often character strings. If you can remove them
from the entity table, the rows become shorter.

Using VARCHAR Strings

An existing database might contain CHAR columns that can be converted
profitably to VARCHAR. VARCHAR columns shorten the average row when
the average value in the CHAR column is at least 2 bytes shorter than the

10-26 Tuning Your Data Model

 Denormalizing

existing fixed width of the column. (When NLS is enabled, NCHAR columns
may replace CHAR columns, and NVARCHAR columns may replace
VARCHAR columns.)

This CHAR-VARCHAR (or NCHAR-NVARCHAR) substitution does not harm

the theoretical characteristics of the model. Furthermore, VARCHAR data is
immediately compatible with most existing programs, forms, and reports.
(Forms must be recompiled. Forms and reports should, of course, be tested
on a sample database.)

Changing Long Strings to TEXT

When the typical string fills half a disk page or more, consider converting it
to a TEXT column in a separate blobspace. The column within the row page
is only 56 bytes long, which should put many more rows on a page. However,
the TEXT data type is not automatically compatible with existing programs.
The code for fetching a TEXT value is more complicated than the code for
fetching a CHARACTER value into a program.

Building a Symbol Table of Repeated Strings

If a column contains strings that are not unique in each row, you can remove
those strings to a table in which only unique copies are stored.
For example, the customer.city column contains city names. Some city names
are repeated down the column, and most rows have some trailing blanks in
the field. Using the VARCHAR data type eliminates the blanks but not the
duplication.
You can create a table named cities, as shown in the following example:

CREATE TABLE cities

(
city_num SERIAL PRIMARY KEY,
city_name VARCHAR(40) UNIQUE
)

You can change the definition of the customer table so that its city column
becomes a foreign key that references the city_num column in the cities table.
You must change any program that inserts a new row into customer to insert
the city of the new customer into cities. The database server return code in
the SQLCODE field can indicate that the insert failed because of a duplicate
key. It is not a logical error; it simply means that some existing customer is

Tuning Your Data Model 10-27

Denormalizing

located in that city. (However, a 4GL program must use the WHENEVER
statement to trap errors; otherwise the negative value in SQLCODE terminates
the program.)
Besides changing programs that insert data, you also must change all
programs and stored queries that retrieve the city name. They must use a join
into the new cities table to obtain their data. The extra complexity in pro-
grams that insert rows and the extra complexity in some queries is the cost of
giving up theoretical correctness in the data model. Before you make the
change, be sure it returns a reasonable savings in disk space or execution
time.

Moving Strings to a Companion Table

Strings less than half a page long waste disk space if you treat them as TEXT,
but you can remove them from the main table to a companion table. The use
of companion tables is the subject of the next section.

Splitting Wide Tables

Consider all the attributes of an entity whose rows are too wide for good
performance. Look for some theme or principle on which they can be divided
into two groups. Split the table into two, a primary table and a companion
table, repeating the primary key in each one. The shorter rows allow each
table to be queried or updated more quickly.

Division by Bulk
One principle on which you can divide an entity table is bulk: Move the
bulky attributes, which are usually character strings, to the companion table.
Keep the numeric and other small attributes in the primary table. In the dem-
onstration database, you can split the ship_instruct column from the orders
table. You can call the companion table orders_ship. It has two columns, a
primary key that is a copy of orders.order_num and the original
ship_instruct column.

Division by Frequency of Use

Another principle for division of an entity is frequency of use. If a few
attributes are rarely queried, they can be moved out to a companion table. In
the demonstration database, it could be that the ship_instruct, ship_weight,
and ship_charge columns are queried only in one program. In that case, you
can move them out to a companion table.

10-28 Tuning Your Data Model

 Denormalizing

Division by Frequency of Update

Updates take longer than queries, and the updating programs lock index
pages and rows of data, slowing down the programs that only query. If cer-
tain attributes of an entity are updated frequently and many others are
changed only rarely or not at all—and if many queries select the latter and do
not need the former—you can split the volatile columns to a companion
table. The updating programs contend for the use of that table, and the query
programs use the primary table.

Costs of Companion Tables

Splitting a table consumes extra disk space and adds complexity. Two copies
of the primary key occurs for each row, one copy in each table. Two primary-
key indexes also exist. You can use the methods described in earlier sections
to estimate the number of pages added.
You must modify existing programs, reports, and forms that use SELECT *
because fewer columns are returned. Programs, reports, and forms that use
attributes from both tables must perform a join to bring them together.
In this case, when you insert or delete a row, two tables are altered instead of
one. If you do not coordinate the alteration of the two tables (by making them
within a single transaction, for example), you lose semantic integrity.

Splitting Tall Tables

A large table poses large management problems. The 400-megabyte table
takes a long time to query, of course, and it also takes a long time to back up
or restore. Sorting becomes radically slower because the sort goes to three or
more merge levels; sorting can even become impossible because too much
disk space is required. Indexes become less effective because the index
structure has three or more intermediate levels.
Consider splitting such a table into segments. Each segment has the same
arrangement of columns, indexes, and constraints but a different subset of
rows. The division into groups of rows should be based on the value of an
attribute that appears in most queries (probably the primary key) so that a
program or user can easily tell in which subtable a row belongs. For example,
if the primary key is a number, rows can be allocated to 10 subtables based on
a middle digit.
If rows include a time or date attribute, it might be useful to split a table into
segments by age, especially if recent rows are used more often than older
ones.

Tuning Your Data Model 10-29

Denormalizing

The benefits of splitting a large table are that the segments can be treated
separately for copying, sorting, indexing, archiving, and restoring. All these
operations become easier. The segments can be assigned to tblspaces in dif-
ferent ways to make the best use of disk devices. And, as long as most queries
are based on the attribute used for segmenting, queries should be
dramatically faster.
The drawback of splitting a table is that operations that apply to the whole
table become much more complicated. If a query cannot be directed to one
subtable, it must be written in the form of a UNION of many subqueries, each
on a different subtable. This greatly complicates the design of reports and
browsing programs.

Redundant and Derived Data

The data model produced by the methods of Chapter 8, “Building Your Data
Model,” contains no redundant data. Every attribute appears in only one
table. The data model also contains no derived data; that is, data that can be
computed from existing attributes is selected as an expression based on those
attributes.
These features minimize the amount of disk space used and make updating
the tables as easy as possible. However, they can force you to use joins and
aggregate functions often, and that may take more time than is acceptable.
As an alternative, you can introduce new columns that contain redundant or
derived data, provided you understand the risks.

Adding Derived Data

In the stores6 database, no column occurs in the orders table for the total
price of an order. The reason is that the information can be derived from the
rows of the items table. A program that wants to report the date and total
price of order number 1009 can obtain it with the following query:

SELECT order_date, SUM (total_price)

FROM orders, items
WHERE orders.order_num = 1009
AND orders.order_num = items.order_num
GROUP BY orders.order_num, orders.order_date

Although the join to the items table reads only three or four additional pages,
that might be too much time for an interactive program. One solution is to
add an order-total column to the orders table.

10-30 Tuning Your Data Model

 Denormalizing

The costs of derived data are in disk space, complexity, and data integrity.
The disk space devoted to an order-total column is wasted because the same
information is stored twice. Also, the presence of the column makes the rows
of the table wider; fewer of them occur in a page and querying the table
becomes slower. Most important, any program that updates the base
attributes must be changed to also update the derived column. Inevitably,
times occur when the derived column is wrong, when it does not contain the
right derived value.
Derived data is not reliable data. In the example of a derived order price, the
unreliability occurs while an order is being entered. When rows are added to
the items table, times occur when the order total in orders is not equal to the
sum of the corresponding rows from items. In general, before you allow
derived data to appear in the database, you must define its accuracy as care-
fully as possible; actually, you should define as closely as possible the
conditions under which the derived column is unreliable.

Adding Redundant Data

A correct data model avoids redundancy by keeping any attribute only in the
table for the entity it describes. If the attribute data is needed in a different
context, you make the connection by joining tables. But joining takes time.
If a frequently used join affects performance, you can eliminate it by
duplicating the joined data in another table.
In the demonstration database, the manufact table contains the names of
manufacturers and their delivery times. (In a real database, it would contain
many other attributes of a supplier, such as address, sale representative
name, and so on.)
The contents of manufact are primarily a supplement to the stock table.
Suppose that a time-critical application frequently refers to the delivery lead
time of a particular product but to no other column of manufact. For each
such reference, the database server must read two or three pages of data to
perform the lookup.
You can add a new column, lead_time, to the stock table and fill it with
copies of the lead_time column from the corresponding rows of manufact.
That eliminates the lookup, which speeds up the application.
As with derived data, redundant data takes space and poses an integrity risk.
In the example, there is not one extra copy of each manufacturer’s lead-time,
there are many. (Each manufacturer can appear in stock many times.) The
programs that insert or update a row of manufact must be changed to update
multiple rows of stock as well.

Tuning Your Data Model 10-31

Maximizing Concurrency

The integrity risk is simply that the redundant copies of the data might not
be accurate. If a lead time is changed in manufact, the stock column is out-
dated until it, too, is updated. As with derived data, you should take pains to
define the conditions under which the redundant data might be wrong.

Maximizing Concurrency
Some databases are used by only one program at a time; others are used
concurrently by multiple programs. The following factors make concurrent
programs inherently slower than doing the same amount of work serially:
• Multiple programs interfere with the use of buffers and disk drives. The
pages read for one program might be driven from the buffers used by a
query from the next program and might have to be read again. Disk I/O
for one query displaces the disk-access arm, slowing the sequential access
of another program.
• Programs that modify data lock pages, delaying all other programs that
use the same data.

Easing Contention
Contention is inevitable between programs using the same resources. You
can deal with it in the following ways:
1. Make programs use fewer resources, either by making them do less work
or by making them work more efficiently.
2. Arrange the resources better, for instance, by allocating tables to dbspaces
to minimize contention.
3. Supply more resources: more memory, more and faster disk drives, and
more and faster computers.
The first point is the subject of Chapter 13, “Optimizing Your Queries,”
which deals with making queries faster. Sometimes you also must consider
making queries do less; that is, reducing the functions available to on-line
users. You can look for any allowed transaction that entails scanning an entire
large table, such as any kind of summation, average, or management report,
especially if it requires a join. Consider removing such transactions from on-
line programs. Instead, offer a new facility that interactively schedules an
off-peak job, with the output returned the next day.

10-32 Tuning Your Data Model

 Maximizing Concurrency

Rescheduling Modifications
To the greatest extent possible, schedule modifications for times when no
interactive users are using the database. One reason for this is that concurrent
modifications must be done with all indexes in place; that means they incur
all the time costs noted earlier. Another reason is that modifications not only
lock table rows they also lock index pages. That increases the number of locks
in use and the number of lock delays for all users.
The means of rescheduling modifications depend entirely on the details of
your application.

Using an Update Journal

Instead of performing updates as the data becomes available, consider
creating an update journal. This is a table that contains rows that represent
pending updates. Its rows contain the following information:
• Data items to be changed in some base table, with null meaning no
change
• Audit information that allows an erroneous update to be traced back to
the transaction that entered the row
It is usually much faster to insert a new row in an update journal than
actually to perform the update on one or more base tables. For one thing, the
journal has, at most, one index to update, but the base tables usually have
several. For another, locking delays are almost nil because inserting a row in
a table with no indexes requires only a lock on the inserted row. Another p
rogram can insert its row even before the first program commits its
transaction.
Also, because no updates are performed on the primary tables, their pages
are not locked. This allows query programs to run without delays.
After peak-usage hours (perhaps after close of business), you run a batch
program to validate and apply the updates, row by row as they were
inserted. You obviously would want to take pains to see that no update is
forgotten or applied twice, regardless of possible system failures.

Tuning Your Data Model 10-33

Maximizing Concurrency

One way to update the base table is to use two cursors. The first cursor is a
hold cursor; it is used to scan the rows of the journal. (See “Hold Cursors” in
Chapter 7.) For each row of the journal, the program goes through the
following steps:
1. Issue the BEGIN WORK statement.
2. Fetch the rows from the tables being updated using an update cursor.
(This locks only these rows.)
3. Validate the update information from the journal against the data in the
target rows.
4. Apply the updates to the target tables.
5. Update the journal row in some way to mark it as finished.
6. Issue the COMMIT WORK statement (or the ROLLBACK WORK statement,
if an error is found).
You run a different program to drop and re-create the journal table only after
every journal row is validated, applied, and marked.
The obvious disadvantage of an update journal is that the base tables do not
reflect the most current data. If it is essential that updates be instantly visible,
the journal scheme does not work.
The great advantages of reduced I/O and reduced locking delays during
peak hours are a powerful argument in favor of an update journal. Deferred
updates are accepted in many applications. For example, no bank promises
to know your account balance more precisely than as of the close of business
the preceding day, and for exactly this reason: the bank records transactions
in a journal during the day and applies the updates overnight.

10-34 Tuning Your Data Model

 Maximizing Concurrency

Isolating and Dispersing Updates

If updates really must be performed interactively during peak hours, you
must find a different way to isolate queries from updates.

Splitting Tables to Isolate Volatile Columns

This idea was covered in an earlier section. If it is necessary to perform
updates during peak hours, examine the structure of the tables. Can you sep-
arate the columns into volatile ones that are updated constantly and static
ones that are rarely updated? If so, consider splitting the tables so that the
volatile columns are isolated in a companion table.
Depending on the mix of operations and on the priorities of the users, you
can then put either the static or the volatile subtable on your fastest disk
drive.

Dispersing Bottleneck Tables

Small tables are sometimes used for summaries, usage records, and audits.
For example, your interactive programs can maintain a table with a row for
each authorized user. Each time the user starts or ends a program, it updates
that user’s row to show time on, time off, number of transactions, or other
work-monitoring data.
Small tables that are only read do not cause performance problems, but small
tables that are the subject of concurrent updates do. Any small table that must
be updated on every transaction can become a bottleneck in which every
on-line program queues up and waits.
To eliminate the problem, either use a journal with off-peak updates as
described earlier, or disperse the bottleneck by creating many tables. To
monitor users, create a one-row table for each user.

Tuning Your Data Model 10-35

Summary

Summary
When tables are moderate in size and only one user at a time accesses the
database, carefully applied relational theory is sufficient to produce good
performance.
When both the number of tables and the number of users become larger and
the response time begins to degrade, you must turn to practical solutions.
The first step is to understand and take advantage of the tools that
INFORMIX-OnLine Dynamic Server offers. It permits you to arrange your
tables on the hardware for maximum benefit. Then, one step at a time and
always measuring, you can begin complicating the structure of the data
model and the programs that use it.

10-36 Tuning Your Data Model

 Chapter

Granting and
Limiting Access to
11
Your Database
Chapter Overview 3
Controlling Access to Databases 4
Securing Database Files 4
Multiuser Systems 4
Single-User Systems 4
Securing Confidential Data 5
Granting Privileges 5
Database-Level Privileges 5
Connect Privilege 6
The Users and the Public 6
Resource Privilege 7
Database Administrator Privilege 7
Ownership Rights 7
Table-Level Privileges 7
Access Privileges 8
Index, Alter, and References Privileges 8
Privileges in the System Catalog 9
Column-Level Privileges 10
Procedure-Level Privileges 12
Automating Privileges 13
Automating with INFORMIX-4GL 13
Automating with a Command Script 14
Controlling Access to Data Using Stored Procedures 15
Restricting Reads of Data 16
Restricting Changes to Data 16
 Monitoring Changes to Data 17
Restricting Object Creation 18
Using Views 19
Creating Views 20
Duplicate Rows from Views 22
Restrictions on Views 22
When the Basis Changes 22
Modifying Through a View 23
Deleting Through a View 24
Updating a View 24
Inserting into a View 25
Using WITH CHECK OPTION 25
Privileges and Views 27
Privileges When Creating a View 27
Privileges When Using a View 28
Summary 30

11-2 Granting and Limiting Access to Your Database

Chapter Overview
In some databases, all data is accessible to every user. In others, this is not the
case; some users are denied access to some or all of the data. You can restrict
access to data at the following levels, which are the subject of this chapter:
• When the database is stored in operating system files, you can sometimes
use the file-permission features of the operating system.
This level is not available when INFORMIX-OnLine Dynamic Server
holds the database. It manages its own disk space and the rules of the
operating system do not apply.
• You can use the GRANT and REVOKE statements to give or deny access to
the database or to specific tables, and you can control the kinds of uses
that people can make of the database.
• You can use the CREATE PROCEDURE statement to write and compile a
stored procedure that controls and monitors which users can read,
modify, or create database tables.
• You can use the CREATE VIEW statement to prepare a restricted or
modified view of the data. The restriction can be vertical, excluding
certain columns, or horizontal, excluding certain rows, or both.
• You can combine GRANT and CREATE VIEW statements to achieve precise
control over the parts of a table a user can modify and with what data.
In addition to these points, INFORMIX-OnLine/Secure offers a type of
automatic security called mandatory access control (MAC). With MAC, all
users and all data are assigned a security label, and OnLine/Secure compares
the labels before allowing access. For example, a TOP SECRET user could view
an UNCLASSIFIED row, but an UNCLASSIFIED user could not view a TOP
SECRET row. This topic, and other topics related to security in OnLine/Secure
are addressed in the INFORMIX-OnLine/Secure Security Features User’s Guide.

Granting and Limiting Access to Your Database 11-3

Controlling Access to Databases

Controlling Access to Databases

The normal database-privilege mechanisms are based on the GRANT and
REVOKE statements. They are covered in the section “Granting Privileges” on
page 11-5. However, you can sometimes use the facilities of the operating
system as an additional way to control access to a database.

Securing Database Files

Database servers other than INFORMIX-OnLine Dynamic Server store
databases in operating system files. Typically, a database is represented as a
number of files: one for each table, one for the indexes on each table, and pos-
sibly others. The files are collected in a directory. The directory represents the
database as a whole.

Multiuser Systems
You can deny access to the database by denying access to the database
directory. The means by which you can do this depend on your operating
system and your computer hardware. Multiuser operating systems have
software facilities such as UNIX file permissions.
Note: In UNIX, the database directory is created with group identity informix, and
the database server always runs under group identity informix. Thus, you cannot
use group permissions to restrict access to a particular group of users. You can, how-
ever, only remove all group permissions (file mode 700) and deny access to anyone
except the owner of the directory.
You also can deny access to individual tables in this way; for example, by
making the files that represent those tables unavailable to certain users, while
leaving the rest of the files accessible. However, the database servers are not
designed for tricks of this kind. When an unauthorized user tries to query one
of the tables, the database server probably returns an error message about not
being able to locate a file. This may confuse users.

Single-User Systems
Typical single-user systems have few software controls on file access; you can
make a database inaccessible to others only by writing it on a disk that you
can detach from the machine and keep locked.
None of these techniques apply when you use the INFORMIX-OnLine
Dynamic Server database server. It controls its own disk space at the device
level, bypassing the file-access mechanisms of the operating system.

11-4 Granting and Limiting Access to Your Database

 Granting Privileges

Securing Confidential Data

No matter what access controls the operating system gives you, when the
contents of an entire database are highly sensitive you might not want to
leave it on a public disk fixed to the machine. You can circumvent normal
software controls when the data must be secure.
When the database is not being used by you or another authorized person, it
does not have to be available on-line. You can make it inaccessible in one of
the following ways, which have varying degrees of inconvenience:
• Detach the physical medium from the machine, and take it away. If the
disk itself is not removable, the disk drive might be removable.
• Copy the database directory to tape, and take possession of the tape.
• Copy the database files using an encryption utility. Keep only the
encrypted version.
In the latter two cases, after making the copies, you must remember to erase
the original database files using a program that overwrites an erased file with
null data.
Instead of removing the entire database directory, you can copy and then
erase the files that represent individual tables. However, do not overlook the
fact that index files contain copies of the data from the indexed column or
columns. Remove and erase the index files as well as the table files.

Granting Privileges
The authorization to use a database is called a privilege. For example, the
authorization to use a database at all is called the Connect privilege, and the
authorization to insert a row into a table is called the Insert privilege. You
control the use of a database by granting these privileges to other users or by
revoking them.
Two groups of privileges exist that control the actions a user can perform on
data. These are database-level privileges, which affect the entire database,
and table-level privileges, which relate to individual tables. In addition to
these two groups, procedure-level privileges exist, which determine who can
execute a procedure.

Database-Level Privileges
The three levels of database privilege provide an overall means of controlling
who accesses a database.

Granting and Limiting Access to Your Database 11-5

Granting Privileges

Connect Privilege
The least of the privilege levels is Connect, which gives a user the basic ability
to query and modify tables. Users with the Connect privilege can perform the
following functions:
• Execute the SELECT, INSERT, UPDATE, and DELETE statements, provided
that they have the necessary table-level privileges.
• Execute a stored procedure, provided that they have the necessary
table-level privileges.
• Create views, provided that they are permitted to query the tables on
which the views are based.
• Create temporary tables and create indexes on the temporary tables.

The Users and the Public

Privileges are granted to single users by name or to all users under the name
of PUBLIC. Any grant to the public serves as a default privilege.
Prior to executing a statement, the database server determines whether the
user has the necessary privileges. (The information is in the system catalog;
see “Privileges in the System Catalog” on page 11-9.)
The database server looks first for privileges granted specifically to the
requesting user. If it finds such a grant, it uses that information. It then
checks to see if less restrictive privileges have been granted to PUBLIC. If so,
the database server uses the less-restrictive privileges. If there has been no
grant to that user, the database server looks for privileges granted to public.
If it finds a relevant one, it uses that.
Thus, you can set a minimum level of privilege for all users by granting
privileges to public. You can override that in specific cases by granting
higher individual privileges to users.

The Connect privilege is necessary before users can access a database.

Ordinarily, in a database that does not contain highly sensitive or private
data, you give the GRANT CONNECT TO PUBLIC privilege shortly after
reating the database.
If you do not grant the Connect privilege to public, the only users who can
access the database through the database server are those to whom you spe-
cifically grant the Connect privilege. If limited users should have access, this
is your means of providing it to them and denying it to all others.

11-6 Granting and Limiting Access to Your Database

 Granting Privileges

Resource Privilege
The Resource privilege carries the same authorization as the Connect
privilege. In addition, users with the Resource privilege can create new, per-
manent tables, indexes, and stored procedures, thus permanently allocating
disk space.

Database Administrator Privilege

The highest level of database privilege is Database Administrator, or DBA.
When you create a database, you are automatically the DBA. Holders of the
DBA privilege can perform the following functions:

• Execute the DROP DATABASE, START DATABASE, and ROLLFORWARD

DATABASE statements.
• Drop or alter any object regardless of who owns it.
• Create tables, views, and indexes to be owned by other users.
• Grant database privileges, including the DBA privilege, to another user.
• Alter the NEXT SIZE (but no other attribute) of the system catalog tables,
and insert, delete, or update rows of any system catalog table except
systables.
Warning: Although users with the DBA privilege can modify the system catalog
! tables, Informix strongly recommends that you do not update, delete, or alter any
rows in the system catalog tables. Modifying the system catalog tables can destroy
the integrity of the database.

Ownership Rights
The database, and every table, view, index, procedure, and synonym in it, has
an owner. The owner of an object is usually the person who created it,
although a user with the DBA privilege can create objects to be owned by
others.
The owner of an object has all rights to that object and can alter or drop it
without needing additional privileges.

Table-Level Privileges
You can apply seven privileges, table by table, to allow nonowners the
privileges of owners. Four of them—the Select, Insert, Delete, and Update
privileges—control access to the contents of the table. The Index privilege

Granting and Limiting Access to Your Database 11-7

Granting Privileges

controls index creation. The Alter privilege controls the authorization to

change the table definition. The References privilege controls the
authorization to specify referential constraints on a table.
In an ANSI-compliant database, only the table owner has any privileges. In
other databases, the database server, as part of creating a table, automatically
grants all table privileges—except Alter and References—to public. This
means that a newly created table is accessible to any user with the Connect
privilege. If this is not what you want—if users exist with the Connect privi-
lege who should not be able to access this table—you must revoke all
privileges on the table from public after you create the table.

Access Privileges
Four privileges govern how users can access a table. As the owner of the
table, you can grant or withhold the following privileges independently:
• The Select privilege allows selection, including selecting into temporary
tables.
• The Insert privilege allows a user to add new rows.
• The Update privilege allows a user to modify existing rows.
• The Delete privilege allows a user to delete rows.
The Select privilege is necessary for a user to retrieve the contents of a table.
However, The Select privilege is not a precondition for the other privileges.
A user can have Insert or Update privileges without having the Select
privilege.
For example, your application might have a usage table. Every time a certain
program is started, it inserts a row into the usage table to document that it
was used. Before the program terminates, it updates that row to show how
long it ran and perhaps to record counts of work performed by its user.
If you want any user of the program to be able to insert and update rows in
this usage table, you grant Insert and Update privileges on it to public.
However, you might grant the Select privilege to only a few users.

Index, Alter, and References Privileges

The Index privilege permits its holder to create and alter indexes on the table.
The Index privilege, similar to the Select, Insert, Update, and Delete
privileges, is granted automatically to public when a table is created.

11-8 Granting and Limiting Access to Your Database

 Granting Privileges

Privileges in the System Catalog

Privileges are recorded in the system catalog tables. Any user with the
Connect privilege can query the system catalog tables to determine what
privileges have been granted and to whom.
Database privileges are recorded in the sysusers table, in which the primary
key is user-id and the only other column contains a single character C, R, or
D for the privilege level. A grant to the keyword of PUBLIC is reflected as a
user name of public (lowercase).
Table-level privileges are recorded in systabauth, which uses a composite
primary key of the table number, grantor, and grantee. In the tabauth
column, the privileges are encoded in a six-letter list as shown in the
following diagram:

unconditional Update Insert

Index
unconditional Select su-idxar References
Alter
Delete
* if column privilege granted
A hyphen means an ungranted privilege, so that a grant of all privileges is
shown as su-idxar, and -u------ shows a grant of only Update. The
code letters are normally lowercase, but they are uppercase when the
keywords WITH GRANT OPTION are used in the GRANT statement.
When an asterisk appears in the third position, some column-level privilege
exists for that table and grantee. The specific privilege is recorded in sysco-
lauth. Its primary key is a composite of the table number, the grantor, the
grantee, and the column number. The only attribute is a three-letter list
showing the type of privilege: s, u, or r.

You can grant the Index privilege to anyone, but to exercise the ability, the
user must also hold the Resource database privilege. So, although the Index
privilege is granted automatically (except in ANSI-compliant databases),
users who have only the Connect privilege to the database cannot exercise
their Index privilege. This is reasonable because an index can fill a large
amount of disk space.

Granting and Limiting Access to Your Database 11-9

Granting Privileges

The Alter privilege permits its holder to use the ALTER TABLE statement on
the table, including the power to add and drop columns, reset the starting
point for SERIAL columns, and so on. You should grant the Alter privilege
only to users who understand the data model very well and whom you trust
to exercise their power very carefully.
The References privilege allows you to impose referential constraints on a
table. As with the Alter privilege, you should grant the References privilege
only to users who understand the data model very well.

Column-Level Privileges
You can qualify the Select, Update, and References privileges with the names
of specific columns. This allows you to grant very specific access to a table:
You can permit a user to see only certain columns, you can allow a user to
update only certain columns, or you can allow a user to impose referential
constraints on certain columns.
Using INFORMIX-OnLine Dynamic Server (so that table data can be
inspected only through a call to the database server), this feature solves the
problem posed earlier: that only certain users should know the salary, perfor-
mance review or other sensitive attributes of an employee. To make the
example specific, suppose a table of employee data is defined as shown in
Figure 11-1.

CREATE TABLE hr_data

(
emp_key INTEGER,
emp_name CHAR(40),
hire_date DATE,
dept_num SMALLINT,
user-id CHAR(18),
salary DECIMAL(8,2)
performance_level CHAR(1),
performance_notes TEXT
)

Figure 11-1 A table of confidential employee information

Because this table contains sensitive data, you execute the following
statement immediately after creating it:

REVOKE ALL ON hr_data FROM PUBLIC

11-10 Granting and Limiting Access to Your Database

 Granting Privileges

For selected persons in the Human Resources department and for all
managers, you execute a statement such as the following one:

GRANT SELECT ON hr_data TO harold_r

In this way, you permit certain users to view all columns. (The final section
of this chapter discusses a way to limit the view of managers to only their
own employees.) For the first-line managers who carry out performance
reviews, you could execute a statement such as the following one:

GRANT UPDATE (performance_level, performance_notes)

ON hr_data TO wallace_s, margot_t

This statement permits the managers to enter their evaluations of their

employees. Only for the manager of the Human Resources department or
whomever is trusted to alter salary levels, would you execute a statement
such as the following one:

GRANT UPDATE (salary) ON hr_data to willard_b

For the clerks in the Human Resources department, you could execute a
statement such as the following one:

GRANT UPDATE (emp_key, emp_name, hire_date, dept_num)

ON hr_data TO marvin_t

This statement gives certain users the ability to maintain the nonsensitive
columns but denies them authorization to change performance ratings or sal-
aries. The person in the MIS department who assigns computer user-ids is the
beneficiary of a statement such as the following one:

GRANT UPDATE (user_id) ON hr_data TO eudora_b

Granting and Limiting Access to Your Database 11-11

Granting Privileges

On behalf of all users who are allowed to connect to the database but who are
not authorized to see salaries or performance reviews, you execute
statements such as the following one to permit them to see the nonsensitive
data:

GRANT SELECT (emp_key, emp_name, hire_date, dept_num, user-id)

ON hr_data TO george_b, john_s

These users can perform queries such as the following one:

SELECT COUNT(*) FROM hr_data WHERE dept_num IN (32,33,34)

However, any attempt to execute a query such as the following one produces
an error message and no data:

SELECT performance_level FROM hr_data

WHERE emp_name LIKE '*Smythe'

Procedure-Level Privileges
You can apply the Execute privilege on a procedure to authorize nonowners
to run a procedure. If you create a procedure in a database that is not ANSI-
compliant, the default procedure-level privilege is Public; you do not need to
grant the Execute privilege to specific users unless you have first revoked it.
If you create a procedure in an ANSI-compliant database, no other users have
the Execute privilege by default; you must grant specific users the Execute
privilege. The following example grants the Execute privilege to the user
orion so that orion can use the stored procedure named read-address:

GRANT EXECUTE ON read_address TO orion;

Procedure-level privileges are recorded in the sysprocauth system catalog

table. The sysprocauth table uses a primary key of the procedure number,
grantor, and grantee. In the procauth column, the execute privilege is indi-
cated by a lowercase letter “e.” If the execute privilege was granted with the
WITH GRANT option, the privilege is represented by an uppercase letter “E.”

For more information on procedure-level privileges see “Privileges on Stored

Procedures” on page 14-13.

11-12 Granting and Limiting Access to Your Database

 Granting Privileges

Automating Privileges
It might seem that this design forces you to execute a tedious number of
GRANT statements when you first set up the database. Furthermore, privi-
leges require constant maintenance as people change their jobs. For example,
if a clerk in Human Resources is terminated, you want to revoke the Update
privilege as soon as possible; otherwise you risk the unhappy employee
executing a statement such as the following one:

UPDATE hr_data
SET (emp_name, hire_date, dept_num) = (NULL, NULL, 0)

Less dramatic, but equally necessary, changes of privilege are required daily,
or even hourly, in any model that contains sensitive data. If you anticipate
this need, you can prepare some automated tools to help maintain privileges.
Your first step should be to specify privilege classes that are based on the jobs
of the users, not on the structure of the tables. For example, a first-line
manager needs the following privileges:
• The Select and limited Update privilege on the hypothetical hr_data table
• The Connect privilege to this and other databases
• Some degree of privilege on several tables in those databases
When the manager is promoted to a staff position or sent to a field office, you
must revoke all those privileges and grant a new set of privileges.
Define the privilege classes you support, and for each class specify the
databases, tables, and columns to which you must give access. Then devise
two automated procedures for each class: one to grant the class to a user and
one to revoke it.

Automating with INFORMIX-4GL

The mechanism you use depends on your operating system and other tools.
If you are a programmer, the most flexible tool is probably INFORMIX-4GL.
4GL makes it quite easy to program a simple user interaction, as shown in the
following fragment:

DEFINE mgr_id char(20)

PROMPT 'What is the user-id of the new manager? ' FOR mgr_id
CALL table_grant ('SELECT', 'hr_data', mgr_id)

Granting and Limiting Access to Your Database 11-13

Granting Privileges

Unfortunately, although INFORMIX-4GL allows you to mix GRANT and

REVOKE statements freely with other program statements, it does not permit
you to create parameters from them from program variables. To customize a
GRANT statement with a user-id taken from user input, the program must
build the statement as a string, prepare it with a PREPARE statement, and exe-
cute it with an EXECUTE statement. (These statements are discussed in detail
in Chapter 5, “Programming with SQL,” where the following example is
analyzed in detail.)
Figure 11-2 shows one possible definition of the 4GL function table_grant()
that is invoked by the CALL statement in the preceding example.

FUNCTION table_grant (priv_to_grant, table_name, user_id)

DEFINE priv_to_grant char(100),{may include column-list}
table_name CHAR(20),
user_id CHAR(20),
grant_stmt CHAR(200)
LET grant_stmt =' GRANT ', priv_to_grant,
' ON ', table_name,
' TO ', user_id
WHENEVER ERROR CONTINUE
PREPARE the_grant FROM grant_stmt
IF status = 0 THEN
EXECUTE the_grant
END IF
IF status <> 0 THEN
DISPLAY 'Sorry, got error #', status, 'attempting:'
DISPLAY ' ', grant_stmt
END IF
WHENEVER ERROR STOP
END FUNCTION

Figure 11-2 A 4GL function that builds, prepares, and executes a GRANT statement

Automating with a Command Script

Your operating system probably supports automatic execution of command
scripts. In most operating environments, DB-Access and INFORMIX-SQL, the
interactive SQL tools, accept commands and SQL statements to execute from
the command line. You can combine these two features to automate privilege
maintenance.

11-14 Granting and Limiting Access to Your Database

 Controlling Access to Data Using Stored Procedures

The details depend on your operating system and the version of DB-Access
or INFORMIX-SQL that you are using. In essence, you want to create a
command script that performs the following functions:
• Takes a user-id whose privileges are to be changed as its parameter
• Prepares a file of GRANT or REVOKE statements customized to contain
that user-id
• Invokes DB-Access or INFORMIX-SQL with parameters that tell it to select
the database and execute the prepared file of GRANT or REVOKE
statements
In this way, you can reduce the change of the privilege class of a user to one
or two commands.

Controlling Access to Data Using Stored Procedures

You can use a stored procedure to control access to individual tables and
columns in the database. You can accomplish various degrees of access con-
trol through a procedure. (Stored procedures are fully described in
Chapter 14, “Creating and Using Stored Procedures.”) A powerful feature of
SPL is the ability to designate a stored procedure as a DBA-privileged proce-
dure. Writing a DBA-privileged procedure, you can allow users who have few
or no table privileges to have DBA privileges when they execute the proce-
dure. In the procedure, users can carry out very specific tasks with their tem-
porary DBA privilege. The DBA-privileged feature allows you to accomplish
the following tasks:
• You can restrict how much information individual users can read from a
table.
• You can restrict all the changes made to the database and ensure that
entire tables are not emptied or changed accidentally.
• You can monitor an entire class of changes made to a table, such as
deletions or insertions.
• You can restrict all object creation (data definition) to occur within a
stored procedure so that you have complete control over how tables,
indexes, and views are built.

Granting and Limiting Access to Your Database 11-15

Controlling Access to Data Using Stored Procedures

Restricting Reads of Data

The procedure in Figure 11-3 hides the SQL syntax from users, but it requires
that users have the Select privilege on the customer table. If you want to
restrict what users can select, you can write your procedure to work in the
following environment:
• You are the DBA of the database.
• The users have the Connect privilege to the database. They do not have
the Select privilege on the table.
• Your stored procedure (or set of stored procedures) is created using the
DBA keyword.
• Your stored procedure (or set of stored procedures) reads from the table
for users.
If you want users to read only the name, address, and telephone number of a
customer, you can modify the procedure as shown in Figure 11-3.

CREATE DBA PROCEDURE read_customer(cnum INT)

RETURNING CHAR(15), CHAR(15), CHAR(18);

DEFINE p_lname,p_fname CHAR(15);

DEFINE p_phone CHAR(18);

SELECT fname, lname, phone

INTO p_fname, p_lname, p_phone
FROM customer
WHERE customer_num = cnum;

RETURN p_fname, p_lname, p_phone;

END PROCEDURE;

Figure 11-3 Stored procedure to restrict reads on customer data

Restricting Changes to Data

Using stored procedures, you can restrict changes made to a table. Simply
channel all changes through a stored procedure. The stored procedure makes
the changes, rather than users making the changes directly. If you want to

11-16 Granting and Limiting Access to Your Database

 Controlling Access to Data Using Stored Procedures

limit users to deleting one row at a time to ensure that they do not
accidentally remove all the rows in the table, set up the database with the
following privileges:
• You are the DBA of the database.
• All the users have the Connect privilege to the database. They may or
may not have the Resource privilege. They do not have the Delete (for this
example) privilege on the table being protected.
• Your stored procedure is created using the DBA keyword.
• Your stored procedure performs the deletion.
Write a stored procedure similar to the one shown in Figure 11-4, which
deletes rows from the customer table using a WHERE clause with the
customer_num provided by the user.

CREATE DBA PROCEDURE delete_customer(cnum INT)

DELETE FROM customer

WHERE customer_num = cnum;

END PROCEDURE;

Figure 11-4 Stored procedure to delete a row

Monitoring Changes to Data

Using stored procedures, you can create a record of changes made to a
database. You can record changes made by a particular user, or you can make
a record of each time a change is made.
You can monitor all the changes made to the database by a single user. Simply
channel all changes through stored procedures that keep track of changes
made by each user. If you want to record each time the user acctclrk modifies
the database, set up the database with the following privileges:
• You are the DBA of the database.
• All other users have the Connect privilege to the database. They may or
may not have the Resource privilege. They do not have the Delete (for this
example) privilege on the table being protected.
• Your stored procedure is created using the DBA keyword.
• Your stored procedure performs the deletion and records that a change
has been made by a certain user.

Granting and Limiting Access to Your Database 11-17

Controlling Access to Data Using Stored Procedures

Write a stored procedure similar to the one in Figure 11-6, which updates a
table using a customer number provided by the user. If the user happens to
be acctclrk, a record of the deletion is put in the file updates.

CREATE DBA PROCEDURE delete_customer(cnum INT)

DEFINE username CHAR(8);

DELETE FROM customer

WHERE customer_num = cnum;

IF username = 'acctclrk' THEN

SYSTEM 'echo Delete from customer by acctclrk >> /mis/records/updates' ;
ENF IF
END PROCEDURE;

Figure 11-5 A procedure to delete rows and record changes made by a certain user

You can monitor all the deletions made through the procedure by removing
the IF statement and making the SYSTEM statement more general. If you
change the procedure in Figure 11-6 to record all deletions, it looks like the
procedure in Figure 11-6.

CREATE DBA PROCEDURE delete_customer(cnum INT)

DEFINE username CHAR(8);

LET username = USER ;
DELETE FROM tbname WHERE customer_num = cnum;

SYSTEM
'echo Deletion made from customer table, by '||username ||'>>/hr/records/deletes';

END PROCEDURE;

Figure 11-6 Stored procedure to delete rows and record user

Restricting Object Creation

To put restraints on what and how objects are built, use stored procedures
within the following setting:
• You are the DBA of the database.
• All the other users have the Connect privilege to the database. They do
not have the Resource privilege.
• Your stored procedure (or set of stored procedures) is created using the
DBA keyword.

11-18 Granting and Limiting Access to Your Database

 Using Views

• Your stored procedure (or set of stored procedures) creates tables,

indexes, and views in the way you defined them. You might use such a
procedure to set up a training database environment.
Your procedure might include the creation of one or more tables and
associated indexes, as shown in Figure 11-7.

CREATE DBA PROCEDURE all_objects()

CREATE TABLE learn1 (intone SERIAL, inttwo INT NOT NULL, charcol CHAR(10) )
CREATE INDEX learn_ix ON learn1 (inttwo).
CREATE TABLE toys (name CHAR(15) NOT NULL UNIQUE,
description CHAR(30), on_hand INT);
END PROCEDURE;

Figure 11-7 DBA-mode procedure that adds tables and indexes to the database

To use the all_objects procedure to control additions of columns to tables,

revoke the Resource privilege on the database from all users. When users try
to create a table, index, or view using an SQL statement outside your proce-
dure, they cannot do so. When users execute the procedure, they have a tem-
porary DBA privilege so the CREATE TABLE statement, for example, succeeds,
and you are guaranteed that every column added has a constraint placed on
it. In addition, objects created by users are owned by that user. For the
all_objects procedure, the two tables and the index are owned by whoever
executed the procedure.

Using Views
A view is a synthetic table. You can query it as if it were a table, and in some
cases you can update it as if it were a table. However, it is not a table, rather,
it is a synthesis of the data that exists in real tables and other views.
The basis of a view is a SELECT statement. When you create a view, you define
a SELECT statement that generates the contents of the view at the time the
view is accessed. A user also queries a view using a SELECT statement. The
database server merges the SELECT statement of the user with the one
defined for the view and then actually performs the combined statements.
The result has the appearance of a table; it is enough like a table that a view
even can be based on other views, or on joins of tables and other views.

Granting and Limiting Access to Your Database 11-19

Using Views

Because you write a SELECT statement that determines the contents of the
view, you can use views for any of the following purposes:
• To restrict users to particular columns of tables
You name only permitted columns in the select list in the view.
• To restrict users to particular rows of tables
You specify a WHERE clause that returns only permitted rows.
• To constrain inserted and updated values to certain ranges
You can use the WITH CHECK OPTION (discussed on page 11-25) to
enforce constraints.
• To provide access to derived data without having to store redundant data
in the database
You write the expressions that derive the data into the select list in the
view. Each time you query the view, the data is derived anew. The derived
data is always up to date, yet no redundancies are introduced into the
data model.
• To hide the details of a complicated SELECT statement
You hide complexities of a multitable join in the view so that neither users
nor application programmers need to repeat them.

Creating Views
The following example creates a view based on a table in the stores6
database:

CREATE VIEW name_only AS

SELECT customer_num, fname, lname FROM customer

The view exposes only three columns of the table. Because it contains no
WHERE clause, the view does not restrict the rows that can appear.

The following example creates a view based on a table that is available when
NLS has been enabled. In it, the view, column, and table names contain
foreign characters.

CREATE VIEW çà_va AS

SELECT numéro, nom FROM abonnés;

11-20 Granting and Limiting Access to Your Database

 Using Views

The following example is based on the join of two tables:

CREATE VIEW full_addr AS

SELECT address1, address2, city, state.sname, zipcode
FROM customer, state
WHERE customer.state = state.code

The table of state names reduces the redundancy of the database; it allows for
storing the full state names only once, which can be very useful for lengthy
state names such as Minnesota. This full_addr view lets users retrieve the
address as if the full state name were stored in every row. The following two
queries are equivalent:

SELECT * FROM full_addr WHERE customer_num = 105

SELECT address1, address2, city, state.sname, zipcode

FROM customer, state
WHERE customer.state = state.code
AND customer_num = 105

However, be careful when you define views that are based on joins. Such
views are not modifiable; that is, you cannot use them with UPDATE, DELETE,
or INSERT statements. (Modifying through views is covered beginning on
page 11-23.)
The following example restricts the rows that can be seen in the view:

CREATE VIEW no_cal_cust AS

SELECT * FROM customer WHERE NOT state = 'CA'

This view exposes all columns of the customer table, but only certain rows.
The following example is a view that restricts users to rows that are relevant
to them:

CREATE VIEW my_calls AS

SELECT * FROM cust_calls WHERE user_id = USER

All the columns of the cust_calls table are available but only in those rows
that contain the user-ids of the users who execute the query.

Granting and Limiting Access to Your Database 11-21

Using Views

Duplicate Rows from Views

It is possible for a view to produce duplicate rows, even when the underlying
table has only unique rows. If the view SELECT statement can return
duplicate rows, the view itself can appear to contain duplicate rows.
You can prevent this problem in two ways. One way is to specify DISTINCT
in the select list in the view. However, that makes it impossible to modify
through the view. The alternative is to always select a column or group of
columns that is constrained to be unique. (You can be sure only unique rows
are returned if you select the columns of a primary key or of a candidate key.
Primary and candidate keys are discussed in Chapter 8, “Building Your Data
Model.”)

Restrictions on Views
Because a view is not really a table, it cannot be indexed, and it cannot be the
object of such statements as ALTER TABLE and RENAME TABLE. The columns
of a view cannot be renamed with RENAME COLUMN. To change anything
about the definition of a view, you must drop the view and re-create it.
Because it must be merged with the user’s query, the SELECT statement on
which a view is based cannot contain any of the following clauses:
INTO TEMP The user’s query might contain INTO TEMP; if the view also
contains it, the data would not know where to go.
UNION The user’s query might contain UNION. No meaning has
been defined for nested UNION clauses.
ORDER BY The user’s query might contain ORDER BY. If the view also
contains it, the choice of columns or sort directions could be
in conflict.

When the Basis Changes

The tables and views on which a view is based can change in several ways.
The view automatically reflects most of the changes.
When a table or view is dropped, any views in the same database that depend
on it are automatically dropped.
The only way to alter the definition of a view is to drop and re-create it.
Therefore, if you change the definition of a view on which other views
depend, you must also re-create the other views (because they all have been
dropped).

11-22 Granting and Limiting Access to Your Database

 Using Views

When a table is renamed, any views in the same database that depend on it
are modified to use the new name. When a column is renamed, views in the
same database that depend on that table are updated to select the proper col-
umn. However, the names of columns in the views themselves are not
changed. For an example of this, recall the following view on the customer
table:

CREATE VIEW name_only AS

SELECT customer_num, fname, lname FROM customer

Now suppose that the customer table is changed in the following way:

RENAME COLUMN customer.lname TO surname

To select last names of customers directly, you must now select the new
column name. However, the name of the column as seen through the view is
unchanged. The following two queries are equivalent:

SELECT fname, surname FROM customer

SELECT fname, lname FROM name_only

When you alter a table by dropping a column, views are not modified. If they
are used, error -217 (Column not found in any table in the query)
occurs. The reason views are not dropped is that you can change the order of
columns in a table by dropping a column and then adding a new column of
the same name. If you do this, views based on that table continue to work.
They retain their original sequence of columns.
INFORMIX-OnLine Dynamic Server permits you to base a view on tables and
views in external databases. Changes to tables and views in other databases
are not reflected in views. Such changes might not be apparent until someone
queries the view and gets an error because an external table changed.

Modifying Through a View

It is possible to modify views as if they were tables. Some views can be
modified and others not, depending on their SELECT statements. The restric-
tions are different, depending on whether you use DELETE, UPDATE, or
INSERT statements.

Granting and Limiting Access to Your Database 11-23

Using Views

No modification is possible on a view when its SELECT statement contains

any of the following features:
• A join of two or more tables
Many anomalies arise if the database server tries to distribute modified
data correctly across the joined tables.
• An aggregate function or the GROUP BY clause
The rows of the view represent many combined rows of data; the
database server cannot distribute modified data into them.
• The DISTINCT keyword or its synonym UNIQUE
The rows of the view represent a selection from among possibly many
duplicate rows; the database server cannot tell which of the original rows
should receive the modification.
When a view avoids all these things, each row of the view corresponds to
exactly one row of one table. Such a view is modifiable. (Of course, particular
users can modify a view only if they have suitable privileges. Privileges on
views are discussed beginning on page 11-27.)

Deleting Through a View

A modifiable view can be used with a DELETE statement as if it were a table.
The database server deletes the proper row of the underlying table.

Updating a View
You can use a modifiable view with an UPDATE statement as if it were a table.
However, a modifiable view can still contain derived columns; that is, col-
umns that are produced by expressions in the select list of the CREATE VIEW
statement. You cannot update derived columns (sometimes called virtual
columns).
When a column is derived from a simple arithmetic combination of a column
with a constant value (for example, order_date+30), the database server
can, in principle, figure out how to invert the expression (in this case, by sub-
tracting 30 from the update value) and perform the update. However, much
more complicated expressions are possible, most of which cannot easily be
inverted. Therefore, the database server does not support updating any
derived column.

11-24 Granting and Limiting Access to Your Database

 Using Views

Figure 11-8 shows a modifiable view that contains a derived column, and an
UPDATE statement that can be accepted against it.

CREATE VIEW call_response(user_id,received,resolved,duration)AS

SELECT user_id,call_dtime,res_dtime, res_dtime -call_dtime
FROM cust_calls
WHERE user_id = USER

UPDATE call_response SET resolved = TODAY

WHERE resolved IS NULL

Figure 11-8 A modifiable view and an UPDATE statement

The duration column of the view cannot be updated because it represents an

expression (the database server cannot, even in principle, decide how to dis-
tribute an update value between the two columns named in the expression).
But as long as no derived columns are named in the SET clause, the update
can be performed as if the view were a table.
A view can return duplicate rows even though the rows of the underlying
table are unique. You cannot distinguish one duplicate row from another. If
you update one of a set of duplicate rows (for example, by using a cursor to
update WHERE CURRENT), you cannot be sure which row in the underlying
table receives the update.

Inserting into a View

You can insert rows into a view provided that the view is modifiable and
contains no derived columns. The reason for the second restriction is that an
inserted row must provide values for all columns, and the database server
cannot tell how to distribute an inserted value through an expression. An
attempt to insert into the call_response view, shown in Figure 11-8, would
fail.
When a modifiable view contains no derived columns, you can insert into it
as if it were a table. However, the database server uses null as the value for
any column that is not exposed by the view. If such a column does not allow
nulls, an error occurs and the insert fails.

Using WITH CHECK OPTION

It is possible to insert into a view a row that does not satisfy the conditions of
the view; that is, a row that is not visible through the view. It is also possible
to update a row of a view so that it no longer satisfies the conditions of the
view.

Granting and Limiting Access to Your Database 11-25

Using Views

If this is improper, you can add the clause WITH CHECK OPTION when you
create the view. This clause asks the database server to test every inserted or
updated row to ensure that it meets the conditions set by the WHERE clause
of the view. The database server rejects the operation with an error if the
conditions are not met.
In Figure 11-8, the view named call_response is defined as shown in the
following example:

CREATE VIEW call_response(user_id,received,resolved,duration)AS

SELECT user_id,call_dtime,res_dtime,res_dtime -call_dtime
FROM cust_calls
WHERE user_id = USER

It is possible to update the user_id column of the view, as shown in the

following example:

UPDATE call_response SET user_id = 'lenora'

WHERE received BETWEEN TODAY AND TODAY-7

The view requires rows in which user_id equals USER. If this update is
performed by a user named tony, the updated rows vanish from the view.
However, you can create the view as shown in the following example:

CREATE VIEW call_response (user_id,received,resolved,duration)AS

SELECT user_id,call_dtime,res_dtime,res_dtime-call_dtime
FROM cust_calls
WHERE user_id = USER
WITH CHECK OPTION

The preceding update by tony is rejected as an error.

11-26 Granting and Limiting Access to Your Database

 Privileges and Views

You can use the WITH CHECK OPTION feature to enforce any kind of data
constraint that can be stated as a Boolean expression. In the following exam-
ple, you can create a view of a table in which all the logical constraints on data
are expressed as conditions of the WHERE clause. Then you can require all
modifications to the table to be made through the view.

CREATE VIEW order_insert AS

SELECT * FROM orders O
WHERE order_date = TODAY -- no back-dated entries
AND EXISTS -- ensure valid foreign key
(SELECT * FROM customer C
WHERE O.customer_num = C.customer_num)
AND ship_weight < 1000 -- reasonableness checks
AND ship_charge < 1000
WITH CHECK OPTION

Because of EXISTS and other tests, all of which are expected to be successful
when retrieving existing rows, this is a most inefficient view for displaying
data from orders. However, if insertions to orders are made only through this
view (and you are not already using integrity constraints to constrain data),
it is impossible to insert a back-dated order, an invalid customer number, or
an excessive shipping weight and shipping charge.

Privileges and Views

When you create a view, the database server tests your privileges on the
underlying tables and views. When you use a view, only your privileges with
regard to the view are tested.

Privileges When Creating a View

When you create a view, the database server tests to make sure that you have
all the privileges needed to execute the SELECT statement in the view
definition. If you do not, the view is not created.
This test ensures that users cannot gain unauthorized access to a table by
creating a view on the table and querying the view.
After you create the view, the database server grants you, the creator and
owner of the view, at least the Select privilege on it. No automatic grant is
made to public as is the case with a newly created table.

Granting and Limiting Access to Your Database 11-27

Privileges and Views

The database server tests the view definition to see if the view is modifiable.
If it is, the database server grants you the Insert, Delete, and Update privi-
leges on the view, provided that you also have those privileges on the under-
lying table or view. In other words, if the new view is modifiable, the
database server copies your Insert, Delete, and Update privileges from the
underlying table or view, and grants them on the new view. If you have only
the Insert privilege on the underlying table, you receive only the Insert
privilege on the view.
This test ensures that users cannot use a view to gain access to any privileges
that they did not already have.
Because you cannot alter or index a view, the Alter and Index privileges are
never granted on a view.

Privileges When Using a View

When you attempt to use a view, the database server tests only the privileges
you are granted on the view. It does not also test your right to access the
underlying tables.
If you created the view, your privileges are the ones noted in the preceding
paragraph. If you are not the creator, you have the privileges that were
granted to you by the creator or someone who had the WITH GRANT OPTION
privilege.
The implication of this is that you can create a table and revoke public access
to it; then you can grant limited access privileges to the table through views.
This can be demonstrated through the previous examples using the hr_data
table. Its definition is repeated in Figure 11-9.

CREATE TABLE hr_data

(
emp_key INTEGER,
emp_name CHAR(40),
hire_date DATE,
dept_num SMALLINT,
user-id CHAR(18),
salary DECIMAL(8,2)
performance_level CHAR(1)
performance_notes TEXT
)

Figure 11-9 A table of confidential employee information (duplicate of Figure 11-1)

11-28 Granting and Limiting Access to Your Database

 Privileges and Views

In Figure 11-9, the example centered on granting privileges directly on this

table. The following examples take a different approach. Assume that when
the table was created, the following statement was executed:

REVOKE ALL ON hr_data FROM PUBLIC

(This is not necessary in an ANSI-compliant database.) Now you create a

series of views for different classes of users. For those who should have
read-only access to the nonsensitive columns, you create the following view:

CREATE VIEW hr_public AS

SELECT emp_key, emp_name, hire_date, dept_num, user_id
FROM hr_data

Users who are given the Select privilege for this view can see nonsensitive
data and update nothing. For the clerks in the Human Resources department
who must enter new rows, you create a different view, as shown in the
following example:

CREATE VIEW hr_enter AS

SELECT emp_key, emp_name, hire_date, dept_num
FROM hr_data

You grant these users both Select and Insert privileges on this view. Because
you, the creator of both the table and the view, have the Insert privilege on
the table and the view, you can grant the Insert privilege on the view to others
who have no privileges on the table.
On behalf of the clerk in the MIS department who enters or updates new
user-ids, you create still another view, as shown in the following example:

CREATE VIEW hr_MIS AS

SELECT emp_key, emp_name, user_id
FROM hr_data

This view differs from the previous view in that it does not expose the
department number and date of hire.
Finally, the managers need access to all columns and need the ability to
update the performance-review data for their own employees only. These
requirements can be met by creating a table, hr_data, that contains a
department number and a computer user-id for each employee. Let it be a

Granting and Limiting Access to Your Database 11-29

Summary

rule that the managers are members of the departments that they manage.
Then the following view restricts managers to rows that reflect only their
employees:

CREATE VIEW hr_mgr_data AS

SELECT * FROM hr_data
WHERE dept_num =
(SELECT dept_num FROM hr_data
WHERE user_id = USER)
AND NOT user_id = USER

The final condition is required so that the managers do not have update
access to their own row of the table. It is, therefore, safe to grant the Update
privilege to managers for this view, but only on selected columns, as shown
in this statement:

GRANT SELECT, UPDATE (performance_level, performance_notes)

ON hr_mgr_data TO peter_m

Summary
In a database that contains public material or is used only by you and trusted
associates, security is not an important consideration and few of the ideas in
this chapter are needed. But as more people are allowed to use and modify
the data, and as the data becomes increasingly confidential, you must spend
more time and be ever more ingenious at controlling the way people can
approach the data.
The techniques discussed here can be divided into the following two groups:
• Keeping data confidential
When the database resides in operating system files you can use features
of the operating system to deny access to the database. In any case, you
control the granting of the Connect privilege to keep people out of the
database.
When different classes of users have different degrees of authorization,
you must allow them all the Connect privilege. You can use table-level
privileges to deny access to confidential tables or columns. Or, you can
use a stored procedure to provide limited access to confidential tables or
columns. In addition, you can deny all access to tables and allow it only
through views that do not expose confidential rows or columns.

11-30 Granting and Limiting Access to Your Database

 Summary

• Controlling changes to data and database structure

You safeguard the integrity of the data model by restricting grants of the
Resource, Alter, References, and DBA privileges. You ensure that only
authorized persons modify the data by controlling the grants of the
Delete and Update privileges, and by granting the Update privilege on as
few columns as possible. You ensure that consistent, reasonable data is
entered by granting the Insert privilege only on views that express logical
constraints on the data. Alternatively, you can control the insertion and
modification of data, or the modification of the database itself, by limiting
access to constrictive stored procedures.

Granting and Limiting Access to Your Database 11-31

Summary

11-32 Granting and Limiting Access to Your Database

 Chapter

Understanding
Informix
Networking
Chapter Overview 3
12
What Is a Network? 4
Database Management System Configurations 4
A Single-User Configuration 4
Advantages and Disadvantages of a Single-User
System 6
A Local Multiuser Configuration 6
Advantages and Disadvantages of Local Multiuser
Systems 7
A Remote Configuration 8
Advantages and Disadvantages of Remote
Network Connections 8
Single-Machine Configuration Using Network
Communication 9
Advantages and Disadvantages of Local
Loopback 10
Distributed Databases 10
Advantages and Disadvantages of Distributed
Databases 11
Distributed Databases Using Multiple Vendor
Servers 11
 Connecting to Data on a UNIX Network 12
Example of Client / Server Connections 13
Environment Variables 14
Connection Information 15
SQL Connection Statements 16
Accessing Tables 16
Using Synonyms with Table Names 17
Synonym Chains 18
Protecting Your Data in a Networked Environment 19
Data Protection with INFORMIX-SE 19
Data Protection with INFORMIX-OnLine Dynamic Server 19
Data Replication 19
Archives 20
Data Integrity for Distributed Data 20
Two-Phase Commit 20
Summary 21

12-2 Understanding Informix Networking

Chapter Overview
This chapter gives an overview of the use of databases on a computer
network. It introduces some commonly used terminology and illustrates var-
ious network configurations. The chapter also presents an overview of how
the components of either a local or network connection fit together so that a
client application can find data on a database server.
We will discuss the following networking configurations that you can use
with Informix databases and point out some of their effects on performance
and usability:
• All on one machine
• A simple network connection
• Multiple connections on a network
• Data managed by non-Informix database servers (INFORMIX-TP/XA)
It is not enough to simply build a computer network; you have to re-examine
how your applications and database servers share data. This chapter also
covers the following topics concerned with managing data shared over a
network:
• Distributed data
• Connecting to data
• Protecting Your data
• Synonym chains
• Network transparency
The final sections of the chapter discuss protecting your data in a networked
environment.

Understanding Informix Networking 12-3

What Is a Network?

What Is a Network?
A computer network is a group of computers, workstations, and other devices
connected together over some kind of communications system to share
resources and data. A network site is simply the location of one of the com-
puters on the network. Sometimes the network sites are widely separated,
but they may also be in the same room or building. Two network sites can
even coexist on the same machine.
To make a practical computer network work, you must master a multitude of
technical details regarding hardware and software. There are far too many of
these details, and they change too fast to cover them in this book. This chap-
ter gives a conceptual discussion of some of the issues that you will encounter
when you use a computer network. For more information, refer to the man-
ual that accompanies the Informix client/server product that you use and/or
to a textbook on computer networks.

Database Management System Configurations

A relational database management system (RDBMS) includes all of the
components necessary to create and maintain a relational database. An Infor-
mix RDBMS has several pieces: the user interface, the database server and the
data itself. It is easiest to think of these pieces as all being located in the same
computer, but many other arrangements are possible. In some arrangements,
the application and the database server are on different computers, and the
data is distributed across several others.
The client applications of an RDBMS do not need to be modified to run on a
network. The communications tools handle the tasks of locating and attach-
ing to the database servers. To the application, a database on a networked
computer appears no different than a database on the computer where the
application resides.

A Single-User Configuration
Figure 12-1 shows a diagram of a simple database management system on a
single computer. The organization in Figure 12-1 is one you would typically
find on a personal computer running the DOS operating system. It is unusual
to have a single-user situation on a UNIX system, but you certainly can. One
example of a single-user UNIX system would be a desktop workstation in a
development environment.

12-4 Understanding Informix Networking

 Database Management System Configurations

personal computer

Connection
INFORMIX-
SE

application

user
databases

Figure 12-1 A database management system on a personal computer

The components of the system in Figure 12-1 are found in the following list:
• An application program
Any program that issues a query can be the application. It could, for
example, be a program written in INFORMIX-4GL, a C language program
with embedded SQL, or compiled screen forms and reports.
• A connection
On a simple system such as this one, the communication component is
frequently so well integrated with the system that it is usually omitted
from diagrams and not discussed. However, it does exist.
• A database server
The database server receives queries from the application, searches the
database, and returns information to the application. The database server
manages or administers the databases that it controls.
Because Figure 12-1 illustrates a personal computer, the database server
must be INFORMIX-SE. The database server in this diagram is a local
server because it resides on the same host computer as the client
application.

Understanding Informix Networking 12-5

Database Management System Configurations

• Databases. The databases are usually stored on a magnetic disk.

If this system were a UNIX system with INFORMIX-OnLine Dynamic
Server, the database might be located on some other medium, such as a
WORM (write-once read-many-times) drive controlled by
INFORMIX-OnLine/Optical.

Advantages and Disadvantages of a Single-User System

A configuration involving only one computer is the easiest configuration to
set up and maintain, and it gives the fastest access to data. However, the data
is not available to users on other computers and the size of the databases or
the amount of processing needed may outgrow the capacity of the computer.

A Local Multiuser Configuration

Figure 12-2 shows another database management system configuration, a
multiuser system with a local database server such as one you might find on
a computer with a UNIX operating system.

UNIX computer

application

user databases
Connection

application
database
server
user

Figure 12-2 A database management system on a UNIX computer

12-6 Understanding Informix Networking

 Database Management System Configurations

The components of the systems in Figure 12-2 are similar to the components
in Figure 12-1, as described in the following list:
• Application programs
Two or more applications use the same database server to access informa-
tion in the databases. You might have two users at individual terminals,
as shown, or you might have multiple windows on a single workstation.
• A connection
On a local UNIX system, the following types of connection are possible:
o Inter-process communication (IPC)
o Network connection
IPC is a UNIX feature that transfers information very quickly between the
application and the database server. It is available only when the applica-
tion and the database server reside on the same computer. INFORMIX-SE
databases use a type of IPC connection called unnamed pipes and
INFORMIX-OnLine Dynamic Server use an IPC connection technique
called shared memory.
• A database server, either INFORMIX-OnLine Dynamic Server or
INFORMIX-SE
• Databases

Advantages and Disadvantages of Local Multiuser Systems

A configuration that allows multiple users gives better access to the data than
does a single-user system. With a local database, it is still easy to set up and
maintain. However, like the single-user system, the data is not available to
users on other computers and the size of the databases or the amount of
processing needed may outgrow the capacity of the computer.
IPC shared memory provides very fast communication between the client
application and the database server. However, IPC shared memory commu-
nication is vulnerable to programming errors if the client application does
explicit memory addressing or overindexes data arrays. Such errors do not
affect the application if you use IPC unnamed pipes or network
communication. (See “Single-Machine Configuration Using Network
Communication” on page 12-9.)

Understanding Informix Networking 12-7

Database Management System Configurations

A Remote Configuration
Figure 12-3 shows a remote configuration, where the application resides on
one computer, and the database server and its associated databases reside on
another computer on the network. In contrast, the database servers in Figure
12-1 and Figure 12-2 are local database servers.

host 1 host 2

application

user
network
connection
database
network
server
application connection databases

user

Figure 12-3 A simple network connection

In Figure 12-3 the applications might be INFORMIX-ESQL/COBOL or

INFORMIX-ESQL/C applications. The database server might be
INFORMIX-OnLine Dynamic Server (UNIX) or INFORMIX-OnLine for Net-
Ware (PC).
Different computers are referred to as sites or host machines. The database
server in Figure 12-3 is a remote database server because it is on a different
host machine, or site, from the application that uses its services. A database
server can be local with respect to one application and remote with respect to
another application, as illustrated in Figure 12-6 on page 12-13.

Advantages and Disadvantages of Remote Network Connections

The configuration shown in Figure 12-3 is an example of distributed processing.
In distributed processing, multiple computers contribute to a single comput-
ing task. In this example, host1 handles the computing requirements of the
application, such as screen display, generation of reports and printing, and

12-8 Understanding Informix Networking

 Database Management System Configurations

host2 handles the computing required to manipulate information in the

databases. Using a network also gives the client application the opportunity
to access data from several computers.
Response from a database server using network communication is not as fast
as response from a database server using local IPC communication, because
of the extra computing required to prepare the data for network communica-
tion and because of the transmit time. A network is somewhat more difficult
to configure and maintain than a local system that uses IPC communication.

Single-Machine Configuration Using Network

Communication
Figure 12-4 shows a configuration that behaves as if there were multiple sites
residing on one machine. In this configuration, which is known as local loop-
back, all the components are on the same computer, but the connections are
made as if they were connected through a network.
Because the connections use network software, the database server appears
to the application as a remote site. The dashed line indicates the division
between the two sites. The database administrator configures the system to
provide local or local loopback connections (or both).
The software for the client application and the software for the database
server can be stored in the same directory, or the software can be stored in
two separate directories.

Understanding Informix Networking 12-9

Database Management System Configurations

computer

client site server site

application

user

network database
connection
server
databases

Figure 12-4 An example of local loopback

Advantages and Disadvantages of Local Loopback

Local loopback allows you to test network operations without requiring an
actual remote machine, but it is not as fast as a local IPC connection. How-
ever, unlike IPC shared-memory communication, local loopback is not vul-
nerable to corruption due to memory-addressing errors or overindexed data
arrays. (See “Advantages and Disadvantages of Local Multiuser Systems” on
page 12-7.)

Distributed Databases
Although a network lets you separate the application from the data, the
application still is limited to the contents of a single database. With most
database servers, you only can query or modify tables in the current
database.
A distributed database has information on multiple databases organized so that
it appears as a single database to the user. The data can be maintained by a
variety of database servers and located on computers supported by different
operating systems and communication networks.

12-10 Understanding Informix Networking

 Database Management System Configurations

The INFORMIX-OnLine Dynamic Server database server allows you to query

data in multiple databases anywhere on the network. When the INFORMIX-
TP/XA feature is added to INFORMIX-OnLine Dynamic Server, you can create
global transactions that span multiple computer systems and even multiple
XA-compliant database systems from different vendors. INFORMIX-SE does
not provide for distributed database processing.

Advantages and Disadvantages of Distributed Databases

You may find distributed databases useful because operations that use
databases are often naturally distributed into separate pieces, either organi-
zationally, geographically, or both. A distributed database system includes
the following advantages:
• Local data can be kept locally where it is most easily maintained and most
frequently used.
• Data from remote sites is available to all users.
• Duplicate copies can be maintained for safety of the data.
A distributed database system has the following disadvantages:
• Management of the distributed system is more involved than
management of a single-host system.
• Network access is slower than local access.

Distributed Databases Using Multiple Vendor Servers

You need INFORMIX-TP/XA if you use database management systems from
multiple vendors to store your data and you want to use one application to
access the data. INFORMIX-TP/XA is a library of functions that allows the
INFORMIX-OnLine Dynamic Server database server to act as a resource man-
ager in a distributed transaction-processing (DTP) environment that follows
an interface standard which was defined by the X/Open Company.
The terminology used by the XA standard is different from the terminology
used by Informix products and documentation. A transaction manager acts as
an intermediary and relays requests from a user interface to a resource man-
ager. Transaction managers are third-party products such as TUXEDO. In this
context, a resource manager corresponds to a database server. Figure 12-5
illustrates a configuration that uses transaction processing.

Understanding Informix Networking 12-11

Connecting to Data on a UNIX Network

transaction
manager
INFORMIX
interface OnLine/XA

user

vendor a’s
DBMS/XA

interface

user vendor b’s

DBMS/XA

client portion server portion resource databases

of each of each managers
application application

Figure 12-5 A configuration using a transaction manager

Connecting to Data on a UNIX Network

When the application and data are moved to separate computers, two
questions immediately arise: What connections can you implement? How do
you instruct your applications to find the data that is now somewhere else?
In fact, connecting a client application to a database server that is on a
networked computer is no different from connecting to a local database
server. To connect a client application with a database, you must consider the
following parts of your configuration:
• Environment variables
• Connection information
• Connection statements

12-12 Understanding Informix Networking

 Connecting to Data on a UNIX Network

This section summarizes how connections are made for Version 6.0 clients
and servers. Detailed instructions for setting up local and network connec-
tions for both Version 6.0 products and earlier products are given in the
INFORMIX-OnLine Dynamic Server Administrator’s Guide and the
INFORMIX-SE Administrator’s Guide.

Example of Client / Server Connections

Figure 12-6 shows both local and networked connections of Version 6.0
products. Although all these connections are possible, you cannot do them all
at once. Client X can make connection 1, 2, or 3, and Client Y can make
connection 4 or 5.

host 1 host 2

4
6.0 6.0 6.0 client Y
OnLine A OnLine B

1 2 sqlexecd

3-1 3-1
6.0
6.0 client X SE
3

Figure 12-6 Local and network connection of version 6.0 products

You can connect clients directly to INFORMIX-OnLine Dynamic Server

database servers using either shared memory (IPC) or a network communi-
cations. Notice also that OnLine database servers (OnLine A and OnLine B)
can communicate with each other.

Understanding Informix Networking 12-13

Connecting to Data on a UNIX Network

Local clients can connect directly to an INFORMIX-SEdatabase server using

unnamed pipes (IPC). Clients using network communications connect to
INFORMIX-SE by connecting first to a daemon, sqlexecd. (A daemon is a back-
ground process that listens for requests from other programs.) The sqlexecd
daemon is started by the system administrator.) When a client requests a con-
nection to an INFORMIX-SE database server, sqlexecd notices the request and
creates a temporary (light gray arrows) connection between the client and SE.
This temporary connection (for example, 3-1) enables the client and INFOR-
MIX-SE database server to establish a direct connection (3). Then sqlexecd
removes itself from the loop, leaving the client attached to the SE database
server.

Environment Variables
The Informix administrator must make sure that each user sets the correct
environment variables. The following list shows the most important
environment variables used by INFORMIX-OnLine Dynamic Server and
INFORMIX-SE:
• PATH
• INFORMIXDIR
• INFORMIXSERVER
• TERM
• DBPATH
The INFORMIXDIR environment variable must be set to the full pathname of
the directory where the Informix files reside. The PATH environment variable
must include the full pathname of the directory where the executables for
INFORMIX-SE and/or INFORMIX-OnLine Dynamic Server reside. These two
environment variables are required. After the Informix administrator has
specified them, you usually do not change them.
The INFORMIXSERVER environment variable is the name of the default
database server. It is also a required variable. You may choose to change
INFORMIXSERVER when you change applications.

The TERM (or TERMCAP and/or INFORMIXTERM) environment variable

enables the clients to recognize and communicate with your terminal. These
variables are system (and terminal) dependent. To set them, you may need
the assistance of your system administrator.
The DBPATH environment variable is optional. If the application does not
fully specify the location of an SE database server, the directories listed in
DBPATH are searched by the database server to find the specified database.

12-14 Understanding Informix Networking

 Connecting to Data on a UNIX Network

The DBPATH environment variable is also used to specify directory names for
reports, forms, and command files used by both INFORMIX-OnLine
Dynamic Server and INFORMIX-SE.

These environment variables are described in detail in Chapter 4 of the

Informix Guide to SQL: Reference.

Connection Information
When the INFORMIX-OnLine Dynamic Server or INFORMIX-SE
administrator installs and configures a database server, he or she must
cooperate with the UNIX system administrator to specify the information
that allows a client application to locate a database server. The application
specifies a database server and the Informix connectivity tools make the
connection. The connection is transparent; the client does not need to know
where the database server resides.
The connection information is specified in the $INFORMIXDIR/etc/sqlhosts
file and two UNIX systems files, /etc/hosts and /etc/services. The three files,
working together, specify the location of the database server and the type of
connection (protocol) for the database server. (See Figure 12-7.) On each com-
puter, every database server that might be accessed by an application must
have an entry in the $INFORMIXDIR/etc/sqlhosts file and corresponding
entries in the /etc/hosts and /etc/services files. These files are described in
detail in the INFORMIX-OnLine Dynamic Server Administrator’s Guide and
INFORMIX-SE Administrator’s Guide.

/etc/hosts
$INFORMIXDIR/etc/sqlhosts address of serverA
serverA
serverB
serverC /etc/services
protocol for serverA

Figure 12-7 Connectivity Information for Networks

Understanding Informix Networking 12-15

Accessing Tables

SQL Connection Statements

Before it can do anything else, the client application must open a database by
executing a CONNECT statement or a DATABASE statement. The CONNECT
statement is preferred because it conforms with both ANSI and
X/Open standards that attempt to specify uniform syntax for networked and
non-networked environments.
Here is an example of the CONNECT statement:

CONNECT TO databasename@servername

If the servername is omitted from the statement, the application connects to

the default database server specified by the INFORMIXSERVER environment
variable.
If the databasename is omitted, the application connects to the database server
servername but does not open a database. Before you can use a database, you
must issue a DATABASE, CREATE DATABASE, or START DATABASE
statement.
Notice that the CONNECT statement does not give any information about the
location of the database server. The location information for the database
server is in the sqlhosts file. (Refer to “Connection Information” on
page 12-15.)
The complete syntax for the CONNECT and DATABASE statements is covered
in Chapter 1 of the Informix Guide to SQL: Syntax.

Accessing Tables
The database that is opened by a CONNECT, DATABASE or CREATE
DATABASE statement is the current database. If you are using
INFORMIX-OnLine Dynamic Server, you can query tables that are not in the
current database. To refer to a table in a database other than the current data-
base, include the database name as part of the table name, as illustrated in the
following SELECT statement:

SELECT name, number FROM salesdb:contacts

12-16 Understanding Informix Networking

 Accessing Tables

The database is salesdb. The table in salesdb is named contacts. You can use
the same notation in a join. When you must specify the database name explic-
itly, the long table names can become cumbersome unless you use aliases to
shorten them. This is shown in the following example:

SELECT C.custname, S.phone

FROM salesdb:contacts C, stores:customer S
WHERE C.custname = S.company

To specify a table in a database that is managed by a different

INFORMIX-OnLine Dynamic Server database server, you must qualify the
database name with a database server name. For example, the following
SELECT statement refers to table customer from database masterdb, which
resides on the database server central:

SELECT O.order_num, C.fname, C.lname

FROM masterdb@central:customer C, sales@boston:orders O
WHERE C.customer_num = O.Customer_num
INTO TEMP mycopy

In the example, two tables are being joined. The joined rows are stored in a
temporary table in the current database. The tables are located in two data-
base servers; central and boston. Informix allows you to over-qualify (to give
more information than is required) table names, so it is possible, in this
example, that the current database is masterdb or sales.

Using Synonyms with Table Names

A synonym is a name that you can use in place of another name. The main use
of the CREATE SYNONYM statement is to make it more convenient to refer to
tables that are not in the current database.
The preceding example has been revised, as shown in the following example,
to use synonyms for the table names:

CREATE SYNONYM mcust FOR masterdb@central:customer;

CREATE SYNONYM bords FOR sales@boston:orders;

SELECT bords.order_num, mcust.fname, mcust.lname

FROM mcust, bords
WHERE mcust.customer_num = bords.Customer_num
INTO TEMP mycopy

Understanding Informix Networking 12-17

Accessing Tables

The CREATE SYNONYM statement stores the synonym name in the system
catalog table syssyntable in the current database. The synonym is available
to any query made within that database.
A short synonym makes it easier to write queries, but synonyms can play
another role. They allow you to move a table to a different database, or even
to a different computer, while leaving your queries the same.
Suppose you have several queries that refer to the tables customer and
orders. The queries are embedded in programs, forms, and reports. The
tables are part of database stores6, which is kept on database server avignon.
Now the decision is made that the same programs, forms, and reports are to
be made available to users of a different computer on the network (database
server nantes). Those users have a database that contains a table named
orders containing the orders at their location, but they need access to the
table customer at avignon.
To those users, the customer table is external. Does this mean you must pre-
pare special versions of the programs and reports, versions in which the
customer table is qualified with a database server name? A better solution is
to create a synonym in the users’ database, as shown in the following
example:

DATABASE stores6@nantes;
CREATE SYNONYM customer FOR stores6@avignon:customer

When the stored queries are executed in your database, the name customer
refers to the actual table. When they are executed in the other database, the
name is translated through the synonym into a reference to the external table.

Synonym Chains
To continue the preceding example, suppose that a new computer is added
to your network. Its name is db_crunch. The customer table and other tables
are moved to it to reduce the load on avignon. You can reproduce the table
on the new database server easily enough, but how can you redirect all
accesses to it? One way is to install a synonym to replace the old table, as
shown in the following example:

DATABASE stores6@avignon EXCLUSIVE;

RENAME TABLE customer TO old_cust;
CREATE SYNONYM customer FOR stores6@db_crunch:customer;
CLOSE DATABASE

12-18 Understanding Informix Networking

 Protecting Your Data in a Networked Environment

When you execute a query within stores6@avignon, a reference to table

customer finds the synonym and is redirected to the version on the new com-
puter. This is also true of queries executed from database server nantes. The
synonym in the database stores6@nantes still redirects references to
customer to database stores6@avignon; however, the new synonym there
sends the query to database stores6@db_crunch.
Chains of synonyms can be useful when, as in this example, you want to redi-
rect all access to a table in one operation. However, you should update all
users’ databases as soon as possible so their synonyms point directly to the
table. You incur extra overhead in handling the extra synonyms and, if any
computer in the chain is down, the table cannot be found.
It is possible to run an application against a local database and later to run the
identical application against a database in another machine. The program
runs equally well in either case (although it may run more slowly on the net-
work database). As long as the data model is the same, a program cannot tell
the difference between a local database server and a remote one.

Protecting Your Data in a Networked Environment

This section gives an overview of data-protection features used on a network.
Chapter 4, “Modifying Data,” has a general discussion of data protection.

Data Protection with INFORMIX-SE

INFORMIX-SE databases use the normal UNIX file structures, so SE databases
can be archived with the usual operating system backup procedures. On a
network, you can write the archives to a device in another location.

Data Protection with INFORMIX-OnLine Dynamic Server

OnLine includes several different tools that make multiple copies of data.
Each one has its own unique task in the area of data protection. The following
discussions give a general description of these tools and highlight the
distinctions between them.

Data Replication
In a general sense, data replication means that a given piece of data has several
distinct representations on several distinct servers. OnLine does data replica-
tion by using two networked computers. Each computer has an OnLine

Understanding Informix Networking 12-19

Protecting Your Data in a Networked Environment

database server and databases that have identical characteristics. One data-
base server is the primary and the other is the secondary. Data is always writ-
ten to the primary server and then transferred to the secondary server.
Applications can read data from either database server. Thus the secondary
site can provide a dual purpose: It provides data protection and it provides
improved performance for users at the secondary site who need to read, but
not write, data.
If the primary server (call it serverA) fails for any reason, the secondary
server (call it serverB) can become an independent server. Users who would
normally use the primary (serverA) can be switched to serverB. Service to all
the users can continue with a minimum of disruption while the original
primary server (serverA) is being repaired.
Data replication provides good data protection and continuity of service, but
it is expensive: Memory must be provided for two complete copies of the
data, shipping the data to a remote site affects performance, and manage-
ment requires the attention of the OnLine administrators on both sites.

Archives
OnLine provides specialized tools for making archives. Archives can be
prepared locally or on a networked computer. Archives should be stored in a
location that is physically removed from the database server, so it may be
convenient to build the archives on a networked computer located at a
different physical site.

Data Integrity for Distributed Data

INFORMIX-OnLine Dynamic Server allows you to update data in several
databases on different database servers. For example, a purchase order (a sin-
gle transaction) might require updating of information in databases on differ-
ent database servers. To maintain the integrity of the data, it is essential that
either all the different databases are updated or none of the databases are
updated.

Two-Phase Commit
Two-phase commit is a protocol that coordinates work performed at multiple
database servers on behalf of a single transaction. Unlike the data-replication
and archiving tools discussed earlier in this section, two-phase commit does
not make two copies of the data. It protects the validity of one transaction,

12-20 Understanding Informix Networking

 Summary

which involves several different databases that may or may not be on a

network. Because two-phase commit involves only one transaction, it is not
usually classed as data protection.
A transaction that involves multiple database servers is called a global
transaction. Two-phase commit is a natural extension of transaction handling,
which is discussed in “Interrupted Modifications” on page 4-22. The
INFORMIX-OnLine Dynamic Server Administrator’s Guide discusses
two-phase commit in detail.
The two-phase commit begins when a user process has completed all of its
work and requests a commit of the global transaction.

Phase 1
The current database server asks each participating database server if it can
commit its local transactions. Each database server responds “Yes” or “No.”

Phase 2
If all the database servers respond affirmatively, the current database server
tells each one to commit its transactions and then the global transaction is
complete. If any database server responds negatively or does not respond, all
database servers are instructed to abort the local transactions.

Summary
A network allows a client application to run in one machine, while the
database server operates in another computer to which the data is physically
attached. This provides distributed processing and the possibility of distrib-
uted database access. Many possible combinations exist of network software,
operating systems, and database servers, and each has subtleties that must be
mastered.

Understanding Informix Networking 12-21

Summary

12-22 Understanding Informix Networking

Section III: Using Advanced SQL
 Chapter

Optimizing Your
Queries
Chapter Overview 3
Optimizing Techniques 4
Verifying the Problem 4
Considering the Total System 5
Understanding the Application 5
Measuring the Application 6
Manual Timing 6
13
Time from Operating System Commands 6
Time from the Programming Language 6
Finding the Guilty Functions 7
Keeping an Open Mind 7
The Query Optimizer 8
How the Optimizer Works 8
How the Optimizer Joins Tables 8
Obtaining Input 9
Assessing Filters 10
Selecting Table-Access Paths 11
Selecting the Query Plan 11
Reading the Plan 12
Time Costs of a Query 14
Activities in Memory 14
Disk-Access Management 14
Page Buffers 16
The Cost of Reading a Row 16
The Cost of Sequential Access 17
The Cost of Nonsequential Access 18
The Cost of Rowid Access 18
The Cost of Indexed Access 18
 The Cost of Small Tables 19
The Cost of NLS Functionality 19
The Cost of Network Access 20
The Importance of Table Order 22
A Join Without Filters 22
A Join with Column Filters 23
Using Indexes 26
The Sort-Merge Join Technique 27
Making Queries Faster 27
Preparing a Test Environment 28
Studying the Data Model 28
Studying the Query Plan 29
Examining the Indexes 29
Replacing Autoindexes with Indexes 29
Creating Data Distributions on Filtered Columns 29
Using Composite Indexes 31
Using oncheck on Suspect Indexes 31
Dropping and Rebuilding Indexes After Updates 31
Refining the Column Filters 31
Avoiding Correlated Subqueries 32
Avoiding Difficult Regular Expressions 32
Avoiding Noninitial Substrings 33
Rethinking the Query 33
Rewriting Joins Through Views 33
Avoiding or Simplifying Sorts 33
Eliminating Sequential Access to Large Tables 34
Using Unions to Avoid Sequential Access 35
Reviewing the Optimization Level 36
Using a Temporary Table to Speed Queries 36
Using a Temporary Table to Avoid Multiple Sorts 36
Substituting Sorting for Nonsequential Access 37
Summary 41

13-2 Optimizing Your Queries

Chapter Overview
How much time should a query take? How many disk operations should the
computer perform while executing a query? For many queries, it does not
matter as long as the computer finds the information faster than a human can.
But some queries must be performed in a limited amount of time or must use
only a limited amount of machine power.
This chapter reviews techniques for making queries more efficient. It assumes
that you work with an existing database and cannot change its arrangement of
tables. (Techniques of designing a new database for reliability and
performance are discussed in Chapters 8 through 11.)
This chapter covers the following topics:
• A discussion of techniques for optimizing software, emphasizing all the
things to look at before you change any SQL statement.
• A discussion of the optimizer, the part of the database server that decides
how to perform a query. When you know what the optimizer looks for
when it forms a query plan, you can help it form more efficient ones.
• A discussion of the operations that take time during a query, so you can
better choose between fast and slow operations.
• An assortment of techniques to help the optimizer choose the fastest way
to accomplish a query.
This chapter concentrates on SQL performance, but performance problems
can arise from other parts of the programs in which SQL statements are
embedded. Two books that address general issues of performance are
The Elements of Programming Style by Kernighan and Ritchie (McGraw-Hill,
1978) and Writing Efficient Programs by Jon Louis Bentley (Prentice-Hall,
1982). Much of Bentley’s advice can be applied to SQL statements and
database design.

Optimizing Your Queries 13-3

Optimizing Techniques

Optimizing Techniques
Before you begin optimizing your SQL statements, begin considering them as
part of a larger system that includes the following components:
• One or more programs
The saved queries, compiled screen forms and reports, and programs in
one or more languages in which SQL statements are embedded
• One or more stored procedures
The compiled procedures, comprising SQL and SPL statements, which are
stored in an executable form in the database
• One or more computers
The machines that store the programs and the databases
• One or more maintainers
The people responsible for maintaining the programs
• One or more users
The people whose work the system is supposed to amplify or simplify
• One or more organizations
The groups that own the computers and choose what work is to be done
You may work in a large corporation where each of these components is
separate, or you may be the proprietor, maintainer, and sole user of a desktop
workstation. In every case, it is important to recognize two points about the
system as a whole. First, the goal of the system is to serve its users. Any effort
you spend optimizing programs is wasted unless the result helps the users in
some way. Second, SQL statements are only a small part of the system:
Effective improvements can be found in many parts of a system.
The following paragraphs outline a general procedure for analyzing any
computer-performance problem. Follow the procedure to help avoid over-
looking possible solutions, including the nontechnical ones that sometimes
provide the best answer to performance problems.

Verifying the Problem

Before you begin optimizing SQL statements, be sure the problem lies with
the program. How much more effective can the user be if you make the
program faster? If the answer is “not much,” look elsewhere for a solution.

13-4 Optimizing Your Queries

 Optimizing Techniques

Considering the Total System

Consider the entire system of programs, computers, and users within an
organization. You might find a better solution by changing schedules or man-
aging resources differently. If you are using INFORMIX-OnLine Dynamic
Server, you might want to make sure that your system is tuned for efficiency.
Maybe the problem operation would be faster if it were done at a different
time, on a different machine, or with different jobs running at the same time;
or maybe not. But you need to consider these possibilities before you decide
the best solution.
For information on tuning your INFORMIX-OnLine Dynamic Server system,
see the INFORMIX-OnLine Dynamic Server Administrator’s Guide.

Understanding the Application

Learn all about the application as a whole. A database application usually has
many parts, such as saved queries, screen forms, report specifications, and
programs. Consider them together with the user’s procedures, and ask
yourself the following questions:
• What is being done?
You may identify excess or redundant steps that can be eliminated. Even
if every step or action is important, it can help you to know them all and
their sequence of use.
• Why is it being done?
Especially in older applications, you may find steps that serve no
purpose; for example, an operation might have been put in long ago for
debugging and never removed.
• For whom is it being done?
Make sure that all output from the application is wanted and used by
someone. You may find output that nobody wants any more, or the
output may be needed infrequently or in a simpler format.
• Where is the slow part?
Isolate as closely as possible which steps are too slow. Your time is limited
too; spend time where it yields the best results.
By taking the time to understand the application, you can know all its parts,
how they are used, whether they are all essential, and which parts are too
slow.

Optimizing Your Queries 13-5

Optimizing Techniques

Measuring the Application

You cannot make meaningful progress on a performance problem until you
have measured it. You must find a way to take repeatable, quantitative
measurements of the slow parts of the application. This is crucial for the
following reasons:
• Without numbers you cannot accurately and specifically describe the
problem to users or to the organization.
To convey the problem appropriately, you need to determine measure-
ments such as “The report runs in 2 hours, 38 minutes,” or “The average
update takes 13 seconds, but during peak hours it takes as long as 49
seconds.”
• Without numbers you cannot set meaningful goals.
When you have specific measurements, you can obtain agreement on
numeric goals for performance.
• Without numbers you cannot measure and demonstrate your progress.
You need measurements to detect and demonstrate small improvements
and to choose between alternative solutions.

Manual Timing
You can obtain repeatable measurements with a hand-operated stopwatch.
With practice, most people can obtain timings that are repeatable within two-
tenths of a second. Manual timing is useful if you are measuring only a few
events that are at least several seconds long. (Very short events should be
measured over several cycles of averaging to reduce the error in
measurement.)

Time from Operating System Commands

Your operating system probably has a command that displays the time. You
can package an operation to be timed between two time commands in a
command script.

Time from the Programming Language

Most programming languages have a library function for the time of day. If
you have access to the source code, you can insert statements to measure the
time of specific actions. For example, if the application is written in

13-6 Optimizing Your Queries

 Optimizing Techniques

INFORMIX-4GL, you can use the CURRENT function to obtain the current
time as a DATETIME value. A 4GL program can perform automated timing
with code similar to that shown in the following example:

DEFINE start_time DATETIME HOUR TO FRACTION(2),

elapsed INTERVAL MINUTE(4) TO FRACTION(2)
LET start_time = EXTEND(CURRENT,HOUR TO FRACTION(2))

{ -- here perform the operation to be timed -- }

LET elapsed = EXTEND(CURRENT,HOUR TO FRACTION(2))-start_time

DISPLAY 'Elapsed time was ',elapsed

Elapsed time, in a multiprogramming system or network environment where

resources are shared among multiple processes, does not always correspond
to execution time. Most C libraries contain a function that returns the CPU
time of a program, and C functions can be called from 4GL programs and ACE
reports.

Finding the Guilty Functions

In most programs, a very small fraction of the code (typically 20 percent or
less) accounts for the bulk of the program execution time (typically 80 percent
or more). The 80-20 rule, as it is called, holds in most cases; often, the propor-
tions are even greater. After you establish a timing mechanism, use it to
define the hot spots in the application, that is, the statements that consume
most of the time. In addition, many computer systems provide profiling tools
that can help determine the location of hot spots.

Keeping an Open Mind

With embedded SQL, in which one statement can trigger thousands of disk
accesses, it is possible that the guilty 20 percent includes some SQL. However,
this is by no means certain. If you do not begin your examination of the problem
with an open mind, it is easy to overlook the part of a program that is
consuming the most time.
If the sluggish operations are not due to the SQL, turn to the books cited on
page 13-3. If the SQL statements need attention, you need to understand how
to structure them so they run more efficiently.

Optimizing Your Queries 13-7

The Query Optimizer

The Query Optimizer

The optimizer component of the database server decides how to perform a
query. Its most important job is to decide the order in which to examine each
table row. To make that decision, it has to decide on the most efficient way to
access each table—by a sequential scan of its rows, by an existing index, when
joining tables by a temporary index built for the occasion, or by sorting—and
it has to estimate how many rows each table contributes to the final result.
Optimizer design is not a science, and the optimizer is one part of a database
server that is under continuous development and improvement.

How the Optimizer Works

The optimizer in Informix database servers formulates all possible query
plans. For each plan, it estimates the number of table rows to be examined,
disk pages to be read, and the network accesses that the plan requires. It
selects the plan with the lowest estimate. The following sections provide a
general overview of some of the steps that the optimizer takes to evaluate
queries. These sections are meant to give you a basic idea of what the opti-
mizer does so that you can understand how to streamline your queries, as
described in later sections.

How the Optimizer Joins Tables

The optimizer in Informix database servers does an exhaustive search on all
joined tables, regardless of whether the joins contain filters or indexes and
removes redundant pairs as it builds the paths. Consider the following
example:

SELECT ... FROM customer c, items i, orders o

WHERE c.customer_num = o.customer_num
AND o.order_num = i.order_num

The optimizer forms all possible join pairs between the tables and then
removes redundant join pairs. It keeps the least-expensive join triples as
shown in the following example.

(c,i) (c,o) (i,c) (i,o) (o,c) (o,i)

13-8 Optimizing Your Queries

 The Query Optimizer

Obtaining Input
The optimizer can be successful only if its estimates are accurate. (They need
not be accurate in absolute terms, but they must be relatively accurate so that
better plans produce lower estimates than worse plans.) However, the opti-
mizer has only limited information. To make its work take a small fraction of
the execution time, it has to make do with the information in the system cat-
alog tables and other summary information. No time exists, for example, for
the optimizer to perform a SELECT COUNT(*) operation to obtain an accurate
count of the rows in a table.
The information available to the optimizer comes from the system catalog
tables. (See Chapter 2 of the Informix Guide to SQL: Reference for information
on system catalog tables.) In all Informix database servers, this information
includes
• The number of rows in a table (as of the most recent UPDATE STATISTICS
command)
• Whether a column is constrained to be unique
• The indexes that exist on a table, including the columns they encompass,
whether they are ascending or descending, and whether they are
clustered
• Data distribution information about filtered columns
The system catalog tables maintained by INFORMIX-OnLine Dynamic
Server supply the following additional input:
• The number of disk pages occupied by row data
• The depth of the index B+ tree structure (a measure of the amount of work
needed to perform an index lookup)
• The number of disk pages occupied by index entries
• The number of unique entries in an index (divided by the number of
rows, this suggests how many rows might match to a given key)
• Second-largest and second-smallest key values in an indexed column
Only the second-largest and second-smallest key values are noted because
the extreme values might be special out-of-range signals. The database server
assumes that key values are distributed smoothly between the second largest
and second smallest. Only the initial four bytes of these keys are stored. In
addition, if the index columns have distributions, the distributions, rather
than the key value, can be used to estimate the number of rows that match a
filter.

Optimizing Your Queries 13-9

The Query Optimizer

Assessing Filters
The optimizer first examines the expressions in the WHERE clause by looking
for filters. The optimizer estimates the selectivity of each filter it finds. The
selectivity is a number between 0 and 1 that indicates the fraction of rows the
optimizer thinks the filter can pass. A very selective filter that passes very few
rows is assigned a selectivity near 0; a filter that passes most rows is assigned
a selectivity near 1. For details on this process, see “Filter Selectivity
Assignments” on page 13-30.
The optimizer also notes other information about the query, such as what
filter columns are indexed and whether an index is sufficient. If only the val-
ues in an index are required (INFORMIX-OnLine Dynamic Server), then no
table read can be required (it is faster to read the index pages without the data
pages).
In addition, the optimizer notes whether an index can be used to evaluate a
filter. For this purpose, an indexed column is a column that has an index, or
one that is named first in a composite index. Consider the following cases:
• When an indexed column is compared to a literal, a host variable, or an
uncorrelated subquery, the database server can look up matching values
in the index instead of reading the rows.
• When an indexed column is compared to a column in another table (a join
expression), the database server can use the index to find matching
values, provided that the query plan calls for reading rows from the other
table first. The following join expression is an example:

WHERE customer.customer_num = orders.customer_num

If rows of customer are read first, values of customer_num can be looked

up in an index on orders.customer_num.
• Whether an index can be used in processing an ORDER BY clause
If all the columns in the clause appear in one index in the same sequence,
the database server can use the index to read the rows in their ordered
sequence, thus avoiding a sort.
• Whether an index can be used in processing a GROUP BY clause
If all the columns in the clause appear in one index in the same sequence,
the database server can read groups with equal keys from the index
without needing a sort.

13-10 Optimizing Your Queries

 The Query Optimizer

Selecting Table-Access Paths

The optimizer next chooses what it estimates to be the most efficient way of
accessing each table named in the query. It has the following choices:
• To read the rows of the table sequentially
• To read one of the indexes for the table and read the rows to which the
index points
• To create and use a temporary index
• To perform a sort merge
The choice between the first two options depends in large part on the
presence of a filter expression. When any filter exists on an indexed column,
the database server chooses the selected rows through the index and
processes only those rows.
When no filter exists, the database server must read all the rows anyway. It is
usually faster to simply read the rows, rather than reading the pages of an
index and then reading the rows. However, if the rows are required in sorted
order, there may be a net savings in reading the index and through it reading
the rows in sorted order.
The optimizer might choose a third option, creating a temporary index, in
two cases. When neither table in a join has an index on the joining column,
and the tables are large enough, the optimizer may decide that it is quicker to
build an index to one table than to read through that table sequentially for
each row of the other table. You can also use a temporary index to generate
rows in sorted or grouped sequence. (The alternative is to write the output
rows to a temporary table and sort that.)
The optimizer has a fourth option—performing a sort merge join—which
does not require a temporary index. This occurs only when at least one filter
of the join is an equality operator, and it has been determined that a
temporary index would be more costly.

Selecting the Query Plan

With all this information available, the optimizer generates all possible query
plans for joining tables in pairs. Then, if more than two tables are joined, the
optimizer uses the best two-table plans to form all plans for joining two tables
to a third, three to a fourth, and so on.

Optimizing Your Queries 13-11

The Query Optimizer

The optimizer adds any final work to the completed join plans. For example,
if the query contains an ORDER BY or GROUP BY clause, and a plan does not
produce rows in ordered sequence, the optimizer adds to the plan an estimate
of the cost of sorting the output. Sorting is discussed in “Time Cost of a Sort”
on page 13-15.
Finally, the optimizer selects a plan that appears to promise the least amount
of work, and passes it to the main part of the database server to be executed.

Reading the Plan

The choice the optimizer makes does not have to remain a mystery; you can
determine exactly what query plan it chooses. Execute the statement SET
EXPLAIN ON before you execute a query. Beginning with the next query, the
optimizer writes an explanation of its query plan to a particular file (the name
of the file and its location depend on the operating system in use). A typical
explanation is shown in Figure 13-1.
After repeating the query, the optimizer shows its estimate of the work to be
done (102 in Figure 13-1) in arbitrary units. A single disk access is one unit,
and other actions are scaled to that. This query plan was chosen over others
because the estimated cost for its execution was the lowest among all the
plans considered.

13-12 Optimizing Your Queries

 The Query Optimizer

QUERY:
------
SELECT C.customer_num, O.order_num, SUM (I.total_price)
FROM customer C, orders O, items I
WHERE C.customer_num = O.customer_num
AND O.order_num = I.order_num
GROUP BY C.customer_num, O.order_num;

Estimated Cost: 102

Estimated # of Rows Returned: 1
Temporary Files Required For: GROUP BY

1) pubs.o: SEQUENTIAL SCAN

2) pubs.c: INDEX PATH

(1) Index Keys: customer_num (Key-Only)

Lower Index Filter: pubs.c.customer_num = pubs.o.customer_num

3) pubs.i: INDEX PATH

(1) Index Keys: order_num

Lower Index Filter: pubs.i.order_num = pubs.o.order_num

Figure 13-1 Typical output produced by the optimizer with SET EXPLAIN ON

The optimizer also reveals its estimate of the number of rows that the query
produces. In Figure 13-1, the optimizer incorrectly estimates 1. In this exam-
ple, the estimate is incorrect because the optimizer has no way to estimate the
number of groups the GROUP BY clause produces.
In the body of the explanation, the optimizer lists the order in which tables
are accessed and the method, or access path, by which it reads each table. The
following breakdown clarifies the plan:
1. The orders table is read first. Because no filter exists on the orders table,
all of its rows must be read, that is, reading the table in physical order is
the “least costly” approach.
2. For each row of orders, a search is made for matching rows in the
customer table. The search uses the index on customer_num.
The notation Key-Only means that because only the c.customer_num
column is used in the join and output, only the index is read; no row is
read from the table.
3. For each row of orders that has a matching customer_num, a search is
made in the items table using the index on order_num.

Optimizing Your Queries 13-13

Time Costs of a Query

It is not always obvious why the optimizer makes its choices. By comparing
the plans produced by several variations in the query, you can usually
deduce some of its logic.

Time Costs of a Query

To execute a query, the database server spends most of its time in performing
two types of operation: reading data from disk, and comparing column val-
ues. Of the two, reading data is, by far, the slower task. This section examines
where the database server spends time. The next section explores the
implications of making queries faster.

Activities in Memory
The database server can process only data in memory. It must read a row into
memory before it can test it with a filter expression. It must read rows from
both tables before it can test a join condition. The database server prepares an
output row in memory by assembling the selected columns from other rows
in memory.
Most of these activities execute very quickly. Depending on the computer, the
database server can perform hundreds or even thousands of comparisons
each second. As a result, the time spent on in-memory work is usually a small
part of the whole execution time.
Two in-memory activities can take a significant amount of time. One is sort-
ing, as described in “Time Cost of a Sort” on page 13-15. The other activity is
processing comparisons using LIKE and MATCHES, when they begin with
“wild card” characters.

Disk-Access Management
It takes much longer to read a row from disk than to examine a row in
memory. The main goal of the optimizer is to reduce the amount of data that
must be read from disk, but it can eliminate only the most obvious
inefficiencies.

Disk Pages
The database server deals with disk storage in units called pages. A page is a
block of fixed size. The same size is used for all databases managed by one
database server. Indexes are stored in page-size units.

13-14 Optimizing Your Queries

 Time Costs of a Query

Time Cost of a Sort

A sort requires in-memory work as well as disk work.
The in-memory work is proportional to c∗w∗n∗log2(n), where
c is the number of columns being ordered, and represents the costs of extracting
column values from the row and concatenating them into a sort key.
w is proportional to the width of the combined sort key in bytes, and stands for the
work of copying or comparing one sort key. A numeric value for w would depend
strongly on the computer hardware in use.
n∗log2(n) is the number of comparisons that are made while sorting a table of n rows.
The disk work is proportional to 2∗n∗m, where n is again the number of rows. The factor m
represents the number of levels of merge the sort must use, a factor that depends on the number of
sort keys that can be held in memory.
When all the keys can be held in memory, m=1 and the disk work is proportional to 2n. In other
words, the rows are read, sorted in memory, and written. (When building an index, only index
pages are written.)
For tables from moderate to large sizes, rows are sorted in batches that fit in memory and then
the batches are merged. When m=2, the rows are read, sorted, and written in batches. Then the
batches are read again and merged and written again, resulting in disk work proportional to 4n.
For extremely large tables, the batches must be broken into batches. The disk work is then pro-
portional to 6n. In short, as table size increases, there are sudden, discontinuous changes in the
amount of disk work in a sort, from 2n to 4n to 6n. The table size at which these steps occur
depends on many factors, only one of which (key size) is in your control.
The best way to reduce the cost of sorting is to find a way to sort fewer rows, because the factor
n dominates both expressions. When that is not possible, look for ways to sort on fewer and nar-
rower columns. Not only does this reduce the factors c and w it also defers the step to the next
merge level.
Sorting involves writing information temporarily to disk. You can direct the disk writes to occur
in a dbspace managed by INFORMIX-OnLine Dynamic Server, or an operating system file. Writ-
ing to a dbspace often is much faster than writing to an operating system file. You use the
PSORT_DBTEMP environment variable to specify one or more directories where temporary infor-
mation is stored in operating system files in a round robin manner. If PSORT_DBTEMP is not set,
temporary information is written to the dbspaces specified by the DBSPACETEMP environment
variable, in a round robin manner. If both PSORT_DBTEMP and DBSPACETEMP are not set, tem-
porary information is written to the dbspaces specified by the DBSPACETEMP ONCONFIG param-
eter, in a round robin manner. If none of these environment variables or ONCONFIG parameter
are set, temporary information is written to operating system files in /tmp. For additional infor-
mation about environment variables, see Chapter 4 of Informix Guide to SQL: Reference. For
additional information about the ONCONFIG file, see the INFORMIX-OnLine Dynamic Server
Administrator’s Guide.

Optimizing Your Queries 13-15

Time Costs of a Query

The size of a page depends on the database server. With INFORMIX-OnLine

Dynamic Server, the page size is set when OnLine is initialized. It is usually
2 kilobytes (2,048 bytes), but you can ask the person who installed OnLine to
tell you what was chosen. Other Informix database servers use the file stor-
age of the host operating system, so their page size is the block size used by
the host operating system. One kilobyte (1,024 bytes) is typical.
It is possible to define tables so wide that one row fills a page (some database
servers permit a row to exceed the size of a page). However, a typical disk
page contains from 5 to 50 rows. An index entry consists of a key value and
a 4-byte pointer, so an index page typically contains from 50 to 500 entries.

Page Buffers
The database server has a set of memory spaces in which it keeps copies of
the disk pages it read most recently. It does this in case these pages are needed
again. If they are, the database server does not have to read them from disk.
As with the size of a disk page, the number of these page buffers depends on
the database server and the host operating system.

The Cost of Reading a Row

When the database server needs to examine a row that is not already in
memory, it must read it from disk. It does not read only one row; it reads the
pages that contain the row. (When a row is larger than a page, it reads as
many whole pages as necessary.) The cost of reading one page is the basic
unit of work that the optimizer uses for its calculations.

13-16 Optimizing Your Queries

 Time Costs of a Query

The actual cost of reading a page is variable and hard to predict. It is a

combination of the following factors:
Buffering The needed page might be in a page buffer already, in which
case the cost of access is near zero.
Contention If more than one application is contending for the use of the
disk hardware, the database server request can be delayed.
Seek time The slowest thing a disk does is to seek; that is, to move the
access arm to the track holding the data. Seek time depends
on the speed of the disk and the location of the disk arm
when the operation starts. Seek time varies from zero to a
large fraction of a second.
Latency The transfer cannot start until the beginning of the page
rotates under the access arm. This latency, or rotational delay,
depends on the speed of the disk and on the position of the
disk when the operation starts. Latency can vary from zero
to a few milliseconds.
The time cost of reading a page can vary from microseconds (for a page in a
buffer), to a few milliseconds (when contention is zero and the disk arm is
already in position), to hundreds of milliseconds.

The Cost of Sequential Access

Disk costs are lowest when the database server reads the rows of a table in
physical order. When the first row on that page is requested, its disk page is
read. Requests for subsequent rows are satisfied from the buffer until the
entire page is used. As a result, each page is read only once.
Provided that the database server is the only program using the disk, the
seek-time cost is also minimized. The disk pages of consecutive rows are usu-
ally in consecutive locations on the disk, so the access arm moves very little
from one page to the next. Even when that is not the case, groups of consec-
utive pages are usually located together so that only a few long seeks are
needed. For example, INFORMIX-OnLine Dynamic Server allocates disk
space to a table in multipage extents. Extents may be separated on the disk,
but within an extent the pages are close together.
Even latency costs may be lowest when pages are read sequentially. This
depends on the hardware and, when operating system files are used, on the
methods of the operating system. Disks are usually set up so that when pages
are read sequentially, latency is minimized. On the other hand, it is possible
to set up a disk so that a full rotation occurs between each sequential page,
drastically slowing sequential access.

Optimizing Your Queries 13-17

Time Costs of a Query

The Cost of Nonsequential Access

Disk costs are higher when the rows of a table are called for in a sequence that
is unrelated to physical order. Practical tables are normally much larger than
the database server page buffers, so only a small part of the table pages can
be held in memory. When a table is read in nonsequential order, only a few
rows are found in buffered pages. Usually one disk page is read for every row
that is requested.
Because the pages are not taken sequentially from the disk, usually both a
seek delay and a rotational delay occur before each page can be read. In short,
when a table is read nonsequentially, the disk-access time is much higher
than when it is read sequentially.

The Cost of Rowid Access

The simplest form of nonsequential access is to select a row based on its rowid
value. (The use of rowid in a SELECT statement is discussed in Chapter 3,
“Composing Advanced SELECT Statements.”) A rowid value specifies the
physical location of the row and its page. The database server simply reads
the page, incurring the costs already noted.

The Cost of Indexed Access

An additional cost is associated with finding a row through an index: The
index itself is stored on disk, and its pages must be read into memory.
The database server uses indexes in two ways. One way is to look up a row
given a key value. This is the kind of look up used when you join two tables,
as shown in the following statement:

SELECT company, order_num

FROM customer, orders
WHERE customer.customer_num = orders.customer_num

One table, probably customer, is read sequentially. Its value of

customer_num is used to search the index on the customer_num column of
orders. When a match is found, that row of orders is read.
An index look up works down from the root page to a leaf page. (See “The
Structure of an Index” on page 13-20.) The root page, because it is used so
often, is almost always found in a page buffer. The odds of finding a leaf page
in a buffer depend on the size of the index; the odds become poorer as the size
of the table increases.

13-18 Optimizing Your Queries

 Time Costs of a Query

If a table is very large, so that the number of index leaf pages is much larger
than the buffer space, almost every search causes a leaf page to be read in
addition to the page containing the row. The number of pages read is approx-
imately r * (index - height - 1), where r is the number of rows to be looked up.
Although it is costly, this kind of access is still a bargain, because the alterna-
tive is to perform each look up by reading the entire orders table, with disk
costs proportional to r 2 .
The other way the database server uses an index is to read it sequentially. It
does this to fetch the rows of a table in a specific order other than the physical
order. For this kind of access, the database server reads the leaf pages in
sequence, treating them as a list of rows in key sequence. Because the index
pages are read only once, the total of disk operations is proportional to r+i; i
is the number of leaf pages in the index.

The Cost of Small Tables

One conclusion you can draw from the preceding paragraphs is that small
tables are never slow. A table is small if it occupies so few pages that it can be
retained entirely in the page buffers. Any table that fits in four or fewer pages
is certainly small in this sense.
In the stores6 database, the state table that relates abbreviations to names of
states has a total size less than 1,000 bytes; it fits in, at most, two pages. It can
be included in any query at little cost. No matter how it is used—even if it is
read sequentially many times—it costs two disk accesses at most.

The Cost of NLS Functionality

The sorting and indexing of data containing foreign characters can cause
significant performance degradation for certain data sets. If you do not need
the functionality of NLS for all your character data, Informix recommends
that you use the CHAR data type instead of NCHAR (or VARCHAR instead of
NVARCHAR) for character columns whenever possible. Because CHAR (and
VARCHAR) data require simple value-based comparison, sorting, and index-
ing, the cost of CHAR (or VARCHAR) columns is thus relatively less expensive
than NCHAR (or NVARCHAR) columns.

Optimizing Your Queries 13-19

Time Costs of a Query

The Structure of an Index

An index is arranged as a hierarchy of pages (technically, a B+ tree), as
depicted in the figure below. At the bottom level, a sequence of leaf pages
contains the index key values of the table rows, each with a pointer to the
location of the row containing that value. The keys in the leaf pages are in
key sequence.
At each higher level, a page lists the highest key value to be found in each
page at the lower next level. The topmost level is a single root page from
which any other page of the index can be found.
root page

...

... ...
leaf page leaf page

... ...
The size of one entry is the width of the indexed columns in bytes, plus 4. If
you know that and the page size, you can estimate the number of entries per
page. This estimate is approximate because first, some data compression is
used in storing keys and second, not all pages are full.
The number of levels depends on the number of rows to be indexed and the
number of entries per page. Suppose a page holds 100 keys. Then a table of
100 rows has a single index page, the root page. A table with between 101
and 10,000 rows has a two-level index consisting of a root page and from 2
to 100 leaf pages. A table with as many as 1,000,000 rows has only a 3-level
index (so the preceding index must belong to a table of considerable size).

The Cost of Network Access

Whenever data is moved over a network, additional delays are imposed.
Networks are used in the following contexts:
• The application sends a query across the network to a database server in
another machine. The database server performs the query using its locally

13-20 Optimizing Your Queries

 Time Costs of a Query

attached disk. The output rows are returned over the network to the
application.
• Using the distributed data capability of INFORMIX-OnLine Dynamic
Server, a database server in one machine can read and update rows from
tables in databases located on other machines.
The data sent over a network consists of command messages and buffer-
sized blocks of row data. Although many differences of detail exist between
the two contexts, they can be treated identically under a simple model in
which one machine, the sender, sends a request to another, the responder,
which responds with a block of data from a table.
Whenever data is exchanged over a network, delays are inevitable in the
following situations:
• If the network is busy, the sender must wait its turn to transmit. Such
delays are typically less than a millisecond. But in a heavily loaded net-
work, they can increase exponentially to tenths of seconds and more.
• The responder may be handling requests from more than one sender, so
when the request arrives, it may be queued for a time that can range from
milliseconds to seconds.
• When the responder acts on the request, it incurs the time costs of disk
access and in-memory operations as described in the preceding topics.
• Transmission of the response is again subject to network delays.
The important point about network access is its extreme variability. In the
best case, when neither the network nor the responder is busy, transmission
and queueing delays are insignificant and the responder sends a row almost
as quickly as a local database server could do it. Furthermore, when the
sender asks for a second row, the page is likely still to be in the responder
page buffers.
Unfortunately, as network load increases, all these factors tend to worsen at
the same rate. Transmission delays rise in both directions. The queue at the
responder gets longer. And the odds of a page remaining in the responder
buffer become worse. Thus, network access costs can change quite suddenly
from very low to extremely high.
The optimizer that INFORMIX-OnLine Dynamic Server uses assumes that
access to a row over the network takes longer than access to a row in a local
database. This estimate includes the cost of retrieving the row from disk and
transmitting it across the network. The optimizer has no way to account for
varying network loads, so it is too pessimistic at times and too optimistic at
others.

Optimizing Your Queries 13-21

Time Costs of a Query

The Importance of Table Order

The order in which tables are examined has an enormous effect on the speed
of a join operation. This concept can be clarified with some examples that
show how the database server works.

A Join Without Filters

The following SELECT statement calls for a three-way join:

SELECT C.customer_num, O.order_num, SUM (I.total_price)

FROM customer C, orders O, items I
WHERE C.customer_num = O.customer_num
AND O.order_num = I.order_num
GROUP BY C.customer_num, O.order_num

For the moment, imagine that indexes were never invented. Without indexes,
the database server has no choice but to perform this operation using a sim-
ple nested loop. One of two practical query plans is displayed in Figure 13-2,
which is expressed in a programming pseudocode. A query plan states the
order in which the database server examines tables and the methods by
which it accesses the tables.

for each row in the customer table do:

read the row into C
for each row in the orders table do:
read the row into O
if O.customer_num = C.customer_num then
let Sum = 0
for each row in the items table do:
read the row into I
if I.order_num = O.order_num then
let Sum = Sum + I.total_price
end if
end for
prepare an output row from C,I,and Sum
end if
end for
end for

Figure 13-2 A query plan in pseudocode

13-22 Optimizing Your Queries

 Time Costs of a Query

This procedure reads the following rows:

• All rows of the customer table once
• All rows of the orders table once for each row of the customer table
• All rows of the items table once for each row of orders that appears in the
output (every row of orders should appear once)
This is not the only possible query plan; another merely reverses the roles of
customer and orders: for each row of orders, it reads all rows of customer
looking for a matching customer_num. It reads the same number of rows, but
in a different sequence.

A Join with Column Filters

In the preceding example, no difference exists in the amount of work that the
two possible query plans need to do. The presence of a column filter changes
things. A column filter is a WHERE expression that reduces the number of
rows that a table contributes to a join. The following example shows the
preceding query with a filter added:

SELECT C.customer_num, O.order_num, SUM (I.total_price)

FROM customer C, orders O, items I
WHERE C.customer_num = O.customer_num
AND O.order_num = I.order_num
AND O.paid_date IS NULL
GROUP BY C.customer_num, O.order_num

Optimizing Your Queries 13-23

Time Costs of a Query

The expression O.paid_date IS NULL filters out some rows, reducing the
number of rows that are used from the orders table. As before, two query
plans are possible. The plan that starts by reading from orders is displayed in
pseudocode in Figure 13-3.

for each row in the orders table do:

read the row into O
if O.paid_date is null then
for each row in the customer table do:
read the row into C
if O.customer_num = C.customer_num then
let Sum = 0
for each row in the items table do:
read the row into I
if I.order_num = O.order_num then
let Sum = Sum + I.total_price
end if
end for
prepare an output row from C,I,and Sum
end if
end for
end if
end for

Figure 13-3 One of two query plans in pseudocode

Let pdnull represent for the number of rows in orders that pass the filter. It is
the number that the following query returns:

SELECT COUNT(*) FROM orders WHERE paid_date IS NULL

Assume that just one customer exists for every order, as there should be. Then
we can say that the plan in Figure 13-3 reads the following rows:
• All rows of the orders table once
• All rows of the customer table, pdnull times
• All rows of the items table, pdnull times

13-24 Optimizing Your Queries

 Time Costs of a Query

An alternative plan is shown in Figure 13-4; it reads from customer first.

for each row in the customer table do:

read the row into C
for each row in the orders table do:
read the row into O
if O.paid_date is null and
O.customer_num = C.customer_num then
let Sum = 0
for each row in the items table do:
read the row into I
if I.order_num = O.order_num then
let Sum = Sum + I.total_price
end if
end for
prepare an output row from C,I,and Sum
end if
end for

Figure 13-4 The alternative query plan in pseudocode

Because the filter is not applied in the first step, shown in Figure 13-4, this
plan reads the following rows:
• All rows of the customer table once
• All rows of the orders table, once for every row of customer
• All rows of the items table, pdnull times
The query plans in Figure 13-3 and Figure 13-4 produce the same output,
although in different sequences. They differ in that one reads a table pdnull
times and the other reads a table SELECT COUNT(*) FROM customer
times. The choice the optimizer makes between them could make a difference
of thousands of disk accesses in a real application.

Optimizing Your Queries 13-25

Time Costs of a Query

Using Indexes
The preceding examples do not use indexes or constraints, which is
unrealistic. Almost all tables have one or more indexes or constraints, and
their presence makes a difference in the query plan. Figure 13-5 shows the
outline of a query plan for the previous query as it might be constructed
using indexes. (Also see “The Structure of an Index” on page 13-20.)

for each row in the customer table do:

read the row into C
look up C.customer_num in index on orders.customer_num
for each matching row in the orders table do:
read the row into O
if O.paid_date is null then
let Sum = 0
look up O.order_num in index on items.order_num
for each matching row in the items table do:
read the row into I
let Sum = Sum + I.total_price
end for
prepare an output row from C,I,and Sum
end if
end for
end for

Figure 13-5 A query plan using indexes in pseudocode

The keys in an index are sorted so that when the first matching key is found,
any other rows with identical keys can be read without further searching.
This query plan reads only the following rows:
• All rows of the customer table once
• All rows of the orders table once
• Those rows of the items table that match to pdnull rows from orders
This is a huge reduction in effort compared with the plans without indexes.
(An inverse plan, reading orders first and looking up rows in customer by its
index, is equally effective.)
However, any plan that uses an index must read index data from disk as well
as read row data. It is difficult to predict the number of index pages that are
read, because some number of recently used index pages are retained in
memory and are found there when they are needed. (See “Disk-Access Man-
agement” on page 13-14.)

13-26 Optimizing Your Queries

 Making Queries Faster

The physical order of a table also affects the cost of index use. The physical
order of the customer table is customer-number order if a clustered index is
on that column. It is approximately that order if customer numbers are
SERIAL values generated by the database server.

However, if the physical order is random with respect to customer numbers,

each search of the index on orders might lead to a different page of index
keys, so that an index page is read for almost every row.
Note that when NLS is enabled, indexes built on NCHAR or NVARCHAR
columns are sorted using a country-specific comparison value. For example,
the Spanish double-l character (ll) may be treated as a single unique character
instead of a pair of l’s.

The Sort-Merge Join Technique

The sort-merge table join provides a cost-effective alternative to constructing
a temporary index for a nested-loop join. It is considered by the optimizer
only when the join is based on an equality filter and no index exists that can
accomplish the join.
The sort-merge join does not change the fundamental strategy of the
optimizer. Rather, it provides a richer set of alternatives to nested-loop joins
using temporary indexes and alters the way the ORDER BY and GROUP BY
paths are analyzed. The path chosen by the optimizer is displayed in the out-
put when the SET EXPLAIN ON statement is issued, as described in Chapter 1
of the Informix Guide to SQL: Syntax and in “Reading the Plan” on page 13-12
of this manual.

Making Queries Faster

In general, you can speed up a query by changing it so that it performs the
following actions:
• Reads fewer rows
• Avoids a sort, sorts fewer rows, or sorts on a simpler key
• Reads rows sequentially instead of nonsequentially
The way to achieve these ends is not always obvious. The specific methods
depend on the details of the application and the database design. The follow-
ing paragraphs suggest a general approach, followed by some techniques
that apply in limited circumstances.

Optimizing Your Queries 13-27

Making Queries Faster

Preparing a Test Environment

First, select a single query that is too slow. Then set up an environment in
which you can take predictable, repeatable timings of that query. Without this
environment, you can never be sure whether a change helps.
If you are using a multiuser system or a network, so that system load varies
widely from hour to hour, you may need to perform your experiments at the
same time each day to obtain repeatable results. Preferably, you should exe-
cute tests when the system load is consistently light so that you are truly
measuring only your query.
If the query is embedded in a complicated program, consider extracting the
SELECT statement and executing it interactively or embedding it in a simpler
program.
If the real query takes many minutes or hours to complete, it may be a good
idea to prepare a scaled-down database in which you can run tests more
quickly. This can be helpful, but you must be aware of the following potential
problems:
• The optimizer can make different choices in a small database than in a
large one, even when the relative sizes of tables are the same. Verify that
the query plan is the same in the real and the model databases.
• Execution time is rarely a linear function of table size. Sorting time, for
example, increases faster than table size, as does the cost of indexed
access when an index goes from two to three levels. What appears to be a
big improvement in the scaled-down environment can be insignificant
when applied to the full database.
Therefore, any conclusion you reach as a result of tests in the model database
must be tentative until verified in the large one.

Studying the Data Model

Study the definitions of all the tables, views, and indexes used in the data-
base. You can examine these details interactively using the Table option of
DB-Access or INFORMIX-SQL. Pay particular attention to the indexes that
exist, to the data types of columns used in join conditions and for ordering,
and to the presence of views. This chapter is written assuming that you can-
not change the data model. Nevertheless, the better you understand it, the
better you can understand the query plans chosen by the optimizer.
For information on data types and views, see Chapter 9, “Implementing Your
Data Model.”

13-28 Optimizing Your Queries

 Making Queries Faster

Studying the Query Plan

Use the output from SET EXPLAIN ON to determine the query plan being
used. The following list contains some things to look for:
Indexes Examine the output to see if indexes are used and if so, how
they are used.
Filters Examine your filters to make sure that they are selective
enough.
Query Examine your query again to make sure that the access plan
is optimal.

Examining the Indexes

You should study the indexes or the lack of indexes on the columns of your
query. Consider using some of the methods that are described in the
following sections to improve your query.

Replacing Autoindexes with Indexes

If the query plan includes an autoindex path to a large table, take it as a
recommendation from the optimizer that an index should be on that column.
It is reasonable to let the database server build and discard an index if you
perform the query only occasionally, but if the query is done even daily, you
save time by creating a permanent index.

Creating Data Distributions on Filtered Columns

Executing the UPDATE STATISTICS statement in MEDIUM or HIGH mode
causes information about the distribution of data values for each column
specified to be placed in a system catalog table called sysdistrib. This infor-
mation is used at optimization time to estimate the number of rows that
match a given query, and from that, to estimate the cost of retrieving those
values. From these cost estimates, the optimizer determines the best path of
execution. Data distributions supercede the colmin and colmax columns in
the syscolumns system catalog table in estimating the number of rows
returned by each part of a WHERE clause.
When using data-distribution statistics for the first time, try updating
statistics in MEDIUM mode on all tables followed by updating statistics in
HIGH mode on columns that head indexes. This will produce single-table
query estimates that, on the average, have an error less than 2.5 percent of the
total number of rows in the table. (For columns with HIGH mode
distributions, the default resolution is 1.0 percent.)
Optimizing Your Queries 13-29
Making Queries Faster

Unless column values change by a considerable margin, it is unnecessary to

regenerate data distributions. You can verify the accuracy of the distribution
by comparing dbschema (-hd option) output with appropriately constructed
SELECT statements.

For additional information about data distributions and the UPDATE

STATISTICS statement, see Chapter 1 of the Informix Guide to SQL: Syntax.

Filter Selectivity Assignments

The following table lists some of the selectivities that the optimizer assigns to filters of different
types. It is not an exhaustive list, and other expression forms may be added in the future.

Expression Form Selectivity (F)

indexed-col = literal-value
indexed-col = host-variable
indexed-col IS NULL F = 1/(number of distinct keys in index)
tab1.indexed-col = tab2.indexed-col F = 1/(number of distinct keys in the larger index)
indexed-col > literal-value F = (2nd-max - literal-value)/(2nd-max - 2nd-min)
indexed-col < literal-value F = (literal-value - 2nd-min)/(2nd-max - 2nd-min)
any-col IS NULL
any-col = any-expression F = 1/10
any-col > any-expression
any-col < any-expression F = 1/3
any-col MATCHES any-expression
any-col LIKE any-expression F = 1/5

EXISTS subquery F = 1 if subquery estimated to return >0 rows, else 0

NOT expression F = 1 - F(expression)

expr1 AND expr2 F = F(expr1) × F(expr2)
expr1 OR expr2 F = F(expr1) + F(expr2) - F(expr1) × F(expr2)

any-col IN list treated as anycol = item1 OR…OR anycol = itemn

any-col relop ANY subquery treated as any-col relop value1 OR…OR anycol relop valuen
for estimated size of subquery n
Key:
indexed-col: first or only column in an index (OnLine only)
2nd-max, 2nd-min: second-largest and -smallest key values in indexed column (OnLine only)
any-col: any column not covered by a preceding formula

Selectivities are more accurate with distributions than through current index
information due to finer granularity statistics. However, if data distributions
are not used, filter selectivities are used as shown in the following chart.

13-30 Optimizing Your Queries

 Making Queries Faster

Using Composite Indexes

The optimizer can use a composite index (one that covers more than one
column) in several ways. In addition to its usual function of ensuring that the
values in columns abc are unique, an index on the columns a, b, and c (in that
order) can be used in the following ways:
• To evaluate filter expressions on column a
• To join column a to another table
• To implement ORDER BY or GROUP BY on columns a, ab, or abc (but not on
b, c, ac, or bc)
When you create a composite index, place the columns in the index in the
order from most frequently used to least frequently used.
If your application is performing several long queries, each of which involves
a sort on the same columns, you might save time by creating a composite
index on those columns. In effect, you perform the sort once and save its
output for use in every query.

Using oncheck on Suspect Indexes

With some database servers, it is possible for an index to become ineffective
because it has been internally corrupted. If a query that uses an index has
slowed down inexplicably, use the oncheck utility to check the integrity of
the index and to repair it if necessary. (The secheck utility does the same job
for INFORMIX-SE.)

Dropping and Rebuilding Indexes After Updates

After extensive amounts of updating (after the replacement of a fourth or
more of the rows of a table), the structure of an index can become inefficient.
If an index seems to be less effective than it should be, yet oncheck reports no
errors, try dropping the index and re-creating it.

Refining the Column Filters

The most efficient queries access only a small body of information. You
control the amount of information returned in the SELECT statement with
WHERE clauses, which are commonly called filters. These sections suggest
the following ways to improve the WHERE clause of your query:
• Avoiding correlated subqueries
• Avoiding difficult regular expressions

Optimizing Your Queries 13-31

Making Queries Faster

• Avoiding noninitial substrings

Avoiding Correlated Subqueries

A correlated subquery is one in which a column label appears in both the
select list of the main query and the WHERE clause of the subquery. Because
the result of the subquery might be different for each row that the database
server examines, the subquery executes anew for every row if the current cor-
relation values are different from the previous ones. The optimizer tries to
use an index on the correlation values to cluster identical values together.
This procedure can be extremely time consuming. Although some queries
cannot be stated in SQL without the use of a correlated subquery, many can
be rewritten as joins that will execute more rapidly.
When you see a subquery in a time-consuming SELECT statement, look to see
if it is correlated. (An uncorrelated subquery, one in which no row values
from the main query are tested within the subquery, is executed only once.)
If so, try to rewrite the query to avoid it. If you cannot, look for ways to
reduce the number of rows that are examined; for instance, try adding other
filter expressions to the WHERE clause, or try selecting a subset of rows into
a temporary table and searching only them.

Avoiding Difficult Regular Expressions

The MATCHES and LIKE keywords support wildcard matches—technically
known as regular expressions. Some regular expressions are more difficult than
others for the database server to process. A wildcard in the initial position, as
in the following example (find customers whose first names do not end in y),
forces the database server to examine every value in the column, as shown in
the following example:

SELECT * FROM customer WHERE fname NOT LIKE '%y'

An index cannot be used with such a filter so the table in this example must
be accessed sequentially.
If a difficult test for a regular expression is essential, avoid combining it with
a join. If necessary, process the single table, applying the test for a regular
expression to select the desired rows. Save the result in a temporary table,
and join that table to the others.
Regular-expression tests with wildcards in the middle or at the end of the
operand do not prevent the use of an index when one exists.

13-32 Optimizing Your Queries

 Making Queries Faster

Avoiding Noninitial Substrings

A filter based on a noninitial substring of a column also requires every value
in the column to be tested, as shown in the following example:

SELECT * FROM customer

WHERE zipcode[4,5] > '50'

An index cannot be used to evaluate such a filter.

The optimizer uses an index to process a filter that tests an initial substring of
an indexed column. However, the presence of the substring test can interfere
with the use of a composite index to test both the substring column and
another column.

Rethinking the Query

After you understand what the query is doing, look for ways to obtain the
same output with less effort. The following suggestions can help you rewrite
your query more efficiently:
• Rewriting joins through views
• Avoiding or simplifying sorts
• Eliminating sequential access to large tables
• Using unions to avoid sequential access

Rewriting Joins Through Views

You might find that the query joins a table to a view that is itself a join. If you
rewrite the query to join directly to fewer tables, you might produce a simpler
query plan.

Avoiding or Simplifying Sorts

A sort is not necessarily negative. The sort algorithm of the database server
is highly tuned and extremely efficient. It is certainly as fast as any external
sort program you might apply to the same data. As long as the sort is per-
formed only occasionally or to a relatively small number of output rows, no
need exists to avoid it.

Optimizing Your Queries 13-33

Making Queries Faster

However, you should avoid or simplify repeated sorts of large tables. The
optimizer avoids a sort step whenever it can produce the output in its proper
order automatically by using an index. The following factors prevent the
optimizer from using an index:
• One or more of the ordered columns is not included in the index.
• The columns are named in a different sequence in the index and the
ORDER BY or GROUP BY clause.
• The ordered columns are taken from different tables.
Another way to avoid sorts is discussed in “Using a Temporary Table to
Speed Queries” on page 13-36.
If a sort is necessary, look for ways to simplify it. As discussed in “Time Cost
of a Sort” on page 13-15, the sort is quicker if you can sort on fewer or
narrower columns.

Eliminating Sequential Access to Large Tables

Sequential access to a table other than the very first one in the plan is
ominous; it threatens to read every row of the table once for every row
selected from the preceding tables. You should be able to judge how many
times that is: perhaps a few, but perhaps hundreds or even thousands.
If the table is small, it can do no harm to read it repeatedly; the table resides
completely in memory. Sequential search of an in-memory table can be faster
than searching the same table through an index, especially if also having its
index pages in memory pushes other useful pages out of the
buffers.
When the table is larger than a few pages, however, repeated sequential
access is deadly to performance. One way to prevent this is to provide an
index to the column being used to join the table.
The subject of indexing is discussed in the context of database design in
“Managing Indexes” on page 10-19. However, any user with Resource
privileges can build additional indexes. Use the CREATE INDEX statement to
make an index.
An index consumes disk space proportional to the width of the key values
and the number of rows. (See “The Structure of an Index” on page 13-20.)
Also, the database server must update the index whenever rows are inserted,
deleted, or updated; this slows these operations. If necessary, you can use the
DROP INDEX statement to release the index after a series of queries, thus
freeing space and making table updates easier.

13-34 Optimizing Your Queries

 Making Queries Faster

Using Unions to Avoid Sequential Access

Certain forms of WHERE clauses force the optimizer to use sequential access
even when indexes exist on all the tested columns. The following query
forces sequential access to the orders table:

SELECT * FROM orders

WHERE (customer_num = 104 AND order_num > 1001)
OR order_num = 1008

The key element is that two (or more) separate sets of rows are retrieved. The
sets are defined by relational expressions that are connected by OR. In the
following example, one set is selected by this test:

(customer_num = 104 AND order_num > 1001)

The other set is selected by the test shown in the following example:

order_num = 1008

The optimizer uses a sequential access path even though indexes exist on the
customer_num and order_num columns.
You can speed queries of this form by converting them into UNION queries.
Write a separate SELECT statement for each set of rows and connect them
with the UNION keyword. The following example shows a rewritten version
of the preceding example:

SELECT * FROM orders

WHERE (customer_num = 104 AND order_num > 1001)

UNION

SELECT * FROM orders

WHERE order_num = 1008

The optimizer uses an index path for each query.

You can also rewrite the WHERE clause as shown in the following example:

WHERE (customer_num = 104 OR order_num = 1008) AND

(order_num > 1001 OR order_num = 1008)

Optimizing Your Queries 13-35

Making Queries Faster

Reviewing the Optimization Level

You normally obtain optimum overall performance with the default
optimization level, HIGH. The time it takes to optimize the statement is
usually unimportant. However, if experimentation with your application
reveals that your query is still taking too long, you may want to set your opti-
mization level to LOW, and then check the SET EXPLAIN output to see if the
optimizer chose the same query plan as before.
You can specify a HIGH or LOW level of database server optimization with the
SET OPTIMIZATION statement. This statement is described in detail in
Chapter 1 of the Informix Guide to SQL: Syntax.

Using a Temporary Table to Speed Queries

Building a temporary, ordered subset of a table can speed up a query. It can
help to avoid multiple-sort operations and can simplify the work of the
optimizer in other ways.

Using a Temporary Table to Avoid Multiple Sorts

For example, suppose your application produces a series of reports on
customers who have outstanding balances, one report for each major postal
area, ordered by customer name. In other words, a series of queries occur
each of the following form (using hypothetical table and column names):

SELECT cust.name, rcvbles.balance, ...other columns...

FROM cust, rcvbles
WHERE cust.customer_id = rcvbles.customer_id
AND rcvbls.balance > 0
AND cust.postcode LIKE '98_ _ _'
ORDER BY cust.name

This query reads the entire cust table. For every row with the right postcode,
the database server searches the index on rcvbles.customer_id and performs
a nonsequential disk access for every match. The rows are written to a
temporary file and sorted.
This procedure is acceptable if the query is done only once, but this example
includes a series of queries, each incurring the same amount of work.

13-36 Optimizing Your Queries

 Making Queries Faster

An alternative is to select all customers with outstanding balances into a

temporary table, ordered by customer name, as shown in the following
example:

SELECT cust.name, rcvbles.balance, ...other columns...

FROM cust, rcvbles
WHERE cust.customer_id = rcvbles.customer_id
AND cvbls.balance > 0
ORDER BY cust.name
INTO TEMP cust_with_balance

Now you can direct queries against the temporary table in this form, as
shown in the following example:

SELECT *
FROM cust_with_balance
WHERE postcode LIKE '98_ _ _'

Each query reads the temporary table sequentially, but the table has fewer
rows than the primary table. No nonsequential disk accesses are performed.
No sort is required because the physical order of the table is the desired order.
The total effort should be considerably less than before.
One possible disadvantage exists: Any changes made to the primary table
after the temporary table has been created are not reflected in the output. This
is not a problem for some applications, but it may be for some.

Substituting Sorting for Nonsequential Access

Nonsequential disk access is the slowest kind. SQL hides this fact and makes
it easy to write queries that access huge numbers of pages nonsequentially.
Sometimes you can improve a query by substituting the sorting ability of the
database server for nonsequential access.The following example
demonstrates this and also demonstrates how to make numerical estimates
of the cost of a query plan.
Imagine a large manufacturing firm whose database includes three tables
shown in schematic form in Figure 13-6. (Table diagrams in this form are
described in Chapter 8, “Building Your Data Model.” Not all the columns are
shown.)
The first table, part, contains the parts used in the products of the company.

Optimizing Your Queries 13-37

Making Queries Faster

The second, vendor, contains data about the vendors who provide the parts.
The third, parven, records which parts are available from which vendors and
at what price.

part vendor
part_num part_desc vendor_num vendor_name
pk pk
100,000 spiral spanner 9,100,000 Wrenchers SA
999,999 spotted paint 9,999,999 Spottiswode

parven
part_num vendor_num price
pk, fk pk, fk
100,000 9,100,000 $14.98
999,999 9,999,999 $0.066

Figure 13-6 Three tables from a manufacturing database

The following query is run regularly against these tables to produce a report
of all available prices:

SELECT part_desc, vendor_name, price

FROM part, vendor, parven
WHERE part.part_num = parven.part_num
AND parven.vendor_num = vendor.vendor_num
ORDER BY part.part_num

Although it appears to be a relatively simple three-table join, the query takes

too much time. As part of your investigation, prepare a table showing the
approximate sizes of the tables and their indexes in pages. The table appears
in Figure 13-7. It is based on a disk page size of 4,096 bytes. Only approximate
numbers are used. The actual number of rows changes often, and only
estimated calculations are made in any case.
On further investigation, you learn that the index on part_num is clustered,
so that the part table is in physical order by part_num. Similarly, the vendor
table is in physical order by vendor_num. The parven table is not in any par-
ticular order. The sizes of these tables indicate that the odds of a successful
nonsequential access from a buffered page are very poor.

13-38 Optimizing Your Queries

 Making Queries Faster

The best query plan for this query (not necessarily the one the optimizer
chooses) is to read the part table sequentially first, then use the value of
part_num to access the matching rows of parven (about 1.5 of them per part),
and then use the value of parven.vendor_num to access vendor through its
index.

Tables Row Size Row Count Rows/Page Data Pages

part 150 10,000 25 400
vendor 150 1,000 25 40
parven 13 15,000 300 50

Indexes Key Size Keys/Page Leaf Pages

part_num 4 500 20
vendor_num 4 500 2
parven (composite) 8 250 60

Figure 13-7 Documenting the sizes of tables and indexes in pages

The resulting number of disk accesses is estimated as shown in the following

list:
• 400 pages read sequentially from part
• 10,000 nonsequential accesses to the parven table, 2 pages each (one index
leaf page, one data page), or 20,000 disk pages
• 15,000 nonsequential accesses to the vendor table, or 30,000 disk pages
Even a simple join on well-indexed tables can require 50,400 disk reads. How-
ever, this can be improved by breaking the query into three steps using tem-
porary tables as shown in Figure 13-8. (The following solution comes from
W.H. Inmon; it is adapted from an example in his book Optimizing
Performance in DB2 Software, Prentice-Hall, 1988.)

Optimizing Your Queries 13-39

Making Queries Faster

DISK PAGES 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000

Sequential and Nonsequential Disk Access

400 pages
part 20,000 pages
parven 30,000 pages

vendor

Disk Access Using Temporary Tables

300 pages
pv_by_vn
892 pages
pvvn_by_pn
580 pages
final query

Figure 13-8 Breaking a query into three steps using temporary tables

The first step is to get the data of the parven table in vendor_num order, as
shown in the following example:

SELECT part_num, vendor_num, price

FROM parven
ORDER BY vendor_num
INTO TEMP pv_by_vn

This statement reads parven sequentially (50 pages), writes a temporary table
(50 pages), and sorts it. The cost of the sort, assuming one merge level, is
about 200 pages, for a total of 300.
Join this temporary table to vendor and put the result in another temporary
table, ordered by part_num, as shown in the following example:

SELECT pv_by_vn.*, vendor.vendor_name

FROM pv_by_vn, vendor
WHERE pv_by_vn.vendor_num = vendor.vendor_num
ORDER BY pv_by_vn.part_num
INTO TEMP pvvn_by_pn;
DROP TABLE pv_by_vn

13-40 Optimizing Your Queries

 Summary

This query reads pv_by_vn sequentially (50 pages). It accesses vendor by

way of its index 15,000 times, but as the keys are presented in vendor_num
sequence, the effect is to read vendor sequentially through its index (42
pages).
If the vendor_name field is 32 bytes long, the output table has about 95 rows
per page and occupies about 160 pages. These pages are written and then
sorted, causing 5×160=800 page reads and writes. Thus, this query reads or
writes 892 pages in all. Now join its output to part for the final result, which
is shown in the following example:

SELECT pvvn_by_pn.*, part.part_desc

FROM pvvn_by_pn, part
WHERE pvvn_by_pn.part_num = part.part_num;
DROP TABLE pvvn_by_pn

This query reads pvvn_by_pn sequentially (160 pages). It accesses part by

way of its index 15,000 times, but again the keys are presented in part_num
sequence so the result is to read part sequentially (400 pages) through its
index (20 pages). The output is produced in sorted order.
By splitting the query into three steps and by substituting sorts for indexed
access, a query that read 50,400 pages is converted into one that reads and
writes approximately 1,772 pages, a 30-to-1 ratio.
Note: Informix database servers earlier than version 4.1 do not permit you to use the
ORDER BY and INTO TEMP clauses in the same query. With these database servers,
you can achieve the preceding solution by selecting INTO TEMP without ordering,
then applying the desired ordering through a clustered index. The improvement is, at
most, 15:1 rather than of 30:1.

Summary
Poor performance can come from several sources, not solely the SQL opera-
tions in a program. Before you focus your attention on the code, take the
following steps:
• Examine the application in its context of machines, people, procedures,
and organization.
• Understand exactly what the application does, for whom, and why.
• Look for nontechnical solutions in the surrounding context.

Optimizing Your Queries 13-41

Summary

• Develop a means of taking repeatable, quantitative measurements of the

performance of the application.
• Isolate the time-consuming parts as narrowly as possible.
The performance of an individual query is determined by the optimizer, a
part of the database server that formulates the query-execution plan. The
optimizer exhaustively considers all possible query plans. The one with the
lowest estimated cost is executed by the database server.
The following operations in a query are time-consuming:
• Reading rows from disk sequentially, nonsequentially by ROWID, and
nonsequentially through an index
• Reading rows over a network
• Sorting
• Performing some difficult matches of regular expressions
• Indexing and sorting NCHAR and NVARCHAR columns
The general method to speed a query is first to study the query plan of the
optimizer as documented by the SET EXPLAIN ON feature, and second to
change the query so that the database server performs the following actions:
• Reads fewer pages of data from disk
• Reads pages sequentially rather than nonsequentially
• Avoids sorting or sorts fewer rows or sorts on narrower columns
You can achieve remarkable time savings in this way. Even if no time is saved,
the necessary study of the application and the query at least make it clear
where the time is being spent.

13-42 Optimizing Your Queries

 Chapter

Creating and Using

Stored Procedures
Chapter Overview 3
Introduction to Stored Procedures and SPL 3
What You Can Do with Stored Procedures 3
Relationship Between SQL and a Stored Procedure 4
Creating and Using Stored Procedures 4
Creating a Procedure Using DB-Access 5
Creating a Procedure Using an SQL API 5
Commenting and Documenting a Procedure 6
Diagnosing Compile-Time Errors 6
Finding Syntax Errors in a Procedure Using DB-
Access 6
Finding Syntax Errors in a Procedure Using an SQL
14
API 7
Looking at Compile-Time Warnings 8
Generating the Text or Documentation 8
Looking at the Procedure Text 9
Looking at the Procedure Documentation 9
Executing a Procedure 9
Executing a Stored Procedure Dynamically 11
Debugging a Procedure 11
Re-creating a Procedure 13
Privileges on Stored Procedures 13
Privileges Necessary at Creation 14
Privileges Necessary at Execution 14
Owner-Privileged Procedures 15
DBA-Privileged Procedures 15
Privileges and Nested Procedures 16
Revoking Privileges 16
 Variables and Expressions 16
Variables 16
Format of Variables 16
Global and Local Variables 17
Defining Variables 17
Data Types for Variables 17
Using Subscripts with Variables 18
Scope of Variables 18
Variable/Keyword Ambiguity 19
SPL Expressions 21
Assigning Values to Variables 21
Program Flow Control 22
Branching 22
Looping 23
Function Handling 24
Calling Procedures Within a Procedure 24
Running an Operating System Command from Within a
Procedure 24
Recursively Calling a Procedure 24
Passing Information into and out of a Procedure 25
Returning Results 25
Specifying Return Data Types 25
Returning the Value 26
Returning More Than One Set of Values from a Procedure 26
Exception Handling 27
Trapping an Error and Recovering 28
Scope of Control of an ON EXCEPTION Statement 29
User-Generated Exceptions 30
Simulating SQL Errors 31
Using RAISE EXCEPTION to Exit Nested Code 32
Summary 33

14-2 Creating and Using Stored Procedures

Chapter Overview
You can write procedures using SQL and some additional statements belong-
ing to the Stored Procedure Language (SPL), and store this procedure in the
database. These stored procedures are effective tools for controlling SQL
activity. This chapter provides instruction on how to write stored procedures.
To help you learn how to write them, examples of working stored procedures
are provided.
The syntax for each of the SPL statements is described in Chapter 2 of the
Informix Guide to SQL: Syntax. Accompanying the syntax for each statement
are usage notes and examples pertinent to that statement.

Introduction to Stored Procedures and SPL

To SQL, a stored procedure is a user-defined function. Anyone who has the
Resource privilege on a database can create a stored procedure. Once the
stored procedure is created, it is stored in an executable format in the data-
base as an object of the database. You can use stored procedures to perform
any function you can perform in SQL as well as to expand what you can
accomplish with SQL alone.
You write a stored procedure using SQL and SPL statements. SPL statements
can be used only inside the CREATE PROCEDURE and CREATE PROCEDURE
FROM statements. The CREATE PROCEDURE statement is available with DB-
Access, and the CREATE PROCEDURE and CREATE PROCEDURE FROM
statements are available with SQL APIs such as INFORMIX-ESQL/C and
INFORMIX-ESQL/COBOL.

What You Can Do with Stored Procedures

You can accomplish a wide range of objectives with stored procedures,
including improving database performance, simplifying the writing of
applications, and limiting or monitoring access to data.

Creating and Using Stored Procedures 14-3

Creating and Using Stored Procedures

Because a stored procedure is stored in an executable format, you can use it

to execute frequently repeated tasks to improve performance. Executing a
stored procedure rather than straight SQL code allows you to bypass repeated
parsing, validity checking, and query optimization.
Because a stored procedure is an object in the database, it is available to every
application running on the database. Several applications can use the same
stored procedure, so development time for applications is reduced.
You can write a stored procedure to be run with the DBA privilege by a user
who does not have the DBA privilege. This allows you to limit and control
access to data in the database. Alternatively, a stored procedure can monitor
which users access certain tables or data. (See “Controlling Access to Data
Using Stored Procedures” on page 11-15 for a discussion of this topic.)

Relationship Between SQL and a Stored Procedure

You can call a procedure within data-manipulation SQL statements and issue
SQL statements within a procedure. See Chapter 1 of the Informix Guide to
SQL: Syntax for a complete list of data manipulation SQL statements.
You use a stored procedure within a data-manipulation SQL statement to
supply values to that statement. For example, you can use a procedure to
perform the following actions:
• Supply values to be inserted into a table.
• Supply a value that makes up part of a condition clause in a SELECT,
DELETE, or UPDATE statement.

These are two possible uses of a procedure within a data manipulation

statement, but others exist. In fact, any expression within a data
manipulation SQL statement can consist of a procedure call.
You can also issue SQL statements within a stored procedure to hide those
SQL statements from a database user. Rather than having all users learn how
to use SQL, one experienced SQL user can write a stored procedure to
encapsulate an SQL activity, and let others know that the procedure is stored
in the database so they can execute it.

Creating and Using Stored Procedures

To write a stored procedure, put the SQL statements that you want run as part
of the procedure inside the statement block in a CREATE PROCEDURE state-
ment. You can use SPL statements to control the flow of the operation within
the procedure. SPL statements include IF, FOR, and others and are described

14-4 Creating and Using Stored Procedures

 Creating and Using Stored Procedures

in the Informix Guide to SQL: Syntax. The CREATE PROCEDURE and CREATE
PROCEDURE FROM statements are also described in Chapter 1 of the Informix
Guide to SQL: Syntax.

Creating a Procedure Using DB-Access

To create a stored procedure using DB-Access, issue the CREATE PROCEDURE
statement, including all the statements that are part of the procedure in the
statement block. For example, to create a procedure that reads a customer
address, you can use a statement such as the one in Figure 14-1.

CREATE PROCEDURE read_address (lastname CHAR(15)) -- one argument

RETURNING CHAR(15), CHAR(15), CHAR(20), CHAR(15),CHAR(2), CHAR(5); -- 6 items

DEFINE p_lname,p_fname, p_city CHAR(15); --define each procedure variable

DEFINE p_add CHAR(20);
DEFINE p_state CHAR(2);
DEFINE p_zip CHAR(5);

SELECT fname, address1, city, state, zipcode

INTO p_fname, p_add, p_city, p_state, p_zip
FROM customer
WHERE lname = lastname;

RETURN p_fname, lastname, p_add, p_city, p_state, p_zip; --6 items

END PROCEDURE

DOCUMENT 'This procedure takes the last name of a customer as', --brief description
'its only argument. It returns the full name and address of',
'the customer.'
WITH LISTING IN '/acctng/test/listfile' -- compile-time warnings go here
; -- end of the procedure read_address

Figure 14-1 Procedure that reads from the customer table

Creating a Procedure Using an SQL API

To create a stored procedure using an SQL API, put the text of the CREATE
PROCEDURE statement in a file. Use the CREATE PROCEDURE FROM state-
ment and refer to that file to compile the procedure. For example, to create a
procedure to read a customer name, you can use a statement such as the one
in Figure 14-1 and store it in a file. If the file is named read_add_source, the
following statement compiles the read_address procedure:

CREATE PROCEDURE FROM 'read_add_source';

Creating and Using Stored Procedures 14-5

Creating and Using Stored Procedures

Figure 14-2 shows how the previous SQL statement looks in an ESQL/C
program.

/* This program creates whatever procedure is in *

* the file 'read_add_source'.
*/
#include <stdio.h>
$include sqlca;
$include sqlda;
$include datetime;
/* Program to create a procedure from the pwd */

main()
{
$database play;
$create procedure from 'read_add_source';
}

Figure 14-2 SQL statement that compiles and stores the read_address procedure in an ESQL/C
program

Commenting and Documenting a Procedure

Observe that the read_address procedure in Figure 14-1 includes comments
and a DOCUMENT clause. The comments are incorporated into the text of the
procedure. Any characters following a double hyphen (--) are considered to
be a comment. The double hyphen can be used anywhere within a line.
The text in the DOCUMENT clause should give a summary of the procedure.
You can extract this text by querying the sysprocbody system catalog table.
See “Looking at the Procedure Documentation” on page 14-9 for more
information about reading the DOCUMENT clause.

Diagnosing Compile-Time Errors

When you issue a CREATE PROCEDURE or CREATE PROCEDURE FROM
statement, the statement fails if a syntax error occurs in the body of the
procedure. The database server stops processing the text of the procedure
and returns the location of the error.

Finding Syntax Errors in a Procedure Using DB-Access

If a procedure created using DB-Access has a syntax error, when you choose
the Modify option of the SQL menu, the cursor sits on the line containing the
offending syntax.

14-6 Creating and Using Stored Procedures

 Creating and Using Stored Procedures

Finding Syntax Errors in a Procedure Using an SQL API

If a procedure created using an SQL API has a syntax error, the CREATE
PROCEDURE statement fails. The database server sets the SQLCODE field of
the SQLCA to a negative number and sets the fifth element of the SQLERRD
array to the character offset into the file. The particular fields of the SQLCA
for each product are shown in the following table:

ESQL/C ESQL/FORTRAN ESQL/COBOL

sqlca.sqlcode sqlcod SQLCODE OF SQLCA
SQLCODE

sqlca.sqlerrd[4] sqlca.sqlerr(5) SQLERRD[5] OF SQLCA

Figure 14-3 shows how to trap for a syntax error when you are creating a
procedure. It also shows how to display a message and character position
where the error occurred.

#include <stdio.h>
$include sqlca;
$include sqlda;
$include datetime;
/* Program to create a procedure from procfile in pwd */

main()
{
long char_num;

$ database play;
$create procedure from 'procfile';
if (sqlca.sqlcode != 0 )
{
printf('\nSqlca.sqlcode = %ld\n', sqlca.sqlcode);
char_num = sqlca.sqlerrd[4];
printf('\nError in creating read_address. Check character position %ld\n',
char_num);
}
.
.
.

Figure 14-3 Checking for failure when creating a procedure using ESQL/C

In Figure 14-3, if the CREATE PROCEDURE FROM statement fails, the program
displays a message in addition to the character position at which the syntax
error occurred.

Creating and Using Stored Procedures 14-7

Creating and Using Stored Procedures

Looking at Compile-Time Warnings

If the database server detects a potential problem but the procedure is
syntactically correct, the database server generates a warning and places it in
a listing file. You can examine this file to check for potential problems before
executing the procedure.
To obtain the listing of compile-time warnings for your procedure, use the
WITH LISTING IN clause in your CREATE PROCEDURE statement, as shown in
the following example:

CREATE PROCEDURE read_address (lastname CHAR(15)) -- one argument

RETURNING CHAR(15), CHAR(15), CHAR(20), CHAR(15), CHAR(2), CHAR(5); -- 6 items
.
.
.
WITH LISTING IN '/acctng/test/listfile' -- compile-time warnings go here
; -- end of the procedure read_address

If you are working on a network, the listing file is created on the machine
where the database resides. If you provide an absolute pathname and file-
name for the file, the file is created where you specify. If you provide a relative
pathname for the listing file, the file is created in your home directory on the
machine where the database resides. (If you do not have a home directory, the
file is created in the root directory.)
After you create the procedure, you can view the file specified in the WITH
LISTING IN clause to see the warnings it contains.

Generating the Text or Documentation

Once you create the procedure, it is stored in the sysprocbody system catalog
table. The sysprocbody system catalog table contains the executable proce-
dure as well as the text of the original CREATE PROCEDURE statement and the
documentation text.

14-8 Creating and Using Stored Procedures

 Creating and Using Stored Procedures

Looking at the Procedure Text

To generate the text of the procedure, select the data column from the
sysprocbody system catalog table. The following SELECT statement reads the
read_address procedure text:

SELECT data FROM informix.sysprocbody

WHERE datakey = 'T' -- find text lines
AND
procid = (SELECT procid FROM informix.sysprocedures
WHERE informix.sysprocedures.procname = 'read_address')

Looking at the Procedure Documentation

If you want to view only the documenting text of the procedure, use the fol-
lowing SELECT statement to read the documentation string. The documenta-
tion lines found in the following example are those in the DOCUMENT clause
of the CREATE PROCEDURE statement.

SELECT data FROM informix.sysprocbody

WHERE datakey = 'D' -- find documentation lines
AND
procid = (SELECT procid FROM informix.sysprocedures
WHERE informix.sysprocedures.procname = 'read_address')

Executing a Procedure
You can execute a procedure in several ways. You can use the SQL statement
EXECUTE PROCEDURE or either the LET or CALL SPL statement. In addition,
you can dynamically execute procedures, as described in the section,
“Executing a Stored Procedure Dynamically” on page 14-11.
The read_address procedure returns the full name and address of a customer.
To run read_address on a customer called “Putnum” using EXECUTE
PROCEDURE, enter the following statement:

EXECUTE PROCEDURE read_address ('Putnum');

Creating and Using Stored Procedures 14-9

Creating and Using Stored Procedures

The read_address procedure returns values; therefore, if you are executing a

procedure from an SQL API or from another procedure, you must use an INTO
clause with host variables to receive the data. For example, executing the
read_address procedure in an ESQL/C program is accomplished with the
code segment shown in Figure 14-4.

#include <stdio.h>
$include sqlca;
$include sqlda;
$include datetime;
/* Program to execute a procedure in the database named 'play' */

main()
{
$ char lname[16], fname[16], address[21];
$ char city[16], state[3], zip[6];

$ connect to play;
$EXECUTE PROCEDURE read_address ('Putnum')
INTO $lname, $fname, $address, $city, $state, $zip;
if (sqlca.sqlcode != 0 )
printf('\nFailure on execute');
}

Figure 14-4 Executing a procedure using ESQL/C

If you are executing a procedure within another procedure, you can use the
SPL statements CALL or LET to run the procedure. To use the CALL statement
with the read_address procedure, you can use the code shown in Figure 14-5.

CREATE PROCEDURE address_list ()

DEFINE p_lname, p_fname, p_city CHAR(15);

DEFINE p_add CHAR(20);
DEFINE p_state CHAR(2);
DEFINE p_zip CHAR(5);
.
.
.
CALL read_address ('Putnum') RETURNING p_fname, p_lname,
p_add, p_city, p_state, p_zip;
.
.
.
-- use the returned data some way
END PROCEDURE;

Figure 14-5 Calling a procedure inside another procedure with the CALL statement

Figure 14-6 provides an example of using the LET statement to assign values
to procedural variables through a procedure call.

14-10 Creating and Using Stored Procedures

 Creating and Using Stored Procedures

CREATE PROCEDURE address_list ()

DEFINE p_lname, p_fname, p_city CHAR(15);

DEFINE p_add CHAR(20);
DEFINE p_state CHAR(2);
DEFINE p_zip CHAR(5);
.
.
.
LET p_fname, p_lname,p_add, p_city, p_state, p_zip = read_address ('Putnum');
.
.
.
-- use the returned data some way
END PROCEDURE;

Figure 14-6 Assigning values from a procedure call with a LET statement

Executing a Stored Procedure Dynamically

You can prepare an EXECUTE PROCEDURE statement in conjunction with the
ALLOCATE DESCRIPTOR and GET DESCRIPTOR statements in an ESQL/C pro-
gram. Parameters to the stored procedure can be passed in the same manner
as the SELECT statement and may be passed at runtime or compile time. For
a detailed example of executing a stored procedure dynamically, see Chapter
10 of the INFORMIX-ESQL/C Programmer’s Manual. For information about
dynamic SQL and using a prepared SELECT statement, see Chapter 5, “Pro-
gramming with SQL.”

Debugging a Procedure
Once you successfully create and run a procedure, you can encounter logical
errors. If logical errors are in the procedure, use the TRACE statement to help
find them. You can trace the values of the following procedural entities:
• Variables
• Procedure arguments
• Return values
• SQL error codes
• ISAM error codes
To generate a listing of traced values, first use the SQL statement SET DEBUG
FILE to name the file that is to contain the traced output. When you create
your procedure, include the TRACE statement in one of its forms.

Creating and Using Stored Procedures 14-11

Creating and Using Stored Procedures

The following ways specify the form of TRACE output:

TRACE ON traces all statements except SQL statements. The
contents of variables are printed before being used.
Procedure calls and returned values are also traced.
TRACE PROCEDURE traces only the procedure calls and returned values.
TRACE expression prints a literal or an expression. If necessary, the
value of the expression is calculated before being
sent to the file.
Figure 14-7 shows how you can use the TRACE statement with a version of
the read_address procedure. Several of the SPL statements shown in this
example have not been discussed, but the entire example demonstrates how
the TRACE statement can help you monitor execution of the procedure.

CREATE PROCEDURE read_many (lastname CHAR(15))

RETURNING CHAR(15), CHAR(15), CHAR(20), CHAR(15),CHAR(2), CHAR(5);

DEFINE p_lname,p_fname, p_city CHAR(15);

DEFINE p_add CHAR(20);
DEFINE p_state CHAR(2);
DEFINE p_zip CHAR(5);
DEFINE lcount, i INT;

LET lcount = 1;

TRACE ON; -- Every expression will be traced from here on

TRACE 'Foreach starts'; -- A trace statement with a literal
FOREACH
SELECT fname, lname, address1, city, state, zipcode
INTO p_fname, p_lname, p_add, p_city, p_state, p_zip
FROM customer
WHERE lname = lastname
RETURN p_fname, p_lname, p_add, p_city, p_state, p_zip WITH RESUME;
LET lcount = lcount + 1; -- count of returned addresses
END FOREACH;

TRACE 'Loop starts'; -- Another literal

FOR i IN (1 TO 5)
BEGIN
RETURN i , i+1, i*i, i/i, i-1,i with resume;
END
END FOR;

END PROCEDURE

Figure 14-7 Using the TRACE statement to monitor execution of a procedure

Each time you execute the traced procedure, entries are added to the file you
specified using the SET DEBUG FILE statement. To see the debug entries, view
the output file using any text editor.

14-12 Creating and Using Stored Procedures

 Privileges on Stored Procedures

The following list contains some of the output generated by the procedure in
Figure 14-7. Next to each traced statement is an explanation of its contents.
TRACE ON echoed TRACE ON statement.
TRACE Foreach starts traced expression, in this case, the literal
string Foreach starts.
start select cursor notification that a cursor is opened to
handle a FOREACH loop.
select cursor iteration notification of the start of each iteration
of the select cursor.
expression: (+lcount, 1) the encountered expression,
(lcount+1) evaluates to 2.
let lcount = 2 each LET statement is echoed with the
value.

Re-creating a Procedure
If a procedure exists in a database, you must drop it explicitly using the DROP
PROCEDURE statement before you can create another procedure with the
same name. If you debug your procedure and attempt to use the CREATE
PROCEDURE statement with the same procedure name again, the attempt
fails unless you first drop the existing procedure from the database.

Privileges on Stored Procedures

A stored procedure resides in the database in which it was created. As with
other database objects, you need appropriate privileges to create a stored pro-
cedure. In addition, you need appropriate privileges to execute a stored
procedure.
Two types of stored procedures exist in a database: DBA-privileged and
owner-privileged. When you create the procedure, you specify which type it
is. You need different privileges to create and execute these two types of pro-
cedures. The differences are described in the sections that follow and are
summarized in Figure 14-8.

Creating and Using Stored Procedures 14-13

Privileges on Stored Procedures

DBA-privileged Owner-privileged
procedure procedure

Can be created by: Any user with the Any user with at least the
DBA privilege Resource privilege

Users having Execute Any user with the Not ANSI-compliant:

privilege by default: DBA privilege Public (any user with Connect
database privilege)

ANSI-compliant:
The procedure owner and any
user with the DBA privilege

Privileges the procedure Execute privilege Execute privilege and

owner or DBA must grant privileges on underlying
another user to enable objects.
that user to run a
procedure: If owner has privileges on
underlying objects with the
GRANT WITH option, only the
Execute privilege is required

Figure 14-8 Differences between DBA-privileged and owner-privileged procedures

Privileges Necessary at Creation

Only users who have the DBA privilege can create a DBA-privileged
procedure. To create an owner-privileged procedure, you need to have at
least the Resource privilege. See “Granting and Limiting Access to Your Data-
base” on page 11-3.

Privileges Necessary at Execution

To run a procedure you always need the Execute privilege for that procedure
or DBA database privileges. Depending on whether the procedure is a DBA-
mode procedure and if the database is ANSI-compliant, certain privileges are
granted to users implicitly by the database server.
If the procedure is owner-privileged, the database server grants the Execute
privilege to PUBLIC. If the database is ANSI-compliant, the database server
grants only the Execute privilege to the owner and users with DBA status.
If the procedure is DBA-privileged, the database server grants the Execute
privilege to all users having the DBA privilege.

14-14 Creating and Using Stored Procedures

 Privileges on Stored Procedures

Owner-Privileged Procedures
When you execute an owner-privileged procedure, the database server
checks the existence of any referenced objects. In addition, the database
server verifies that you have the necessary privileges on the referenced
objects.
If you execute a procedure that only references objects that you own, no
privilege conflicts occurs. If you do not own the referenced objects and you
execute a procedure that contains, say, SELECT statements, for example, you
need the Select privilege on the appropriate tables to run the procedure
without generating errors.
If the owner has the necessary privileges with the WITH GRANT option, those
privileges are automatically conferred to you when the owner issues a
GRANT EXECUTE statement.

Unqualified objects created in the course of the procedure are owned by the
owner of the procedure, not the user running the procedure. For example, the
following example shows lines in an owner-privileged stored procedure that
create two tables. If this procedure is owned by tony and a user marty runs
the procedure, the first table, gargantuan, is owned by tony. The second table,
tiny, is owned by libby. The table gargantuan is an unqualified name; there-
fore, tony owns the table gargantuan. The table tiny is qualified by the owner
libby, so libby owns the table tiny.

CREATE PROCEDURE tryit()

.
.
.
CREATE TABLE gargantuan (col1 INT, col2 INT, col3 INT);
CREATE TABLE libby.tiny (col1 INT, col2 INT, col3 INT);

END PROCEDURE

DBA-Privileged Procedures
When you execute a DBA-privileged procedure, you assume the privileges of
a DBA for the duration of the procedure. A DBA-privileged procedure acts as
if the user running the procedure is first granted DBA privilege, then executes
each statement of the procedure manually, and finally has DBA privilege
revoked.
Objects created in the course of running a DBA procedure are owned by the
user running the procedure, unless the data-definition statement in the
procedure explicitly names the owner to be someone else.
Creating and Using Stored Procedures 14-15
Variables and Expressions

Privileges and Nested Procedures

DBA-privileged status is not inherited by a called procedure. For example, if
a DBA-privileged procedure executes an owner-privileged procedure, the
owner-privileged procedure does not run as a DBA procedure. If an owner-
privileged procedure calls a DBA-privileged procedure, the statements
within the DBA-privileged procedure execute as they would within any
DBA-privileged procedure.

Revoking Privileges
The owner of a procedure can revoke Execute privilege from a user. If a user
loses the Execute privilege on a procedure, the Execute privilege also is
revoked from all users who were granted the Execute privilege by that user.

Variables and Expressions

This section discusses how to define and use variables in SPL. The differences
between SPL and SQL expressions also are covered here.

Variables
You can use a variable in a stored procedure in several ways. You can use a
variable in a database query or other SQL statement wherever a constant is
expected. You can use a variable with SPL statements to assign and calculate
values, keep track of the number of rows returned from a query, and execute
a loop as well as handle other tasks.
The value of a variable is held in memory; the variable is not a database
object. Hence, rolling back a transaction does not restore values of procedural
variables.

Format of Variables
A variable follows the rules of an SQL identifier. (See Chapter 1 of the Informix
Guide to SQL: Syntax.) Once you define a variable, you can use it anywhere in
the procedure as appropriate.
If you are using an SQL API, you do not have to set off the variable with a
special symbol (unlike host variables in an SQL API).

14-16 Creating and Using Stored Procedures

 Variables and Expressions

Global and Local Variables

You can define a variable to be either local or global. A variable is local by
default. The following outline provides the differences between the two
types:
Local A local variable is available only within the procedure in
which it is defined. Local variables do not allow a default
value to be assigned at compile time.
Global A global variable is available to other procedures run by the
same user session in the same database. The values of global
variables are stored in memory. The global environment is
the memory used by all the procedures run within a given
session on a given database server, such as all procedures
run by an SQL API or in a DB-Access session. The values of
the variables are lost when the session ends.
Global variables require a default value to be assigned at
compile time.
The first definition of a global variable puts the variable into
the global environment. Subsequent definitions of the same
variable, in different procedures, simply bind the variable to
the global environment.

Defining Variables
You define variables using the DEFINE statement. If you list a variable in the
argument list of a procedure, the variable is defined implicitly, and you do
not need to define it formally with the DEFINE statement. You must assign a
value, which may be null, to a variable before you can use it.

Data Types for Variables

You can define a variable as any of the data types available for columns in a
table except SERIAL. The following example shows several cases of defined
procedural variables:

DEFINE x INT;
DEFINE name CHAR(15);
DEFINE this_day DATETIME YEAR TO DAY ;

If you define a variable for TEXT or BYTE data, the variable does not actually
contain the data; instead, it serves as a pointer to the data. However, you use
this procedural variable as you would use any other procedural variable.

Creating and Using Stored Procedures 14-17

Variables and Expressions

When you define a TEXT or BYTE variable, you must use the word
REFERENCES, which emphasizes that these variables do not contain the data;
they simply reference the data. The following example shows the definition
of a TEXT and a BYTE variable:

DEFINE ttt REFERENCES TEXT;

DEFINE bbb REFERENCES BYTE;

Using Subscripts with Variables

You can use subscripts with variables having CHAR, VARCHAR, NCHAR,
NVARCHAR, BYTE or TEXT data types, just as you can with SQL column
names. The subscripts indicate the starting and ending character positions of
the variable. Subscripts must always be constants. You cannot use variables
as subscripts. The following example illustrates the usage:

DEFINE name CHAR(15);

LET name[4,7] = 'Ream';
SELECT fname[1,3] INTO name[1,3] FROM customer
WHERE lname = 'Ream';

The portion of the variable contents that is delimited by the two subscripts is
referred to as a substring.

Scope of Variables
A variable is valid within the statement block in which it is defined. It is valid
within statement blocks nested within that statement block as well, unless it
is masked by a redefinition of a variable with the same name.
In the beginning of the procedure in Figure 14-9, the integer variables x, y,
and z are defined and initialized. The BEGIN and END statements mark a
nested statement block in which the integer variables x and q are defined, as

14-18 Creating and Using Stored Procedures

 Variables and Expressions

well as the CHAR variable z. Within the nested block, the redefined variable
x masks the original variable x. After the END statement, which marks the
end of the nested block, the original value of x is accessible again.

CREATE PROCEDURE scope()

DEFINE x,y,z INT;
LET x = 5; LET y = 10;
LET z = x + y; --z is 15
BEGIN
DEFINE x, q INT; DEFINE z CHAR(5);
LET x = 100;
LET q = x + y; -- q = 110
LET z = 'silly'; -- z receives a character value
END
LET y = x; -- y is now 5
LET x = z; -- z is now 15, not 'silly'
END PROCEDURE

Figure 14-9 Masking of variables with nested statement blocks

Variable/Keyword Ambiguity
If you define a variable as a keyword, ambiguities can occur. The following
rules for identifiers help you avoid ambiguities for variables, procedure
names, and system function names:
• Defined variables take the highest precedence.
• Procedures defined as such in a DEFINE statement take precedence over
SQL functions.
• SQL functions take precedence over procedures that exist but are not
identified as procedures in a DEFINE statement.
In some cases, you must change the name of the variable. For example, you
cannot define a variable with the name count or max, because these are the
names of aggregate functions. Refer to Chapter 1 of the Informix Guide to SQL:
Syntax for a list of the keywords that can be used ambiguously.

Variables and Column Names

If you use the same identifier for a procedural variable as you use for a col-
umn name, the database server assumes that each instance of the identifier is
a variable. Qualify the column name with the table name to use the identifier

Creating and Using Stored Procedures 14-19

Variables and Expressions

as a column name. In the following example, the procedure variable lname is

the same as the column name. In the following SELECT statement,
customer.lname is a column name and lname is a variable name:

CREATE PROCEDURE table_test()

DEFINE lname CHAR(15);

LET lname = 'Miller';
.
.
.
SELECT customer.lname FROM customer INTO lname
WHERE customer_num = 502;
.
.
.

Variables and SQL Functions

If you use the same identifier for a procedural variable as for an SQL function,
the database server assumes that an instance of the identifier is a variable and
disallows the use of the SQL function. You cannot use the SQL function within
the block of code in which the variable is defined. The following example
shows a block within a procedure in which the variable called user is defined.
This definition disallows the use of the USER function while in the
BEGIN...END block.

CREATE PROCEDURE user_test()

DEFINE name CHAR(10);
DEFINE name2 CHAR(10);
LET name = user; -- the SQL function

BEGIN
DEFINE user CHAR(15); -- disables user function
LET user = 'Miller';
LET name = user; -- assigns 'Miller' to variable name

END
.
.
.
LET name2 = user; -- SQL function again

14-20 Creating and Using Stored Procedures

 Variables and Expressions

Procedure Names and SQL Functions

For information about ambiguities between procedure names and SQL
function names, see the Informix Guide to SQL: Syntax.

SPL Expressions
You can use any SQL expression in a stored procedure except for an aggregate
expression. The complete syntax and notes for SQL expressions are described
in Chapter 1 of the Informix Guide to SQL: Syntax.
The following examples contain SQL expressions:

var1
var1 + var2 + 5
read_address('Miller')
read_address(lastname = 'Miller')
get_duedate(acct_num) + 10 UNITS DAY
fname[1,5] || ''|| lname
'(415)' || get_phonenum(cust_name)

Assigning Values to Variables

The following ways can assign a value to a procedure variable:
• Using a LET statement
• Using a SELECT...INTO statement
• Using a CALL statement with a procedure that has a RETURNING clause
• Using an EXECUTE PROCEDURE...INTO statement
Use the LET statement to assign a value to one or more variables. The
following example illustrates several forms of the LET statement:

LET a = b + a;
LET a, b = c, d;
LET a, b = (SELECT fname, lname FROM customer
WHERE customer_num = 101);
LET a, b = read_name(101);

Creating and Using Stored Procedures 14-21

Program Flow Control

Use the SELECT statement to assign a value directly from the database to a
variable. The statement in the following example accomplishes the same task
as the third LET statement in the previous example:

SELECT fname, lname into a, b FROM customer

WHERE customer_num = 101

Use the CALL or EXECUTE PROCEDURE statements to assign values returned

by a procedure to one or more procedural variables. Both statements in the
following example return the full address from the procedure read_address
into the specified procedural variables:

EXECUTE PROCEDURE read_address('Smith')

INTO p_fname, p_lname, p_add, p_city, p_state, p_zip;

CALL read_address('Smith')
RETURNING p_fname, p_lname, p_add, p_city, p_state, p_zip;

Program Flow Control

SPL contains several statements that enable you to control the flow of your
stored procedure and to make decisions based on data obtained at run time.
These program-flow-control statements are described briefly in this section.
Their syntax and complete descriptions are provided in Chapter 2 of the
Informix Guide to SQL: Syntax.

Branching
Use an IF statement to form a logic branch in a stored procedure. An IF
statement first evaluates a condition and, if the condition is true, the
statement block contained in the THEN portion of the statement is executed.

14-22 Creating and Using Stored Procedures

 Program Flow Control

If the condition is not true, execution falls through to the next statement,
unless the IF statement includes an ELSE clause or ELIF (else if) clause. Figure
14-10 shows an example of an IF statement.

CREATE PROCEDURE str_compare (str1 CHAR(20), str2 CHAR(20))

RETURNING INT;
DEFINE result INT;

IF str1 > str2 THEN

result = 1;
ELIF str2 > str1 THEN
result = -1;
ELSE
result = 0;
END IF
RETURN result;
END PROCEDURE -- str_compare

Figure 14-10 Using an IF statement

Looping
Three methods exist of looping in SPL, accomplished with one of the
following statements:
FOR initiates a controlled loop. Termination is guaranteed.
FOREACH allows you to select and manipulate more than one row
from the database. It declares and opens a cursor
implicitly.
WHILE initiates a loop. Termination is not guaranteed.
Four ways exist to leave a loop, which can accomplished with one of the
following statements:
CONTINUE skips the remaining statements in the present, identified
loop and starts the next iteration of that loop.
EXIT exits the present, identified loop. Execution resumes at the
first statement after the loop.
RETURN exits the procedure. If a return value is specified, that
value is returned upon exit.
RAISE EXCEPTION exits the loop if the exception is not trapped (caught) in
the body of the loop.
See Chapter 2 of the Informix Guide to SQL: Syntax for more information
concerning the syntax and use of these statements.

Creating and Using Stored Procedures 14-23

Program Flow Control

Function Handling
You can call procedures as well as run operating system commands from
within a procedure.

Calling Procedures Within a Procedure

Use a CALL statement or the SQL statement EXECUTE PROCEDURE to execute
a procedure from a procedure. Figure 14-11 shows a call to the read_name
procedure using a CALL statement.

CREATE PROCEDURE call_test()

RETURNING CHAR(15), CHAR(15);

DEFINE fname, lname CHAR(15);

CALL read_name('Putnum') RETURNING fname, lname;

IF fname = 'Eileen' THEN RETURN 'Jessica', lname;

ELSE RETURN fname, lname;
END IF
END PROCEDURE

Figure 14-11 Calling a procedure with a CALL statement

Running an Operating System Command from Within a Procedure

Use the SYSTEM statement to execute a system call from a procedure. Figure
14-12 shows a call to the echo command.

CREATE DBA PROCEDURE delete_customer(cnum INT)

DEFINE username CHAR(8);

DELETE FROM customer

WHERE customer_num = cnum;

IF username = 'acctclrk' THEN

SYSTEM 'echo ''Delete from customer by acctclrk'' >> /mis/records/updates' ;
END IF
END PROCEDURE -- delete_customer

Figure 14-12 Making a system call from a procedure with a SYSTEM statement

Recursively Calling a Procedure

You can call a procedure from itself. No restrictions apply on calling a
procedure recursively.

14-24 Creating and Using Stored Procedures

 Passing Information into and out of a Procedure

Passing Information into and out of a Procedure

When you create a procedure, you determine whether it expects information
to be passed to it by specifying an argument list. For each piece of informa-
tion that the procedure expects, you specify one argument and the data type
of that argument.
For example, if a procedure needs to have a single piece of integer
information passed to it, you can provide a procedure heading as shown in
the following example:

CREATE PROCEDURE safe_delete(cnum INT)

Returning Results
A procedure that returns one or more values must contain two lines of code
to accomplish the transfer of information: one line to state the data types that
are going to be returned and one line to return the values explicitly.

Specifying Return Data Types

Immediately after you specify the name and input parameters of your
procedure, you must include a RETURNING clause with the data type of each
value you expect to be returned. The following example shows the header of
a procedure (name, parameters, and RETURNING clause) that expects one
integer as input and returns one integer and one 10-byte character value:

CREATE PROCEDURE get_call(cnum INT)

RETURNING INT, CHAR(10);

Creating and Using Stored Procedures 14-25

Passing Information into and out of a Procedure

Returning the Value

Once you use the RETURNING clause to indicate the type of values that are to
be returned, you can use the RETURN statement at any point in your proce-
dure to return the same number and data types as listed in the RETURNING
clause. The following example shows how you can return information from
the get_call procedure:

CREATE PROCEDURE get_call(cnum INT)

RETURNING INT, CHAR(10);
DEFINE ncalls INT;
DEFINE o_name CHAR(10);
.
.
.
RETURN ncalls, o_name;
.
.
.
END PROCEDURE

If you neglect to include a RETURN statement, you do not get an error

message at compile time.

Returning More Than One Set of Values from a Procedure

If your procedure performs a select that can return more than one row from
the database or if you return values from inside a loop, you must use the
WITH RESUME keywords in the RETURN statement. Using a RETURN...WITH
RESUME statement causes the value or values to be returned to the calling
program or procedure. After the calling program receives the values, execu-
tion returns to the statement immediately following the RETURN...WITH
RESUME statement.

Figure 14-13 is an example of a cursory procedure. It returns values from a

FOREACH loop and a FOR loop. This procedure is called a cursory procedure
because it contains a FOREACH loop.

14-26 Creating and Using Stored Procedures

 Exception Handling

CREATE PROCEDURE read_many (lastname CHAR(15))

RETURNING CHAR(15), CHAR(15), CHAR(20), CHAR(15),CHAR(2),
CHAR(5);

DEFINE p_lname,p_fname, p_city CHAR(15);

DEFINE p_add CHAR(20);
DEFINE p_state CHAR(2);
DEFINE p_zip CHAR(5);
DEFINE lcount INT ;
DEFINE i INT ;

LET lcount = 0;
TRACE ON;
CREATE VIEW myview AS SELECT * FROM customer;
TRACE 'Foreach starts';
FOREACH
SELECT fname, lname, address1, city, state, zipcode
INTO p_fname, p_lname, p_add, p_city, p_state, p_zip
FROM customer
WHERE lname = lastname
RETURN p_fname, p_lname, p_add, p_city, p_state, p_zip
WITH RESUME;
LET lcount = lcount +1;
END FOREACH;

FOR i IN (1 TO 5)
BEGIN
RETURN 'a', 'b', 'c', 'd', 'e' WITH RESUME;
END
END FOR;
END PROCEDURE

Figure 14-13 Procedure that returns values from a FOREACH loop and a FOR loop

When you execute this procedure, it returns the name and address for each
person with the specified last name. It also returns a sequence of letters. The
calling procedure or program must be expecting multiple returned values,
and it must use a cursor or a FOREACH statement to handle the multiple
returned values.

Exception Handling
You can trap any exception (or error) returned by the database server to your
procedure, or raised by your procedure, with the ON EXCEPTION statement.
The RAISE EXCEPTION statement enables you to generate an exception within
your procedure.

Creating and Using Stored Procedures 14-27

Exception Handling

Trapping an Error and Recovering

The ON EXCEPTION statement provides a mechanism to trap any error.
To trap an error, enclose a group of statements in a statement block and
precede the statement block with an ON EXCEPTION statement. If an error
occurs in the block that follows the ON EXCEPTION statement, you can take
recovery action.
Figure 14-14 shows an example of an ON EXCEPTION statement within a
BEGIN...END block.

BEGIN
DEFINE c INT;
ON EXCEPTION IN
(
-206, -- table does not exist
-217 -- column does not exist
) SET err_num

IF err_num = -206 THEN

CREATE TABLE t (c INT);
iNSERT INTO t VALUES (10);
-- continue after the insert statement
ELSE
ALTER TABLE t ADD(d INT);
LET c = (SELECT d FROM t);
-- continue after the select statement.
END IF
END EXCEPTION WITH RESUME

INSERT INTO t VALUES (10); -- will fail if t does not exist

LET c = (SELECT d FROM t); -- will fail if d does not exist

END

Figure 14-14 An ON EXCEPTION statement within a BEGIN...END block

When an error occurs, the SPL interpreter searches for the innermost ON
EXCEPTION declaration that traps the error. Note that the first action after
trapping the error is to reset the error. When execution of the error action
code is complete, and if the ON EXCEPTION declaration that was raised
included the WITH RESUME keywords, execution resumes automatically
with the statement following the statement that generated the error. If the ON
EXCEPTION declaration did not include the WITH RESUME keywords,
execution exits the current block completely.

14-28 Creating and Using Stored Procedures

 Exception Handling

Scope of Control of an ON EXCEPTION Statement

An ON EXCEPTION statement is valid for the statement block that follows the
ON EXCEPTION statement, all of the statement blocks nested within that fol-
lowing statement block, and all of the statement blocks that follow the ON
EXCEPTION statement. It is not valid in the statement block that contains the
ON EXCEPTION statement.

The pseudocode in Figure 14-15 shows where the exception is valid within
the procedure. That is, if error 201 occurs in any of the indicated blocks, the
action labeled a201 occurs.

Creating and Using Stored Procedures 14-29

Exception Handling

CREATE PROCEDURE scope()

DEFINE i INT;
.
.
.
BEGIN -- begin statement block A
.
.
.
ON EXCEPTION IN (201)
-- do action a201
END EXCEPTION
BEGIN -- statement block aa
-- do action, a201 valid here
END
BEGIN -- statement block bb
-- do action, a201 valid here
END
WHILE i < 10
-- do something, a201 is valid here
END WHILE

END
BEGIN -- begin statement block B
-- do something
-- a201 is NOT valid here
END
END PROCEDURE

Figure 14-15 Pseudocode showing the ON EXCEPTION statement is valid within a procedure

User-Generated Exceptions
You can generate your own error using the RAISE EXCEPTION statement, as
shown in the following pseudocode example. In this example, the ON
EXCEPTION statement uses two variables, esql and eisam, to hold the error
numbers returned by the database server. If an error occurs and the SQL error

14-30 Creating and Using Stored Procedures

 Exception Handling

number is -206, the action defined in the IF clause is taken. If any other SQL
error is caught, it is passed out of this BEGIN...END block to the block that con-
tains this block.

BEGIN
ON EXCEPTION SET esql, eisam -- trap all errors
IF esql = -206 THEN -- table not found
-- recover somehow
ELSE
RAISE exception esql, eisam ; -- pass the error up
END IF
END EXCEPTION
-- do something
END

Simulating SQL Errors

You can generate errors to simulate SQL errors, as shown in the following
example. Here, if the user is pault, then the stored procedure acts as if that
user has no update privileges, even if he really does have that privilege.

BEGIN
IF user = 'pault' THEN
RAISE EXCEPTION -273; -- deny Paul update privilege
END IF
END

Creating and Using Stored Procedures 14-31

Exception Handling

Using RAISE EXCEPTION to Exit Nested Code

You can use the RAISE EXCEPTION statement to break out of a deeply nested
block, as shown in Figure 14-16. If the innermost condition is true (if aa is
negative), then the exception is raised and execution jumps to the code fol-
lowing the END of the block. In this case, execution jumps to the TRACE
statement.

BEGIN
ON EXCEPTION IN (1)
END EXCEPTION WITH RESUME -- do nothing significant (cont)

BEGIN
FOR i IN (1 TO 1000)
FOREACH select ..INTO aa FROM t
IF aa < 0 THEN
RAISE EXCEPTION 1 ; -- emergency exit
END IF
END FOREACH
END FOR
RETURN 1;
END

--do something; -- emergency exit to

-- this statement.
TRACE 'Negative value returned'
RETURN -10;
END

Figure 14-16 Breaking out of nested loop with a RAISE EXCEPTION statement

Remember that a BEGIN...END block is a single statement. When an error

occurs somewhere inside a block and the trap is outside the block, when exe-
cution resumes, the rest of the block is skipped and execution resumes at the
next statement.
Unless you set a trap for this error somewhere in the block, the error
condition is passed back to the block that contains the call and back to any
blocks that contain the block. If no ON EXCEPTION statement exists that is set
to handle the error, execution of the procedure stops, creating an error for the
program or procedure that is executing the procedure.

14-32 Creating and Using Stored Procedures

 Summary

Summary
Stored procedures provide many opportunities for streamlining your
database process, including enhanced database performance, simplified
applications, and limited or monitored access to data. See the Informix Guide
to SQL: Syntax for syntax diagrams of SPL statements.

Creating and Using Stored Procedures 14-33

Summary

14-34 Creating and Using Stored Procedures

 Chapter

Creating and Using

Triggers
Chapter Overview 3
When to Use Triggers 3
How to Create a Trigger 4
Assigning a Trigger Name 5
Specifying the Trigger Event 5
Defining the Triggered Actions 5
A Complete CREATE TRIGGER Statement 6
Using Triggered Actions 7
Using BEFORE and AFTER Triggered Actions 7
Using FOR EACH ROW Triggered Actions 8
Using the REFERENCING Clause 9
Using the WHEN Condition 10
Using Stored Procedures as Triggered Actions 10
Passing Data to a Stored Procedure 11
Using the SPL 11
15
Updating Nontriggering Columns with Data from a
Stored Procedure 12
Tracing Triggered Actions 12
Generating Error Messages 14
Applying a Fixed Error Message 15
Generating a Variable Error Message 16
Summary 17
15-2 Creating and Using Triggers
Chapter Overview
An SQL trigger is a mechanism that resides in the database. It is available to
any user who has permission to use it. It specifies that when a particular
action—an insert, a delete, or an update—occurs on a particular table, the
database server should automatically perform one or more additional
actions. The additional actions can be INSERT, DELETE, UPDATE, or EXECUTE
PROCEDURE statements.

This chapter describes the purpose of each component of the CREATE

TRIGGER statement, illustrates some uses for triggers, and describes the
advantages of using a stored procedure as a triggered action.

When to Use Triggers

Because a trigger resides in the database and anyone who has the required
privilege can use it, a trigger enables you to write a set of SQL statements that
multiple applications can use. It enables you to prevent redundant code
when multiple programs need to perform the same database operation.
Among other uses, you can use triggers to perform the following actions:
• Create an audit trail of activity in the database. For example, you can track
updates to the orders table by updating corroborating information to an
audit table.
• Implement a business rule. For example, you can determine when an
order exceeds a customer’s credit limit and display a message to that
effect.
• Derive additional data that is not available within a table or within the
database. For example, when an update occurs to the quantity column of
the items table, you can calculate the corresponding adjustment to the
total_price column.
• Enforce referential integrity. When you delete a customer, for example,
you can use a trigger to delete corresponding rows (that is, rows having
the same customer number) in the orders table.

Creating and Using Triggers 15-3

How to Create a Trigger

How to Create a Trigger

You use the CREATE TRIGGER statement to create a trigger. The CREATE
TRIGGER statement is a data definition statement that associates SQL state-
ments with a precipitating action on a table. When the precipitating action
occurs, the associated SQL statements, which are stored in the database, are
triggered. The following diagram illustrates the relationship of the
precipitating action, or trigger event, to the triggered action.

UPDATE

item_numb quantity total_price

xcxcxc xcxcxc xcxcxcx EXECUTE PROCEDURE
xcxcxc xcxc xcxc
xcxcx xcxc xcxcxcxc
upd_items
xcxcxcxc xcxcx xcxc

trigger event

Figure 15-1 A trigger consists of a trigger event and a triggered action

The CREATE TRIGGER statement consists of clauses that perform the

following actions:
• Assign a trigger name.
• Specify the trigger event— that is, the table and the type of action that
initiate the trigger.
• Define the SQL actions that are triggered.
An optional clause, called the REFERENCING clause is discussed later under
“Using FOR EACH ROW Triggered Actions” on page 15-8.
You can create a trigger using the DB-Access utility or one of the SQL APIs,
INFORMIX-ESQL/C or INFORMIX-ESQL/COBOL. This section describes the
CREATE TRIGGER statement as you would enter it using the interactive
Query-language option in DB-Access. In an SQL API, you simply precede the
statement with the symbol or keywords that identify it as an embedded
statement.

15-4 Creating and Using Triggers

 How to Create a Trigger

Assigning a Trigger Name

The trigger name identifies the trigger. It follows the words CREATE TRIGGER
in the statement. It can be up to 18 characters inclusive, in length, beginning
with a letter and consisting of letters, the digits 0-9, and the underscore. In the
following example, the portion of the CREATE TRIGGER statement that is
shown assigns the name upqty to the trigger:

CREATE TRIGGER upqty -- assign trigger name

Specifying the Trigger Event

The trigger event is the type of statement that activates the trigger. When a
statement of this type is performed on the table, the database server executes
the SQL statements that make up the triggered action. The trigger event can
be an INSERT, DELETE, or UPDATE statement. When you define an UPDATE
trigger event, you can name one or more columns in the table to activate the
trigger. If you do not name any columns, then an update of any column in the
table activates the trigger. You can create only one INSERT and one DELETE
trigger per table, but you can create multiple UPDATE triggers as long as the
triggering columns are mutually exclusive.
In the following excerpt of a CREATE TRIGGER statement, the trigger event is
defined as an update of the quantity column in the items table:

CREATE TRIGGER upqty

UPDATE OF quantity ON items -- an UPDATE trigger event

Notice that this portion of the statement identifies the table on which you
create the trigger. If the trigger event is an insert or delete, only the type of
statement and the table name are required, as shown in the following
example:

CREATE TRIGGER ins_qty

INSERT ON items -- an INSERT trigger event

Creating and Using Triggers 15-5

How to Create a Trigger

Defining the Triggered Actions

The triggered actions are the SQL statements that are performed when the
trigger event occurs. The triggered actions can consist of INSERT, DELETE,
UPDATE, and EXECUTE PROCEDURE statements. In addition to specifying
what actions are to be performed, however, you must also specify when they
are to be performed in relation to the triggering statement. You have the
following choices:
• Before the triggering statement executes
• After the triggering statement executes
• For each row that is affected by the triggering statement
A single trigger can define actions for each of these times.
You define a triggered action by specifying when it occurs and then
providing the SQL statement or statements to execute. You specify when the
action is to occur with the keywords BEFORE, AFTER, or FOR EACH ROW. The
triggered actions follow, enclosed in parentheses. The following triggered
action definition specifies that the stored procedure upd_items_p1 is to be
executed before the triggering statement:

BEFORE(EXECUTE PROCEDURE upd_items_p1) -- a BEFORE action

A Complete CREATE TRIGGER Statement

If you combine the trigger name clause, the trigger event clause, and the
triggered action clause, you have a complete CREATE TRIGGER statement.
The following CREATE TRIGGER statement is the result of combining the
components of the statement from the preceding examples. This trigger
executes the stored procedure upd_items_p1 whenever the quantity column
of the items table is updated.

CREATE TRIGGER upqty

UPDATE OF quantity ON items
BEFORE(EXECUTE PROCEDURE upd_items_p1)

Note that if a database object in the trigger definition, such as the stored
procedure upd_items_p1 in this example, does not exist when the database
server processes the CREATE TRIGGER statement, it returns an error.

15-6 Creating and Using Triggers

 Using Triggered Actions

Using Triggered Actions

To use triggers effectively, you need to understand the relationship between
the triggering statement and the resulting triggered actions. You define this
relationship when you specify the time that the triggered action occurs —that
is, BEFORE, AFTER, or FOR EACH ROW.

Using BEFORE and AFTER Triggered Actions

Triggered actions that occur before or after the trigger event execute only
once. A BEFORE triggered action executes before the triggering statement —
that is, before the occurrence of the trigger event. An AFTER triggered action
executes after the action of the triggering statement is complete. BEFORE and
AFTER triggered actions execute even if the triggering statement does not
process any rows.
Among other uses, you can use BEFORE and AFTER triggered actions to
determine the effect of the triggering statement. For example, before updat-
ing the quantity column in the items table, you could call the stored proce-
dure upd_items_p1, shown in the following example, to calculate the total
quantity on order for all items in the table. Notice that the procedure stores
the total in a global variable called old_qty.

CREATE PROCEDURE upd_items_p1()

DEFINE GLOBAL old_qty INT DEFAULT 0;
LET old_qty = (SELECT SUM(quantity) FROM items);
END PROCEDURE;

After the triggering update completes, you can calculate the total again to see
how much it has changed. The following stored procedure, upd_items_p2,
calculates the total of quantity again and stores the result in the local variable
new_qty. Then it compares new_qty to the global variable old_qty to see if

Creating and Using Triggers 15-7

Using Triggered Actions

the total quantity for all orders has increased by more than 50 percent. If so
the procedure uses the RAISE EXCEPTION statement to simulate an SQL
error.

CREATE PROCEDURE upd_items_p2()

DEFINE GLOBAL old_qty INT DEFAULT 0;
DEFINE new_qty INT;
LET new_qty = (SELECT SUM(quantity) FROM items);
IF new_qty > old_qty * 1.50 THEN
RAISE EXCEPTION -746, 0, 'Not allowed - rule violation;'
END IF
END PROCEDURE;

The following trigger calls upd_items_p1 and upd_items_p2 to prevent an

extraordinary update on the quantity column of the items table:

CREATE TRIGGER up_items

UPDATE OF quantity ON items
BEFORE(EXECUTE PROCEDURE upd_items_p1())
AFTER(EXECUTE PROCEDURE upd_items_p2());

If an update raises the total quantity on order for all items by more than 50
percent, the RAISE EXCEPTION statement in upd_items_p2 terminates the
trigger with an error. When a trigger fails in INFORMIX-OnLine Dynamic
Server and the database has logging, the database server rolls back the
changes made by both the triggering statement and the triggered actions. See
CREATE TRIGGER in Chapter 1 of the Informix Guide to SQL: Syntax for more
information on what happens when a trigger fails.

Using FOR EACH ROW Triggered Actions

A FOR EACH ROW triggered action executes once for each row that the
triggering statement effects. For example, if the triggering statement is has
the following syntax:

UPDATE items SET quantity = quantity * 2 WHERE manu_code = 'KAR'

A FOR EACH ROW triggered action executes once for each row in the items
table in which the manu_code column has a value of ‘KAR’.
If the triggering statement does not process any rows, a FOR EACH ROW
triggered action does not execute.

15-8 Creating and Using Triggers

 Using Triggered Actions

Using the REFERENCING Clause

When you create a FOR EACH ROW triggered action, you must usually
indicate in the triggered action statements whether you are referring to the
value of a column before or after the effect of the triggering statement. For
example, imagine that you want to track updates to the quantity column of
the items table. To do this, you create the following table to record the
activity:

CREATE TABLE log_record

(item_num SMALLINT,
ord_num INTEGER,
username CHARACTER(8),
update_time DATETIME YEAR TO MINUTE,
old_qty SMALLINT,
new_qty SMALLINT);

To supply values for the old_qty and new_qty columns in this table, you
must be able to refer to the old and new values of quantity in the items table-
that is, the values before and after the effect of the triggering statement. The
REFERENCING clause enables you to do this.

The REFERENCING clause lets you create two prefixes that you can combine
with a column name, one to reference the old value of the column and one to
reference its new value. These prefixes are called correlation names. You can
create one or both correlation names, depending on your requirements. You
indicate which one you are creating with the keywords OLD and NEW. The
following REFERENCING clause creates the correlation names pre_upd and
post_upd to refer to the old and new values in a row:

REFERENCING OLD AS pre_upd NEW AS post_upd

The following triggered action creates a row in log_record when quantity is

updated in a row of the items table. The INSERT statement refers to the old
values of the item_num and order_num columns and to both the old and
new values of the quantity column.

FOR EACH ROW(INSERT INTO log_record

VALUES (pre_upd.item_num, pre_upd.order_num, USER, CURRENT,
pre_upd.quantity, post_upd.quantity));

The correlation names defined in the REFERENCING clause apply to all rows
affected by the triggering statement.

Creating and Using Triggers 15-9

Using Triggered Actions

If you refer to a column name in the triggering table without using a correla-
tion name, the database server makes no special effort to search for the col-
umn in the definition of the triggering table. You must always use a
correlation name with a column name in SQL statements within a FOR EACH
ROW triggered action, unless the statement is valid independent of the trig-
gered action. See CREATE TRIGGER in Chapter 1 of the Informix Guide to SQL:
Syntax.

Using the WHEN Condition

As an option, you can precede a triggered action with a WHEN clause to make
the action dependent on the outcome of a test. The WHEN clause consists of
the keyword WHEN followed by the condition statement, given in parenthe-
ses. In the CREATE TRIGGER statement, the WHEN clause follows the key-
words BEFORE, AFTER, or FOR EACH ROW and precedes the triggered action
list.
When a WHEN condition is present, if it evaluates to true, the triggered
actions execute in the order in which they appear. If the WHEN condition
evaluates to false or unknown, the actions in the triggered action list do not
execute. If the trigger specifies FOR EACH ROW, the condition is evaluated for
each row also.
In the following trigger example, the triggered action executes only if the
condition in the WHEN clause is true — that is, if the post-update unit price
is greater than two times the pre-update unit price:

CREATE TRIGGER up_price

UPDATE OF unit_price ON stock
REFERENCING OLD AS pre NEW AS post
FOR EACH ROW WHEN(post.unit_price > pre_unit_price * 2)
(INSERT INTO warn_tab VALUES(pre.stock_num, pre.order_num,
pre.unit_price, post.unit_price, CURRENT))

See CREATE TRIGGER in Chapter 1 of the Informix Guide to SQL: Syntax for
more information on the WHEN condition.

Using Stored Procedures as Triggered Actions

Probably the most powerful feature of triggers is the ability to call a stored
procedure as a triggered action. The EXECUTE PROCEDURE statement, which
calls a stored procedure, enables you to pass data from the triggering table to
the stored procedure and also to update the triggering table with data

15-10 Creating and Using Triggers

 Using Triggered Actions

returned by the stored procedure. SPL also lets you define variables, assign
data to them, make comparisons, and use procedural statements to
accomplish complex tasks within a triggered action.

Passing Data to a Stored Procedure

You can pass data to a stored procedure in the argument list of the EXECUTE
PROCEDURE statement. The EXECUTE PROCEDURE statement in the follow-
ing trigger example passes values from the quantity and total_price columns
of the items table to the stored procedure calc_totpr:

CREATE TRIGGER upd_totpr

UPDATE OF quantity ON items
REFERENCING OLD AS pre_upd NEW AS post_upd
FOR EACH ROW(EXECUTE PROCEDURE calc_totpr(pre_upd.quantity,
post_upd.quantity, pre_upd.total_price) INTO total_price)

Passing data to a stored procedure allows you to use it in the operations that
the procedure performs.

Using the SPL

The EXECUTE PROCEDURE statement in the preceding trigger calls the stored
procedure that is shown in the following example. The procedure uses the
SPL to calculate the change that needs to be made to the total_price column
when quantity is updated in the items table. The procedure receives both the
old and new values of quantity and the old value of total_price. It divides the
old total price by the old quantity to derive the unit price. It then multiplies
the unit price by the new quantity to obtain the new total price.

CREATE PROCEDURE calc_totpr(old_qty SMALLINT, new_qty SMALLINT,

total MONEY(8)) RETURNING MONEY(8);
DEFINE u_price LIKE items.total_price;
DEFINE n_total LIKE items.total_price;
LET u_price = total / old_qty;
LET n_total = new_qty * u_price;
RETURN n_total;
END PROCEDURE;

In this example, SPL enables the trigger to derive data that is not directly
available from the triggering table.

Creating and Using Triggers 15-11

Tracing Triggered Actions

Updating Nontriggering Columns with Data from a Stored Procedure

Within a triggered action, the INTO clause of the EXECUTE PROCEDURE
statement lets you update nontriggering columns in the triggering table. The
EXECUTE PROCEDURE statement in the following example calls the calc_-
totpr stored procedure that contains an INTO clause, which references the
column total_price:

FOR EACH ROW(EXECUTE PROCEDURE calc_totpr(pre_upd.quantity,

post_upd.quantity, pre_upd.total_price) INTO total_price);

The value that is updated into total_price is returned by the RETURN

statement at the conclusion of the stored procedure. The total_price column
is updated for each row that the triggering statement affects.

Tracing Triggered Actions

If a triggered action does not behave as you expect, place it in a stored
procedure and use the SPL TRACE statement to monitor its operation. Before
starting the trace, you must direct the output to a file with the SET DEBUG
FILE TO statement. Figure 15-2 shows TRACE statements that have been
added to the stored procedure items_pct. The SET DEBUG FILE TO statement
directs the trace output to the file /usr/mydir/trig.trace. The TRACE ON
statement begins tracing the statements and variables within the procedure.

15-12 Creating and Using Triggers

 Tracing Triggered Actions

CREATE PROCEDURE items_pct(mac CHAR(3))

DEFINE tp MONEY;
DEFINE mc_tot MONEY;
DEFINE pct DECIMAL;
SET DEBUG FILE TO '/usr/mydir/trig.trace';
TRACE 'begin trace';
TRACE ON
LET tp = (SELECT SUM(total_price) FROM items);
LET mc_tot = (SELECT SUM(total_price) FROM items
WHERE manu_code = mac);
LET pct = mc_tot / tp;
IF pct > .10 THEN
RAISE EXCEPTION -745;
END IF
TRACE OFF;
END PROCEDURE;

CREATE TRIGGER items_ins

INSERT ON items
REFERENCING NEW AS post_ins
FOR EACH ROW(EXECUTE PROCEDURE items_pct (post_ins.manu_code));

Figure 15-2 Using a stored procedure to trace a triggered action

Creating and Using Triggers 15-13

Generating Error Messages

Figure 15-3 shows sample trace output from the items_pct procedure as it
appears in the file /usr/mydir/trig.trace. The output reveals the values of
procedure variables, procedure arguments, return values, and error codes.

trace expression :begin trace

trace on
expression:
(select (sum total_price)
from items)
evaluates to $18280.77 ;
let tp = $18280.77
expression:
(select (sum total_price)
from items
where (= manu_code, mac))
evaluates to $3008.00 ;
let mc_tot = $3008.00
expression:(/ mc_tot, tp)
evaluates to 0.16
let pct = 0.16
expression:(> pct, 0.1)
evaluates to 1
expression:(- 745)
evaluates to -745
raise exception :-745, 0, ''
exception : looking for handler
SQL error = -745 ISAM error = 0 error string = = ''
exception : no appropriate handler

Figure 15-3 Trace output from the items.pct stored procedure

See Chapter 14, “Creating and Using Stored Procedures” for more
information on using the TRACE statement to diagnose logic errors in stored
procedures.

Generating Error Messages

When a trigger fails because of an SQL statement, the database server returns
the SQL error number that applies to the specific cause of the failure.
When the triggered action is a stored procedure, you can generate error
messages for other error conditions by using one of two reserved error num-
bers. The first one is error number -745, which has a generalized and fixed
error message. The second one is error number -746, which allows you to
supply the message text, up to a maximum of 71 characters.

15-14 Creating and Using Triggers

 Generating Error Messages

Applying a Fixed Error Message

You can apply error number -745 to any trigger failure that is not an SQL error.
The following fixed message is for this error:
-745 Trigger execution has failed.
You can apply this message with the RAISE EXCEPTION statement in SPL. The
following example generates error -745 if new_qty is greater than
old_qty multiplied by 1.50:

CREATE PROCEDURE upd_items_p2()

DEFINE GLOBAL old_qty INT DEFAULT 0;
DEFINE new_qty INT;
LET new_qty = (SELECT SUM(quantity) FROM items);
IF new_qty > old_qty * 1.50 THEN
RAISE EXCEPTION -745;
END IF
END PROCEDURE

If you are using DB-Access, the text of the message for error -745 displays on
the bottom of the screen, as seen in Figure 15-4.

SQL: New Run Modify Use-editor Output Choose Save Info Drop Exit
Modify the current SQL statements using the SQL editor.

----------------------- stores6@myserver --------- Press CTRL-W for Help --------

INSERT INTO items VALUES( 2, 1001, 2, 'HRO', 1, 126.00);

745: Trigger execution has failed.

Figure 15-4 Error message -745 with fixed message

Creating and Using Triggers 15-15

Generating Error Messages

If you trigger the erring procedure through an SQL statement in your SQL API,
the database server sets the SQL error status variable to -745 and returns it to
your program. To display the text of the message, follow the procedure that
your Informix application development tool provides for retrieving the text
of an SQL error message.

Generating a Variable Error Message

Error number -746 allows you to provide the text of the error message. As
with the preceding example, the following one also generates an error if
new_qty is greater than old_qty multiplied by 1.50. However, in this case the
error number is -746 and the message text - Too many items for Mfr. -
is supplied as the third argument in the RAISE EXCEPTION statement. See the
RAISE EXCEPTION statement in Chapter 14, “Creating and Using Stored Pro-
cedures” for more information on the syntax and use of this statement.

CREATE PROCEDURE upd_items_p2()

DEFINE GLOBAL old_qty INT DEFAULT 0;
DEFINE new_qty INT;
LET new_qty = (SELECT SUM(quantity) FROM items);
IF new_qty > old_qty * 1.50 THEN
RAISE EXCEPTION -746, 0, 'Too many items for Mfr.';
END IF
END PROCEDURE

15-16 Creating and Using Triggers

 Summary

If you use DB-Access to submit the triggering statement, and if new_qty is

greater than old_qty, the result of this stored procedure is as shown in Figure
15-5.

SQL: New Run Modify Use-editor Output Choose Save Info Drop Exit
Modify the current SQL statements using the SQL editor.

---------------------- store6@myserver --------- Press CTRL-W for Help --------

INSERT INTO items VALUES( 2, 1001, 2, 'HRO', 1, 126.00);

746: Too many items for Mfr.

Figure 15-5 Error number -746 with user-specified message text

If you invoke the trigger through an SQL statement in an SQL API, the
database server sets sqlcode to -746 and returns the message text in the
sqlerrm field of the SQL Communications Area (SQLCA). See Chapter 5, “Pro-
gramming with SQL” for your SQL API for a description of the SQLCA.

Summary
In introducing triggers, this chapter explains the following topics:
• The purpose of each component of the CREATE TRIGGER statement
• How to create BEFORE and AFTER triggered actions and how to use them
to determine the impact of the triggering statement
• How to create a FOR EACH ROW triggered action and how to use the
REFERENCING clause to refer to the values of columns both before and
after the action of the triggering statement
• The advantages of using stored procedures as triggered actions
• How to trace triggered actions if they are behaving unexpectedly
• How to generate two types of error messages within a triggered action

Creating and Using Triggers 15-17

Summary

15-18 Creating and Using Triggers

 Index

Index
This index covers the Tutorial, Reference, and Syntax manuals.
Page numbers that end in T can be found in the Tutorial, those that
end in R can be found in the Reference Manual, and those that end
in S can be found in the Syntax manual. Special symbols are listed
in ASCII order at the end of the index.

A
ABS function
syntax in expression 1-444S
use in expression 1-445S
ABSOLUTE keyword
syntax in FETCH 1-194S
use in FETCH 1-196S
Access control. See Privilege.
ACCESS FOR keywords, in INFO statement 1-243S
Accessing tables 12-16T
ACOS function
syntax in expression 1-458S
use in expression 1-460S
Action clause
AFTER 1-117S
definition of 1-116S
FOR EACH ROW 1-116S
subset, syntax 1-121S
syntax 1-116S
Action statements
in triggered action clause 1-122S
list of 1-122S
order of execution 1-122S
Active set
constructing with OPEN 1-264S, 1-265S
definition of 2-29T
of a cursor 5-24T
retrieving data with FETCH 1-194S
 ADD CONSTRAINT keywords, syntax use
in ALTER TABLE 1-15S in Condition subquery 1-415S
AFTER in expression 1-463S
action 1-117S in GRANT 1-236S
keyword 1-117S in SELECT 1-313S
Aggregate function with UNION operator 1-344S
ALL keyword, syntax 1-462S ALLOCATE DESCRIPTOR statement
and GROUP BY clause 3-5T syntax 1-9S
AVG function, syntax 1-462S with concatenation operator 1-432S
description of 2-53T Allocating memory with the ALLOCATE
DISTINCT keyword, syntax 1-462S DESCRIPTOR statement 1-9S
in ESQL 1-467S, 5-14T ALTER INDEX statement
in EXISTS subquery 1-414S creating clustered index 1-12S, 10-25T
in expressions 1-314S dropping clustered index 1-13S
in SELECT 1-315S locks table 7-7T
in SPL expressions 14-21T syntax 1-12S
in subquery 3-36T See also Index Name segment.
MAX function, syntax 1-462S ALTER keyword
MIN function, syntax 1-462S syntax
null value signalled 5-12T in GRANT 1-231S
restrictions in modifiable view 11-24T in REVOKE 1-301S
restrictions with GROUP BY 1-334S use
SUM function, syntax 1-462S in GRANT 1-236S
summary 1-465S in REVOKE 1-302S
Algebraic functions Alter privilege 1-236S, 11-8T
ABS function 1-445S
MOD function 1-446S ALTER TABLE statement
POW function 1-446S ADD clause 1-17S
ROOT function 1-446S ADD CONSTRAINT clause 1-30S
ROUND function 1-446S adding a column 1-17S
SQRT function 1-447S adding a column constraint 1-29S
syntax 1-444S and NCHAR column 1-19R
TRUNC function 1-447S and NVARCHAR column 1-19R
cascading deletes 1-22S
Alias changing column data type 1-27S,
for a table in SELECT 1-323S 9-19T
for table name 2-79T changing table lock mode 1-33S
to assign column names in temporary CHECK clause 1-25S
table 3-12T DEFAULT clause 1-18S
use in ORDER BY clause 1-338S DROP clause 1-26S
use with GROUP BY clause 1-334S DROP CONSTRAINT clause 1-32S
with self-join 3-11T dropping a column 1-26S
See also Synonym. dropping a column constraint 1-32S
ALL keyword LOCK MODE clause 1-33S
beginning a subquery 1-330S, 3-33T MODIFY NEXT SIZE clause 2-9R,
DISCONNECT statement 1-169S 1-33S
syntax NEXT SIZE clause 10-8T
in expression 1-462S ON DELETE CASCADE keyword
in GRANT 1-235S 1-21S
in REVOKE 1-301S PAGE keyword 1-33S
in SELECT 1-312S privilege for 1-231S, 11-10T
with UNION operator 1-310S reclustering a table 1-13S

2 Index
 REFERENCES clause 1-21S SQL statements allowed 1-14R
ROW keyword 1-33S table privileges 1-85S, 11-7T
rules for primary key constraints using with INFORMIX-SE 1-61S
1-31S with BEGIN WORK 1-36S
rules for unique constraints 1-31S ANY keyword
American National Standards Institute. beginning a subquery 1-330S, 3-34T
See ANSI. in WHENEVER 1-398S
AND keyword use in Condition subquery 1-415S
syntax in Condition segment 1-404S Application
use common features 1-18T
in Condition segment 1-417S description of 1-16T
with BETWEEN keyword 1-327S design of order-entry 4-23T
AND logical operator 1-417S, 2-36T handling errors 5-16T
ANSI 1-15T performance analysis of 13-5T
report generator 1-17T
ANSI compliance screen forms 1-17T
determining 1-10R
table naming 1-296S Archives
with INFORMIX-SE 12-19T
ANSI-compliance
-ansi flag 4-15R, Intro-7S, 1-78S, 1-85S, Archiving
1-137S description of 1-10T, 4-25T
described 1-9R INFORMIX-OnLine Dynamic Server
list of SQL statements 1-7S methods 4-26T
reserved words 1-470S transaction log 4-26T
updating rows 1-385S ARC_DEFAULT environment variable
ANSI-compliant database 4-14R
buffered logging restricted in 9-23T ARC_KEYPAD environment variable
create with START DATABASE 4-14R
1-376S Arithmetic functions. See Algebraic
description of 1-59S, 1-15T functions.
designating 1-10R Arithmetic operator, in expression
effect on 1-431S, 2-46T
cursor behavior 1-13R Array, moving rows into with FETCH
decimal data type 1-13R 1-199S
default isolation level 1-13R
escape characters 1-13R AS keyword
object privileges 1-12R in SELECT 1-312S
owner-naming 1-12R syntax
SQLCODE 1-14R in CREATE VIEW 1-136S
transaction logging 1-12R in GRANT 1-231S
transactions 1-11R use
FOR UPDATE not required in 1-148S, in CREATE VIEW 1-137S
6-16T in GRANT 1-239S
index naming 1-419S, 1-484S, 1-505S with display labels 1-316S
owner-naming 1-12R with table aliases 1-325S
privileges 1-12R ASC keyword
procedure naming 1-495S syntax
reason for creating 1-9R in CREATE INDEX 1-63S
Repeatable Read isolation standard in in SELECT 1-337S
7-13T use
signalled in SQLAWARN 5-12T in CREATE INDEX 1-66S
in SELECT 1-338S

Index 3
 Ascending order in SELECT 2-14T syntax in Condition segment 1-405S
ASCII collating order 1-15R, 1-501S use
ASIN function in Condition segment 1-408S
syntax in expression 1-458S in SELECT 1-327S
use in expression 1-460S BETWEEN keyword, used to test for
Asterisk equality in WHERE clause 2-29T
wild card character in SELECT 2-12T BETWEEN operator 2-32T
Asterisk (*) Binary Large Object (BLOB)
arithmetic operator 1-431S choosing location for 10-18T
use in SELECT 1-312S disk storage for 10-4T, 10-6T
At (@) sign, in database name 1-421S effect of isolation on retrieval 1-368S
estimating disk space for 10-17T
ATAN function in a LOAD statement 1-257S
syntax in expression 1-458S in an UNLOAD statement 1-379S
use in expression 1-460S See also BYTE data type.
ATAN2 function See also TEXT data type.
syntax in expression 1-458S blobspace 10-4T, 10-18T
use in expression 1-461S
BLOB. See Binary Large Object.
Attribute
identifying 8-14T Boolean expression
important qualities of 8-15T and logical operator 2-36T
nondecomposable 8-15T in Condition segment 1-404S
Audit trail Bourne shell
applying with RECOVER TABLE how to set environment variables
1-292S 4-5R
dropping with DROP AUDIT 1-171S .profile file 4-4R
manipulating audit trail file 1-293S BUFFERED keyword, syntax in SET LOG
no clustered index 1-64S 1-372S
starting with CREATE AUDIT 1-55S BUFFERED LOG keywords
Automatic type conversion. See Data syntax in CREATE DATABASE 1-57S
type conversion. use in CREATE DATABASE 1-59S
AVG function Buffered logging 1-57S, 9-22T
as aggregate function 2-53T Building your data model 8-3T to 8-33T
syntax in expression 1-462S BYTE data type
use in expression 1-464S choosing location for 10-18T
considerations for UNLOAD
B statement 1-379S
description of 3-5R, 9-19T
Bachman, C.R. 8-17T disk storage for 10-4T
Backslash (\) estimating disk space for 10-17T
as escape character with LIKE 1-410S inserting data 3-5R
as escape character with MATCHES requirements for LOAD statement
1-411S 1-257S
Backup. See Archive. restrictions
BEFORE keyword 1-116S in Boolean expression 3-5R
with GROUP BY 3-5R
BEGIN WORK statement with LIKE or MATCHES 3-5R
locking in a transaction 1-35S with ORDER BY 3-5R
specifies start of a transaction 4-24T restrictions with GROUP BY 3-7T
syntax 1-35S restrictions with LIKE or MATCHES
BETWEEN keyword 2-37T
4 Index
 restrictions with relational expression CHAR data type 9-15T
2-29T changing data types 3-23R
selecting a BYTE column 3-5R description of 3-6R
syntax 1-425S in INSERT 1-498S
with stored procedures 2-9S, 2-13S in relational expressions 2-29T
BYTE value, displaying 2-11T replacing with TEXT 10-27T
replacing with VARCHAR 10-26T
subscripting 2-44T
C substrings of 2-27T
C shell syntax 1-425S
how to set environment variables truncation signalled 5-12T
4-5R using as default value 1-19S, 1-89S
.cshrc file 4-4R versus NCHAR data type 1-19R
.login file 4-4R CHARACTER data type. See CHAR data
Calculated expression type.
restrictions with GROUP BY 1-334S Character mapping files 5-5R
See also Expression segment. Character string
CALL keyword, in the WHENEVER as DATE values 3-29R
statement 1-398S, 1-401S as DATETIME values 3-10R, 3-29R
CALL statement as INTERVAL values 3-15R
assigning values with 14-21T processing with NLS 1-15R
executing a procedure 14-9T See also Quoted String segment.
syntax 2-3S Character-position form of FILE and
call_type table in stores6 database, INSERT statements 5-28R
columns in A-5R Check constraint
Candidate key adding with ALTER TABLE 1-25S
defined 8-23T definition of 1-98S, 4-19T
Cardinality 8-9T specifying at column level 1-98S
specifying at table level 1-98S
Cardinality in relationship 8-13T
CHECK keyword
Caret (^) wildcard in Condition segment use in ALTER TABLE 1-25S
1-411S use in CREATE TABLE 1-98S
Cartesian product CHECK TABLE statement, syntax and
basis of any join 2-71T use 1-37S
description of 2-69T
Checking contents of environment
Cascading deletes configuration file 5-4R
defined 1-22S, 1-97S, 4-21T
locking associated with 1-24S, 1-98S, Checking for corrupted tables 1-37S
4-21T chkenv utility
logging 1-24S, 1-98S, 4-21T description of 5-4R
restriction 1-24S, 1-98S, 4-22T error message for 5-4R
syntax 1-21S, 1-94S Chunk
Cascading triggers description of 10-4T
and triggering table 1-127S, 1-131S mirrored 10-5T
description of 1-130S Client/server environment 1-133S
maximum number of 1-130S CLOSE DATABASE statement
scope of correlation names 1-125S effect on database locks 7-7T
triggered actions 1-118S prerequisites to close 1-41S
Catalog. See System catalog. syntax 1-41S
Chaining synonyms 12-18T CLOSE statement

Index 5
 closing in relational model 1-12T, 8-20T
a select cursor 1-38S in stores6 database A-2R to A-6R
closing an insert cursor 1-39S inserting into 1-246S
cursors affected by transaction end label on 3-48T
1-40S modifying with ALTER TABLE 1-27S
syntax 1-38S naming conventions 1-87S, 1-102S,
with concatenation operator 1-432S 1-294S
CLUSTER keyword naming, allowable characters 1-469S
syntax naming, in ALTER TABLE 1-17S
in ALTER INDEX 1-12S naming, in CREATE TABLE 1-87S
in CREATE INDEX 1-63S number allowed when defining
use constraint 1-86S
in ALTER INDEX 1-12S putting a constraint on 1-86S
in CREATE INDEX 1-64S referenced and referencing 1-22S,
Clustered index 1-94S
creating with CREATE INDEX 1-64S renaming 1-294S
description of 10-25T specifying check constraint for 1-98S
with ALTER INDEX 1-12S specifying with CREATE TABLE
with audit trails 1-64S 1-87S
virtual 1-137S
COBOL 5-6T See also Constraint.
Codd, E. F. 1-11T, 8-4T, 8-32T See also Data Type segment.
Code set Column expression
for crtcmap text file 5-6R in SELECT 1-314S
mapping non-standard to standard syntax 1-433S
5-5R See also Expression segment.
mapping with crtcmap utility 5-5R Column filter. See Filter expression.
Collation Column name
COLLCHAR environment variable allowable characters 1-469S
4-45R in UPDATE clause 1-114S
LC_COLLATE environment variable naming in ALTER TABLE 1-17S
4-50R using functions as names 1-472S
simultaneous, and performance 1-20R using keywords as names 1-473S
with NLS activated 1-15R when qualified 1-124S
Collation order and NLS 2-25T Column number 2-24T
COLLCHAR environment variable effect on triggers 1-115S
1-16R, 4-45R Column value
Colon (:) in triggered action 1-125S
as delimiter in DATETIME 3-9R qualified vs. unqualified 1-126S
as delimiter in INTERVAL 3-15R when unqualified 1-125S
Color, setting INFORMIXTERM for Column-level privilege 1-236S, 11-10T
4-38R COLUMNS FOR keywords, in INFO
Column statement 1-242S
changing data type 3-23R Command file, dbload 5-23R
creating with NLS 1-19R
defined 2-5T Command script, creating database
defining 8-20T 9-26T
defining as foreign key 1-93S Commit, two-phase 12-20T
defining as primary key 1-93S COMMIT WORK statement
description of 1-12T closes cursors 7-18T
displaying information for 1-242S releases locks 7-8T, 7-18T

6 Index
 sets SQLCODE 6-5T kinds of locks 7-6T
syntax 1-43S lock duration 7-8T
Committed Read isolation level 1-367S, lock scope 7-6T
7-10T maximizing 10-32T, 10-35T
COMMITTED READ keywords, syntax Repeatable Read isolation 1-367S,
in SET ISOLATION 1-366S 7-12T
SERIAL values 9-7T
Comparison condition table lock 7-7T
description of 2-29T
syntax and use 1-405S Condition segment
See also Boolean expression. ALL, ANY, SOME subquery 1-415S
boolean expressions 1-405S
Compiler comparison condition 1-405S
setting environment variable for C description of 1-404S
4-31R join conditions 1-331S
setting environment variable for null values 1-405S
COBOL 4-32R, 4-33R relational operators in 1-407S
specifying storage mode for COBOL subquery in SELECT 1-413S
4-33R syntax 1-404S
Complex use of functions in 1-405S
relationship 8-28T wildcards in searches 1-410S
Complex condition. See Condition with BETWEEN keyword 1-408S
segment. with ESCAPE keyword 1-411S
Complex query with EXISTS keyword 1-414S
example of 1-364S with IN keyword 1-408S
Composite column list, multiple-column with IS keyword 1-409S
restrictions 1-30S, 1-31S with MATCHES keyword 1-409S
with NOT keyword 1-410S
Composite index
column limit 1-65S Configuring a database server 12-15T
creating with CREATE INDEX 1-63S CONNECT keyword
definition of 1-65S in GRANT 1-232S
order of columns 13-31T in REVOKE 1-303S
use 13-31T Connect privilege 1-232S, 1-303S, 11-6T
Composite key 8-23T CONNECT statement
Compound assignment 2-28S and INFORMIXSERVER
Compound query 3-43T environment variable 4-37R,
1-46S
Computer network 12-4T connection context 1-45S
Concatenation operator (||) 1-432S connection identifiers 1-45S
Concurrency database environment 1-49S
Committed Read isolation 1-367S, DEFAULT option 1-46S
7-10T implicit connections 1-46S
Cursor Stability isolation 1-367S, opening a database 12-16T
7-11T syntax 1-44S
database lock 7-6T use 1-44S
deadlock 7-14T USER clause 1-52S
defining with SET ISOLATION WITH CONCURRENT
1-366S TRANSACTION option 1-47S
description of 4-27T, 7-3T CONNECT TO statement
Dirty Read isolation 1-367S, 7-10T with concatenation operator 1-432S
effect on performance 7-3T Connecting to data 12-12T
isolation level 7-9T

Index 7
 Connection exiting a loop 14-23T
context 1-45S syntax 2-6S
dormant 1-169S Conventions
identifiers 1-45S example code Intro-8R, Intro-10S, 5T
setting the INFORMIXCONRETRY for naming tables 1-84S
environment variable 4-34R syntax Intro-5S
setting the INFORMIXCONTIME typographical Intro-5R, Intro-5S, 5T
environment variable 4-35R Converting data types 3-23R
Connectivity in relationship 8-8T, 8-11T, Coordinated deletes 6-6T
8-17T
Correlated subquery
Constant expression definition of 1-413S, 3-32T
in SELECT 1-314S restriction with cascading deletes
inserting with PUT 1-286S 4-22T
restrictions with GROUP BY 1-334S
syntax 1-436S Correlation name
See also Expression segment. and stored procedures 1-125S
See also Literal Number. in COUNT DISTINCT clause 1-125S
in DELETE REFERENCING clause
Constraint 1-119S
adding with ALTER TABLE 1-29S, in GROUP BY clause 1-125S
1-86S in INSERT REFERENCING clause
cardinality 8-9T 1-118S
checking 1-131S in SET clause 1-125S
defining domains 9-3T in stored procedure 1-128S
definition of 1-85S in UPDATE REFERENCING clause
dropping with ALTER TABLE 1-32S, 1-120S
1-86S new 1-120S
enforcing 1-86S old 1-120S
modifying a column that has rules for 1-124S
constraints 1-27S scope of 1-125S
number of columns allowed 1-86S, table of values 1-126S
1-92S using 1-124S
optimizer uses 13-9T when to use 1-125S
privileges needed to create 1-31S
rules for unique constraints 1-30S COS function
setting checking mode 1-349S syntax in expression 1-458S
specifying at table level 1-92S use in expression 1-459S
with DROP INDEX 1-174S COUNT DISTINCT clause 1-125S
See also Primary key constraint. COUNT field
See also Referential constraint. getting contents with GET
See also Unique constraint. DESCRIPTOR 1-210S
CONSTRAINT keyword setting value for WHERE clause
in ALTER TABLE 1-29S 1-354S
in CREATE TABLE 1-91S use in GET DESCRIPTOR 1-212S
Contention COUNT function
for bottleneck tables 10-35T and GROUP BY 3-6T
for disk access arms 10-7T as aggregate function 2-53T
reducing 10-32T, 10-35T count rows to delete 4-5T
CONTINUE keyword, in the use in a subquery 4-6T
WHENEVER statement 1-398S, use in expression 1-463S, 1-465S
1-402S with DISTINCT 2-54T
CONTINUE statement
8 Index
COUNT keyword, use in SET synonym for a table 1-80S
DESCRIPTOR 1-355S synonym for a view 1-80S
CREATE AUDIT statement syntax 1-80S
need for archive 1-55S use of dbschema 5-34R
starts audit trail 1-55S with CREATE SCHEMA 1-77S
syntax 1-55S CREATE TABLE
CREATE DATABASE statement use of dbschema 5-34R
and dbspace 10-5T CREATE TABLE statement
ANSI compliance 1-59S and COLLCHAR environment
in command script 9-26T variable 4-46R
logging with OnLine 1-59S and NCHAR column 1-19R
sets shared lock 7-6T and NVARCHAR column 1-19R
SQLAWARN after 5-12T cascading deletes 1-97S
syntax 1-57S CHECK clause 1-98S
using with creating temporary table 1-100S
CREATE SCHEMA 1-77S DEFAULT clause 1-88S
INFORMIX-SE 1-60S, 9-23T defining constraints
OnLine 9-21T at column level 1-91S
PREPARE 1-58S at table level 1-92S
CREATE INDEX statement description of 9-24T
composite indexes 1-65S EXTENT SIZE clause 10-8T
implicit table locks 1-63S in command script 9-26T
locks table 7-7T IN dbspace clause 1-105S
syntax 1-62S locating BLOB column 10-18T
using LOCK MODE clause 1-108S
with CREATE SCHEMA 1-77S naming conventions 1-84S
using with NEXT SIZE clause 10-8T
ASC keyword 1-66S ON DELETE CASCADE keyword
CLUSTER keyword 1-64S 1-94S
DESC keyword 1-66S REFERENCES clause 1-94S
UNIQUE keyword 1-64S rules for primary keys 1-94S
See also Index Name segment. rules for referential constraints 1-94S
CREATE PROCEDURE FROM rules for unique constraints 1-94S
statement sets initial SERIAL value 9-8T
in embedded languages 14-5T setting columns NOT NULL 1-20S,
syntax and use 1-76S 1-90S
specifying extent size 1-107S
CREATE PROCEDURE statement specifying table columns 1-87S
inside a CREATE PROCEDURE storing database tables 1-105S
FROM 14-5T syntax 1-84S
syntax 1-68S with BLOB data types 1-91S
use of dbschema 5-34R with CREATE SCHEMA 1-77S
using 14-5T
CREATE TRIGGER statement
CREATE SCHEMA statement in ESQL/C 1-111S
defining a trigger 1-111S in ESQL/COBOL 1-111S
syntax 1-77S privilege to use 1-111S
with CREATE sequences 1-78S purpose 1-110S
with GRANT 1-78S syntax 1-110S
CREATE SYNONYM statement triggered action clause 1-121S
ANSI-compliant naming 1-80S use 1-111S
chaining synonyms 1-83S CREATE VIEW statement

Index 9
 column data types 1-137S associating with prepared statements
privileges 1-137S 1-154S
restrictions on 11-22T characteristics 1-149S
syntax 1-136S closing 1-38S, 7-18T
use of dbschema 5-34R closing with ROLLBACK WORK
using 11-20T 1-307S
virtual column 1-137S declaring 1-145S, 5-20T
WITH CHECK OPTION 1-138S definition of types 1-148S
with CREATE SCHEMA 1-77S for insert 6-8T
with SELECT * notation 1-136S for update 6-15T, 7-9T
Creating hold 7-18T
a database from ASCII files 5-14R manipulation statements 1-6S
a dbload command file 5-23R opening 1-264S, 1-265S, 5-20T, 5-24T
crtcmap utility retrieving values with FETCH 1-194S,
creating mapping files 5-5R 5-21T
description of 5-5R scroll 1-149S, 5-23T
error messages for 5-7R sequential 1-149S, 5-23T, 5-25T
formats for mapping files 5-6R statement, as trigger event 1-112S
using with transactions 1-155S
Currency, representing with NLS 1-15R with
Current database DELETE 1-159S
specifying with DATABASE 1-140S INTO keyword in SELECT 1-319S
CURRENT function prepared statements 1-148S, 5-32T
4GL example 13-6T WITH HOLD 7-18T, 10-34T
comparing column values 2-56T See also CLOSE statement.
syntax See also Hold cursor.
in Condition segment 1-405S See also Insert cursor.
in expression 1-436S See also OPEN statement.
in INSERT 1-250S See also Scroll cursor.
use See also Select cursor.
in ALTER TABLE 1-18S See also Sequential cursor.
in CREATE TABLE 1-88S See also Update cursor.
in expression 1-440S Cursor Stability isolation level 1-367S,
in INSERT 1-252S 7-11T
in WHERE condition 1-441S CURSOR STABILITY keywords, syntax
input for DAY function 1-441S in SET ISOLATION 1-366S
CURRENT keyword Cursory procedure 2-21S
DISCONNECT statement 1-168S
syntax in FETCH 1-194S customer table in stores6 database,
use in FETCH 1-196S columns in A-2R
CURRENT OF keywords cust_calls table in stores6 database,
syntax columns in A-5R
in DELETE 1-159S Cyclic query 4-22T
in UPDATE 1-383S
use
in DELETE 1-161S
D
in UPDATE 1-390S Daemon 12-14T
Cursor Data
activating with OPEN 1-263S connecting to 12-12T
active set of 5-24T containing foreign characters 1-19R
affected by transaction end 1-40S inserting with the LOAD statement
1-255S
10 Index
 integrity 12-20T floating-point 3-12R, 9-9T
Data access statements 1-6S in SPL variables 14-17T
Data definition statements 1-5S, 5-34T INT 3-13R
INTEGER 3-13R, 9-7T
Data distributions INTERVAL 3-13R, 9-14T
confidence 1-397S MONEY 3-16R, 9-10T
creating on filtered columns 13-9T, NCHAR 3-17R
13-29T NCHAR, mapping to CHAR 4-46R
on temporary tables 1-397S NUMERIC 3-18R
RESOLUTION 1-396S, 1-397S numeric 9-7T
use with optimizer 13-9T NVARCHAR 3-18R
DATA field REAL 3-19R, 9-9T
setting with SET DESCRIPTOR 1-357S requirements for referential
Data integrity 4-22T to 4-25T, 12-20T constraints 1-22S, 1-96S
statements 1-6S segment 1-424S
Data manipulation statements 1-6S SERIAL 3-19R, 9-7T
Data model SMALLFLOAT 3-20R, 9-9T
attribute 8-14T SMALLINT 3-20R
building 8-3T to 8-33T specifying with CREATE VIEW
defining relationships 8-8T 1-137S
denormalizing 10-26T to 10-32T summary table 3-4R
description of 1-3T, 8-3T syntax 1-425S
entity-relationship 8-4T TEXT 3-21R, 9-17T
many-to-many relationship 8-11T VARCHAR 3-22R, 9-15T
one-to-many relationship 8-11T See also Data Type segment.
one-to-one relationship 8-11T Data types
See also Relational model. creating NCHAR columns 1-19R
telephone-directory example 8-6T creating NVARCHAR columns 1-19R
Data protection OnLine specific 1-4R
with OnLine 12-19T Database
Data replication 12-19T ANSI-compliant 1-15T
application 1-16T
Data type archiving 1-10T
automatic conversions 5-15T closing with CLOSE DATABASE
BYTE 3-5R, 9-19T 1-41S
changing with ALTER TABLE 1-28S concurrent use 1-8T
CHAR 3-6R, 9-15T creating ANSI-compliant 1-376S
CHARACTER 3-7R creating in NLS mode 1-20R
character data 9-15T creating with CREATE DATABASE
CHAR, mapping to NCHAR 4-46R 1-57S
choosing 9-19T data types 3-4R
chronological 9-12T default isolation levels 1-368S
considerations for INSERT 1-251S, defined 1-3T, 1-11T
1-498S dropping 1-172S
conversion 3-23R, 4-8T implicit connection 1-168S
DATE 3-7R, 9-12T lock 1-142S
DATETIME 3-7R, 9-12T management of 1-9T
DEC 3-10R map of
DECIMAL 3-10R, 9-10T stores6 A-6R
DOUBLE PRECISION 3-12R system catalog tables 2-33R
fixed-point 9-10T mission-critical 1-10T
FLOAT 3-12R naming conventions 1-422S

Index 11
 naming unique to engine 9-21T for database outside DBPATH 1-141S
naming with variable 1-423S locking 7-6T
NLS 1-16T specifying current database 1-140S
NLS versus non-NLS 1-20R SQLAWARN after 1-141S, 5-12T
NLS, accessing 1-20R syntax 1-140S
opening in exclusive mode 1-142S using with program variables 1-141S
optimizing queries 1-394S See also Database Name segment.
populating new tables 9-27T Database-level privilege
relation to dbspace 10-5T description of 1-232S, 4-15T
relational, defined 1-11T granting 1-232S
remote 1-422S passing grant ability 1-237S
restoring 1-308S revoking 1-303S
server 1-8T, 1-16T See also Privilege.
server, definition of 1-16T DATE data type
stopping logging on 1-376S converting to DATETIME 3-25R
stores6 Intro-9R, Intro-12S, 7T description of 3-7R, 9-12T
stores6 description of A-1R functions in 1-454S, 2-56T
table names 12-16T in ORDER BY sequence 2-14T
See also Database Name segment. range of operations 3-25R
Database Administrator (DBA) 1-233S, representing DATE values 3-29R
11-7T syntax 1-425S
Database application. See Application. using with DATETIME and
Database lock 7-6T INTERVAL values 3-28R
Database management system 12-4T DATE function
Database Name segment as time function 2-56T
database outside DBPATH 1-423S syntax in expression 1-454S
for remote database 1-422S use in expression 1-455S, 2-61T
naming conventions 1-421S DATE value
naming with variable 1-423S setting DBDATE environment
syntax 1-421S variable 4-16R
using quotes, slashes 1-423S specifying European format with
Database server DBDATE 4-17R
choosing OnLine or SE 1-3R DATETIME data type
configuration 12-15T 4GL example 13-6T
definition of 1-16T adding or subtracting INTERVAL
effect of server type on values 3-27R
available data types 1-4R as quoted string 1-498S
isolation level 1-6R character string values 3-10R
locking 1-5R converting to DATE 3-25R
rolling back transactions 1-4R description of 9-12T
SQL statements supported 1-7R displaying format 2-61T, 9-16T
system catalog tables 1-7R field qualifiers 3-8R, 1-428S
transaction logging 1-5R functions on 2-56T
local 12-5T in
NLS versus non-NLS 1-20R expression 1-441S
remote 12-8T INSERT 1-498S
specifying default for connection ORDER BY sequence 2-14T
4-36R relational expressions 2-29T
DATABASE statement multiplying values 3-26R
determining database type 1-140S precision and size 3-8R, 9-13T
exclusive mode 1-142S, 7-6T range of expressions 3-26R

12 Index
 range of operations with DATE and destination options 5-10R
INTERVAL 3-25R Interrupt key 5-9R
representing DATETIME values specifying field delimiter with
3-29R DBDELIMITER 4-18R
syntax 1-425S, 1-488S unloading a database 5-8R
using the DBTIME environment using with NLS 5-12R
variable 4-27R dbimport utility
with EXTEND function 3-26R, 3-28R create options 5-16R
See also Literal DATETIME. creating a database 5-14R
DATETIME Field Qualifier segment description of 5-13R
1-428S input file location options 5-15R
DATETIME formats, using the DBTIME Interrupt key 5-14R
environment variable 4-27R using with NLS 5-13R, 5-18R
Date, representing with NLS 1-15R DBINFO function
DAY function syntax in expression 1-448S
as time function 2-57T use in expression 1-448S
syntax in expression 1-454S DBLANG environment variable 4-19R
use dbload utility
as time function 2-56T creating a command file 5-23R
in expression 1-455S description of 5-19R
DAY keyword INSERT statement, compared to SQL
syntax INSERT statement 5-30R
in DATETIME data type 1-428S Interrupt key 5-21R
in INTERVAL data type 1-485S loading data from a command file
use 5-20R
as DATETIME field qualifier 3-8R, loading data into a table 9-28T, 10-12T
1-488S options
as INTERVAL field qualifier 3-14R, bad-row limits 5-23R
1-491S batch size 5-22R
DBA keyword command-file syntax check 5-21R
in GRANT 1-233S load start point 5-22R
in REVOKE 1-303S specifying field delimiter with
DBDELIMITER 4-18R
DB-Access writing a command file in
creating database with 5-34T, 9-26T character-position form 5-31R
UNLOAD statement 9-28T writing a command file in delimiter
DBANSIWARN environment variable form 5-26R
4-15R, 1-78S, 1-85S, 1-137S, 5-10T DBMONEY environment variable 4-20R,
DBAPICODE environment variable 9-12T
1-16R, 4-47R DBMS. See Database management system.
relation to crtcmap utility 5-5R
DBNLS environment variable 1-16R,
DBA-privileged procedure 14-13T 4-48R
DBA. See Database Administrator. DBPATH environment variable 4-21R,
DBDATE environment variable 4-16R, 1-141S, 1-423S, 12-14T
4-8T, 9-13T DBPRINT environment variable 4-24R
DBDELIMITER environment variable DBREMOTECMD environment variable
4-18R, 1-258S 4-25R
DBEDIT environment variable 4-18R Directory, extension,.dbs extension
dbexport utility 1-101S
description of 5-8R .dbs extension Intro-11R, 1-58S, 1-141S

Index 13
 dbschema utility as delimiter in INTERVAL 3-15R
create schema for a database 5-34R DECLARE statement
description of 5-33R cursor characteristics 1-149S
options cursor types 1-148S
obtaining privilege schema 5-36R cursors with prepared statements
obtaining synonym schema 5-35R 1-154S
specifying a table, view, or cursors with transactions 1-155S
procedure 5-37R definition and use
owner conventions 5-35R hold cursor 1-150S
use of 9-26T insert cursor 1-149S, 1-157S
DBSERVERNAME function procedure cursor 1-148S
returning servername 1-438S scroll cursor 1-149S
use select cursor 1-148S
in ALTER TABLE 1-18S sequential cursor 1-149S
in CREATE TABLE 1-88S update cursor 1-148S, 1-152S
in expression 1-438S description of 5-20T
in SELECT 2-63T, 2-65T, 3-20T FOR INSERT clause 6-8T
dbspace FOR UPDATE 6-15T
definition of 10-4T insert cursor 1-148S
division into extents 10-8T insert cursor with hold 1-157S
for temporary tables 10-6T procedure cursor 1-148S
mirrored 10-5T restrictions with SELECT with
multiple access arms in 10-7T ORDER BY 1-340S
on dedicated device 10-6T SCROLL keyword 5-23T
relation to tblspace 10-7T syntax 1-145S
root 10-5T update cursor 1-148S
selecting with CREATE DATABASE updating specified columns 1-153S
1-57S, 9-22T use
DBSPACETEMP environment variable with concatenation operator
4-26R, 1-100S 1-432S
with FOR UPDATE keywords
DBTEMP environment variable 4-27R 1-148S
DBTIME environment variable 4-27R WITH HOLD clause 7-19T
DBUPSPACE environment variable with SELECT 1-321S
4-30R Default assumptions for your
DDL statements, summary 1-5S environment 4-6R
Deadlock detection 1-371S, 7-14T DEFAULT keyword 1-169S
DEALLOCATE DESCRIPTOR statement in the CONNECT statement 1-46S
syntax 1-143S Default value
with concatenation operator 1-432S description of 4-19T
DEC data type. See DECIMAL data type. specifying
DECIMAL data type with ALTER TABLE 1-19S
changing data types 3-23R with CREATE TABLE 1-88S
description of 3-11R Deferred checking 1-349S
fixed-point 9-10T DEFERRED keyword, in the SET
floating-point 3-11R, 9-10T CONSTRAINTS statement 1-349S
signalled in SQLAWARN 5-12T DEFINE statement
syntax 1-425S in stored procedures 14-17T
using as default value 1-19S, 1-89S placement of 2-8S
Decimal point (.) syntax 2-7S
as delimiter in DATETIME 3-9R
14 Index
DELETE keyword copying Intro-10R, Intro-13S, 8T
syntax installation script Intro-9R, Intro-12S,
in GRANT 1-235S 7T
in REVOKE 1-301S map of A-6R
use overview Intro-9R, Intro-12S, 7T
in GRANT 1-236S structure of tables A-2R
in REVOKE 1-302S tables in A-2R to A-6R
Delete privilege 1-235S, 11-8T, 11-28T See also stores6 database.
DELETE REFERENCING clause Denormalizing 10-26T
and FOR EACH ROW section 1-121S Derived data
correlation name 1-119S introduced for performance 10-30T
syntax 1-119S produced by view 11-20T
DELETE statement DESC keyword 1-338S
all rows of table 4-4T syntax
and end of data 6-14T in CREATE INDEX 1-63S
applied to view 11-24T in SELECT 1-337S
as triggering statement 1-112S use
cascading 1-160S in CREATE INDEX 1-66S
coordinated deletes 6-6T in SELECT 1-338S
count of rows 6-4T Descending order in SELECT 2-14T
CURRENT OF clause 1-161S DESCRIBE statement
description of 4-4T and COLLCHAR environment
embedded 5-6T, 6-3T to 6-8T variable 4-46R
in trigger event 1-111S and the USING SQL DESCRIPTOR
in triggered action 1-122S clause 1-164S
number of rows 5-12T describing statement type 1-163S,
preparing 5-30T 5-33T
privilege for 1-235S, 11-6T, 11-8T INTO sqlda pointer clause 1-165S
syntax 1-159S relation to GET DESCRIPTOR 1-213S
time to update indexes 10-20T syntax 1-162S
transactions with 6-5T using with concatenation operator
using subquery 4-6T 1-432S
WHERE clause restricted 4-6T values returned by SELECT 1-163S
with Condition segment 1-404S
with cursor 1-152S, 6-7T Descriptor 1-164S
with select..for update 1-340S Determining ANSI-compliance 1-10R
within a transaction 1-159S Device
Delete, cascading optical 12-6T
See Cascading deletes storage 12-6T
Delimiter Directory, extension, .dbs 1-58S, 1-141S
for DATETIME values 3-9R Dirty Read isolation level 1-367S, 7-10T
for INTERVAL values 3-15R DIRTY READ keywords, syntax in SET
for LOAD input file 1-258S ISOLATION 1-366S
specifying with UNLOAD 1-380S DISCONNECT statement 1-169S
Delimiter form of FILE and INSERT ALL keyword 1-169S
statements 5-24R CURRENT keyword 1-168S
DELIMITER keyword with
in LOAD 1-258S concatenation operator 1-432S
in UNLOAD 1-380S Disk access
Demonstration database chunk 10-4T

Index 15
 cost to read a row 13-16T DML statements, summary 1-6S
dbspace 10-4T DOCUMENT keyword, use in stored
latency of 13-17T procedures 14-6T
nonsequential 13-18T Documentation notes Intro-8R,
nonsequential avoided by sorting Intro-11S, 6T
13-37T
performance 13-17T to 13-19T, 13-34T Dominant table 3-21T
reducing contention 10-7T Dormant connection 1-169S
seek time 13-17T DOS operating system 12-4T
sequential 13-17T, 13-34T DOUBLE PRECISION data type. See
sequential forced by query 13-32T, FLOAT data type.
13-33T, 13-35T DROP AUDIT statement 1-171S
using rowid 13-18T
DROP CONSTRAINT keywords
Disk buffer. See Page buffer. syntax in ALTER TABLE 1-15S
Disk contention use in ALTER TABLE 1-32S
effect of 13-17T DROP DATABASE statement 1-172S
multiple arms to reduce 10-7T
DROP INDEX statement
Disk extent 10-8T locks table 7-7T
Disk mirroring 10-5T releasing an index 13-34T
Disk page syntax 1-174S
buffer for 13-16T, 13-18T DROP keyword
size of 10-4T, 13-14T syntax in ALTER TABLE 1-15S
Display label use in ALTER TABLE 1-26S
in ORDER BY clause 2-52T DROP SYNONYM statement 1-177S
syntax in SELECT 1-312S DROP TABLE statement 1-179S
with SELECT 2-49T
DROP TRIGGER statement
Display schema for a database 5-34R syntax 1-181S
DISTINCT keyword use of 1-181S
relation to GROUP BY 3-5T DROP VIEW statement 1-183S
restrictions in modifiable view 11-24T
syntax Duplicate index keys 10-22T
in CREATE INDEX 1-63S Duplicate values
in expression 1-462S finding 3-16T
in SELECT 1-312S in a query 1-313S
use Dynamic management statements 1-6S
in CREATE INDEX 1-64S Dynamic SQL
in SELECT 1-313S, 2-19T cursor use with 5-32T
no effect in subquery 1-414S description of 5-5T, 5-28T
with COUNT function 2-54T freeing prepared statements 5-33T
Distributed deadlock 7-15T
Distributed processing 12-8T E
Distributions
dropping with DROP Editor, specifying with DBEDIT 4-18R
DISTRIBUTIONS clause 1-396S Effective checking 1-349S
privileges required to create 1-396S Ellipses (...), wildcard in Condition
using the HIGH keyword 1-396S segment 1-411S
using the MEDIUM keyword 1-397S Embedded SQL
Division (/) symbol, arithmetic operator defined 5-3T
1-431S languages available 5-4T
See also ESQL.
16 Index
End of data defining in environment
signal in SQLCODE 5-11T, 5-17T configuration file 4-4R
signal only for SELECT 6-14T definition of 4-3R
when opening cursor 5-20T ENVIGNORE 4-30R
Entity ENVIGNORE, and chkenv utility
attributes associated with 8-15T 5-4R
business rules 8-5T how to set in Bourne shell 4-5R
criteria for choosing 8-7T how to set in C shell 4-5R
defined 8-4T how to set in Korn shell 4-5R
important qualities of 8-5T INFORMIX environment variables,
in telephone-directory example 8-7T listing 4-8R
integrity 4-18T INFORMIXC 4-31R
naming 8-4T INFORMIXCOB 4-32R
represented by a table 8-22T INFORMIXCOBDIR 4-32R
Entity occurrence, defined 8-16T INFORMIXCOBSTORE 4-33R
INFORMIXCONRETRY 4-34R
Entity-relationship diagram INFORMIXCONTIME 4-34R
connectivity 8-17T INFORMIXDIR 4-36R, 12-14T
discussed 8-17T INFORMIXSERVER 4-36R, 12-14T
meaning of symbols 8-17T INFORMIXSHMBASE 4-37R
reading 8-18T INFORMIXSTACKSIZE 4-38R
ENVIGNORE environment variable INFORMIXTERM 4-38R
4-30R LANG 4-49R
relation to chkenv utility 5-4R LANG, and crtcmap utility 5-5R
Environment configuration file LC_COLLATE 4-50R
debugging with chkenv 5-4R LC_CTYPE 4-51R
example 4-4R LC_MONETARY 4-52R
where stored 4-5R LC_NUMERIC 4-53R
Environment variable LC_TIME 4-53R
and case sensitivity 4-6R listed 4-8R
ARC_DEFAULT 4-14R listed, by topic 4-9R
ARC_KEYPAD 4-14R listed, for NLS 4-9R
COLLCHAR 4-45R listed, for UNIX 4-9R
DBANSIWARN 4-15R NLS environment variables, listing
DBAPICODE 4-47R 4-9R
DBAPICODE, and crtcmap utility ONCONFIG 4-39R
5-5R overriding a setting 4-4R, 4-30R
DBDATE 4-16R PATH 4-54R, 12-14T
DBDELIMITER 4-18R PSORT_DBTEMP 4-40R
DBEDIT 4-18R PSORT_NPROCS 4-41R
DBLANG 4-19R rules of precedence 4-7R
DBLANG, and crtcmap utility 5-5R setting at the command line 4-4R
DBMONEY 4-20R setting in a shell file 4-4R
DBNLS 4-48R SQLEXEC 4-41R
DBPATH 4-21R, 12-14T SQLRM 4-41R, 4-42R
DBPRINT 4-24R SQLRMDIR 4-43R
DBREMOTECMD 4-25R TERM 4-55R
DBSPACETEMP 4-26R TERMCAP 4-56R, 12-14T
DBTEMP 4-27R TERMINFO 4-56R
DBTIME 4-27R UNIX environment variables, listing
DBUPSPACE 4-30R 4-9R
default assumptions 4-6R where to set 4-4R

Index 17
 Equals (=) relational operator 2-30T, indicator variable 5-15T
2-71T INSERT in 6-8T
Equi-join 2-71T NLS errors in SQLERRM field 1-19R
ERROR 1-400S overview 5-3T to 5-37T, 6-3T to 6-17T
preprocessor 5-4T
Error checking scroll cursor 5-23T
continuing after error in stored selecting single rows 5-13T
procedure 2-34S SQL Communications Area 5-8T
error status with ON EXCEPTION SQLCODE 5-11T
2-32S SQLERRD fields 5-12T
exception handling 14-27T static embedding 5-5T
in stored procedures 14-27T UPDATE in 6-14T
simulating errors 14-31T
with SYSTEM 2-40S Estimating
blobpages 10-17T
ERROR keyword, in the WHENEVER maximum number of extents 10-10T
statement 1-398S size of index 10-16T
Error messages table size with fixed-length rows
for NLS 1-21R 10-13T
for trigger failure 15-14T table size with variable-length rows
generating in a trigger 15-14T 10-14T
retrieving trigger text in a program Example database. See Demonstration
15-16T, 15-17T database.
Errors EXCLUSIVE keyword
after DELETE 6-4T syntax
at compile time 14-6T in DATABASE 1-140S
codes for 5-11T in LOCK TABLE 1-260S
dealing with 5-16T use
detected on opening cursor 5-20T in DATABASE 1-142S
during updates 4-22T in LOCK TABLE 1-262S
in stored procedure syntax 14-7T
inserting with a cursor 6-11T Exclusive lock 7-6T
ISAM error code 5-12T EXECUTE IMMEDIATE statement
using to identify NLS database server description of 5-34T
1-19R restricted statement types 1-191S
ESCAPE keyword syntax and usage 1-190S
syntax in Condition segment 1-405S using with concatenation operator
use 1-432S
in Condition segment 1-409S EXECUTE ON keywords
with LIKE keyword 1-328S, 1-411S syntax
with MATCHES keyword 1-329S, in GRANT 1-231S
1-412S in REVOKE 1-300S
with WHERE keyword 1-328S, use
2-43T in GRANT 1-237S
ESQL in REVOKE 1-300S
cursor use 5-19T to 5-28T EXECUTE PROCEDURE statement
DELETE statement in 6-3T 1-192S
delimiting host variables 5-6T assigning values with 14-21T
dynamic embedding 5-5T, 5-28T associating cursor with DECLARE
error handling 5-16T 1-148S
fetching rows from cursor 5-21T in FOREACH 2-20S
host variable 5-6T, 5-8T in triggered action 1-122S
using 14-9T
18 Index
 with .dbs Intro-11R
INTO keyword 1-198S .idx Intro-11R
EXECUTE statement .lok 1-370S
and sqlca record 1-186S Extension checking, specifying with
description of 5-31T DBANSIWARN 4-15R
parameterizing a statement 1-186S Extension, to SQL
syntax 1-184S symbol for Intro-7S
USING DESCRIPTOR clause 1-188S with ANSI-compliant database 1-14R
with concatenation operator 1-432S Extent
with USING keyword 1-186S changing size of for system table 2-9R
Existence dependency 8-9T description of 10-8T
EXISTS keyword next extent doubles 10-10T
beginning a subquery 1-330S sizes of 10-8T
in a WHERE clause 3-33T upper limit on 10-10T
use in condition subquery 1-414S, EXTENT SIZE keywords 1-107S, 10-8T
11-27T
EXIT statement
exiting a loop 14-23T
F
syntax 2-14S FETCH statement
EXP function ABSOLUTE keyword 5-23T
syntax in expression 1-451S as affected by CLOSE 1-39S
use in expression 1-451S checking results with SQLCA 1-202S
Explicit temporary table 1-100S description of 5-21T
fetching a row for update 1-201S
Exponential function locking for update 1-201S
EXP function 1-451S relation to GET DESCRIPTOR 1-210S
Exponential function. See EXP function. sequential 5-23T
Expression specifying a value’s memory location
date-oriented 2-56T 1-198S
description of 2-46T syntax 1-194S
display label for 2-49T with
in UPDATE 1-387S concatenation operator 1-432S
ordering by 1-339S INTO keyword 1-319S
Expression segment program arrays 1-199S
aggregate expressions 1-462S scroll cursor 1-196S
column expressions 1-433S sequential cursor 1-195S, 5-25T
combined expressions 1-467S X/Open mode 1-195S
constant expressions 1-436S Field qualifier
expression types 1-431S for DATETIME 3-8R, 1-428S
function expressions 1-443S for INTERVAL 3-13R, 1-485S, 1-491S
in SPL expressions 1-468S File
syntax 1-431S compared to database 1-3T
EXTEND function environment configuration 4-4R
syntax in expression 1-454S environment configuration, checking
use in expression 1-456S with chkenv 5-4R
with DATE, DATETIME and extension
INTERVAL 3-26R, 3-28R, 2-56T, .lok 1-370S
2-61T mapping, and COLLCHAR 4-45R
extension mapping, and crtcmap utility 5-5R
.dat Intro-11R mapping, and DBAPICODE 4-47R
mapping, format for 5-6R

Index 19
 permissions in UNIX 11-4T specifying multiple ranges 2-18S
sending output with the OUTPUT syntax 2-16S
statement 1-271S using expression lists 2-18S
shell 4-4R using increments 2-17S
temporary for OnLine 4-26R FOR TABLE keywords, in UPDATE
temporary for SE 4-27R STATISTICS 1-393S
FILE statement FOR UPDATE keywords
character-position form 5-28R conflicts with ORDER BY 6-8T
delimiter form 5-24R not needed in ANSI-compliant
syntax for character-position form database 6-16T
5-29R relation to UPDATE 1-390S
syntax for delimiter form 5-25R specific columns 6-16T
with dbload 5-23R syntax in DECLARE 1-145S
Filter expression use
effect on performance 13-23T, 13-32T, in DECLARE 1-148S, 1-152S,
13-33T 1-155S
evaluated from index 13-10T, 13-31T in SELECT 1-340S
optimizer uses 13-10T, 13-23T with column list 1-153S
selectivity estimates 13-30T FOREACH keyword
finding location of row 1-435S in CONTINUE statement 2-6S
FIRST keyword in EXIT 2-14S
syntax in FETCH 1-194S FOREACH keyword. See also FOR EACH
use in FETCH 1-196S ROW action.
First normal form 8-29T FOREACH statement
Fixed point 9-10T looping in a stored procedure 14-23T
FLOAT data type syntax 2-20S
changing data types 3-23R Foreign characters, using NLS 1-14R
description of 3-12R, 9-9T Foreign key 1-22S, 1-93S, 1-94S, 4-19T
syntax 1-425S See also Referential constraint.
using as default value 1-19S, 1-89S FOREIGN KEY keywords
Floating point 9-9T in ALTER TABLE 1-29S
FLUSH statement in CREATE TABLE 1-92S
count of rows inserted 6-11T Format
syntax 1-204S for crtcmap mapping file 5-6R
with concatenation operator 1-432S specifying for DATE value with
writing rows to buffer 6-10T DBDATE 4-16R
FOR EACH ROW action specifying for DATETIME value with
SELECT statement in 1-117S DBTIME 4-27R
triggered action list 1-116S specifying for MONEY value with
FOR EACH ROW action.See also DBMONEY 4-20R
FOREACH keyword. FRACTION keyword
FOR keyword syntax
in CONTINUE 2-6S in DATETIME data type 1-428S
in CREATE AUDIT 1-55S in INTERVAL data type 1-485S
in CREATE SYNONYM 1-80S use
in EXIT 2-14S as DATETIME field qualifier 3-8R,
1-488S
for locating temporary tables 1-100S as INTERVAL field qualifier 3-14R,
FOR statement 1-491S
looping in a stored procedure 14-23T FREE statement

20 Index
 effect on cursors 1-269S syntax 1-217S
freeing prepared statements 5-33T Global transaction 12-21T
syntax 1-207S GOTO keyword, in the WHENEVER
with concatenation operator 1-432S statement 1-398S
FROM keyword GRANT statement
alias names 2-79T automated 11-13T
syntax changing grantor 1-238S
in PUT 1-284S creating a privilege chain 1-238S
in REVOKE 1-300S database-level privileges 1-232S,
in SELECT 1-310S 11-5T
use default table privileges 1-237S
in PUT 1-287S in 4GL 11-13T
in SELECT 1-323S in embedded SQL 5-34T to 5-37T
Function passing grant ability 1-237S
aggregate 2-53T privileges on a view 1-239S
algebraic 1-444S syntax 1-231S
date-oriented 2-56T table-level privileges 1-235S, 11-7T
in SELECT statements 2-53T use of dbschema 5-34R
within a stored procedure 14-24T with CREATE SCHEMA 1-77S
Function expression GROUP BY clause 1-125S
DBINFO function 1-448S GROUP BY keywords
description of 1-443S column number with 3-7T
in SELECT 1-314S composite index used for 13-31T
Functional dependency 8-31T description of 3-4T
indexes for 13-10T, 13-34T
G restrictions in modifiable view 11-24T
sorting rows 13-12T
GET DESCRIPTOR statement syntax in SELECT 1-310S
syntax 1-210S use in SELECT 1-334S
the COUNT keyword 1-212S
use with FETCH statement 1-200S
with concatenation operator 1-432S H
X/Open mode 1-213S HAVING keyword
GET DIAGNOSTICS statement description of 3-9T
CLASS_ORIGIN keyword 1-225S syntax in SELECT 1-310S
CONNECTION_NAME keyword use in SELECT 1-336S
1-227S Header, of a procedure 14-25T
exception clause 1-223S HEX function 1-435S
MESSAGE_LENGTH keyword syntax in expression 1-452S
1-225S use in expression 1-452S
MESSAGE_TEXT keyword 1-225S
MORE keyword 1-222S HIGH keyword 1-374S
NUMBER keyword 1-223S Hold cursor
purpose 1-217S definition of 1-149S, 7-18T
RETURNED_SQLSTATE keyword insert cursor with hold 1-157S
1-225S use of 1-150S
ROW_COUNT keyword 1-223S Host machine 12-8T
SERVER_NAME keyword 1-225S Host variable
statement clause 1-222S delimiter for 5-6T
keywords 1-222S description of 5-6T
SUBCLASS_ORIGIN keyword 1-225S dynamic allocation of 5-33T

Index 21
 fetching data into 5-21T SET DESCRIPTOR statement 1-357S
in DELETE statement 6-4T with X/Open programs 1-213S
in INSERT 6-8T IMMEDIATE keyword, in the SET
in UPDATE 6-14T CONSTRAINTS statement 1-349S
in WHERE clause 5-14T Implicit connection
INTO keyword sets 5-13T defined 1-46S
null indicator 5-15T with DISCONNECT 1-168S
restrictions in prepared statement
5-29T Implicit temporary table 1-100S
truncation signalled 5-12T IN keyword
using non-English characters in 1-22R locating BLOB column 10-18T
with EXECUTE 5-31T syntax
HOUR keyword in CREATE AUDIT 1-55S
syntax in CREATE DATABASE 1-57S
in DATETIME data type 1-428S in CREATE TABLE 1-105S
in INTERVAL data type 1-485S in LOCK TABLE 1-260S
use use
as DATETIME field qualifier 3-8R, in Condition segment 1-408S
1-488S in Condition subquery 1-414S
as INTERVAL field qualifier 3-14R, in CREATE DATABASE 10-5T
1-491S in CREATE TABLE 10-6T, 10-8T
with WHERE keyword 1-327S
Hyphen (-)
as delimiter in DATETIME 3-9R IN keyword, used to test for equality in
as delimiter in INTERVAL 3-15R WHERE clause 2-29T
IN relational operator 3-33T
Index
I adding for performance 10-21T,
Icon, explanation of Intro-6S 13-34T
IDATA field allowable characters in index name
SET DESCRIPTOR statement 1-357S 1-469S
with X/Open programs 1-213S building with NLS 1-20R
Identifier segment clustered 10-25T
column names 1-476S composite 13-31T
column naming, allowable characters creating with CREATE INDEX 1-62S
1-469S disk space used by 10-16T, 10-20T,
cursor name 1-482S 13-34T
cursor names 1-479S displaying information for 1-242S
database objects that use 1-469S dropping 10-24T
stored procedures 1-479S dropping with DROP INDEX 1-174S
syntax 1-469S duplicate entries 10-16T, 10-22T
table names 1-475S, 1-477S effect of physical order 13-27T
used in column naming 1-88S in GROUP BY 13-10T
using keywords as column names in ORDER BY 13-10T
1-473S locks table 7-7T
variable name 1-481S managing 10-19T
naming conventions 1-419S, 1-469S,
IF statement 1-484S, 1-505S
branching 14-22T optimizer 13-29T
syntax 2-24S optimizer uses 13-9T, 13-10T, 13-26T
syntax and use 2-24S ordering columns in composite
with null values 2-25S 13-31T
ILENGTH field performance effects 13-18T
22 Index
 sharing with constraints 1-86S STATUS variable 5-11T
sorting with NLS 1-15R terminates on errors 5-36T, 6-14T
time cost of 10-20T timing operations in 13-6T
updating affects 13-31T using SQLCODE with 5-11T
utility to test or repair 13-31T WHENEVER ERROR statement 5-36T
when not used by optimizer 13-32T, INFORMIXC environment variable
13-33T, 13-35T 4-31R
with temporary tables 1-343S INFORMIXCOB environment variable
INDEX keyword 4-32R
syntax INFORMIXCOBDIR environment
in GRANT 1-235S variable 4-32R
in REVOKE 1-301S
use INFORMIXCOBSTORE environment
in GRANT 1-236S variable 4-33R
in REVOKE 1-302S INFORMIXCOBTYPE environment
Index Name segment variable 4-33R
syntax 1-419S, 1-495S INFORMIXCONRETRY environment
use 1-504S variable 4-34R
Index privilege 1-236S, 11-8T INFORMIXCONTIME environment
INDEXES FOR keywords, in INFO variable 4-34R
statement 1-242S INFORMIXDIR environment variable
Indexes, with dbload utility 5-21R 4-36R, 12-14T
INDICATOR field INFORMIX-OnLine
setting with SET DESCRIPTOR 1-358S disk access by 13-17T
INDICATOR keyword, in SELECT INFORMIX-OnLine Dynamic Server
1-319S, 1-320S, 1-321S, 1-432S allows views on external tables 11-23T
and triggering statement 1-112S
Indicator variable archiving 4-26T
definition of 5-15T characteristics of 1-10T
in EXECUTE 1-186S creating demonstration database
in expression 1-467S Intro-10R, Intro-13S
in SELECT 1-319S, 1-320S, 1-321S, disk page size 13-16T
1-432S disk storage methods 10-3T to 10-12T
using non-English characters in 1-22R is NLS-ready 1-15R
INFO statement optimizer input with 13-9T
displaying privileges and status signalled in SQLAWARN 5-12T
1-242S when tables are locked 7-7T
displaying tables, columns, and INFORMIX-OnLine/Optical 12-6T
indexes 1-241S list of statements 1-7S
syntax 1-241S
INFORMIX-SE
informix characteristics of 1-10T
environment configuration file 4-4R creating database 9-23T
privileges associated with user 1-234S creating demonstration database
Informix extension checking, specifying Intro-10R, Intro-13S
with DBANSIWARN 4-15R is NLS-ready 1-15R
INFORMIX-4GL 12-5T INFORMIXSERVER environment
detecting null value 5-16T variable 4-36R, 1-46S, 12-14T
example of dynamic SQL 11-13T INFORMIXSHMBASE environment
indicator variable not used 5-16T variable 4-37R
program variable 5-5T
INFORMIX-SQL

Index 23
 creating database with 5-34T, 9-26T effect of transactions 1-249S
UNLOAD statement 9-27T embedded 6-8T to 6-14T
INFORMIXSTACKSIZE environment filling insert buffer with PUT 1-285S
variable 4-38R in dynamic SQL 1-253S
INFORMIXTERM environment variable in trigger event 1-111S
4-38R in triggered action 1-122S
inserting
informix.rc file 4-4R multiple rows 4-9T
Insert buffer nulls with the VALUES clause
counting inserted rows 1-205S, 1-290S 1-252S
filling with constant values 1-286S rows 4-6T
inserting rows with a cursor 1-248S rows through a view 1-246S
storing rows with PUT 1-285S rows with a cursor 1-248S
triggering flushing 1-289S single rows 4-7T
Insert cursor 1-148S values into SERIAL columns 3-19R
closing 1-39S null values in 4-7T
definition of 1-148S, 6-8T number of rows 5-12T
in INSERT 1-248S privilege for 11-6T, 11-8T
in PUT 1-286S SELECT statement in 4-9T
opening 1-266S SERIAL columns 1-251S
reopening 1-267S specifying values to insert 1-250S
result of CLOSE in SQLCA 1-39S syntax 1-245S
use of 1-149S, 6-11T for character position form 5-29R
with hold 1-157S for delimiter form 5-25R
writing buffered rows with FLUSH time to update indexes 10-20T
1-204S use with insert cursor 1-157S
INSERT INTO keywords VALUES clause 4-7T
in INSERT 1-245S with
in LOAD 1-258S a view 11-25T
INSERT keyword DECLARE 1-145S
syntax SELECT 1-253S
in GRANT 1-235S with dbload 5-23R
in REVOKE 1-301S Inserting rows of constant data 6-11T
use Installation directory, specifying with
in GRANT 1-236S INFORMIXDIR 4-36R
in REVOKE 1-302S INT data type. See INTEGER data type.
Insert privilege 11-8T, 11-28T INTEGER data type
INSERT REFERENCING clause changing data types 3-23R
and FOR EACH ROW section 1-121S description of 3-13R, 9-7T
correlation name 1-118S syntax 1-425S
syntax 1-118S using as default value 1-19S, 1-89S
INSERT STATEMENT Integrity. See Data integrity.
using functions in the VALUES clause Integrity. See Referential integrity.
1-252S Integrity. See Semantic integrity.
INSERT statement Intensity attributes, setting
and end of data 6-14T INFORMIXTERM for 4-38R
character-position form 5-28R
constant data with 6-11T Interrupt key
count of rows inserted 6-11T with dbexport 5-9R
delimiter form 5-24R with dbimport 5-14R
duplicate values in 4-7T with dbload 5-21R

24 Index
Interrupted modifications 4-22T IS NOT keywords, syntax in Condition
INTERVAL data type segment 1-405S
adding or subtracting from 3-30R IS NULL keywords 1-327S
adding or subtracting from ISAM error code 2-31S, 2-36S, 5-12T
DATETIME values 3-27R Isolation level
as quoted string 1-498S Committed Read 1-367S, 7-10T
description of 3-13R, 9-14T Cursor Stability 1-367S, 7-11T
display format 9-16T default in ANSI-compliant database
field delimiters 3-15R 1-13R
field qualifier, syntax 1-485S definitions 1-367S
in expression 1-441S description of 7-9T
in INSERT 1-498S Dirty Read 1-367S, 7-10T
in relational expressions 2-29T in external tables 1-368S
multiplying or dividing values 3-31R Repeatable Read 1-367S, 7-12T
precision and size 9-14T setting 7-9T
range of expressions 3-26R use with FETCH statement 1-201S
range of operations with DATE and
DATETIME 3-25R items table in stores6 database, columns
syntax 1-425S, 1-491S in A-3R
with EXTEND function 3-26R, 3-28R ITYPE field
See also Literal INTERVAL. SET DESCRIPTOR statement 1-357S
INTERVAL Field Qualifier segment setting with SET DESCRIPTOR 1-359S
1-485S with X/Open programs 1-213S
INTO keyword
choice of location 5-22T J
in FETCH statement 5-22T Join
in SELECT 1-318S associative 2-76T
mismatch signalled in SQLAWARN creating 2-71T
5-12T definition of 2-8T
restrictions in INSERT 4-10T dominant table 3-21T
restrictions in prepared statement effect of large join on optimization
5-29T 13-36T
retrieving multiple rows 5-20T equi-join 2-71T
retrieving single rows 5-13T in Condition segment 1-331S
syntax multiple-table join 1-332S, 2-77T
in FETCH 1-194S natural 2-75T
in SELECT 1-310S nested outer 3-28T
use nested simple 3-25T
in FETCH 1-199S outer 3-21T
in SELECT 1-318S outer join 1-333S
INTO TEMP keywords restrictions in modifiable view 11-24T
description of 2-83T self-join 1-333S, 3-11T
restrictions in view 11-22T sort merge 13-27T
syntax in SELECT 1-310S subservient table 3-21T
use two-table join 1-332S
in SELECT 1-341S Join column. See Foreign key.
with UNION operator 1-344S
Join condition. See Condition segment.
IS keyword
in Condition segment 1-409S Journal updates 10-33T
with WHERE keyword 1-327S

Index 25
 K LET statement
assigning values 14-21T
Key lock 7-8T executing a procedure 14-9T
Keywords, using in triggered action syntax 2-28S
1-123S LIKE keyword
Key, candidate. See Candidate key. syntax in Condition segment 1-405S
Key, composite 8-23T use in SELECT 1-328S
Key, foreign. See Foreign key. wildcard characters 1-328S
Key, primary 8-22T LIKE keyword, used to test for equality
in WHERE clause 2-29T
Key, primary. See Primary key
constraint. LIKE relational operator 2-37T
Korn shell LIKE test 13-32T
how to set environment variables Literal
4-5R DATETIME
.profile file 4-4R in Condition segment 1-405S
in expression 1-436S, 1-441S
segment 1-487S
L syntax 1-488S
Label 2-49T, 3-48T syntax in INSERT 1-250S
LANG environment variable 1-16R, use in ALTER TABLE 1-18S
4-49R use in CREATE TABLE 1-88S
with IN keyword 1-327S
Language supplement for added NLS DATE, using as a default value 1-19S,
functionality 1-22R 1-89S
LAST keyword INTERVAL
syntax in FETCH 1-194S in Condition segment 1-405S
use in FETCH 1-196S in expression 1-436S, 1-441S
Latency 13-17T segment 1-490S
LC_COLLATE environment variable syntax 1-491S
1-16R, 4-50R syntax in INSERT 1-250S
LC_CTYPE environment variable 1-16R, using as default value 1-19S, 1-89S
4-51R Number
in Condition segment 1-405S
LC_MONETARY environment variable in expression 1-436S, 1-439S
1-17R, 4-52R segment 1-493S
LC_NUMERIC environment variable syntax 1-493S
1-17R, 4-53R syntax in INSERT 1-250S
LC_TIME environment variable 1-17R, with IN keyword 1-409S
4-53R LOAD statement
LENGTH field DELIMITER clause 1-258S
setting with SET DESCRIPTOR 1-357S input formats for data 1-256S
with DATETIME and INTERVAL INSERT INTO clause 1-258S
types 1-358S loading VARCHAR, TEXT, or BYTE
with DECIMAL and MONEY types data types 1-257S
1-358S specifying field delimiter with
LENGTH function DBDELIMITER 4-18R
in expression 1-314S specifying the table to load into 1-258S
on TEXT 2-64T syntax 1-255S
on VARCHAR 2-64T the LOAD FROM file 1-256S
syntax in expression 1-443S, 1-453S Loading data from a command file into a
use in expression 1-453S, 2-63T table 5-20R
26 Index
Local loopback 12-9T update cursors effect on 1-152S
Local server 12-5T update locks 1-385S
LOCK MODE keywords waiting period 1-370S
syntax with
in ALTER TABLE 1-15S DELETE 6-4T
in CREATE TABLE 1-108S FETCH 1-201S
use scroll cursor 1-369S
in ALTER TABLE 1-33S SET ISOLATION 1-366S
in CREATE TABLE 1-108S SET LOCK MODE 1-370S
UNLOCK TABLE 1-381S
LOCK TABLE statement update cursor 7-9T
in databases with transactions 1-261S within transaction 1-35S
in databases without transactions See also Table locking.
1-262S
locking a table explicitly 7-7T LOG IN keywords, syntax in CREATE
syntax 1-260S DATABASE 1-57S
Locking LOG10 function
and concurrency 4-27T syntax in expression 1-452S
and integrity 7-3T use in expression 1-452S
deadlock 7-14T Logarithmic function
description of 7-5T syntax
during LOG10 function 1-451S
delete 1-159S LOGN function 1-451S
inserts 1-249S use
updates 1-152S, 1-385S LOG10 function 1-452S
granularity 7-6T LOGN function 1-452S
in OnLine 1-5R Logging
in SE 1-6R buffered 9-22T
lock duration 7-8T buffered vs. unbuffered 1-372S
lock mode 7-13T cascading deletes 1-24S, 1-98S
not-wait 7-14T changing mode with SET LOG 1-372S
wait 7-13T choosing for OnLine database server
locks released at end of transaction 9-22T
7-18T choosing for SE database server 9-23T
mode 1-5R finding log file location 1-60S
overriding row-level 1-261S renaming log 1-377S
releasing with COMMIT WORK 1-43S setting with CREATE TABLE 1-104S
releasing with ROLLBACK WORK starting with START DATABASE
1-306S 1-60S, 1-376S
scope 1-5R, 7-6T stopping 1-377S
scope of lock 7-6T stopping with START DATABASE
setting lock mode 7-13T 1-376S
shared locks 1-6R unbuffered 9-22T
types of locks with INFORMIX-OnLine 1-59S
database lock 7-6T with INFORMIX-SE 1-60S
exclusive lock 7-6T with triggers 1-134S
key lock 7-8T Logical log
page lock 1-108S, 7-8T definition of 4-25T
promotable lock 7-6T, 7-9T Logical operator
row lock 1-108S, 7-8T AND 2-36T
shared lock 7-6T in Condition segment 1-417S
table lock 7-7T NOT 2-36T

Index 27
 OR 2-36T MIN function
LOGN function as aggregate function 2-53T
syntax in expression 1-452S syntax in expression 1-462S
use in expression 1-452S use in expression 1-464S
.lok extension 1-370S Minus (-) sign, arithmetic operator
Loop 1-431S
controlled 2-16S MINUTE keyword
creating and exiting in SPL 14-23T syntax
exiting using RAISE exception 14-32T in DATETIME data type 1-428S
indefinite with WHILE 2-46S in INTERVAL data type 1-485S
Loopback, local 12-9T use
as DATETIME field qualifier 3-8R,
LOW keyword 1-374S 1-488S
as INTERVAL field qualifier 3-14R,
M 1-491S
Machine notes Intro-9R, Intro-11S, 6T Mirror. See Disk mirroring.
Mandatory, entity in relationship 8-9T MOD function
syntax in expression 1-444S
Many-to-many relationship 8-9T, 8-11T, use in expression 1-446S
8-26T
MODE ANSI keywords
Mapping files for non-standard code sets ANSI-compliant database 1-15T
5-5R ANSI-compliant logging 9-23T
MATCHES keyword specifying ANSI-compliance 1-11R
syntax in Condition segment 1-405S specifying transactions 4-24T
use syntax
in Condition segment 1-409S in CREATE DATABASE 1-57S
in SELECT 1-328S in START DATABASE 1-376S
with NLS 2-42T use
used to test for equality in WHERE in CREATE DATABASE 1-60S
clause 2-29T in START DATABASE 1-377S
wildcard characters 1-329S Model. See Data model.
MATCHES relational operator MODIFY keyword
in WHERE clause 2-37T syntax in ALTER TABLE 1-15S
MAX function use in ALTER TABLE 1-27S
as aggregate function 2-53T MODIFY NEXT SIZE keywords
syntax in expression 1-462S syntax in ALTER TABLE 1-15S
use in expression 1-464S use in ALTER TABLE 1-33S
MDY function MONEY data type 9-12T
as time function 2-56T changing data types 3-23R
syntax in expression 1-454S description of 3-16R, 9-10T
Memory display format 9-12T
allocating for a system sqlda structure in INSERT 4-8T
1-9S syntax 1-425S
shared 12-7T using as default value 1-19S, 1-89S
Message files See also DECIMAL data type.
error messages Intro-9R, Intro-11S, 7T MONEY value
setting LANG for NLS 4-49R setting DBMONEY environment
specifying subdirectory for NLS 1-22R variable 4-20R
specifying subdirectory with setting LC_MONETARY
DBLANG 4-19R environment variable 4-52R

28 Index
 specifying European format with connection information 12-15T
DBMONEY 4-20R data sent over 13-21T
specifying for NLS 4-53R performance of 13-20T
Money, representing with NLS 1-15R simple model of 13-21T
MONTH function site 12-4T
as time function 2-56T Network environment variable
syntax in expression 1-454S DBPATH 4-21R
use in expression 1-456S SQLRM 4-42R
MONTH keyword SQLRMDIR 4-43R
syntax NEW keyword
in DATETIME data type 1-428S in DELETE REFERENCING clause
in INTERVAL data type 1-485S 1-119S
use in INSERT REFERENCING clause
as DATETIME field qualifier 3-8R, 1-118S
1-488S in UPDATE REFERENCING clause
as INTERVAL field qualifier 3-13R, 1-120S
1-491S NEXT keyword
Multiple triggers syntax in FETCH 1-194S
column numbers in 1-115S use in FETCH 1-196S
example 1-114S NEXT SIZE keywords
order of execution 1-115S specifying size of extents 10-8T
preventing overriding 1-132S use in CREATE TABLE 1-107S
Multiple-table join 2-77T use in GRANT 1-233S
Multi-row query NLS 1-14R
destination of returned values 1-198S activating 2-25T
managing with FETCH 1-195S activating in Informix products 1-16R
and collation order 2-25T
and dbexport utility 5-12R
N and dbimport utility 5-13R
Naming convention and MATCHES keyword 2-42T
column 1-87S, 1-102S, 1-294S and ORDER BY keywords 2-25T,
database 1-422S 2-42T
index 1-419S, 1-484S, 1-505S checking products for functionality
table 1-84S, 1-470S, 1-507S 1-18R
See also Database Name segment. COLLCHAR environment variable
See also Index Name segment. 4-45R
See also Table Name segment. creating character mapping files for
Naming conventions 5-5R
tables 12-16T data types for 1-19R
database server compatibility 1-19R
Native Language Support. See NLS. DBAPICODE environment variable
Natural join 2-75T 4-47R
NCHAR data type DBNLS environment variable 4-48R
description of 3-17R, 9-15T environment variables listed 4-9R
syntax 1-425S error messages for incompatibility
versus CHAR data type 1-19R 1-21R
Nested ordering, in SELECT 1-339S, functionality in Informix products
2-15T 1-22R
Network functionality listed 1-15R
computer 12-4T installation notes for language
supplement 1-22R

Index 29
 LANG environment variable 4-49R with LIKE, MATCHES keywords
LC_COLLATE environment variable 1-328S
4-50R NOT logical operator 2-36T
LC_CTYPE environment variable NOT NULL keywords
4-51R syntax
LC_MONETARY environment in ALTER TABLE 1-17S
variable 4-52R in CREATE TABLE 1-88S
LC_NUMERIC environment variable use
4-53R in ALTER TABLE 1-27S
LC_TIME environment variable 4-53R in CREATE TABLE 1-90S, 9-24T
NCHAR data type 3-17R with IS keyword 1-327S
NVARCHAR data type 3-18R
populating with dbimport 5-18R NOT operator
setting environment variables 1-16R, condition 1-405S
4-44R NOT relational operator 2-32T
specifying a language environment NOT WAIT keywords, in SET LOCK
2-25T MODE 1-370S
viewing characteristics of 1-18R NULL keyword, ambiguous as
NLS Database procedure variable 1-480S
description of 1-16T NULL relational operator 2-35T
NLS database NULL value
versus non-NLS database 1-19R testing in BYTE expression 3-5R
Nondecomposable attributes 8-15T testing with TEXT data type 3-21R
Nonsequential access. See Disk access, Null value
nonsequential. checking for in SELECT 1-184S, 1-318S
Normal form 8-29T defined 8-23T
Normalization detecting in ESQL 5-15T
benefits 8-29T in INSERT statement 4-7T
first normal form 8-29T in SPL IF statement 2-25S
of data model 8-29T inserting with the VALUES clause
rules 8-29T 1-252S
rules, summary 8-32T restrictions in primary key 8-22T
second normal form 8-31T returned implicitly by stored
third normal form 8-32T procedure 2-38S
specifying as default value 1-20S
NOT CLUSTER keywords testing for 2-35T
syntax in ALTER INDEX 1-12S updating a column 1-387S
use in ALTER TABLE 1-13S used in Condition with NOT operator
NOT FOUND keywords, in the 1-405S
WHENEVER statement 1-398S, used in the ORDER BY clause 1-339S
1-400S with logical operator 2-36T
NOT IN keywords, use in Condition with SPL WHILE statement 2-46S
subquery 1-414S NUMERIC data type. See DECIMAL data
NOT keyword type.
syntax NVARCHAR data type
in Condition segment 1-404S, description of 3-18R, 9-15T
1-405S syntax 1-425S
with BETWEEN keyword 1-327S versus VARCHAR data type 1-19R
with IN keyword 1-329S
use
in Condition segment 1-410S

30 Index
 O constructing the active set 1-264S
opening a procedure cursor 1-265S
OF keyword opening an insert cursor 1-266S
syntax in DECLARE 1-145S opening select or update cursors
use in DECLARE 1-153S 1-264S, 5-20T
OLD keyword reopening a cursor 1-267S
in DELETE REFERENCING clause substituting values for ? parameters
1-119S 1-268S
in INSERT REFERENCING clause syntax 1-263S
1-119S with concatenation operator 1-432S
in UPDATE REFERENCING clause with FREE 1-269S
1-120S Opening a cursor 5-20T, 5-24T
ON DELETE CASCADE keyword Operating system
DELETE trigger event 1-112S DOS 12-4T
ON EXCEPTION statement UNIX 12-4T
placement of 2-32S Optical device 12-6T
scope of control 14-29T Optimization, specifying a high or low
syntax 2-31S level 1-374S
trapping errors 14-27T
user-generated errors 14-30T Optimizer
and GROUP BY 13-10T, 13-12T,
ON keyword 13-27T
syntax and ORDER BY 13-10T, 13-12T,
in CREATE INDEX 1-63S 13-27T
in GRANT 1-231S and SET OPTIMIZATION statement
in REVOKE 1-300S 1-374S, 13-36T
use autoindex path 13-29T
in CREATE INDEX 1-64S composite index use 13-31T
in GRANT 1-237S data distributions 13-29T
ON-Archive utility, using the disk access 13-14T
ARC_DEFAULT environment display query plan 13-12T
variable 4-14R filter selectivity 13-30T
ON-Archive utility, using the index not used by 13-32T, 13-35T
ARC_KEYPAD environment index used by 13-9T, 13-10T
variable 4-15R methods of 13-8T
oncheck utility 10-7T, 10-12T, 13-31T query plan 13-22T
ONCONFIG environment variable 4-39R sort merge join 13-27T
sorting 13-12T
onconfig file, specifying with specifying high or low level of
ONCONFIG 4-40R optimization 13-36T
One-to-many relationship 8-9T, 8-11T system catalog use 13-9T
One-to-one relationship 8-9T, 8-11T when index not used 13-33T
OnLine 5-10T with UPDATE STATISTICS 1-394S
On-line Optimizing
files Intro-8R, Intro-11S, 6T a query 1-360S
help Intro-9R, Intro-11S, 7T a server 1-374S
onload utility 4-27T, 10-11T across a network 1-374S
techniques 13-3T
onstat utility 10-12T
Optional, entity in relationship 8-9T
onunload utility 4-27T, 10-11T
OR keyword
OPEN statement syntax in Condition segment 1-404S
activating a cursor 5-20T

Index 31
 use in Condition segment 1-417S P
OR logical operator 2-36T
Page buffer
OR relational operator 2-33T cost of nonsequential access 13-18T
ORDER BY keywords description of 13-16T
and NLS 2-25T effect on performance 13-17T
ascending order 1-338S, 2-14T restrictions with BLOB data 10-19T
DESC keyword 2-15T, 2-25T PAGE keyword
descending order 1-338S use in ALTER TABLE 1-33S
display label with 2-52T use in CREATE TABLE 1-108S
indexes for 13-10T, 13-34T
multiple columns 2-15T Page lock 7-8T
relation to GROUP BY 3-7T Page, definition of 10-4T
restrictions in INSERT 1-253S, 4-10T Parameter
restrictions in view 11-22T BYTE or TEXT in SPL 2-13S
restrictions with FOR UPDATE 6-8T in CALL statement 2-4S
select columns by number 1-339S, to a stored procedure 14-25T
2-24T Parameterizing a statement
sorting rows 2-13T, 13-12T with SQL identifiers 1-279S
syntax in SELECT 1-310S Parent-child relationship 1-21S, 1-94S
use
in SELECT 1-337S Parts explosion 5-26T
with UNION operator 1-344S PATH environment variable 4-54R,
Order of execution, of action statements 12-14T
1-122S Pathname
orders table in stores6 database, columns including in SQLEXEC 4-42R
in A-3R specifying with DBPATH 4-21R
specifying with PATH 4-54R
Outer join
forming 1-324S Percent (%) sign, wildcard in Condition
segment 1-410S
OUTER keyword, with FROM keyword
in SELECT 1-323S PERFORM keyword, in the WHENEVER
statement 1-398S
Output from TRACE command 15-14T
Performance
OUTPUT statement, syntax and use adding indexes 13-34T
1-271S assigning table to dedicated disk
Owner 10-6T
in ALTER TABLE 1-16S bottleneck tables 10-35T
in CREATE SYNONYM 1-80S buffered log 9-22T
in dbschema 5-35R clustered index 10-25T
in Index Name segment 1-419S, depends on concurrency 7-3T
1-484S, 1-495S, 1-505S disk access 13-14T, 13-17T, 13-34T
in RENAME COLUMN 1-294S disk access by rowid 13-18T
in RENAME TABLE 1-296S disk arm motion 10-8T
in Table Name segment 1-469S, 1-507S disk latency 13-17T
in View Name segment 1-510S dropping indexes to speed
of view in CREATE VIEW 1-511S modifications 10-24T
Owner-privileged procedure 14-13T duplicate keys slow modifications
Ownership 11-7T 10-22T
effect of
BLOB location 10-18T
correlated subquery 13-32T

32 Index
 filter expression 13-23T, 13-32T, Permission, with SYSTEM 2-40S
13-33T Phantom row 1-367S
index damage 13-31T PIPE keyword, in the OUTPUT
indexes 10-21T to 10-22T, 13-26T statement 1-272S
regular expressions 13-32T
table size 13-19T, 13-34T Pipe symbol. See concatenation operator.
updates 13-31T Pipes, unnamed 12-7T
filter selectivity 13-30T Plus (+) sign, arithmetic operator 1-431S
“hot spots,” finding 13-7T Populating tables 9-27T
improved by specifying optimization POW function
level 13-36T syntax in expression 1-444S
improved with temporary table use in expression 1-446S
13-36T
improving 13-27T Precedence, rules for environment
increasing with stored procedures variables 4-7R
14-4T PRECISION field
index time during modification setting with SET DESCRIPTOR 1-358S
10-20T with GET DESCRIPTOR 1-214S
interleaved dbspaces 10-10T PREPARE statement
journal updates 10-33T description of 5-29T
measurement 13-6T error return in SQLERRD 5-12T
multiple access arms per table 10-7T executing 1-184S
network access 13-20T increasing performance efficiency
nonsequential access 13-18T 1-283S
optimizing 13-3T to 13-42T missing WHERE signalled 5-10T
references to other books 13-3T multiple SQL statements 5-30T
reorganizing a dbspace 10-10T multi-statement text 1-277S, 1-281S
row access 13-16T parameterizing a statement 1-278S
seek time 13-17T parameterizing for SQL identifiers
sequential access 13-17T, 13-34T 1-279S
sorting replaces nonsequential access preparing GRANT 11-13T
13-37T question (?) mark as placeholder
splitting tall tables 10-29T 1-273S
splitting wide tables 10-28T releasing resources with FREE 1-208S
time costs of query 13-14T restrictions with SELECT 1-276S
use of derived data 10-30T statement identifier use 1-274S
use of redundant data 10-31T syntax 1-273S
using CHAR instead of NCHAR valid statement text 1-275S
1-19R with concatenation operator 1-432S
Performance analysis Prepared statement
“80-20 rule” 13-7T describing returned values with
measurement 13-6T DESCRIBE 1-162S
methods 13-27T to 13-41T executing 1-184S
nonsequential access 13-37T to 13-41T prepared object limit 1-274S
optimizing techniques 13-4T valid statement text 1-275S
setting up test environment 13-28T See also PREPARE statement.
timing PREVIOUS keyword
from 4GL program 13-6T syntax in FETCH 1-194S
from command script 13-6T use in FETCH 1-196S
from watch 13-6T Primary key
using query plan 13-29T definition of 8-22T
verifying problem 13-4T

Index 33
 restrictions with 8-22T overview 1-8T
Primary key constraint 1-22S Resource 1-233S, 11-7T, 13-34T
composite 8-23T Select 11-8T, 11-10T, 11-27T
data type conversion 1-28S Update 1-236S, 11-8T, 11-10T, 11-28T
defining column as 1-93S when privileges conflict 1-232S
definition of 4-19T with DBA-privileged procedures
dropping 1-32S 14-13T
enforcing 1-86S with owner-privileged procedures
modifying a column with 1-28S 14-13T
referencing 1-22S See also Database-level privilege.
requirements for 1-20S, 1-93S See also Table-level privilege.
rules of use 1-31S, 1-94S PRIVILEGES FOR keywords, in INFO
PRIMARY KEY keywords statement 1-243S
in ALTER TABLE 1-29S PRIVILEGES keyword
in ALTER TABLE statement 1-20S syntax
in CREATE TABLE 1-91S, 1-92S in GRANT 1-235S
Primary site 12-20T in REVOKE 1-301S
Printing, specifying print program with use
DBPRINT 4-24R in GRANT 1-236S
in REVOKE 1-302S
PRIOR keyword
syntax in FETCH 1-194S Procedure cursor
use in FETCH 1-196S opening 1-265S
reopening 1-267S
Privilege
Alter 1-236S, 11-8T procedure cursor 1-148S
and views 11-27T to 11-30T Procedure name
ANSI-compliant databases 1-12R conflict with function name 1-495S
automating grants of 11-13T naming conventions 1-495S
column-level 11-10T Procedure, stored. See Stored procedure.
Connect 1-232S, 11-6T Processing, distributed 12-8T
DBA 1-233S, 11-7T Projection, described 2-7T
default for stored procedures 14-14T
default for table using CREATE Project, description of 1-13T
TABLE 1-85S Promotable lock 1-152S, 7-6T, 7-9T
Delete 1-236S, 11-8T, 11-28T PSORT_DBTEMP environment variable
displaying 4-16T 4-40R
displaying with the INFO statement PSORT_NPROCS environment variable
1-242S 4-41R
encoded in system catalog 2-28R, PUBLIC keyword
2-32R, 11-9T privilege granted to all users 11-6T
Execute 11-12T, 14-14T syntax
for triggered action 1-129S in GRANT 1-231S
granting 11-5T to 11-15T in REVOKE 1-300S
Index 1-236S, 11-8T use
Insert 1-236S, 11-8T, 11-28T in GRANT 1-234S
needed in REVOKE 1-302S
to create a view 1-239S, 11-27T
to drop an index 1-174S PUT statement
to modify data 1-236S, 4-15T constant data with 6-11T
on a synonym 1-80S count of rows inserted 6-11T
on a view 1-137S, 11-28T impact on trigger 1-112S
on stored procedures 14-13T sends returned data to buffer 6-10T

34 Index
 source of row values 1-285S with LIKE, MATCHES keywords
syntax 1-284S 1-328S
use in transactions 1-285S See also Quoted String segment.
with concatenation operator 1-432S Quoted String segment
with FLUSH 1-285S DATETIME, INTERVAL values as
strings 1-498S
Q syntax 1-497S
wildcards 1-498S
Qualifier, field with LIKE in a condition 1-498S
for DATETIME 3-8R, 1-428S, 1-488S Quotes, single and double Intro-6S
for INTERVAL 3-13R, 1-485S, 1-491S
Query
cyclic 4-22T R
improving performance of 13-27T to RAISE EXCEPTION statement
13-41T exiting a loop 14-23T
performance of 13-3T to 13-42T syntax 2-36S
piping results to another program REAL data type. See SMALLFLOAT data
1-272S type.
self-referencing 4-22T
sending results to an operating RECOVER TABLE statement
system file 1-271S archiving a database with audit trails
sending results to another program 1-292S
1-272S manipulating audit trail file 1-293S
stated in terms of data model 1-7T syntax 1-292S
time costs of 13-14T Recursion, in a stored procedure 14-24T
Query optimization information Recursive relationship 8-11T, 8-28T
statements 1-6S Redundant data, introduced for
Query optimizer. See Optimizer. performance 10-31T
Query plan Redundant relationship 8-29T
autoindex path 13-29T REFERENCES FOR keywords, in INFO
chosen by optimizer 13-11T statement 1-243S
description of 13-22T REFERENCES keyword
display with SET EXPLAIN 13-12T in ALTER TABLE 1-21S
indexes in 13-26T in CREATE TABLE 1-94S, 1-96S
use in analyzing performance 13-29T syntax
Question (?) mark in GRANT 1-235S
as placeholder in PREPARE 1-273S in REVOKE 1-301S
naming variables in PUT 1-287S use
replacing with USING keyword in GRANT 1-236S
1-268S in REVOKE 1-302S
wildcard in Condition segment 1-411S References privilege
Quoted string definition of 1-236S
in expression 1-436S displaying with the INFO statement
syntax 1-243S
in Condition segment 1-405S REFERENCING clause
in expression 1-462S DELETE REFERENCING clause
in INSERT 1-250S 1-119S
use INSERT REFERENCING clause
in expression 1-438S 1-118S
in INSERT 1-498S UPDATE REFERENCING clause
1-120S

Index 35
 using referencing 1-125S, 15-9T cardinality 8-9T, 8-13T
Referential constraint complex 8-28T
and a DELETE trigger 1-112S connectivity 8-8T, 8-11T
data type restrictions 1-96S defining in data model 8-8T
definition of 1-21S, 1-94S, 4-19T entity 8-5T
dropping 1-32S existence dependency 8-9T
enforcing 1-86S mandatory 8-9T
modifying a column with 1-28S many-to-many 8-9T, 8-11T
rules of use 1-94S many-to-many, resolving 8-26T
Referential integrity 4-19T one-to-many 8-9T, 8-11T
defining primary and foreign keys one-to-one 8-9T, 8-11T
8-24T optional 8-9T
recursive 8-28T
Regular expression redundant 8-29T
effect on performance 13-32T using matrix to discover 8-10T
evaluating with NLS 1-15R
RELATIVE keyword
Relational calculus. See also relational syntax in FETCH 1-194S
model. use in FETCH 1-196S
Relational database, defined 1-11T Relay Module
Relational model SQLRM environment variable 4-42R
attribute 8-14T SQLRMDIR environment variable
denormalizing 10-26T 4-43R
description of 1-11T, 8-3T to 8-33T Release notes Intro-9R, Intro-11S, 6T
entity 8-4T
join 2-8T Remote database server 12-8T
many-to-many relationship 8-11T RENAME COLUMN statement
normalizing data 8-29T restrictions 1-294S
one-to-many relationship 8-11T syntax 1-294S
one-to-one relationship 8-11T RENAME TABLE statement
projection 2-6T ANSI-compliant naming 1-296S
resolving relationships 8-26T syntax 1-296S
rules for defining tables, rows, and REPAIR TABLE statement, syntax and
columns 8-20T use 1-299S
selection 2-5T Repeatable Read isolation level
Relational operation 2-5T description of 1-367S, 7-12T
Relational operator emulating during update 1-202S
BETWEEN 2-32T REPEATABLE READ keywords, syntax
equals 2-30T in SET ISOLATION 1-366S
EXISTS 3-33T Replication of data 12-19T
IN 3-33T
in a WHERE clause 2-29T to 2-45T Report generator 1-17T
in Condition segment 1-405S Reserved words 1-470S
LIKE 2-37T Resolution
NOT 2-32T in UPDATE STATISTICS 1-396S,
NULL 2-35T 1-397S
OR 2-33T with data distributions 1-396S
segment 1-500S RESOURCE keyword
with WHERE keyword in SELECT use in GRANT 1-233S
1-326S use in REVOKE 1-303S
Relationship Resource manager 12-11T
attribute 8-15T Resource privilege 1-233S, 11-7T, 13-34T

36 Index
Restricting access, using file system defining 8-20T
11-4T deleting 1-159S, 4-4T
Result of triggering statement 1-123S description of 1-12T
RETURN statement engine response to locked row 1-370S
exiting a loop 14-23T in relational model 1-12T, 8-20T
returning insufficient values 2-38S inserting 4-6T
returning null values 2-38S inserting through a view 1-246S
syntax 2-38S inserting with a cursor 1-248S
multi-row queries with FETCH
REVOKE statement 1-195S
column-specific privileges 1-303S order, guaranteeing independence of
database-level privileges 1-303S 1-117S
granting privileges 11-5T to 11-15T phantom 1-367S
in embedded SQL 5-34T to 5-37T retrieving with FETCH 1-197S
privileges needed 1-301S rowid definition 1-197S
syntax 1-300S size of fixed-length 10-13T
table-level privileges 1-301S updating through a view 1-384S
with a view 11-29T writing buffered rows with FLUSH
ROLLBACK WORK statement 1-204S
cancels a transaction 4-25T ROW keyword
closes cursors 7-18T use in ALTER TABLE 1-33S
releases locks 7-8T, 7-18T use in CREATE TABLE 1-108S
sets SQLCODE 6-5T
syntax 1-306S Row lock 7-8T
use with WHENEVER 1-36S, 1-42S, Rowid
1-307S locating internal row numbers 3-17T
with DROP DATABASE 1-173S use in a column expression 1-435S
with DROP INDEX statement 1-175S use in a join 3-16T
with DROP PROCEDURE statement use in ORDER BY clause 1-338S
1-176S used as column name 1-474S
with DROP SYNONYM statement Rowid function 13-18T
1-177S ROWID keyword 1-435S
with DROP TABLE statement 1-179S Rules for stored procedures 1-128S
with DROP TRIGGER statement
1-181S Run-time program
with DROP VIEW statement 1-183S setting DBANSIWARN 4-15R
setting INFORMIXCOBDIR 4-32R
ROLLFORWARD DATABASE
statement
applies log to restored database 4-26T S
exclusive locking 1-308S SCALE field
syntax 1-308S setting with SET DESCRIPTOR 1-358S
Root dbspace, definition of 10-5T with GET DESCRIPTOR 1-214S
ROOT function Schema. See Data Model.
syntax in expression 1-444S Scroll cursor
use in expression 1-446S active set 5-25T
ROUND function definition of 1-149S, 5-23T
syntax in expression 1-444S use of 1-149S
use in expression 1-446S with FETCH 1-196S
Row with hold, in a transaction 1-369S
cost of reading from disk 13-16T SCROLL keyword
defined 2-5T syntax in DECLARE 1-145S

Index 37
 use in DECLARE 1-149S, 5-23T selecting all columns 2-12T
Search condition. See Condition segment. selecting specific columns 2-18T
secheck utility 13-31T specifying a substring in 2-27T
SECOND keyword Select privilege
syntax column level 11-10T
in DATETIME data type 1-428S definition of 1-236S, 11-8T
in INTERVAL data type 1-485S with a view 11-27T
use SELECT statement
as DATETIME field qualifier 3-8R, active set 2-29T
1-488S aggregate functions in 1-462S, 2-53T
as INTERVAL field qualifier 3-14R, alias names 2-79T
1-491S and COLLCHAR environment
Second normal form 8-31T variable 4-46R
and LC_COLLATE environment
Secondary site 12-20T variable 4-51R
Security and NLS collation order 1-15R
constraining inserted values 11-20T, as an argument to a stored procedure
11-25T 2-4S
database-level privileges 11-5T assigning values with 14-21T
making database inaccessible 11-5T associating with cursor with
restricting access to columns 11-20T DECLARE 1-148S
restricting access to rows 11-20T, BETWEEN condition 1-327S
11-21T column numbers 1-339S
restricting access to view 11-27T compound query 3-43T
table-level privileges 11-10T cursor for 5-19T, 5-20T
using host file system 11-4T date-oriented functions in 2-56T
using operating system facilities describing returned values with
11-4T DESCRIBE 1-162S
with stored procedures 11-3T description of advanced 3-4T to 3-55T
Seek time 13-17T description of simple 2-3T to 2-83T
Select display label 2-49T
description of 1-13T DISTINCT keyword 2-19T
Select cursor embedded 5-13T to 5-16T
definition of 1-148S for joined tables 2-69T to 2-83T
opening 1-264S, 5-20T for single tables 2-11T to 2-66T
reopening 1-267S FROM Clause 1-323S
use of 1-148S, 5-20T functions 2-53T to 2-66T
GROUP BY clause 1-334S, 3-4T
SELECT keyword HAVING clause 1-336S, 3-9T
ambiguous use as procedure variable IN condition 1-327S
1-480S in FOR EACH ROW section 1-117S
syntax in modifiable view 11-24T
in GRANT 1-235S INTO clause with ESQL 1-318S
in REVOKE 1-301S INTO TEMP clause 1-341S, 2-83T
use IS NULL condition 1-327S
in GRANT 1-236S join 2-71T to 2-79T
in REVOKE 1-302S joining tables in WHERE clause
Select list 1-331S
display label 2-49T LIKE or MATCHES condition 1-328S
expressions in 2-46T multiple-table 2-69T
functions in 2-53T to 2-66T natural join 2-75T
labels in 3-48T

38 Index
 null values in the ORDER BY clause definition of 1-149S, 5-23T
1-339S use of 1-149S
ORDER BY clause 1-337S, 2-13T with FETCH 1-195S
ORDER BY clause and NLS 1-15R SERIAL data type
outer join 3-21T to 3-31T description of 3-19R, 9-7T
privilege for 11-6T, 11-8T generated number in SQLERRD 5-12T
relational-operator condition 1-326S in ALTER TABLE 1-17S
restrictions with INTO clause 1-276S in INSERT 1-251S
rowid 3-16T, 3-21T inserting a starting value 4-8T
ROWID keyword 1-435S inserting values 3-19R
SELECT clause 1-312S, 2-12T to 2-28T resetting values 3-19R
select numbers 1-339S syntax 1-425S
selecting a substring 2-27T treatment by dbschema 5-34R
selecting expressions 2-46T with stored procedures 2-8S
selection list 2-12T Server. See Database server.
self-join 3-11T
single-table 2-11T SET clause 1-125S, 4-14T
singleton 1-318S, 2-29T SET CONNECTION statement
subquery 3-32T to 3-42T syntax and use 1-346S
subquery with WHERE keyword with concatenation operator 1-432S
1-326S SET CONSTRAINTS statement
syntax 1-310S syntax and use 1-349S
UNION operator 1-344S, 3-43T use with CREATE TRIGGER 1-131S
use of expressions 1-313S SET DEBUG FILE TO statement
using syntax and use 1-351S
for join 2-8T with TRACE 2-42S
for projection 2-7T SET DESCRIPTOR statement
for selection 2-5T syntax 1-353S
WHERE clause and NLS 1-15R the VALUE option 1-355S
with with concatenation operator 1-432S
Condition segment 1-404S X/Open mode 1-357S
DECLARE 1-145S
FOREACH 2-20S Set difference 3-53T
INSERT 1-253S SET EXPLAIN statement
INTO keyword 1-198S interpreting output 1-361S, 13-29T
LET 2-29S MERGE JOIN information 1-362S
writing rows retrieved to an ASCII file optimizer access paths 1-361S
1-378S output examples 1-362S
Selection, described 2-5T SORT SCAN information 1-362S
syntax 1-360S
Self-join writes query plan 13-12T
assigning column names with INTO
TEMP 3-12T Set intersection 3-51T
description of 1-333S, 3-11T SET ISOLATION statement
See also Join. controlling the effect of locks 4-28T
Self-referencing query 3-11T, 4-22T default database levels 1-368S
definition of isolation levels 1-367S
Self-referencing query. See also Self-join. effects of isolation 1-368S
Semantic integrity 4-18T, 9-3T restrictions 7-9T
Sequential access. See Disk access, syntax 1-366S
sequential. SET keyword
Sequential cursor syntax in UPDATE 1-383S
use in UPDATE 1-386S, 4-12T

Index 39
 SET LOCK MODE statement description of 3-20R, 9-9T
controlling the effect of locks 4-28T syntax 1-425S
description of 7-13T SMALLINT data type
kernel locking 1-370S changing data types 3-23R
setting wait period 1-371S description of 3-20R, 9-7T
syntax 1-370S syntax 1-425S
SET LOG statement using as default value 1-19S, 1-89S
buffered vs. unbuffered 1-372S, 9-22T SOME keyword
syntax 1-372S beginning a subquery 1-330S, 3-33T
SET OPTIMIZATION statement, syntax use in Condition subquery 1-415S
and use 1-374S Sort merge join 13-27T
Setting environment variables 4-5R Sorting
SHARE keyword, syntax in LOCK avoiding nonsequential access 13-37T
TABLE 1-260S avoiding with temporary table 13-36T
Shared memory effect on performance 13-33T
network connection 12-7T in a combined query 1-344S
Shared memory parameters, specifying in SELECT 1-337S
file with ONCONFIG 4-39R nested 2-15T
optimizer estimates cost 13-12T
Shell PSORT_DBTEMP environment
setting environment variables in a file variable 4-40R
4-4R PSORT_NPROCS environment
specifying with DBREMOTECMD variable 4-41R
4-25R sort merge join 13-27T
Simple assignment 2-28S time costs of 13-15T
SIN function when there are multiple active NLS
syntax in expression 1-458S locales 1-20R
use in expression 1-459S with NLS activated 1-15R, 2-25T
Single-precision floating-point number, with ORDER BY 2-14T
storage of 3-12R See also ORDER BY keywords.
Singleton SELECT statement 1-318S, Space ( )
2-29T as delimiter in DATETIME 3-9R
Site as delimiter in INTERVAL 3-15R
network 12-4T Specifying ANSI-compliance 1-11R
primary 12-20T SPL
secondary 12-20T flow control statements 14-22T
SITENAME function program variable 5-5T
returns servername 1-438S relation to SQL 14-3T
syntax See also Stored procedure.
in expression 1-436S SQL
in INSERT 1-250S ANSI standard 1-15T
use cursor 5-19T
in ALTER TABLE 1-18S description of 1-14T
in CREATE TABLE 1-88S error handling 5-16T
in expression 1-438S history 1-14T
in INSERT 1-252S in NLS-ready products 1-22R
in SELECT 2-63T, 2-65T, 3-20T Informix SQL and ANSI SQL 1-15T
Slash (/), arithmetic operator 1-431S interactive use 1-17T
SMALLFLOAT data type optimizing. See Optimizer.
changing data types 3-23R standardization 1-14T

40 Index
 statement types 1-5S in OPEN 1-269S
SQL Communications Area (SQLCA) in PUT 1-288S
altered by end of transaction 6-5T use with EXECUTE statement 1-186S
description of 5-8T SQLERRD array
effect of setting DBANSIWARN 4-15R count of deleted rows 6-4T
inserting rows 6-11T count of inserted rows 6-11T
result after CLOSE 1-39S count of rows 6-14T
result after DATABASE 1-140S description of 5-12T
result after DESCRIBE 1-163S syntax of naming 5-11T
result after FETCH 1-202S SQLERROR keyword, in the
result after FLUSH 1-204S WHENEVER statement 1-398S
result after OPEN 1-264S, 1-265S SQLEXEC environment variable 4-41R
result after PUT 1-290S
result after SELECT 1-321S sqlexecd 12-14T
returning NLS error messages to sqlhosts. See
1-19R $INFORMIXDIR/etc/sqlhosts.
See also SQLAWARN, SQLCODE, and SQLNOTFOUND
SQLERRD. error conditions with EXECUTE
SQL DESCRIPTOR clause. See statement 1-188S
DESCRIBE statement. SQLRM environment variable 4-42R
SQL Descriptor. See SQL SQLRMDIR environment variable 4-43R
Communications Area (SQLCA). SQLSTATE
SQL statements in databases that are not
FILE 5-23R ANSI-compliant 5-18T
INSERT 5-23R use with a cursor 5-21T
SQLAWARN array SQLWARNING keyword, in the
description of 5-12T WHENEVER statement 1-400S
syntax of naming 5-11T SQRT function
with PREPARE 5-30T syntax in expression 1-444S
sqlca record and EXECUTE statement use in expression 1-447S
1-186S START DATABASE statement
SQLCA. See SQL Communications Area. adding a transaction log 9-24T
SQLCODE field syntax and use 1-376S
after opening cursor 5-20T state table in stores6 database, columns
description of 5-11T in A-6R
end of data on SELECT only 6-14T Statement
end of data signalled 5-17T naming with NLS 1-15R
set by DELETE 6-4T SQL, ANSI compliance and
set by DESCRIBE 5-33T DBANSIWARN 4-15R
set by PUT, FLUSH 6-11T SQL, CONNECT and
sqlda structure INFORMIXSERVER 4-37R
syntax SQL, CREATE TABLE and
in DESCRIBE 1-162S COLLCHAR 4-46R
in EXECUTE 1-184S SQL, DESCRIBE and COLLCHAR
in FETCH 1-194S 4-46R
in OPEN 1-263S SQL, editing and DBEDIT 4-18R
in PUT 1-284S SQL, LOAD and DBDELIMITER
use 4-18R
in DESCRIBE 1-164S SQL, SELECT and COLLCHAR 4-46R
in FETCH 1-201S

Index 41
 SQL, SELECT and LC_COLLATE definition of 14-3T
4-51R displaying contents 14-9T
SQL, UNLOAD and DBDELIMITER displaying documentation 14-9T
4-18R executing 14-9T
SQL, UPDATE STATISTICS and general programming 1-17T
DBUPSPACE 4-30R granting privileges on 1-237S, 11-12T,
Statement identifier 14-15T
associating with cursor 1-148S handling multiple rows 2-39S
definition of 1-274S header 2-8S, 14-25T
releasing 1-274S in SELECT statements 1-315S, 2-67T
syntax in WHEN condition 1-122S
in DECLARE 1-145S introduction to 14-3T
in DESCRIBE 1-162S looping 14-23T
in EXECUTE 1-184S name confusion with SQL functions
in FREE 1-207S 14-21T
in PREPARE 1-273S naming output file for TRACE
use statement 1-351S
in DECLARE 1-154S owner-privileged 1-129S, 14-13T
in FREE 1-208S privileges 1-129S
in PREPARE 1-274S privileges necessary at execution
Statement types 1-5S 14-14T
program flow control 14-22T
Statement variable name, definition receiving data from SELECT 1-318S
1-190S recursion 14-24T
Static SQL 5-5T REFERENCES clause 14-18T
STATUS FOR keywords, in INFO returning values 14-25T
statement 1-244S revoking privileges on 1-301S, 14-16T
STATUS variable (4GL) 5-11T security purposes 11-3T
Status, displaying with INFO statement simulating errors 2-36S
1-244S tracing triggered actions 15-12T
use 14-3T
stock table in stores6 database, columns variable 14-16T
in A-4R
Stored Procedure Language. See SPL.
STOP keyword, in the WHENEVER
statement 1-398S, 1-402S stores6 database
call_type table columns A-5R
Storage device 12-6T catalog table columns A-4R
Stored procedure copying Intro-10R, Intro-13S, 8T
altering 14-13T creating Intro-10R, Intro-13S, 8T
as triggered action 1-128S, 15-10T customer table columns A-2R
branching 14-22T cust_calls table columns A-5R
BYTE and TEXT data types 2-9S, 2-13S data values A-15R
checking references 1-129S description of A-1R
comments in 14-6T items table columns A-3R
creating from an embedded language manufact table columns A-5R
14-5T map of A-6R
creating from DB-Access 14-5T orders table columns A-3R
cursors with 2-20S overview Intro-9R, Intro-12S, 7T
DBA-privileged, use with triggers primary-foreign key relationships
1-129S, 14-13T A-8R to A-15R
debugging 2-42S, 14-11T state table columns A-6R
default privileges 14-14T stock table columns A-4R
DEFINE statement 14-17T structure of tables A-2R
42 Index
Structured Query Language. See SQL. altering contents 2-9R
Subquery character column data type when
beginning with ALL/ANY/SOME NLS activated 1-19R
keywords 1-330S database entries 1-57S
beginning with EXISTS keyword description of 2-3R
1-330S map of tables 2-33R
beginning with IN keyword 1-329S NCHAR columns in 1-19R
correlated 1-413S, 3-32T, 4-22T, 13-32T privileges in 4-16T, 11-9T
definition of 1-326S querying 4-16T
in Condition segment 1-413S sysblobs 2-11R
in DELETE statement 4-6T syschecks 2-11R
in SELECT 3-32T to 3-42T syscolauth 2-12R, 1-302S, 11-9T
in UPDATE-SET 4-13T syscoldepend 2-13R
in UPDATE-WHERE 4-12T syscolumns 2-13R
performance of 13-32T sysconstraints 2-16R
restrictions with UNION operator sysdefaults 2-17R
1-344S sysdepend 2-18R
with DISTINCT keyword 1-313S sysdistrib 2-18R
See also Condition segment. sysindexes 2-20R
Subscripting sysopclstr 2-22R
in a WHERE clause 2-44T sysprocauth 2-23R
on character columns 1-434S sysprocbody 2-24R, 14-8T
SPL variables 14-18T sysprocedures 2-25R
sysprocplan 2-26R
Subservient table 3-21T sysreferences 2-26R
Substring 2-27T, 14-18T syssynonyms 2-27R
SUM function syssyntable 2-27R
as aggregate function 2-53T systabauth 2-28R, 1-239S, 1-302S,
syntax in expression 1-462S 4-16T, 11-9T
use in expression 1-465S systables 2-29R
Symbol table 10-27T systrigbody 2-31R
Synonym systriggers 2-32R
ANSI-compliant naming 1-80S sysusers 2-32R, 11-9T
chaining 1-83S sysviews 2-33R
chains 12-18T updating 2-9R
creating with CREATE SYNONYM updating statistics 2-9R
1-80S used by optimizer 13-9T
difference from alias 1-80S System catalog tables. See System
dropping 1-177S catalog.
in ANSI-compliant database 1-14R System descriptor area 1-164S
synonym behavior assigning values to 1-354S
in ANSI-compliant database 1-14R modifying contents 1-354S
Synonyms for table names 12-17T resizing 1-355S
use with EXECUTE statement 1-187S
Syntax diagram
conventions Intro-5S System name, in database name 1-422S
elements of Intro-9S SYSTEM statement
sysdepend system catalog table 1-183S syntax 2-40S
syssyntable 12-18T
System catalog
accessing 2-8R

Index 43
 T represents an entity 8-22T
restoring with audit trail 1-292S
Table structure in stores6 database A-2R
adding a constraint 1-29S system catalog tables 2-11R to 2-33R
alias in SELECT 1-323S unlocking 1-381S
ANSI-compliant naming 1-507S variable-length rows 10-14T
bottleneck 10-35T See also ALTER TABLE statement.
candidate keys, defined 8-23T See also CREATE TABLE statement.
changing the data type of a column See also Join.
3-23R See also Table Name segment.
checking with the CHECK TABLE TABLE keyword, syntax in UPDATE
statement 1-37S STATISTICS 1-393S
composite key, defined 8-23T
contained in one dbspace 10-5T Table Name segment 1-506S
creating Table size
a synonym for 1-80S calculating 10-12T, 10-19T
a table 1-84S, 9-24T cost of access 13-19T, 13-34T
a temporary table 1-100S with fixed-length rows 10-13T
dedicated device for 10-6T with variable-length rows 10-14T
description of 1-11T Table-level privilege
dropping column-specific privileges 1-303S,
a constraint 1-32S 11-10T
a synonym 1-177S default with GRANT 1-237S
a table 1-179S definition and use 1-236S, 11-7T
engine response to locked table 1-370S granting 1-235S
extent sizes of 10-8T passing grant ability 1-237S
fixed-length rows 10-13T revoking 1-301S
in mirrored storage 10-6T See also ALTER TABLE statement.
in relational model 1-11T, 8-20T TABLES keyword, in INFO statement
interleaved dbspaces 10-10T 1-242S
joins in Condition segment 1-331S tabtype 2-29R, 2-30R, 1-60S
loading data with the LOAD
statement 1-255S TAN function
lock 7-7T syntax in expression 1-458S
locking use in expression 1-460S
changing mode 1-33S tblspace
with ALTER INDEX 1-13S description of 10-7T
with LOCK TABLE 1-260S used for BLOB data 10-18T
logging 1-104S TEMP keyword
multiple access arms for 10-7T syntax in SELECT 1-310S
names 12-16T use in SELECT 1-341S
names, synonyms 12-17T TEMP TABLE keywords, syntax in
naming conventions 1-84S, 1-470S, CREATE TABLE 1-84S
1-507S Temporary
naming with NLS 1-15R files, specifying directory with
optimizing queries 1-394S DBTEMP 4-27R
ownership 11-7T tables, specifying dbspace with
primary key in 8-22T DBSPACETEMP 4-26R
primary key, defined 8-22T
relation to dbspace 10-6T Temporary table
repairing with REPAIR TABLE and active set of cursor 5-24T
statement 1-299S assigning column names 3-12T
building distributions 1-397S
44 Index
 creating constraints for 1-102S syntax 1-425S
DBSPACETEMP environment use in Boolean expression 3-21R
variable 1-100S used for performance 10-27T
example 4-11T with control characters 3-21R
explicit 1-100S with LENGTH function 2-64T
implicit 1-100S with stored procedures 2-9S, 2-13S
location of 1-100S Text editor, specifying with DBEDIT
naming 1-100S 4-18R
shared disk space for 10-6T TEXT value, displaying 2-11T
updating statistics 1-394S
using to speed query 13-36T Third normal form 8-32T
when deleted 1-100S Time function
TERM environment variable 4-55R restrictions with GROUP BY 1-334S
use in SELECT 1-314S
TERMCAP environment variable 4-56R,
12-14T Time, representing with NLS 1-15R
termcap file TO CLUSTER keywords, in ALTER
and TERMCAP environment variable INDEX 1-12S
4-56R TO keyword
selecting with INFORMIXTERM in expression 1-462S
4-39R in GRANT 1-231S
Terminal handling TODAY function
and TERM environment variable syntax
4-55R in Condition segment 1-405S
and TERMCAP environment variable in expression 1-436S
4-56R in INSERT 1-250S
and TERMINFO environment use
variable 4-56R in ALTER TABLE 1-18S
terminfo directory in constant expression 1-439S,
and TERMINFO environment 2-63T, 4-8T
variable 4-56R in CREATE TABLE 1-88S
selecting with INFORMIXTERM in INSERT 1-252S
4-39R TP/XA. See Transaction manager.
TERMINFO environment variable 4-56R TRACE command
TEXT data type output from 15-14T
choosing location for 10-18T TRACE statement
description of 3-21R, 9-17T debugging a stored procedure 14-11T
disk storage for 10-4T syntax 2-42S
estimating disk space for 10-17T Transaction
inserting values 3-21R and CREATE DATABASE 1-60S
requirements for LOAD statement ANSI-compliant database, effects
1-257S 1-11R
restrictions committing with COMMIT WORK
with aggregate functions 3-21R 1-43S
with GROUP BY 3-21R, 3-7T cursors closed at end 7-18T
with IN clause 3-21R description of 4-22T
with LIKE or MATCHES 3-21R, example with DELETE 6-5T
2-37T global 12-21T
with ORDER BY 3-21R locks held to end of 7-9T
with relational expression 2-29T locks released at end 7-8T, 7-18T
selecting a column 3-22R logging 1-376S
recovering transactions 1-308S

Index 45
 rolling back 1-4R, 1-306S syntax 1-121S
scroll cursor and data consistency WHEN condition 1-121S
1-369S correlation name in 1-125S, 1-128S
starting with BEGIN WORK 1-35S FOR EACH ROW 15-8T
stopping logging 1-376S generating an error message 15-14T
transaction log 4-24T, 4-26T in client/server environment 1-133S
transaction log required 9-22T in relation to triggering statement
use signalled in SQLAWARN 5-12T 15-6T
using cursors in 1-155S list
Transaction logging AFTER 1-117S
ANSI-compliant database, effects BEFORE 1-116S
1-12R FOR EACH ROW 1-116S
buffered 9-22T for multiple triggers 1-117S
effect on database server type 1-5R sequence of 1-116S
establishing with CREATE merged 1-117S
DATABASE 9-21T preventing overriding 1-132S
OnLine methods of 9-22T statements 15-3T
renaming log 1-377S tracing 15-12T
stopping 1-377S using 15-7T
turning off for faster loading 9-28T using stored procedures 15-10T
turning off not possible 9-24T WHEN condition 1-121S, 15-10T
Transaction log, contents of 4-25T Triggering statement
Transaction manager 12-11T affecting multiple rows 1-117S
execution of 1-113S
Transfer of database files 5-19R guaranteeing same result 1-112S
Transitive dependency 8-32T result of 1-123S
Trigger UPDATE 1-115S
creating 15-4T Triggering table
definition of 15-3T action on 1-127S
in client/server environment 1-133S and cascading triggers 1-131S
number on a table 1-112S Trigonometric function
preventing overriding 1-132S ACOS function 1-460S
when to use 15-3T ASIN function 1-460S
Trigger event ATAN function 1-460S
definition of 1-111S, 15-5T ATAN2 function 1-461S
example of 15-5T COS function 1-459S
in CREATE TRIGGER statement SIN function 1-459S
1-111S TAN function 1-460S
INSERT 1-119S TRUNC function
privileges on 1-113S syntax in expression 1-444S
with cursor statement 1-112S use in expression 1-447S
Trigger name Truncation, signalled in SQLAWARN
assigning 15-5T 5-12T
syntax 1-113S
Two-phase commit 12-20T
Triggered action
action on triggering table 1-127S TYPE field
anyone can use 1-129S changing from BYTE or TEXT 1-359S
BEFORE and AFTER 15-7T setting in SET DESCRIPTOR 1-356S
cascading 1-118S setting in X/Open programs 1-357S
clause 1-121S with X/Open programs 1-213S
action statements 1-122S

46 Index
Typographical conventions Intro-5R, TERMINFO environment variable
Intro-5S, 5T 4-56R
UNIX operating system 12-4T, 12-6T
U UNLOAD statement
DELIMITER clause 1-380S
Unbuffered logging 9-22T exporting data to a file 9-27T
Underscore (_), wildcard in Condition specifying field delimiter with
segment 1-410S DBDELIMITER 4-18R
UNION operator syntax 1-378S
description of 3-43T UNLOAD TO file 1-378S
display labels with 3-48T unloading VARCHAR, TEXT, or
restrictions in view 11-22T BYTE columns 1-379S
restrictions on use 1-344S UNLOAD TO file 1-378S
syntax in SELECT 1-310S Unloading a database 5-8R
use in SELECT 1-344S
UNLOCK TABLE statement, syntax and
Unique constraint use 1-381S
dropping 1-32S
modifying a column with 1-28S Unnamed pipes 12-7T
rules of use 1-30S, 1-92S, 1-94S Updatable view 1-139S
UNIQUE keyword UPDATE clause, syntax 1-114S
constraint in CREATE TABLE 9-24T Update cursor 1-148S
restrictions in modifiable view 11-24T definition of 1-148S, 6-15T
syntax locking considerations 1-152S
in CREATE INDEX 1-63S opening 1-264S
in CREATE TABLE 1-91S restricted statements 1-152S
in SELECT 1-312S use in UPDATE 1-390S
use using 1-152S
in ALTER TABLE 1-29S Update journal 10-33T
in CREATE INDEX 1-64S UPDATE keyword
in CREATE TABLE 1-92S syntax
in expression 1-462S in GRANT 1-235S
in SELECT 1-313S, 2-19T in REVOKE 1-301S
no effect in subquery 1-414S use
UNITS keyword in GRANT 1-236S
syntax in expression 1-436S in REVOKE 1-302S
use in expression 1-442S Update privilege
UNIX column level 11-10T
BSD definition of 1-236S, 11-8T
default print capability 4-7R, 4-24R with a view 1-384S, 11-28T
viewing environment settings 4-5R UPDATE REFERENCING clause
environment variables listed 4-9R and FOR EACH ROW section 1-121S
PATH environment variable 4-54R correlation name 1-120S
specifying directories for syntax 1-120S
intermediate writes 4-40R
System V UPDATE statement
default print capability 4-7R, 4-24R and end of data 6-14T
terminfo library support 4-39R and transactions 1-384S
viewing environment settings 4-5R applied to view 11-24T
TERM environment variable 4-55R as triggered action 1-122S
TERMCAP environment variable as triggering statement 1-112S, 1-114S,
4-56R 1-115S

Index 47
 description of 4-12T in expression 1-438S, 2-63T, 2-64T,
embedded 6-14T to 6-17T 3-19T
in trigger event 1-111S in INSERT 1-252S
locking considerations 1-385S User informix, privileges associated with
missing WHERE signalled 5-10T 1-234S
multiple assignment 4-14T Using correlation names 1-124S
number of rows 5-12T
preparing 5-30T USING DESCRIPTOR keywords
privilege for 11-6T, 11-8T information from DESCRIBE 1-164S
restrictions on columns for update syntax
1-153S in EXECUTE 1-184S
restrictions on subqueries 4-13T in FETCH 1-194S
rolling back updates 1-385S in OPEN 1-263S
syntax 1-383S in PUT 1-284S
time to update indexes 10-20T use
updating a column to null 1-387S in FETCH 1-201S
updating through a view 1-384S in OPEN 1-269S
updating with cursor 1-390S in PUT 1-188S, 1-288S, 1-289S
use of expressions 1-388S USING keyword
with syntax
Condition segment 1-404S in EXECUTE 1-186S
FETCH 1-201S in OPEN 1-263S
SET keyword 1-386S use
WHERE CURRENT OF keywords in EXECUTE 1-186S, 5-31T
1-390S in OPEN 1-268S
WHERE keyword 1-388S USING SQL DESCRIPTOR keywords
with a select..for update 1-340S in DESCRIBE 1-164S
with an update cursor 1-152S in EXECUTE 1-187S
UPDATE STATISTICS statement Utility program
affect on sysdistrib 2-19R chkenv 5-4R
and DBUPSPACE environment crtcmap 5-5R
variable 4-30R dbexport 5-8R
creating distributions 1-396S dbimport 5-13R
dropping data distributions 1-396S dbload 5-19R, 9-28T, 10-12T
examining index pages 1-394S dbschema 5-33R, 9-26T
optimizing search strategies 1-394S oncheck 10-7T, 10-12T
syntax 1-392S onload 4-27T, 10-11T
using the LOW keyword 1-395S onstat 10-12T
when to execute 1-395S onunload 4-27T, 10-11T
Update trigger, defining multiple 1-114S
USER function V
as affected by ANSI compliance
1-231S, 1-300S, 1-438S VALUE clause
syntax after NULL value is fetched 1-215S
in Condition segment 1-405S relation to FETCH 1-214S
in expression 1-436S use in GET DESCRIPTOR 1-212S
in INSERT 1-250S use in SET DESCRIPTOR 1-355S
use VALUES clause
in ALTER TABLE 1-18S effect with PUT 1-286S
in CREATE TABLE 1-88S syntax in INSERT 1-245S
use in INSERT 1-250S, 4-7T

48 Index
VARCHAR data type virtual column 1-137S, 11-24T
considerations for UNLOAD with SELECT * notation 1-136S
statement 1-379S View Name segment 1-510S
description of 3-22R, 9-15T
effect on table size 10-14T
requirements for LOAD statement W
1-257S WAIT keyword, in the SET LOCK
syntax 1-425S MODE statement 1-370S
used for performance 10-26T WARNING keyword, in the
using as default value 1-19S, 1-89S WHENEVER statement 1-398S
versus NVARCHAR data type 1-19R
with LENGTH function 2-64T Warnings, with stored procedures at
compile time 14-8T
VARCHAR value, displaying 2-11T
WEEKDAY function
Variable as time function 2-56T, 2-60T
default values in SPL 2-11S, 2-12S syntax in expression 1-454S
define in SPL 2-7S use in expression 1-456S
global, in SPL 2-10S, 14-17T
in SPL 14-16T WHEN condition
local, in SPL 2-12S, 14-17T in triggered action 1-122S
scope of SPL variable 2-8S restrictions 1-122S
unknown values in IF 2-25S use of 1-122S
with same name as a keyword 14-19T WHENEVER statement, syntax and use
View 1-398S
creating a view 1-136S, 11-20T WHERE clause, subscripting 2-44T
creating synonym for 1-80S WHERE CURRENT OF clause, impact
deleting rows in 11-24T on trigger 1-112S
description of 11-19T WHERE CURRENT OF keywords
display description with dbschema syntax
5-34R in DELETE 1-159S
dropped when basis is dropped in UPDATE 1-383S
11-22T use
dropping 1-183S in DELETE 6-7T
effect of changing basis 11-23T in UPDATE 1-390S, 6-15T
effect on performance 13-33T WHERE keyword
inserting rows in 11-25T Boolean expression in 2-36T
modifying 11-23T to 11-27T comparison condition 2-29T to 2-45T
naming with NLS 1-15R date-oriented functions in 2-60T
null inserted in unexposed columns enforcing data constraints 11-27T
11-25T host variables in 5-14T
privilege when accessing 11-28T in DELETE 4-4T to 4-6T
privilege when creating 1-137S joining tables 1-331S
privileges 11-27T to 11-30T null data tests 2-35T
privileges with GRANT 1-239S prevents use of index 13-32T, 13-33T,
produces duplicate rows 11-22T 13-35T
restrictions with UNION operator range of values 2-32T
1-344S relational operators 2-29T
system catalog table 2-33R selecting rows 2-28T
updatable 1-139S setting descriptions of items 1-354S
updating 1-384S subqueries in 3-33T
updating duplicate rows 11-25T syntax
using CHECK OPTION 11-25T in DELETE 1-159S

Index 49
 in SELECT 1-310S warnings in a stored procedure 14-8T
in UPDATE 1-383S WITH LOG IN keywords, syntax in
testing a subscript 2-44T START DATABASE 1-376S
use WITH MAX keywords
in DELETE 1-160S relationship with COUNT field 1-354S
in UPDATE 1-388S
with a subquery 1-326S WITH NO LOG keywords
with ALL keyword 1-330S syntax
with ANY keyword 1-330S in CREATE TABLE 1-84S
with BETWEEN keyword 1-327S in SELECT 1-341S
with IN keyword 1-327S use
with IS keyword 1-327S in CREATE TABLE 1-104S
with LIKE keyword 1-328S in SELECT 1-343S
with MATCHES keyword 1-328S WITH NO LOG keywords, syntax in
with NOT keyword 2-32T START DATABASE 1-376S
with OR keyword 2-33T WITH RESUME keywords, in RETURN
with relational operator 1-326S 2-39S
with SOME keyword 1-330S WITHOUT HEADINGS keywords, in
wildcard comparisons 2-37T the OUTPUT statement 1-271S
See also Condition segment. WORM drive 12-6T
WHILE keyword Write-once read-many-times device. See
in CONTINUE statement 2-6S WORM drive.
in EXIT 2-14S
Writing a dbload command file
WHILE statement in character-position form 5-31R
looping in a stored procedure 14-23T in delimiter form 5-26R
syntax 2-46S
with NULL expressions 2-46S
Wildcard character X
asterisk 2-12T X/Open
Wildcard characters, with LIKE or and Informix implementation of NLS
MATCHES 1-498S 1-15R
Wildcard comparison setting NLS environment variables
in WHERE clause 2-37T to 2-44T 4-43R
WITH APPEND keywords, in the SET setting the LC_COLLATE category
DEBUG FILE TO statement 1-351S 4-50R
setting the LC_CTYPE category 4-51R
WITH CHECK keywords setting the LC_MONETARY category
syntax in CREATE VIEW 1-136S 4-52R
use in CREATE VIEW 1-138S setting the LC_NUMERIC category
WITH CHECK OPTION keywords 4-53R
11-25T setting the LC_TIME category 4-53R
WITH GRANT keywords specifications, icon for Intro-8S
syntax in GRANT 1-231S X/Open mode
use in GRANT 1-237S FETCH statement 1-195S
WITH HOLD keywords GET DESCRIPTOR 1-213S
declaring a hold cursor 7-19T, 10-34T SET DESCRIPTOR statement 1-357S
syntax in DECLARE 1-145S X/Open specification of NLS 1-15R
use in DECLARE 1-150S, 1-157S
WITH keyword, syntax in CREATE
DATABASE 1-57S
WITH LISTING IN keywords
50 Index
 Y as placeholder in PREPARE 1-273S,
5-29T
YEAR function naming variables in PUT 1-287S
as time function 2-56T replacing with USING keyword
syntax in expression 1-454S 1-268S
use in expression 1-456S wildcard in Condition segment 1-411S
YEAR keyword @, at sign, in database name 1-421S
syntax \, backslash, as escape character
in DATETIME data type 1-428S with LIKE 1-410S
in INTERVAL data type 1-485S with MATCHES 1-411S
use
as DATETIME field qualifier 3-8R, ^, caret, wildcard in Condition segment
1-488S 1-411S
as INTERVAL field qualifier 3-13R, _, underscore, wildcard in Condition
1-491S segment 1-410S
||, concatenation operator 1-432S
Symbols
$INFORMIXDIR/etc/sqlhosts 12-15T
%, percent sign, wildcard in Condition
segment 1-410S
( ), space, as delimiter
in DATETIME 3-9R
in INTERVAL 3-15R
*, asterisk
arithmetic operator 1-431S
use in SELECT 1-312S
+, plus sign, arithmetic operator 1-431S
-, hyphen, as delimiter
in DATETIME 3-9R
in INTERVAL 3-15R
-, minus sign, arithmetic operator 1-431S
. 4-4R
., decimal point, as delimiter
in DATETIME 3-9R
in INTERVAL 3-15R
..., ellipses, wildcard in Condition
segment 1-411S
/etc/hosts 12-15T
/etc/services 12-15T
/, division symbol, arithmetic operator
1-431S
/, slash, arithmetic operator 1-431S
:, colon, as delimiter
in DATETIME 3-9R
in INTERVAL 3-15R
=, equals, relational operator 2-30T,
2-71T
?, question mark

Index 51
52 Index
 Reader Response Card
Informix Guide to SQL: Tutorial, Version 6.0

Dear Reader,
At Informix Software, we think documentation is very important. After all, manuals are
an integral part of our product line. Because documentation is important, we want to know
what you think about our manuals. You can tell us by filling out and returning the Reader
Response Card.
Thanks for your help!
1. Overall, how do you rate this manual?
Outstanding Very good Good Average Poor

2. Is the manual effective?

• Is it organized so you can find things? Yes No
• Is the index adequate? Yes No
• Is the table of contents easy to use? Yes No
• If applicable, do you find the statement syntax readable? Yes No
• Are topics presented in a useful sequence? Yes No
• Are the examples useful? Yes No

3. What did you think of the level of description?

• Writing quality: Very good Good Average Poor
• Clarity: Very clear Average Hard to understand
• Level: Too technical Just right Oversimplified

4. When you need to find information quickly, which part of the documentation
do you use?
• Index: Often Sometimes Rarely Never
• Table of Contents: Often Sometimes Rarely Never
• Quick Reference Card: Often Sometimes Rarely Never
• Chapter summaries: Often Sometimes Rarely Never
• On-line help: Often Sometimes Rarely Never
• Browse through manuals: Often Sometimes Rarely Never

5. How long have you been developing database applications?

• How many years have you been programming? 0 1-3 4-5 >5
• How many years have you been using databases? 0 1-3 4-5 >5
• How many years have you been using Informix products? 0 1-3 4-5 >5

6. Have you used other programming languages? If so, which ones?

7. Have you used other database products? If so, which ones?

January 1993
9. What other Informix products do you use?

10. What is your job title and area of responsibility?

11. Additional comments:

Your Name: Company Name:

Address:
City: State/Province:

Country: Zip/Postal Code:

Telephone Number:

INFORMIX is a registered trademark of Informix Software, Inc.

To mail this card, please fold on the dotted line and staple or tape the open end.

NO POSTAGE
BUSINESS REPLY MAIL NECESSARY
IF MAILED
FIRST CLASS PERMIT NO. 25 MENLO PARK, CA. IN THE
UNITED STATES
POSTAGE WILL BE PAID BY ADDRESSEE

Informix Software, Inc.

Technical Publications Department
4100 Bohannon Drive
Menlo Park, California 94025