Amiga C For Advanced Programmers

,[
PO
(jfT
P()
N 0
66 11V
CK PEN
61 MME
Amiga C
for
Advanced
Programmers
Bleek
Jennrich
Schulz
EIDWINDOUJ e SI,"1MEZERflZERO BLOCI(PEN RRSPOR
8T TOPED6E., FOLLOUlMOUSE. 6RD61MMEDIRTE 1'1
DSETI(E'YMRP() L
DRRUlMDDE
IE OFFGRDGT
11'16 COSET
EUlWINDOUJ 61 .
CI(PEN RRSPOR
8T. TOPED6
IMMEDIRTE N.
DSETI(EVMRP(
EN DRRUJMDDE 4
'E OFF680
'1RL
DRR661N6 CDSETIC
EUlUlINDOUJ
OZERO BLOCI(PEN RRSPOR
~T. TOPED6E
USE
61MMEDIRTE 1'1
DSETKE'IMRP()
1'1 DRRUJMODE 4
lE OFF6RDGET
R661N6 CDSETIC
fUJUJINDOUJ GI
OCI(PEN RRSPOR
FIT TOPED6E
D61MMEDIRTE 1'1,
DSETI(EVMRP() L
TPEN DRRUJMODE
IE OFF6RDGET DISPLR'IRL RT. DRRG6'
A Data Becker Book

Published by
Second Printing, December 1989

Printed in U.S.A.
Copyright 1989
Copyright 1987, 1988
Abacus
5370 52nd Street, SE
Grand Rapids, MI 49512
Data Becker GmbH
Merowingerstrasse 30
4000 Duesseldorf, West Germany
This book is copyrighted. No part of this book may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, electronic, mechanical, photocopying,
recording or otherwise. without the prior written permission of Abacus or Data Becker
GmbH.
Every effort has been made to ensure complete and accurate information concerning the
material presented in this book. However, Abacus can neither guarantee nor be held legally
responsible for any mistakes in printing or faulty instructions contained in this book. The
authors always appreciate receiving notice of any errors or misprints.
Amiga 500, Amiga 1000, Amiga 2000 and Amiga are trademarks or registered trademarks
of Commodore-Amiga. Inc. AC/BASIC Compiler and AC/BASIC are trademarks or
registered trademarks of Absoft Corporation. Cray is a trademark or registered trademark of
Cray Incorporated. AmigaBASIC is a trademark or registered trademark of Microsoft
Corporation.
:r::SBN
ii
1-55755-046-8
Table of Contents
Preface
1.
1.1
1.1.1
1.1.2
1.1.3
1.1.4
1.2
1.2.1
1.2.2
1.2.3
2.
2.1
2.1.1
2.1.2
2.1.3
2.1.4
2.2
2.3
2.3.1
2.3.2
2.4
2.5
2.5.1
2.5.2
2.5.3
3.
3.1
3.1.1
3.1.2
3.1.2.1
3.1.2.2
3.1.2.3
3.1.2.4
3.1.3
3.1.3.1
3.1.3.2
3.1.3.3
3.2
3.2.1
3.2.1.1
3.2.1.2
3.2.1.3
The C language
Two Amiga C compilers

The Aztec compiler
The professional version
The developer's version
The commercial version
The compiler in action
Comparing files
File hex dumps
A monitor
4
4
8
9
13
14
14
20
25
C Compiler Operations
35
How a C compiler works

Internal organization of variables and structures
Software organization of variables and structures
Translating control statements
Function calls
The assembler
The linker
"Large data" and "small data"
"Large code" and "small code"
The debugger
Tips and tricks
Accessing absolute memory locations
Dynamic arrays
Function tables
Intuition and C
Windows
Window parameters and selecting them
How Intuition manages windows
Accessing the Intuition library
The NewWindow structure
The Window structure
A summary of window functions
Example window application programs
All-purpose windows
Program routines for text editing
Window for a new CLI
Screen Fundamentals
Creating custom screens
The NewScreen structure
The first screen listing
The Screen structure
39
40
49
54
59
63
67
70
70
74
77
77
78
79
83
84
84
89
89
90
101
106
110
111
117
120
123
123
124
127
130
iii
3.2.1.4
3.2.2
3.2.2.1
3.2.2.2
3.3
3.3.1
3.3.1.1
3.3.1.2
3.3.1.3
3.3.1.4
3.3.2
3.3.2.1
3.3.2.2
3.3.2.3
3.3.2.4
3.3.3
3.3.3.1
3.3.3.2
3.3.3.3
3.3.3.4
3.3.4
3.3.4.1
3.3.4.2
3.3.4.3
3.4
3.4.1
3.4.1.1
3.4.1.2
3.4.1.3
3.4.2
3.4.2.1
3.4.2.2
3.5
3.5.1
3.5.2
3.5.3
3.5.3.1
3.5.3.2
3.5.3.3
3.5.4
3.6
3.6.1
3.6.2
3.6.3
3.6.4
3.7
3.7.1
3.7.1.1
3.7.1.2
3.7.1.3
3.7.2
iv
The screen functions

Examples of using screens
General examples
Program sections for the text editor
Output
Text output
Position, color, character mode
The IntuiText structure
Fonts and type styles
PrintIText
Drawing lines
Color, position, etc.
The Border structure
Coordinate table for line drawing
DrawBorder function
Graphic output
Size, position, color, etc.
The Image structure
The graphic data in the Image structure
The DrawImage function
Examples of graphic applications
Easy text definition
Borders for gadgets
Symbols say more than words
Gadgets
Different gadgets, different applications
Boolean gadgets
Proportional gadgets
The string gadget
Gadgets in action
Gadget functions
A new sizing gadget
Requesters
Automatic requesters
System requesters
Making our own requester
Requester structure
Establishing a requester structure
The requester functions
Requesters and accessing the user
Alerts
Applications for alerts
Creating and configuring an alert
Alerts for program projects
Understanding the Guru Meditation
Checking the IDCMP
Configuration and receiving
The IntuiMessage structure
MsgPort structures
Waiting for a message
Reading the message
134
138
138
141
143
143
144
144
145
147
152
152
152
153
154
158
158
159
160
160
165
166
170
175
177
177
178
189
195
203
204
227
229
229
233
234
235
236
239
242
251
252
252
255
258
262
263
267
268
269
271
3.7.2.1
3.7.2.2
3.7.2.3
3.7.2.4
3.7.2.5
3.7.2.6
3.7.2.7
3.8
3.8.1
3.8.2
3.8.2.1
3.8.2.2
3.8.2.3
3.8.2.4
3.8.2.5
3.8.2.6
3.8.3
3.8.4
3.8.4.4
3.9
3.9.1
3.9.2
3.9.3
3.9.4
3.9.5
3.9.6
3.10
3.10.1
3.10.2
3.10.3
3.11
3.11.1
3.11.2
3.11.3
3.11.4
4.
4.1
4.2
4.3
4.3.1
4.3.2
4.3.3
4.3.4
4.3.5
4.3.6
4.3.7
4.4.8
4.3.9
4.4
4.4.1
Gadgets
Window messages
Requesters and execution
Menu flag analysis
Mouse reading
Recognizing keyboard input
The disk drive
Menus
General construction of a menu
Making a menu
Filling the menu strip
The Menultem structure
Adding extras to menu design
Submenus
Style menu tutorial
Graphic menus
Reading the menu strip
Working with source code utilities
Fine-tuning the source text
The Console Device
Designing Communication Direction
Receiving the First Data
Displaying the Characters
Console Device Control Sequences
Reading Messages from the Console
Finishing the CLI Editor
User-Defined Keyboard Tables
Keyboard Table Design
Working with the Keyboard Tables
Creating Your Own Keyboard Table
Memory Management
Memory Organization of the Amiga
First Steps with AlIocMemO
Improvement using AllocEntryO
The Final Solution
271
Operating System Programming
381
Planning the Editor

Development Stages
Step by Step
Opening a Window
From RAWKEY to ASCII
Memory Management
Testing the new functions
Deleting Lines
Text Output in a Window
The Cursor
Text Input
Command Line
Editor Comand Set
Editor Keyborad Commands
272
273
274
276
278
281
283
283
284
287
290
298
302
305
305
321
325
335
337
337
339
340
341
343
349
361
361
363
367
369
369
370
372
375
381
389
398
398
406
408
419
422
427
445
474
512
558
558
v
4.4.2
Editor Command Line Commands
Appendix A Editor Listings

Version 1 makefile
Version 1 Editor.h
Version 1 Editor.c
Version 4 TestTexl
Version 6 makefile
Version 6 Editor.h
Version 6 Editor.c
Version 6 Edit.c
Version 6 Cursor.c
Version 6 Output.c
Version 6 Memory.c
Version 6 Command.c
Index
558
561
561
562
564
568
570
571
574
586
606
624
636
644
645
Preface
Now that the initial excitement over the Amiga has died down, people
are discovering that Amiga can also be used for "real" applications as
well as fantastic graphics and sound. Users want more and more
literature about programming the Amiga for these serious applications.
This book shows you how to make use of the many Amiga operating
system functions in your own programs.
This book is not a guide on learning C: It's a book on advanced C programming. If you need a guide for the beginner, please read Abacus'
Amiga C for Beginners before continuing with this book.
Since this book shows you how to program using the C language,
we'll begin with author B. Jennrich demonstrating the actual operation
of a C compiler. Then we'll get to the actual "instructional" part of the
book. You'll learn what makes type casting and other aspects of the
language necessary. We hope this saves you time and aggravation in
developing larger projects. Maybe you tried C before and spent more
time creating errors and system crashes than working programs. This
book will hopefully give you a better understanding of the language and
what you can accomplish in this language.
The next segment of the book describes the subject of Intuition in
detail. W. Bleek, a specialist in this area, leads you into the depths of
the Intuition user interface. Mter reading this section you'll be able to
easily add screens, windows, gadgets, menus and more to your C
programs.
The third section shows you how to develop an Intuition-based text
editor. Peter Schulz, known for his AssemPro assembler software and
the book Amiga 3D Graphic Programming applies your newlyacquired knowledge of Intuition to a practical project You'll learn how
to wode with devices and other libraries.
All in all, this book is one which the serious C programmer will
always want to have within reach of his Amiga.
Bleek, Jennrich, Schulz
February 1988
vii
1.
The C
language
1.
ABACUS
1.
THE
LANGUAGE
The C language
In 1972 Dennis Ritche began work on improving the B language, an
improved version of the typeless BCPL language (Basic Combined
Programming Language.) The result was the C language.
The C language was developed to ease the implementation of the UNIX
operating system on PDP series computers. Dennis Ritche probably
never dreamed that his language would become so popular among personal computer users.
C's popularity stems from its nature-an extremely efficient high-level
language. This efficiency is especially important in home computing.
Home computers contain limited amounts of memory, may require
large amounts of time for program execution and have low clock
frequencies, therefore a compiler should make a program as small and as
fast as possible.
Code created by a C compiler executes quickly and complies into a
compact form. These two factors are very important to the programmer
who wants to write fast-running programs without the hassle of
learning machine language.
In the early days any C source code could be run on any C compiler.
This was another factor favoring C-portability. As time went on,
however, different operating systems required different compilers (and
different operating system interfaces). This made complete compatibility almost impossible.
The 68000-based Amiga, Atart ST and MAC contain many crucial
system differences. For example, any Atari ST system routines (GEM,
VDI and AES) must be replaced with the equivalent Amiga routines if
you want an ST source code to run on the Amiga. This requires a
detailed knowledge of both operating systems.
This incompatibility problem becomes worse when you transfer a C
source code from an IBM PC or compatible to the Amiga. IBM
systems use segmented addressing, requiring the use of FAR declarations of variables and functions. In addition, MS-DOS access occurs
almost exclusively through interrupts. This means that the C
programmer needs detailed knowledge of both operating systems when
converting MS-DOS programs to the Amiga.
1.
THE
C LANGUAGE
1.1
AMIGA C FOR ADVANCED PROGRAMMERS
Two Amiga C compilers

Let's leave the compatibility problem for now and turn to the Amiga.
You have a choice of basically two complete C compilers for the
Amiga-Lattice C and Aztec (Manx) C. Dyed-in-the-wool Lattice
programmers swear to the reliability of the Lattice compiler (some parts
of the Amiga operating system were written with the Lattice compiler).
Meanwhile, Aztec programmers love the speed of their compiler.
The compilers The earlier versions of the Lattice compiler were considerably better
than the early versions of Aztec C. The Lattice compiler could handle
arrays which occupied more than 65,535 bytes, while Aztec could noL
In addition, problems arose when trying to compile programs written
for Lattice on the Aztec compiler. On the other hand, Aztec programs
always compiled fine with Lattice. Thus programmers had to stick with
the Lattice compiler if they wanted to be able to use their old programs.
The new Aztec version (Version 3.4a) has no problems with Lattice
compatibility, at least that we could find. We tested a number of programs developed with Lattice and they all compiled successfully with
Aztec (with the help of a compiler option).
Thus we decided to use the new Aztec compiler exclusively in this
book. The programs will require only minor, if any, modifications to
compile with the Lattice C compiler.
1.1.1
The Aztec compiler

Now we come to the Aztec compiler package. The developer's version
of the package contains four disks. In addition to the compiler, assembler and linker, these disks contain all of the include files and various
libraries which can be linked to compiled source codes and assembler
programs (see Section 2.3). These disks also feature several useful
utility programs which, for example, allow you to turn your own
routines into libraries.
Let's look at what you need for using the compiler. Later versions of
the compiler may be different than the descriptions presented here,
which are for Version 3.6A. Compilers are continually being updated
and improved upon. On disk "SYS1:" of the package you will find three
subdirectories in addition to a complete Workbench disk. These are
accessed using the CLI:
1.1 THE AZTEC COMPILER
ABACUS
bini
include/
lib/
The SYS1:bin! directory contains all the programs necessary for using
the compiler. These include the compiler itself (cc) which creates
assembler source. The assembler (as) which assembles this source and
the linker (In) which links the object code of the assembled source
with the libraries found in the SYS1:lib directory.
The SYS 1 :include directory contains the individual header or include
files which must be included in each C program. These include and
header files access necessary parts of the operating system.
The second disk, "SYS2:" contains assembler include files and
additional libraries which you can link with your programs, as
necessary (see Section 2.3 for more information). This disk also
contains some example programs.
You can start programming in C with only these two disks-the
standard version of the Aztec compiler. But the developer's version of
the compiler, which adds a few utility programs, makes compilation
much easier.
The third disk "SYS3:" contains various utility programs (e.g., make,
and more program examples).
Disk "Library Source:", contains the source code of the assembler
and C calls for the individual operating system routines in compressed
(ARCed) format. These assembler and C files are assembled, compiled
and combined into c.lib and m.lib. which can and/or must be linked
to your programs.
First we'll look at the standard (and most cost-effective) version of the
Aztec compiler. This contains just the disks "SYS1:" and "SYS2:".
1.
THE
C LANGUAGE
AMIGA
Dlsk IsysJ
C FOR ADVANCED PROGRAMMERS
pi sk
.1
asml
binI
as
cc
In
mclk
set
setdat
(Assembler)
(Compiler)
(Linker)
(Manx Clock)
(Tool for system
variables)
(Date and time)
(CLI ColT'lllands)
(Devices)
(Fonts)
(C Include-Files)
(Hardware-Handler)
cf
devsl
fontsl
includel
11
sys2-
(Assembler Include Files)
binI
Ib
examples
libl
cl.lib
rnal.lib
c32.lib
bits)
ma32.lib
cl32.lib
ma132.lib
IcrtO.o
(Library hangler)
(examples)
ml.lib (Large data)
sl.lib
m32.lib (int=32
s32.lib
m132.lib(Large data)
s132.libint=32 bits)
libl
c.lib
rna. lib
m.lib
libsl
(Arniga Librarys)
(Ed's Backup)
tl
sl
Figure 1.1
Startup sequence
.dbint
(for Debugger db)
crt scrl
crtO.a68
clipaesr.c
vars.c
wbparse.c
exist.c
rnain.c
Iff
(IFF file-demos
lintl
rnanx.c (function definitions)
systeml
Disk 1 is enough to compile short programs. You may have to write a

startup sequence to inform the compiler where to fmd the correct fIles:
;l.l.A.startup-one
stack 10240
;stack should be at least 10K

ior more
copy dfO:lib/c.lib to ram:

i most common library
;in RAM disk
bin/set
INCLUDE=dfO:include
bin/set
CLIB=ram:!dfO:lib
bin/set
CCTEMP=ram:
;where are the include files?

;where are the libraries?
bin/setdat
run bin/mclk
;where to put the temporary files?

;set time and date
; (important for developer's!)
;start time and memory display
stack overflow This startup-sequence first increases the stack to 10240 bytes. This is a
preventative measure which should protect you from a "stack overflow"
when developing programs (you will learn another way of preventing
stack overflow later).
linker library
Then the startup-sequence copies the linker library (which must be

linked to every C program) to the RAM disk. This speeds up the access
to the linker library during linking. This library isn't the same as the
graphics.library, for example, which contains the addresses of each
graphic function in the operating system. The libraries which the linker
binds to the C program consist of small machine language segments
which call the operating system routines when the system library is
used.
1.1 THE A ZTEC COMPILER
ABACUS
But back to the startup-sequence. Next the environment variables of the

compiler are set. bini set lNCLUDE=df 0:incl ude tells the compiler to look in the dfO: include directory for include fIles specified
in your C program.
CLIB variable The CLlB variable performs a similar function. This variable specifies
which directory will be searched to find the libraries for linking.

bin/set CLlB'" ram:!dfO:lib tells the linker to look for the
libraries on the RAM disk first. If it cannot find the fIles there, the
dfO:lib directory is searched next. If multiple directories are to be
searched, the individual directories are separated by"!".
The environment variable CCTEMP specifies where the temporary
assembly language files are stored. The compiler creates temporary
assembly language source files in the specified directory which are later
combined into a complete source fIle and assembled by the assembler.
Next the startup-sequence asks for the current date and time. Those with
battery-backed realtime clocks should set the time and date in the startup
sequence in the manner appropriate for your clock.
Those who don't own a realtime clock should always enter the correct
time and date. You only need to do this when you tum on the computer
or first start the compiler. After a reset you can bypass the time and date
prompts by pressing <Return>, because the time and date remain
somewhat intact on a reset. This only puts the clock behind by a few
seconds.
Finally, the mclk program is started. This program displays the current
time, the free chip and fast memory in a small window in the CLI
window's menu line. This is used to check whether the program you are
developing actually releases all the memory it allocates. You can get
more information about allocated and free memory from the utility
program avail (sys3:bin).
After initialization it's possible to compile small programs with just
one disk drive. But you will need two disk drives to compile larger programs. Two disk drives allow you to copy the include files and the
linker libraries onto a second disk, for example. You then have to adapt
the startup sequence accordingly:
;l.l.B.startup-two
stack 10240
copy df1:lib/c.lib to ram:
; stack should be at least 10K

lor more
;most common library
lin RAM disk
bin/set INCLUDE=df1:include
bin/set CLIB=ram:!df1:lib
bin/set CCTEMP=ram:
bin/setdat
run bin/mclk
;where are the include files?

;where are the libraries?
;where to put the temporary files?
;set time and date
; (important for developer's version! )
;start time and memory display
1.
THE
LANGUAGE
AMIGA
Now you have enough room on the disk in drive dfO:, disk I, to write
large programs (approx. 30004000 lines).
Another possibility would be to leave the startup sequence intact and
store the programs on the disk in dfl:. This depends on the programmer's preferences. Amiga owners using a hard disk will have to alter
their startup sequence accordingly.
1.1.2
The professional version

The professional version of the compiler allows you to create a script
file that can be executed with execute. This allows compiling,
assembling and linking with one line of entry
A very simple script file, which could be named comp, looks like this:
.key file
cc <file>
In <file>.o -IRAM:c
The source file is fIrst passed to the compiler, which automatically calls
the assembler. The assembler then assembles the assembly language
source created by the compiler. The object file created by the assembler
is then linked with c.lib to form an executable program.
This script file assumes that the compiler (co), the assembler (as) and
the linker (In) are contained in the CLI command directory C:
(commands), and can be treated like normal CLI commands. To do this,
copy these fIles from the SYS1:bin directory to SYS1:c. The path
command may also be used to set the search path, path SYS1 :BIN.
The program to be compiled must be in the current directory.
Assuming the name of the script fIle is comp, then a C program can be
compiled with execute oomp cJ>rog. You can interrupt the
compiler and linker by pressing <Ctrl><C>.1f you call the assembler
separately, you can also interrupt script file processing when assembling.
This is because the FAILAT level defaults to 10 after loading the
Workbench. The compiler and linker interrupt the compilation or link
process after <Ctrl><C> with exi t (10). This stops the script file
processing because it determined through the exit () command that a
serious "error" occurred. But since the compiler normally calls the
assembler, the command exi t(l 0) goes to the compiler when <Ctrl>
<C> is pressed, so the script file cannot be interrupted when
assembling.
1.1
ABACUS
THE AZTEC
COMPILER
Another method for interrupting a script file is pressing <Ctrl><D>.

This key combination interrupts the script file without stopping any of
the programs started from the script file. If the compiler starts and you
press <Ctrl><D>, you must wait until the compiler has finished the
processing.
1.1.3
The developer's version

Script files are impractical for developing larger programming projects,
which normally consist of several modules. For example, if you write a
script file which compiles and links all the modules of a program, you
have to wait until all the modules have been compiled and linked, even
if you made a change in only one of the modules.
You can compile the modified module by hand through a direct call to
the compiler, then link it to the other modules "by hand." But this gets
rather annoying. This is why the developer's version of the Aztec C
compiler contains the ma ke tool in addition to the sample programs
and libraries.
Disk sys3;
libl
m8.lib
m832.lib
mx32.lib
m81.lib
mx1.lib
m8132.1ib mx132.1ib
(FFP functions for 6881 and

FFP for lEE/Manx
(lnt=32 bit)
(Large Data)
(Large Data and lnt=32 bit)
binI
(tools)
mx.lib
make
examples/
db
(C examples)
Figure 1.2
A makefile lets you easily compile large programs consisting of
multiple modules. You do not have to manually compile a modified
module and then link it with the old modules. make handles all this.
You should copy the make utility into the C: subdirectory of the
"SYS1:" disk. Be sure you are working with copies and not the original
disks!
Lees look at the following four modules and create a makefile to
compile them. The makefile must be in the same directory as the
modules to be compiled and it must be named makef ile.:
1.
THE
AMIGA
LANGUAGE
/***********************************/
/* modl.c
*/
/***********************************/
funcl()
{
printf ("Function 1 calls function 3!\n");

func3 ();
)
1***********************************/
1* mod2.c
*/
1***********************************/
func2 ()
{
printf ("Function 2 is called by function 3!\n");

)
1***********************************/
1* mod3.c
*/
/***********************************/
func3()
printf ("Function 3 calls function 2!'\n");

func20;
/***********************************/
1* main.c
*/
1***********************************/
main 0
{
printf ("MAIN calls function l!\n");

funcl() ;
}
The makefile shows how the individual files depend on eaCh other.
For example, the frrst object file depends on the C source file mod1.c.
This is expressed in the make file in the line mod1.o: mod1.c, which
says that the commands in the line after mod1.o: mod1.c are executed
if modl.c is older than mod1.o or if mod1.o does not exist.
In order to determine which file is older than another, the time and date
must be set correctly. The startup sequence calls setdat. Since the
date of the last modification of a file is always stored in the file info
block, make can always determine how old a flle is.
Back to the makefile. The first line of your makefile looks like
this:
10
ABACUS
mod1. 0: modl. c
The next line shows what should be done if modl.o is older than
mod1.e:
cc modI
If modl.o is older than modl.e (i.e., if a change was made to modl.e

since the last time it was compiled), then make creates the new object
file mod1.o (mod1.o is updated). The command to be executed according to a file-to-file relationship (mod1.o: mod1.e) must be indented by
at least one character. The makef ile which creates the corresponding
object files from the four C sources looks like this:
modI. 0 : modI. c
cc modI
mod2.0: mod2.c
cc mod2
mod3.0: mod3.c
cc mod3
main.o: main.c
cc main
Now to create the object file for main.e, you only need to call make
main.o. But you can't use the individual object files if they are not
linked. Therefore you must add the following lines to the make file:
"func: modLo mod2.0 mod3.o main.o
In -0 func modl.o mod2.0 mod3.o main.o -IRAM:c n
The func file These two lines indicate that the fune file, which represents the executable program relies on the four object files. If one of these files is
more recent than func, the func file must be relinked. The make
program checks each object file, and if it determines that a modification
has been made to one of them, it creates the new object file from the
modified C source. This keeps the fune program up to date after a
make.
Here is the complete makefile which creates the executable function
from the modules listed before with make fune:
modI. 0 : modl. c
cc modI
mod2.0: mod2.c
cc mod2
mod3.0: mod3.c
cc mod3
main.o: main.c
cc main
func: modl.o mod2.o mod3.0 main.o
In -0 func mod1.o mod2.0 mod3.0 main.o -IRAM:c
These are make's simplest functions. If you modify the makefile so

that the linker call (the last two lines) are the first two lines, you can
11
1.
THE
LANGUAGE
AMIGA
perfonn the make with just make instead of make func. When the
call is simply make, the fIrst command of the makefile executes.
If you want to execute multiple commands after a flle-to-flle relationship, you must ensure that each command is on a separate line and is
indented by at least one character. In addition, you must pay attention to
the correct calling sequence of commands, whereby you can use any
CLI command (e.g., cd, delete, etc.):
func: modl.o mod2.o mod3.o main.o
ln -0 func modl.o mod2.o mod3.o main.o -lRAM:c
delete modl.o
delete mod2.o
Another option of make is the macro defInition. In your makefile

the string mod1.o mod2.0 mod3.0 main.o occurs twice. If you
define this at the start of the makefile you only have to write the
string once. This is particularly useful for programs with many
modules:
OBJECT = modl.o mod2.o mod3.o main.o
func: $ (OBJECT)
In -0 func $(OBJECT) -lRAM:c
You don't have to enter the same string repeatedly. If the string doesn't
fIt on one line, you can "extend" the line with the backslash (\):
OBJECT = modl.o mod2.0 mod3.o mod4.0 modS.o mod6.0\
mod7.o modS.o mod9.0 modlO.o main.o
Now you can start writing your own C programs using the make
utility.
12
ABACUS
1.1.4
The commercial version

For the sake of completeness we will briefly describe the commercial
version of the Aztec compiler. The only difference between this and the
developer's version is that a fourth disk contains the assembly language
source for all of the functions. Some support functions for operating
devices are written as C procedures.
Most of the machine language calls appear in a form similar to the
following:
movem.l 4(sp), dO
move.l #[BasePointer],a6
jsr _LVOFunction(a6)
rts
;get parameter from stack

;return address!
;library base in register
;call function
There are also routines which perform more complex computations and
manipulations with arguments.
A little tip for those of you who are thinking of purchasing the Aztec
compiler: Consider exactly the extent to which you want to program
the operating system, and whether it is necessary to buy the commercial
version for the function sources. Normally the developer's version will
do everything you want, and the large price differential is something to
seriously consider.
13
1.
THE
1.2
LANGUAGE
The compiler in action

Before we describe the operation of the compiler, we should use it a
little and comment on its features.
1.2.1
Comparing files
As an introduction to Aztec programming, we will use a program
which compares two files with each other and displays the differences as
well as the positions at which they occur. To do this, the program must
be told which fIles to compare. This is done through the command line
of the CLI. The arguments of the command line pass to the program as
the arguments of the main () function:
main (arge, argv)
int
iirge;
ehar
**argv;
{
...
argc (argument count) contains the number of arguments. It should be

noted that the program name is interpreted as the first argument. If
argc equals 2, then actually only one argument was specified.
You can access the arguments (strings) with the pointer argv. Let's
take a look at this pointer:
char **argv;
This structure represents a pointer to a pointer of type cha r. A pointer

to type char is nothing more than a pointer to a string. Because of
this close relationship between pointers and arrays, you can access the
individual parameters as argv [1], argv [2] , etc. argv [1] is the
address of the first argument string. argv [0] contains the address of
the program name string. (You can replace char **argv; with char
*argv [];, which verifies array access to the argument strings).
Now let's test to see if the correct number of arguments have been
passed. In the example you need two arguments (the names of the two
files to be compared). a rgc must have the value 3 (remember the
program name!). If it doesn't, a message goes to the user telling him
which arguments must be passed.
14
1.2 THE COMPILER IN ACTION
ABACUS
i f (argc != 3)
{
printf ("USAGE: compare FILE1 FILE2 \n");

exit (10);
It would be a better programming style to react to a "1" as an argument,

printing out some instructions and entering the required parameters
direct from the keyboard. Many CLI commands support this option, but
it's really not worth the effort to add this to our little example program.
The argument strings are available to you in argv [1] and argv [2 ]
(argv = argument values). You can use these strings as arguments to
the C library function fopenO.
fopenO
fopenO requires a second argument in addition to the filename to be

opened. This argument specifies how the file is accessed. The function
fopen (name, "r'') opens an existing file for reading, while the
function fopen (name, "w") creates a new file which can only be
written.
In the example you only need be concerned with existing files:
File1 = fopen (argv[l],"r");
File2 = fopen (argv[2],"r");
Now let's try to open the two files. But since you can't be sure that the
user hasn't made a mistake in specifying the files, or that the files even
exist, you should test to see if the files actually exist:
i f (File1 == OL)
{
printf ("%s cannot be opened!! !\n",argv[l]);

CloseIt (10);
if (File2 == OL)
{
printf ("%s cannot be opened !!!\n",argv[2]);

Close It (10);
You may wonder why the letter L follows the 0 in the condition
Filel == OL. This tells the compiler that it should compare the
pointer Filel with a long value. You could also replace this OL with
(long) O.
There is another way to ensure that constants are treated as "long
variables": the +L compiler option. Here you should note that both
constants and int variables convert to long. When using the +L
option you should link the c32.lib to the program, which ensures
that the argument counter (argc) of main(argc, argv) is actually a
long variable.
15
1.
THE
LANGUAGE
AMIGA
FOR ADVANCED PROGRAMMERS
If you access this counter using c.lib and the +L option (e.g., if
(argc == 3)), then four bytes specify the value of this variable instead
of just two. However, this completely invalidates the result because the
two bytes of the previous int variable now become the two high bytes
of the current long variable (+ L option). In addition, the pointers to the
strings shift by two bytes.
But back to the real problem: If the file descriptor returned by f openO
equals zero, then an error occurred in opening the file. The CloseItO
routine closes the open files, sends a message to the user and exits the
program. A tip: Programs which allocate memory or perform other
operations which must be undone should have a routine which automatically releases everything previously allocated
If the files were opened properly, you can now compare them. Now
comes the main loop of the program, which compares the two files
byte for byte for similarities or differences.
The next byte is read from each of the files, and these two bytes are
then compared. If the bytes are identical, then the program reads the
next two bytes. If the bytes are different, the two bytes and the position
in the file at which they occurred are displayed.
It is best to go through the two files inside a while loop, which
terminates when one of the two files has been completely read. The
feofO function determines the end of the file, and the next byte of the
file can be read with fgetc.
filepos = 0;
while (!feof(File1) && !feof(File2
{
By tel = fgetc(File1);
Byte2 = fgetc(File2);
if (By tel != Byte2)
printf ("$%08x %02x <> %02x\n", filepos,Bytel,Byte2);
filepos++;
In this loop, the two file bytes (Bytel and Byte2) are compared with
each other and if they differ, a message like the following appears on
the screen:
$00000154
$12 <> $3a
In addition to the two different bytes, their location in the file also
appears. This necessitates a counter variable (filepos) which contains the number of bytes read, or the read position within the file.
After the comparison, this counter variable increments by one. If both
files still contain data, the comparison continues. But if one of the files
is empty, the loop ends.
16
ABACUS
The file attribute which indicates how many bytes the individual files
contain comes from the following program segments:
if (feof(File1) && !feof(File2
1* file1 is empty *1
1* but File2 is not. *1
printf ("Comparison terminated !!! \n") ;

printf .("\"%s\" out of data at $%08Ix\n",argv[1],filepos-1);
while (!feof(File2
{
fgetc (File2);
filepos++;
1* read to end of file *1

printf ("\"%s\" out of data at $%08Ix\n",argv[2],filepos-1);
}
else
if (!feof(File1) && feof(File2
1* File2 is empty *1
1* but File1 is not. *1
printf ("Comparison terminated !!! \n");

printf ("\"%s\" out of data at $%08lx\n",argv[2],filepos-
1);
while (!feof(File1
{
fgetc (File1);
filepos++;
1* read to end of file *1

printf ("\"%s\" out of data at $%08lx\n",argv[1],filepos-1);
}
The two if instructions test which file is empty. The name of the
empty file and its length are then displayed. The other file is read to the
end and for each byte read the filepos counter increments. After the
end of the file has been reached, this filename and the length of the file
are displayed:
Comparison terminated I!!
File1 out of data at $00000170
File2 out of data at $00000200
A few words about the filepos counter. In the ftrst while loop, in
which the comparison between the two file bytes is performed,
filepos always increments by one when one of the two mes still has
bytes available. But if it detects the end of a me with feofO, then the
fgetc inside the loop always reads one more byte than is actually
present. This is why when comparing files of different lengths, one
extra byte is "compared" and printed. You must ensure that this "extra
byte" is subtracted from the size of filepos printed.
If both files are the same length, this is determined by the following if
statement.
17
1.
THE C LANGUAGE
else
if (feof(File1) && feof(File2
printf ("\"%s\" and \"%s\" have the size: %08lx\n",
argv[1],argv[2],filepos-1);
The following output then results:

FILE1 and FILE2 have the same size: $00000123
Now you have to close the open files and exit the program. The
CloseItO routine performs this task:
CloseIt (Error Code)
int
Error-Code;
{
if (File1 != 0) fclose (File1);
exi t (Error_Code);
This routine closes the open files (file descriptor != 0) then terminates
the program with exi
The Error Code which exi passes to
the CLI, and can be used to terminate sCript files.
to.
to
Here is the complete program, which also defines all of the variables
used:
1*******************************************************1
1* 1.2.1.A. compare
*1
1*
COMPARE
*I
1*
(c) Bruno Jennrich
*1
1*
*1
1*
*1
1* This program compares two files with each other.
*1
1*******************************************************1
iinclude "stdio.h"
1* File descriptors *1
FILE *File1,
*File2;
1* current file position *1
long filepos;
unsigned char Byte1,
Byte2;
1* byte read *1
main (argc, argv)

int argc;
char
**argv;
{
i f (argc != 3)
{
printf ("USAGE: compare FILE1 FILE2 \n");

exit (10);
File1 = fopen (argv[1],"r n );
File2 = fopen (argv[2],"r");
i f (File1 =
18
OL)
1* open files *1
1* file open () error *1
1.2
ABACUS
THE COMPILER IN ACTION
printf ("%s cannot be opened !!!\n",argv[l]);

Closelt (10);
i f (File2 =
OL)
/* file open() error */
printf ("%s cannot be opened!! !\n",argv[2]);

Closelt (10);
filepos = 0;
/* current file position = 0 */
while (!feof(Filel) && !feof(File2
{
By tel = fgetc (Filel);

/* read bytes */
Byte2 = fgetc (File2);
if (Bytel != Byte2)
/* output difference */
printf ("$%OBlx $%02x <> $%02x\n",
filepos,Bytel,Byte2);
filepos++;
/* increment file position */
if (feof(Filel) && !feof(File2
/* filel is empty */
/* but File2 is not. */

printf ("\"%s\" out of data at $%OBlx\n",argv[l],filepos-l);
while (!feof(File2
{
fgetc (File2);
filepos++;
/* read to end of file */

printf ("\"%s\" out of data at $%08lx\n",argv[2),filepos-l);
}
else
if (!feof(Filel) && feof(File2
/* File2 is empty */
1* but Filel is not. *1

printf ("\"%s\" out of data at $%08Ix\n",argv[2],filepos-l);
while (!feof(Filel
{
fgetc (Filel);
filepos++;
1* read to end of file */

printf ("\"%s\" out of data at $%08lx\n",argv[1],filepos1);
}
else
if (feof(Filel) && feof(File2
printf ("\"%s\" and \"%s\" have the size: %OBlx\n",
argv[1],argv[2),filepos-l);
CloseIt (0);
Closelt (Error Code)
int
Error=Code;
{
19
1. THE C LANGUAGE
AMIGA
if (Filel != 0) fclose (Filel);

exit (Error_Code);
Since this program consists of just one module, it doesn't make much
sense to create a makefile for it as well. A script file works just as
well:
.key file
cc <file>
In +Cb <file> -Lram:c
This script file can (and should) be used for all of the programs in
Chapters I and 2, if the libraries are in the RAM disk.
The programs created by this script file are usually loaded into the
computer's chip memory (+Cb). Even if you have more than SI2K of
memory, the program usually goes into the lower SI2K if that memory
is available.
If you write a program which doesn't fit in the remaining chip memory
(about 320K on the Amiga 1000), you can link your program without
the +Cb option. However, you must ensure that the structures which
lie below the SI2K boundary are stored there (such as memory
assignments for structures with *pointer_to_struct =
AllocMem(sizeof (struct ),MEMF CHIP.
1.2.2
File hex dumps

The program above can be easily rewritten so that it outputs a hex
dump of a file. A hex dump is program output which appears in the
following form:
$00000000
$00000009
$61 $62 $63 $64 $65 $65 $66 $67 $68

$69
abcdefghl
j
The display is the current read position (first column), hexadecimal

notation of the characters read (second column), and the characters read
in ASCII form.
The program must first be modified so that only one file opens. Here
again you must test for the correct number of parameters:
main (argc, argyl
int
argc;
char
*argv[];
{
i f (argc != 2)
{
20
t.Z
ABACUS
THE COMPILER IN ACTION
printf ("USAGE: dump FILE \n");

exit (10);
}
_File = fopen (argv[l], "r");

i f eFile == 0)
{
printf ("\"%s\" cannot be opened!!! \n",

argv[l]);
Closelt (10);
}
This should be familiar from the last program. What is new is the addition of a conversion table:
for (i=0;i<32 ;i++)
;i<128;i++)
for (
for (
;i<160;i++)
for (
;i<256;i++)
Conversion_Table[i]
conversion_Table!i]
Conversion_Table[i]
conversion_Table!i]
. ,.
(char)i;
(char)i;
This conversion table displays the bytes read as characters. You can't
display all of the ASCII codes. For example, ASCII code "12" clears
the screen. To prevent this sort of thing, all of the printable characters
are stored in a table. Unprintable characters (ASCII codes 0-31 and 128159) are represented by periods "." (see your AmigaBASIC handbook,
Appendix A for a list of ASCII characters).
We used four for loops to construct this table. Note that the
initialization of the loop variable i is missing from the last three for
loops. This is unnecessary. After each for loop the loop variable i
has the same value as the variable used in starting the next for loop.
Now the individual bytes can be read, after setting the file position to 0:
filepos = 0;
while ok = BRead(
!= 0)
printf (1I$%08lx
filepos += ok;
.. , filepos);
for (i=O;i<ok;i++) printf ("$%02x ",Byte!i]);

printf ("
");
if (ok < WIDTH) for \
;~<WIDTH;i++)
printf ("
");
for (i=O;l<ok;i++)
printf ("%C" ,Conversion_Table [Byte til ]);
printf ("\n");
21
1.
THE
AMIGA
LANGUAGE
The condition in the while loop ensures that bytes are read from the
file. To do so, the routine BRead is called, which causes either WIDTH
= 12 bytes or the number of bytes remaining in the me to be read (Plus
one, because feofO doesn't become true until one character has been
read past the end).
The variable z contains the number of variables read:
int BRead()
{
int i;
int z;
for (i=O;i<WIDTH;i++)
if (lfeof(_File
{
Byte[ij
z++;
fgetc(_File);
return (z) ;
WIDTH is a constant defined at the beginning of the program. It speci-
fies the number of bytes to be displayed per line. The sample output at
the start of this section was created using a WIDTH of 9. When printing
the hex dump to an 80-column CLI window, WIDTH can be set to 12.
This utilizes the entire width of the window.
When the hex dump is displayed, the file position is printed first
(printf ($%08lx ,filepos);). Next the file position increments
by the number of bytes read (filepos += ok;). The result of
BReadO is the number of bytes actually read (while ok = BReadO)
!= 0).
Mter the file position (and some spaces) has been printed, each byte
must be printed in hexadecimal notation. This is done in a f or loop
(for (i=O; i<ok; i++)printf("$%02x ",Byte[iJ);). Here the
hexadecimal numbers occupy two characters. A leading zero is placed in
front of the numbers "0" through "f'.
If the number of bytes actually read is less than WIDTH, spaces are
printed for the missing bytes (if (ok < WIDTH) for (i=ok;
i <WIDTH; i ++) prin t f(" ");). It can only occur at the end of the
file that fewer than WIDTH characters could be read, but it still looks
unprofessional if the hexadecimal and ASCII characters are not kept in
their respective columns, such as:
$00000000 $12 $23 $10 $07 $06 $07 $c2 $82 $85
$00000009 $12 $23 $10 $07 $06
Therefore the for loop prints as many spaces as would be used by a

hexadecimal number.
22
ABACUS
Now you just have to print the bytes as ASCn characters (for (i=O;
i<ok; i++) printf("%c",Conversion Table [Byte[i]]);).
The previously constructed Conversion Table is indexed by the

previously read bytes. After the ASCn characters have been printed, a
carriage return is printed (printf("\n");) so that subsequent output
appears on the next line.
The file must be closed when the while loop exits (all bytes of the
file have been read). The modified version of CloseltO used in this
program looks like this:
CloseIt (Error_Code)
int
Error_Code;
if _File == 0) fclose (_File);
exit (Error_Code);
Here is the entire program, again with the definitions of all variables
and arrays:
/*******************************************************/
/*
*/
/* 1.2.2.A. dump.c
*/
/*
DUMP
*/
/*
(c) Bruno Jennrich
*/
/*
*/
/*
*/
/* This program creates a hex dump of a file.
*/
/*******************************************************/
#include "stdio.h"
#define WIDTH 12L
1* 12 numbers per line */

1* aO-column screen */
FILE *_File;
/* file descriptor
long filepos;
/* current file position
unsigned char Byte [WIDTH);
/* buffer for bytes read
unsigned char Conversion_Table[256);
/* ASCII - printable conversion
int ok;
1* number of bytes read
*1
*/
*1
*1
*1
int BRead()
{
int i;
int z = 0;
for (i=O;i<WIDTH;i++)
if (!feof(_File
{
Byte[i)
z++;
= fgetc(_File);
return (z);
23
1.
THE
AMIGA
LANGUAGE
main (argc, argv)

int
argc;
char
*argv[];
{
/* loop variable *1
int i;
1* too few parameters *1
i f (argc != 2)
{
printf ("USAGE: dump FILE \n");

exit (10);
File = fopen (argv[l],"r");
i f (]ile =
/* open file *1
/* file open error *1
0)
printf ("\"%s\" cannot be opened!! !\n",argv[1]);

Closelt (10);
for
for
for
for
(i=0;i<32 ;i++)
(
;i<128;i++)
(
;i<160;i++)
(
;i<256;i++)
filepos=O;
while ok
Conversion_Table[i]
'.';
Conversion_Table[i]
(char)i;
Conversion_Table[i]
'.';
Conversion_Table[i]
(char)i;
1* Conversion_Table aufbauen *1
1* current file position equals 0 *1

BRead (
!=
0)
printf ("$%08lx
filepos+=ok;
",filepos);
1* print current file position *1
1* increment current file position *1
for (i=O;i<ok;i++) printf ("$%02x ",Byte[i]);

/* hexadecimal numbers */
i f (ok < WIDTH) for (i=ok;i<WIDTH;i++) printf ("
");
/* "blank numbers" *1
printf (" 0);
/* column divider */
for (i=O;i<ok;i++) printf

("%c",Conversion_Table[Byte[i]]);
1* "ASCII" representation *1
printf ("\n");
1* carriage return *1
24
ABACUS
CloseIt (0);
/* bye */
CloseIt (Error_Code)
int
Error_Code;
if (_File != 0) fclose (_File);
/* if file open */
/* close it */
exit (Error_Code);
1.2.3
A monitor
With a few changes, the hexadecimal dump program can also be used as
a machine language monitor. A monitor lets you view the contents of
the Amiga's memory. The monitor displays the various memory
contents.
To do this you must tell the program the starting and ending addresses
of the area of memory you want to see. This is done again with command arguments. Two arguments are passed:
main (argc, argyl
int
argc;
char
**argv;
{
i f (argc != 3)
{
printf ("USAGE
exit (10);
mon START END \n");
After you have determined that the correct number of arguments have
been passed, you can start converting them. The starting and ending
addresses are specified as ASCII strings, and must be converted to
numbers. We decided to convert the strings to hexadecimal numbers.
Almost every monitor on the market works with hexadecimal numbers.
so your monitor will too.
The Aztec compiler offers a routine which converts an ASCII string
into a hex number. The following is our conversion routine:
unsigned char *atoh (String)
char
*String;
{
int i;
unsigned char *hex;
25
1.
THE
LANGUAGE
unsigned long factor;

hex = (unsigned char *)OL;
factor = lL;
if (strlen (String) <= 8)

for (i=(strlen(String)-l);i>=O;i--)
{
if string[i] >= '0') && (String[i] <= '9'

{
hex += (string[i] -'O')*factor;

factor *= Oxl0;
if (String[i] & MASK) >= 'A') &&
string[i] & MASK) <= 'F'
hex += String[i] & MASK) - 'A'+Oxa)*factor;
factor *= Oxl0;
return (hex);
This routine ftrst sets the return value hex to 0 and factor to 1. The
following steps only occur when the string length doesn't exceed eight
bytes. This is because the address range of four gigabytes, which the
Amiga can theoretically address, is represented by the addresses from 0
to FFFFFFFF. The highest address is therefore FFFFFFFF. As a
string, this number is eight characters long.
If the string passed to the routine has eight or fewer characters, the
main loop of this function executes. In this loop the string is exam-
ined character by character-starting with the last character of the string

and working toward the ftrst.
In this loop, the value 1 is subtracted from the length of the string in
the initialization of the loop variable (for (i=(strlen(String)l);i>=O;i--. This is done because the first character of the string is
String[O] and not String[l]. So you don't have to subtract one
from subsequent accesses to characters within the loop, you do it once
and for all at the initialization of the loop variable.
The loop distinguishes between two cases:
1.
The character of the string is a number (0-9).
2.
The character of the string is a letter (A-F or a-f).
When calculating the result you have to test to see if the current
character is a digit (0-9) or a hex character (O-F).
In the ftrst case the ASCII value of the digit 0 is subtracted from the
ASCII value of the digit O. This is done to obtain the numerical value
26
ABACUS
which the digit represents. "1 - 0" yields the number 1. "2 - 0" yields
the number 2, etc. These relationships can be easily seen from an
ASCII table. The ASCII code for the digit 0 is 48, the digit 1 is 49, and
so on, up to the digit 9, which has the ASCII code 57.
This number is multiplied by factor and added to the previous result
factor is then multiplied by the value "Oxl0 = 16". This causes a
place shift
In the decimal system the number 123 means:
3 ones + 2 tens + 1 hundred
This is similar for hexadecimal numbers. except that the number 16 is

used as the base instead of 10. The hexadecimal number "123" means:
3 ones + 2 sixteens + 1 two-hundred-fifty-sixes
Or written differently:
3
Oxl + 2
OxlO + 1
OxlOO
factor runs through Oxl, OxlO, OxlOO. etc. Now it may be clear
why the string processing goes from back to front. This guarantees that
the last character of the string represents the ones place of the number.
The second-to-Iast character is in the sixteens place, the third at the
256's place, etc.
Now we come to the second case. Here you have a hex digit A-F. Since
you do not assume that you use only upper or lowercase, you fIrst have
to convert the character to uppercase. This is done by masking out bit
4. (String[i] & MASK" = "String[i] & (255L-32L (see ASCII
table). The ASCII value of the letter "A" is subtracted from the hex
digit in order to get the numerical values 0. 1,2 etc. Since the digit A
corresponds to the value Oxa or 10, you have to add Oxa to this result
You can continue the calculation as above.
The starting and ending addresses which determine the memory area can
now be calculated with this function:
lowadress = atoh (argv[l]);
highadress = atoh (argv[2]);
You should note that characters which aren't valid hex digits are
ignored. The string 12$$5 returns the same value as 125.
Now test to see if the starting address is lower than the ending address
of the memory range. The memory is processed in ascending order:
27
1.
THE
LANGUAGE
if (lowadress > highadress)

{
printf ("START must be less than END I \n")

exit (10);
}
If the starting address is larger than the ending address, the program
exits immediately. If the starting address is less than the ending address,
then you can use the conversion table from the previous program
example:
, ,.,
for (i=0;i<32 ;i++) Conversion_Table[ij
for (
for (
for (
;i<128;i++) Conversion_Table[ij
;i<160;i++) Conversion Table[ij
;i<256;i++) Conversion:Table[il
.
, .,.,
(char)i;
(char) i;
Similar to the way you used filepos in the previous programs to

determine the current file position, use pas here to store the value of
the memory location to display. At the start this contains the value of
lowadress:
pos
= lowadress;
Now you can start displaying the memory contents. This should be
done in a separate routine because you have to test when the output
should stop.
In the hex dump of a file it wasn't so hard because the ReadO function
would tell you if you could continue to read from the file or DOL
ReadO also told you how many bytes had to be printed.
You have to calculate the number of characters to display yourself:

width = WIDTH;
while (pos < highadress)
if highadress-pos) >= width) Show (WIDTH);
else Show (highadress-pos);
As long as the value of pos has not reached the address of the highest
memory location to display, bytes are displayed. You also need to check
to see if you have enough bytes to fit in a line (Show(WIDTH or
fewer.
If the number of bytes to print is less than the number of bytes which
can be displayed on a line (WIDTH), then only the remaining bytes are
displayed (Show (highadress-pos.
To test to see if fewer than WIDTH bytes are remaining to be printed,
you must compare the number of bytes to be printed (highadresspo s) with the number of bytes which can be displayed on a line.
Unfortunately, the Aztec compiler does not understand long constants
(like 12L) in a comparison of long constants (an exception is "=" in
28
ABACUS
which long constants are not processed). Long constants are correctly
handled in assignments and as parameters to procedures.
We used a little trick: We assigned the value of the long constant to a
long variable and used this variable in the comparison. The compiler
understands this (you don't have to use this trick with the +L option
and the c32.1ib library).
Now on to the routine responsible for printing the bytes on the screen.
It should look familiar to you-we made a few small changes:
Show (ok)
long ok;
{
int i;
printf ("$%08lx
",pos);
for (i=O;i<ok;i++)
printf("$%02x ",*(pos+i;
i f (ok < WIDTH) for (
; i<WIDTH;i++) printf ("
");
printf (" ");

for (i=O;i<ok;i++)
printf ("%c",Conversion_Table[*(pos++)];
printf ("\n");
This routine first outputs the address of the first byte to be displayed on
the line. Then each byte is printed in hexadecimal notation. The method
used to make sure that the ASCII versions of the bytes are always
aligned when there are fewer then WIDTH characters to be printed is also
familiar.
The access to pos is something new. This variable (a pointer) is
accessed with *(pos+i). This means: Add the value i to the current
address stored in pos, and return the contents of the memory location
to which this result points.
The same goes for *(pos++). This means: Increment the value ofpos
by one and return the contents of the memory location to which this
value points.
The parentheses are necessary in both cases. * has a higher precedence
than ++. If the parentheses were omitted, *pos++ would mean: Get
the contents of the memory location to which pes points and then
increment the value of pes.
Here again is the complete program:
29
1.
THE
C LANGUAGE
AMIGA
/**************************************************************/
/*
1.2.3.A.mon.c
*/
/*
*/
/*
MON
*/
/*
(c) Bruno Jennrich
*/
/*
*/
/*
*/
/* This program outputs a memory range
*/
1* as a hex dump.
*1
/**************************************************************/
idefine WIDTH llL
#define MASK (255L-32L)
/* number of bytes to display */

/* for conversion to uppercase */
unsigned char Conversion_Table[2561;
1* conversion table */
unsigned char *lowadress,
/* lowest address to be displayed *1
*highadress; /* highest address to be displayed */
unsigned char *pos;
1* current address or position *1

/* in memory
unsigned char *atoh (String)

char
*String;
*/
/* digit string to hex *1
int i;
unsigned char *hex;
unsigned long factor;
hex = (unsigned char *) OL;
factor = 1L;
/* result *1
/* place factor */
1* result = 0 *1
1* factor equals 1 (one place) *1
if (strlen (String) <= 8)

1* more than eight characters */
for (i=(strlen(String)-l);i>=O;i--)
{
if String[ij >= '0') && (String[il <= '9'

{
/* figures */
hex += (String[i] - 'O')*factor;
factor *= OxlO;
if CCCString[i] & MASK) >= 'Ao) &&
CCString[i] & MASK) <= 'F'
/* Hex-figures *1
hex += CCString[i] & MASK)-'A'+Oxa)*factor;
factor *= Ox10;
}
return Chex);
main Cargc, argv)
int argc;
char
**argv:
{
int i;
i f (arge 1= 3)
{
/* number of parameters is wrong *1
printf C"USAGE: mon START END\n");

exit (10);
30
ABACUS
lowadress : atoh (argv[l):

highadress = atoh (argv[2):
1* calculate lowest and *1

1* highest addresses *1
if (lowadress > highadress)
1* error in input *1
printf ("START must be less than END !!! \n");

exit (10):
for
for
for
for
(i%0;i<32 ;i++)
( ;i<128;i++)
(
;i<160;i++)
(
:i<256;i++)
Conversion Tab1e[i)
'.';
Conversion-Table [i) = (char)!;
Conversion=Table[i]
'.';
Conversion Table[i]
(char)i:
- 1* build conversion table *1
1* current position equals *1

I*lowest address *1
pos = lowadress;
while (pos < highadress)

1* output loop *1
if (long) (highadress-pos) >= WIDTH) Show (WIDTH):
else Show long) (highadress-pos;
Show (ok)
long ok;
{
int i;
printf (1I$%OBlx
H,
1* output position *1
pos) ;
for (i=O;i<ok;i++)
printf ("$%02x ",*(pos+i;
if (ok < WIDTH) for
printf ("
");
printf ("
1* output byte as hex *1
; i<WIDTH;i++)
1* fill space with *1

1* blanks *1
It):
for (i=O;i<ok;i++)
printf (II%cll,Conversion Table[*(pos++)]);
printf("\n") :
1* bytes as hex *1
1* carriage return *1
Note that the contents of the specified ending address are not printed.
mon 1 00 200 displays the contents of memory locations hex 100 to
up to and including hex 1ff.
31
2.
C Compiler
Operations
2.
ABACUS
2.
C COMPILER OPERATIONS
C Compiler Operations
The compiler, assembler and linker are all common to the two popular
C compiler implementations. We will now look at these three and their
options and take a closer look at the compiler during compilation.
First let's look at how the compiler is called. It follows this form:
cc [>output_file] [options] prog[.c]
Everything in square brackets is optional. You must specify the name

of the file to be compiled. It doesn't matter to the compiler if you enter
ee program.e or ee program.
The output_file and options arguments can be optionally

specified. The output_file is especially useful during program
development. All messages displayed by the compiler (including error
messages) are written to this file. For example, if you enter ee >prt:
program the following is sent to the printer, assuming that no errors
occur intlle program:
"Aztec C68K 3.6a
12-18-87 (C) 1982-1987 by Manx Software

Systems, Inc.
Aztec 68000 Assembler 3.6a 12-18-87"
You can also send the messages to a disk file named errors.e and
then read this file with ED (ee >errors.e program).
Redirection
The ability to send the output to another file or device is called

redirection. It can be used with almost any CLI command, including the
assembler and linker. For example, dir >prt: sends the current
directory to the printer.
When redirecting compiler messages, you should bear in mind that
when five errors (not warnings) have occurred, the compiler asks the
user if he wants to continue compiling. This has the advantage that
error messages don't scroll quickly off the screen. You can see all of the
error messages, even if there are too many to fit on the screen at a time.
This is unnecessary when sending the messages to the printer or a file.
You should disable this feature using the compiler option -B.
An error message always has the following format
Listing_of_the_line_in_which_the_error_occurred
,.
program_name.c:error_line: ERROR number: description:
35
2. C COMPILER OPERATIONS
1* example of error message *1

}
,..
test.c:23: ERROR 69: missing semicolon
The caret (1\) shows where the compiler believes the error occurred. The
compiler doesn't notice the missing semicolon error, number 69, until
one line later. Since it is an unwritten law that only one function is
written per line (excluding if statements where commands only appear
in the "then" segment of the function), this error always occurs one line
later.
Now we come to the compiler options:
Aztec
Compiler
Options
-A
The assembler doesn't automatically start when the

compiler is fmished.
-Dsymbol[=value]
This option assigns the optional value to the

specified symbol. -DNAME=value is the same as
#define NAME value, while -DNAME causes
the symbol to represent the value 1 (ltdefine
NAME 1).
36
-Idirectory
The include files are read from the specified

directory. If you want to search multiple directories,
you can separate directories with exclamation
points (-Idirl! dir2! dir3).
-0 filename
The object file (or the assembler source, if the -A

option is used) is written to the specified file.
-s
Warning messages are suppressed (silent option).
-T
The lines of C source code are included in the

assembler source as comments. This option only
makes sense with the -A option.
-B
This option suppresses the halting of message output after every five messages. When directing error
messages to the printer, for example, it doesn't
make sense for the compiler to stop after every five
messages to ask you if you want to continue.
-Enumber
This option allocates an expression table with

numeric entries. This expression table is used when
the compiler evaluates for loops, swi tch/ case
statements, etc.
- Lnumber
This option reserves numeric entries for the local

symbol table. This local symbol table is used for
variable definitions and declarations within func-
ABACUS
2. C
COMPILER OPEk"TIONS
tions or blocks (program segments are enclosed in

curly bmces).
-Ynumber
This option allocates a case table with numeric

entries. This ca se table is used when compiling
swi tch statements.
-Znumber
This option allocates a string table with number

bytes. The string table is used for storing strings
during compilation.
+B
This option suppresses creation of pub 1 i c

.begin in the assembly language file. This is
important when creating multiple modules to
prevent the command jrnp .begin from being
included in every module.
+c
Creates "large code" (see Section 2.3).
+D
Creates "large data" (see Section 2.3).
+Hfilename
This option saves the symbol table created by the

compiler for a progmm.
+Ifilenarne
This option loads the symbol table previously

saved with +H. This allows you to compile, save
and reload the include files used by a program.
This increases the speed of later compilations
because the incl ude files don't need recompiling
each time.
+L
This option instructs the compiler to interpret all

int variables and constants as long. Together
with the c32.lib library, this configures Aztec C
to compile source codes originally written for
Lattice.
+Q
The strings used in the program are written to the

data segment.
+ff
Calculations are performed in single precision.

m.lib must be linked with the object file.
+fi
Calculations are performed in double precision.

You must link in rna.lib ormx.lib when using
the double precision floating point libraries.
+f8
This option enables the 68881 math coprocessor

for calculations. When using this coprocessor,
which must naturally be installed, you must link in
thern8.lib.
37
2.
+m
This option checks the stack for overflow. If overflow occurs, a runtime error is produced.
-n
Debugging information is added to the assembler

source, which is further processed by the assembler.
+P
This option is identical to the options +C +D +L.

In addition, the registers D2 and D3 are saved before
each function call. This option requires that the
e132.lib be linked with the main program.
+r
The A4 register can be used as a register variable.

This option requires using the large code and data
model and that the el.lib or e132.lib be
linked with the main program.
Some of these options are self-explanatory, such as the -A option

which you can remember as "minus assembler". It tells the compiler
not to start the assembler as it usually does.
When the -A option is used, you still have to tell the compiler where
to write the assembly language source file which it normally creates on
the RAM disk (see CCTEMP variable) and deletes later. You must
specify the name of the assembly source file with the -0 option. The
compiler and assembler can then be called separately:
cc -A -0 program.asm program.c
as program.asm
It's more than just passing additional arguments to ee. You also have
to call the assembler "by hand," which creates the object file
program.o from program. asm. If you specify only the -0 option
(without -A) when compiling, the object file created by the assembler
(now called by the compiler) has the specified name.
Normally you don't have to specify the name for the object file because
it automatically has the same root name as the C source file, but with
an extension of .0. If your program is large enough that you have to
conserve every byte of space on the disk, you should remove the object
file after each link with delete. This can be done with the professional version in the script file or with the developer's version in the
makefile. This is useful only when you have multiple programs on
the disk, each of which consists of only one module.
38
2.1 How A C
ABACUS
2.1
COMPILER
WORKS
How a C compiler works

In order to explain the additional compiler options, lets take a closer
look at the compiler itself. We want to see what the compiler does with
a source code, which is really only an ASCII me.
First, the compiler looks for obvious errors. The compiler knows, for
example, that a semicolon must follow each statemenL If the compiler
does not find the semicolon after a statement, it displays an error message. If the condition is missing in an if statement, the compiler tells
the user. However, the compiler cannot recognize all other errors. These
are the errors which only you as the programmer can determine. These
errors usually manifest themselves in system crashes, or when the
compiled program doesn't do what you expected it to do. Then the
debugger comes into play (see Section 2.5).
A typical error of this type is exchanging "=" for "-", using "=" in the
condition of an if statement when "==" is intended. Another common
error is accessing a non-existent array element
int array(4);
This array contains the elements 0, 1, 2 and 3, for a total of four

elements. But if you assign a value to array [4] , the "fifth" element,
you may change another program variable, or even cause the program to
crash.
When hunting down these errors, it helps to have a good eye and a feel
for where such errors can hide. Failing good eyes and programming
know-how, the debugger can help.
Mter the error test comes the translation of statements and expressions
into assembly language. This translation can take place immediately
after checking a statement for errors, or after checking the entire program, depending on the compiler.
39
2.1.1
Internal organization of variables and

structures
The program translation can begin if the program is free of syntax
errors, etc. Generally a C program consists of variable definitions and
declarations in addition to statements.
Before looking at actual C statements, let's look at how variables and
structures are compiled:
First the individual variable types must be determined. The C language
recognizes the following scalar types:
int, long, char, short, unsigned char, unsigned int,
unsigned long, float, double and pointers
Structured typeS are arrays (or vectors), structs, enumerated types,

bit fields and unions.
The compiler encodes each type and stores the variable name and its
type code in the global symbol table. It should be pointed out here that
the following symbol tables are for an ideal C compiler and not the
Aztec compiler. But the Aztec comoiler works according to the same
principle as the ideal compiler.
int i_val;
char c_val;
Symbol table:
It looks similar for structures. Here you have the name of the structure,
followed by a code (a single byte) for the start of the structure. Then the
variables contained in the structure are listed:
struct nonsense
{
int hello;
char na;
unsigned char good_day;
};
Symbol table:
"nonsense"/"code_structure_type/"hello"/code_int/
"na"/code_char/"good_day"/code_unsigned_char/end_code
This is a structure declaration. The individual codes are represented here

with text (e.g., code_char instead of a byte value). The structure
40
2.1 How
ABACUS
COMPILER
WORKS
cannot be used until a suitable variable has been defined (struct

nonsense nonsense structure). This is also possible with
typedef (typedef struct {... } structure type name).
Scalar types can also be renamed with typedef:
"typedef unsigned char BYTE"
Symbol table:
"BYTE"/code_type_begin/code_unsigned/code_char/end_code
The include me exec/types.h contains a number of new types

which are used in almost every incl ude file. They are required for
using incl ude files and must be included before any of the other
include files can be used.
But back to structures. The symbol table for a structure definition looks
like this:
struct nonsense nonsense_structure;
Symbol table:
"nonsense_structure"/code_structure_begin/
"nonsense"/code_structure_type/end_code
You can directly define a nonsense_structure. The structure is

then no longer available for future definitions:
struct
{
int hello;
char na;
unsigned char good_day;
nonsense_structure;
Symbol table:
"nonsense structure"/"code structure begin/
"hello"/code_int/"na"/code=char/
"good_day"/code_unsigned_char/end_code
A definition of the form struct nonsense structure name is

no longer allowed here. The declaration of bit fields, arrays, enumerated
types and unions is similar to that of structures:
ARRAY;
int digits [6] ;
Symbol table:
"digits"/code_array/code_int/6/end_code
41
ENUMERATED:
enum color
{red, green, blue};
Symbol table:
"color"/code_enum/"red"/O/"green"/1/"blue"/2/end_code
Compiler assigned integer numbers represent the individual enumerated

types in the program. The same effect can be achieved with a set of
#define instructions. Instead of enum color = {red, green,
bl ue} you could write:
#define red 0
#define green 1
#define blue 2
int color;
It should be noted that a #defined symbol simply replaces it with the

string which follows it. In the symbol table, the values of #defined
symbols are therefore stored as strings:
Symbol table for defines:
"red"/code_define/"O"/end_code/"green"/code_define/
"1 "/end_code/"blue"/code_define/"2"/end_code
Caution should be used especially for mathematical expressions:

#define sum a+b
printf("%d
%d",sum,sum*lO);
The printf function fIrst prints the sum of the two variables a+b.1f
the second output is supposed to be the sum multiplied by ten, you
will probably be disappointed here. The compiler sees sum*10 and
generates a +b * 1 O. It adds the value of a to ten times the value of b.
You should put the equation in parentheses to avoid this:
#define sum (a+b)
The #ifdef, #ifndef and #endif directives should also be

mentioned in the context of #define. These occur often in include
files and are usually used to ensure that an include file compiles only
once, since otherwise this would lead to errors (redefined symbols, etc.).
Since most include files are built from others, includes often appear
within include files:
/* **** Include-File **** */
/* if EXEC_TYPES_H is not */
/* defined, */
42
2.1 How
ABACUS
#include "exec/types.h"
#endif
C COMPILER
WORKS
/* then include exec/types.h */
#ifndef GFX_H
/* if GFX_H is not defined, */
#include "graphics/gfx.h" /* then include graphics/gfx.h */
#endif
The compiler can use the symbol table to check to see if a symbol is
defined (#ifdef) or not (#ifnde). The compiler can then determine
the fate of the statements enclosed in the #if... - #endif block. Each
include file defines a symbol at the start with the name
"IncludeFile H".
Bit fields
Now let's look at bit fields. These are a new feature added to Aztec
compiler version 3.4a. As the name implies. bit fields provide a way of
organizing individual bits of information. They are declared similarly to
structures:
struct bitfield
{
unsigned
unsigned
unsigned
unsigned
unsigned
2:;
1: a;
1: b;
1: c;
3:;
/* bits 0 and 1 */
/* bit 2 */
/* bit 3 */
/* bit 4 */
/* bits 5-7 *1
11
struct bit field bf;
This declaration creates an eight-bit-wide bit field. You can now access
bits 2. 3 and 4 with bf.a. bf.b and bf.c. You can only set these
variables to 0 or 1. This is logical. since a single bit can have only one
of two values.
You cannot access bits O. 1.5.6 and 7 because they were not assigned
names. These bits were "skipped" as told by the unsigned 2:; and
unsigned 3:; directives.
The symbol table for bit field declaration and definition above looks
like this:
Symbol table:
"bit field" / code _type_begin/"" /code_bitO/'''' /code_bit1/
"a"/code_bit2/"b"/code_bit3/"c"/code_bit4/""/code_bit5 I
""/code_bit6/""/code_bit7/end_code
Definition:
"bf"/code_structure_start/code_type/"bitfield"/end_code
union
Now let's look at the last of the non-scalar data types-the union
union designates a variable which can consist of more than one data
type:
43
union number
{
int i;
float f;
union number n;
Symbol table:
"number"/code_union_type/"i"/code_int/"f"/code_float/
end_code
Definition:
The variables i and f occupy the same storage space. Therefore, this
variable can be interpreted as either type.
There are also pointers to the individual variables, structures and arrays,
so we must introduce the code for pointers:
int *pointer;
struct nonsense *sensible;
Symbol table:
"pointer"/codeyointer/"sensible"/code_int/end_code/
"sensible"/codeyointer/"nonsense"/code_structure_type/
end_code
The symbol table constructed during the compilation is needed whenever a variable is accessed. IT you want to increment the variable i
(i ++), for example, the compiler must know the variable type of i.
Assuming that i is an int object, the compiler then knows that this
variable is 16 bits wide and that it must generate the statement
add.w #l,address of i and not add.b #l,address of i or
add.l #l,address_of_1 for the assembly language source. It is similar when accessing the values of structure elements, e.g.,
sensible->hello = O. The compiler must first find out if
sensible is really a pointer. From the symbol table it learns:
sensible/code-pointer and knows that sensible is a pointer.
Now it must determine what kind of object sensible points to. If
sensible points to an int variable, than the line sensible>hello doesn't make sense, because int variables don't contain
additional elements, as do structures.
From the symbol table the compiler discovers that sensible points
to a structure. The "->" is therefore allowed. sensible->hello can
also be replaced with (* sensible).hello, whereby it should be
44
2.1 How
ABACUS
C COMPILER
WORKS
noted that the period (.) has higher priority than the star (*) in front of
sensible, which is why the parentheses are necessary.
If you wrote sensible.hello, the compiler would respond with an
error message (incorrect pointer usage).

Now the compiler must check to see if the structure to which
sensible points has an element named hello. It also gets this
information from the symbol table. First sensible points to a
nonsense object, and in the symbol table entry for nonsense the
element hello occurs. Now the compiler just has to check to see if
the assignment of 0 is allowed to the variable hello of the
nonsense structure, and if so, it must detennine what this assignment should look like. After the compiler has detennined from the
symbol table entry for nonsense that hello is an int object, it
can produce clr.w address_of_sensible->hello.
During the compilation, the compiler checks the type of each variable
whenever it is accessed. If you try to assign a variable to a variable of a
different type, the compiler produces a warning which tells you that an
illegal type assignment is being performed.
The compiler automatically perfonns the appropriate type conversions
when assigning standard types like char, int and long. In the
assignment of an int object to a char variable, for example, only the
lower eight bits of the int object are passed to the char variable,
while the missing bits in the assignment of char to long are fIlled
with zeroes (signs are added as needed).
Type conversion becomes problematical in connection with pointers:
int *Inputl;
char *string = "The Amiga is tops!!";
string = Inputl;
The compiler isn't certain if the programmer knows that he has

assigned two pointers with different types. Therefore, the compiler
displays a warning which warns the user about a potentially incorrect
pointer assignment
It can be desirable to assign an int pointer to a char pointer, perhaps
to access the high-order byte of the int variable through the int
pointer. This is possible with a char pointer because char objects
are only one byte wide and after string++ the pointer string
points to the next address.
If you want an assignment like this to occur, you should let the compiler know this with the help of a type cast:
string
(char *) input;
4S
AMIGA.C FOR ADVANCED PROGRAMMERS
The program segment above illustrates two features beside the cast:
First, a variable is initialized in its definition, and second, strings are
used.
When initializing a variable the location at which the variable definition

rests is important. If the variable was defined outside a function, the
storage space occupied by this variable is assigned with dc.<b/w /1>
value:
1**************************1
1* C code
*1
1**************************1
char *string
"The Amiga is tops!!";
1**************************1
1* Assembly language
*1
1**************************1
dseg
ds 0
public _String
_String:
dc.l .1+0
cseg
;data segment
;reserve 0 bytes
;global variable
;allocate memory
;with address of string
;code segment
.1
dc.b 84,104,101,32,65,109,105,103,97,32,105,115,32,84
dc.b 111, 112,115, 33,33,0
;ASCII Codes
ds 0
;null byte
public _main
_main:
;here we start
When initializing within a function, things look different:

1**************************1
1* C code
*1
1**************************1
proc ()
{
char *string
"The Amiga is tops!!";
1**************************1
1* Assembly language
*1
1**************************1
public yroc
yroc:
link a5,#.2
movem.l .8,-(sp)
46
;procedure is 'global'
;allocate stack storage
;save registers
2.1 How
ABACUS
lea .1,aO
move.l aO,-4(a5)
C COMPILER
WORKS
;address of the string -> aO

;aO -> variable '_string'
movem.l (sp)+,.8
unlk a5
rts
.2 equ -4
.3 reg
.1
dc.b 84,104,101,32,65,109,105,103,97,32,105,115,32,84
dc.b 111, 112,115, 33,33,0
ds 0
;null byte
As you see, the strings are in an unusual form. The ASCII codes of the
characters, not the characters themselves, go into the assembler file (see
Section 2.2 for information about the common assembler directives
global, public, dseg and cseg).
Here we should probably explain the width of the individual variables:
Char objects are always one byte wide and therefore have a value range
from -128 to 127 or 0-255, if you use "unsigned char". Int
variables occupy two bytes with a value range of -32768 to 32767 or 065535. Long objects are four bytes wide with a value range of 2147483648 to 2147483647 or 0-4294967296. Float variables
contain four bytes and double variables 8 bytes.
Pointers are variables which contain the starting address of a variable or

structure. Because the 68000 can theoretically address four gigabytes,
pointers must have a range of 0 to 4294967296. Therefore a pointer
always occupies four bytes, regardless of what type they point to.
If you want to increment or decrement a pointer to a long object, you
should note that four bytes are added to or subtracted from the old
address. The pointer points to the next long object. The same applies
to pointers to structures. Assuming a structure holds 95 bytes, then
incrementing a pointer to such a structure with <pointer>++ or
<pointer>=<pointer> + 1 increments the address by 95.
All this time you have assumed that your program only declares and
defines variables outside functions. But you can also define variables
within a function which are available only for that function. Take a
look at the following program:
int i;
main () { ... }
prroc ()
{
int i;
i
= . ;
47
The variable i appears twice. If you access i in any line of the prroc
function, the compiler knows that you mean the i defined within this
routine. The compiler needs a second symbol table to recognize this. If
the compiler wrote the symbol code for the variable i in the global
symbol table, then it is unable to distinguish between the global i
dermed before mainO and the i in procO.
The local symbol table is used whenever variables are declared and used
within a function. When the function is compiled, the next function can
access the entire memory range of the local symbol table. When a variable is accessed in the routine, the compiler looks for the variable in the
local symbol table first. If the variable could not be found there, then
the compiler searches the global symbol table. When using libraries,
defining the library base pointer inside or outside the function is important. For example, if you define the pointer to the Intuition library
(struct IntuitionBase *IntuitionBase) within a function
(e.g., mainO), then you cannot use the Intuition functions.
They require the variable _IntuitionBase, which can only be made
available for other modules and linker libraries with the assembler
directive global Intui tionBase,4. If IntuitionBase is
dermed within a function, then only stack space is reserved for this
pointer, which cannot be accessed with the label_Intui tionBase.
But back to the local symbol table. Naturally it also needs space in
memory. The compiler automatically allocates 1040 bytes for it. If the
compiler gives you the error message Local table full (Use -L),
you should follow its advice and increase the memory for the local
symbol table with the -L option. You should note that an entry is 26
bytes long. -L40 reserves 1040 bytes.
The global symbol table can also be saved-the local symbol table
cannot. This is especially useful when you write programs which use
many include files.
You can separately compile the include file block (which may contain
only include files) and then store the symbol table by using the +H
option (cc +Hcompiler includes include block.h)
whereby the include block looks something like:
finclude "exec/types.h"
finclude "graphics/gfx.h"
#include "graphics/gels.h"
Mterward you can read this symbol table back in with the +I option
(cc +Icompiled_includes program.c). This saves the time of
reading the individual include files and constructing the symbol table.
48
ABACUS
2.1.2
2.1 How
COMPILER
WORKS
Software organization of variables and

structures
The contents of variables and structures must be stored in memory
locations. '.fhe functions request the required memory space from the
system stack if no other storage class (extern, static, auto.
register) is specified for the variable. To do this, the compiler first
uses the symbol table to calculate the number of bytes which must be
reserved.
Since the compiler knows which variables are being used, and it can
find out exactly how much memory is required by each, this calculation
is simple. The compiler also accesses the symbol table extensively
when the sizeofO instruction is used.
But it can lead to serious problems if the variables require more space
than the stack has available. You can enable the stack checking with the
+m option. Before each routine executes. the routine _stkchk#O is
called. which calls the routine stkoverO if a stack overflow is
detected. On a stack overflow, - stkover() immediately halts the
program and sometimes displays the reset requester (Finish all
disk activity ...).
According to the Aztec manual, _stkoverO supposedly informs the
user of a stack overflow. Unfortunately it doesn't always work that
way. Sometimes the guru appears. But you can write your own
s t k 0 v e r() routine. This routine should be written in machine
language so that it can use an alternate stack. It should be noted that
address register a7, which is the stack pointer. points to the highest
address of the alternate stack area, since a stack "grows" down.
When linking this new _stkoverO routine, you should note that a
routine by the same name is already stored in the c.lib. The linker
also has a second machine language routine by the name of
_stkover: available-the first is yours and the second is that from
c.lib (an underscore precedes the name of the C procedure in assembly
language). The linker uses your routine without the use of a linker
option, however, since it appears before that in c.lib.
You check for stack overflow only in the development phases.
Programs which have to call_stkchk#O every time they enter a
routine become slower and longer.
Now we come to variable organization in assembly language. Here is a
short C program:
49
1*********************1
1* C language
*1
1*********************1
lnt c;
main ()
(
lnt j;
j = 0;
I
proc()
{
lnt i;
c = 0;
. ..,.
For the sake of simplicity, we are using only int variables here. But
the principle is the same for allocating or reserving memory for structures,arrays, etc.
Let's look at the assembly source created by the compiler:
;:ts = 8
global _0,2
; senseless! ? !
;tells the assembler to reserve
;two bytes for c. Same as
;" c: ds.b 2" ; main is known to all modules
;"as is _cIt
main:
link
movem.l
as,#.2
.3,-(sp)
clr.w -2(aS)
;as to -(sp), sp to as, sp + #.2

;registers on stack (no register
;list specified!) variable stack
;access variable j
.4
rnovem.l (sp)+,.3
unlnk
rts
as
.2 equ -2
.3 reg
public ...Froc
link as,#.9
rnovem.l .IO,-(sp)
clr.w c
;saved register contents from stack

;back to register
;as to sp,-(sp) to as
;2 bytes for int object
;registers to be saved (list empty)
:'proc' is known to all modules.
;allocate variable stack
:save registers
;c
.11
rnovem.l (sp)+,.lO
unlnk as
rts
.9 equ -2
.10 reg
public .begin
dseg
end
so
;2 bytes for i
;no registers to be saved
;start of the data segment
2.1 How A C COMPILER
ABACUS
WORKS
The variable storage for variables in functions comes from the stack.
while the storage for global variables, which are defined outside the
functions, occupies "normal" memory. These declarations look similar
for structures and arrays, except more bytes are reserved. For unions,
sufficient storage is reserved for the largest variant in the union. This
memory space is then accessed when a union element is accessed:
/*********************/
/* C language
*/
/*********************/
union num
{
int i;
float f;
union num z;
main 0
{
z.f
= 1.1;
z.i
= 10;

global _z,4
public main
_main:
link as,#.2
movem.l .3,-(sp)
move.l #$8ccccd41,_z
representation);
;reserve bytes for the union

; (global)
;z.f = 1.1 (float
;z.i
movem.l -(sp),.3
unlk as
rts
.2 equ 0
.3 reg
public .begin
dseg
end
=0
;no local variables
When accessing a structure or array element, the offset of the element in

question must be computed in addition to the base address of the structure or array:
51
/*********************/
/* C language
*/
/*********************/
main ()
{
struct Structure
{
int Elementl;
long Element2;
float Element3;
struct Structure s;
struct Structure t;
s.Elementl
0;
s.Element2
s.Element3;
s=t;

public _main
_main:
link as,t.2
movem.l .3,-(sp)
clr.w -lO(aS)
;s.Elementl
=0
move.l -4(aS),dO
jsr Ffid
move.l do,-8(aS)
;s.Element3 -> dO
;convert float to long
;dO -> s.Element2
lea -lO(aS),aO
lea -20(aS),al
move.l (al) +, (aO) +
move.l (al) +, (aO) +
move.w (al) +, (aO) +
;&s to aO
;&t to al
;convert bytes
.4
movem.l (sp) +, .3
unlk as
rts
.2 equ -20
.3 reg
public begin
dseg
end
As you can see, this program uses variables of type float. It must
therefore be compiled with the +fi options. When using double
variables, the +ff option must be used. When using the 68881 math
coprocessor you must compile your programs with the +f8 option. In
all three cases math routines are needed which are not in the c.lib
52
2.1 How
ABACUS
COMPILER
WORKS
library, therefore rn.lib, rna.lib ormx.lib must also be linked in

with the object file.
Now we'll look at the memory classes we mentioned earlier: With the
memory class extern you are informing the compiler that the
variable is found in a different module. The compiler must not reserve
any space for this variable here since this must be done in another
module. If this does not happen, the linker displays an unresolved
reference error.
If you use the extern class in a variable declaration in a function, the
global symbol table searches for the variable. If you want to use a
global variable, you should always specify the storage class extern in
the declaration. This way you always know which variables in the function are being used; which are local and which are global:
int i;
proc ()
{
extern int i;
int j;
/* access to global i (declaration) */

/* local (auto) j (definition) */
The static storage class gives the compiler information about the
type of storage which should be used for the variable. Normally all of
the variables in a function are auto. This means that memory space is
allocated on the stack when the function is entered, and then released
when control leaves it. If you declare a variable static, then the
memory space is taken from the normal program storage. When the
function is left, the value is retained so that you can use the old value
when the function is re-entered. The bss directive of the assembler is
used to assign static storage space. This is similar to the global
directive, except that the storage space for the routine in which the
sta tic variable is defined is valid. This results because the variable is
not accessed by its name, as in global_c,2, but with a normal label
(like .4).
static variables l!Ie particularly useful in recursive functions. They

can be used as counters or to quickly pass arguments without having to
pass them to the function every time (see Section 2.1.4).
With the register storage class, a CPU register is used to hold the
value. You can declare any scalar variable type as a register variable. Note that double variables which are declared as register
must have some bytes truncated. double variables are eight bytes
wide, but the address and data registers of the 68000 can hold only four
bytes each.
The registers DO, 01, 02, 03, A2 and A3 can be used by the compiler
for storing values.
S3
2.
2.1.3
Translating control statements

Let's look at control statements. They determine the order in which
various routines and commands are processed.
After the error search has produced no (apparent) errors and all variables
and structures have been defined and allocated, the actual compiler
process begins, translating the individual C instructions and function
calls into assembly language. First we'll compile the control structures
(for, while, switch, if):
"for (i=0;i<5;i++) j += 2;".
Compiled, this program segment looks like:

clr.l -4 (a5)
.1:
add.l #2,-8(a5)
;
;
i=O (intialization of
i')
j += 2 (loop body)
(increment)
add.l #1,-4(a5)
i++
cmp #5,-4(a5)
; i<5
(termination condition)
blt .1
; yes, then back to the beginning
Once the compiler recognizes this instruction, the initialization of the

variable i is checked. The compiler determines that the variable i is
being set to O. This is represented by the assembly language instruction
clr.l -4(a5). It is assumed that the storage space for the variable i is
reserved in -4(a5) (stack storage). The variable j is located at -S(a5).
After the initialization of the variable i the compiler must then look at
the loop body (j += 2). The loop condition and incrementing of the
variable i are ignored for the moment. The compiler goes to the statement j += 2. Since this is the first (and only) statement in the loop
body, a label must precede this statement so that you can jump to it.
After the label is set, the actual statement is translated (add.l
:fI:2,8(a5. Now the compiler takes care of incrementing the variable i
(add.l :fI:l,-4(a5 and the loop condition (cmp.l :fI:5,-4 (a5); bIt
.1).
Here you can see that the linearity of the C commands is retained in
assembly language, only in reverse order.
The while loop compiles in a manner similar to the for loop. In a
while loop. the condition specified immediately after the keyword
while is translated immediately. The loop condition is tested before
the loop body executes. The compiler notes that it must set a label at
the end of the while body so that if the while condition is not fulfilled, it can jump past the body:
S4
2.1 How
ABACUS
C COMPILER
WORKS
1*******************1
1* c language
*1
1*******************1
1* Have to do the *1
1* initialization ourselves *1
1* while loop
*1
i = 0;
while (i<S)
{
1* loop body
j += 2;
*1
i ++;
Here is the assembly source created by the compiler:
.1 :
elr.l -4(aS)
i = 0;
crop #5,-4 (a5)

bge .2
add.l #2,-8(a5)
add.l #1, -4 (as)
bra .1:
*
*
*
*
*
i<5
no
{ j += 2;
i ++; }
back to the start of the loop
.2 :
There is a second type of while loop:

1*********************1
1* C language
*1
1*********************1
i = 0;
do {
j+=2;
i++;
while (i<5);
In this type of while loop the body executes at least once. The loop
condition is checked after each execution.
Another control structure is the switch statement. It provides a way
of choosing from several alternatives. A typical switch statement
looks like this:
switch (i)
{
case 3:
i++;
break;
case 4:
i--;
break;
ss
default:
i=O;
break;
The compiler turns this into:

move.l -4(aS),dO
bra .4
.6:
add.l #1,-4(aS)
bra .S
* case 3: i++;
* break;
.7:
sub.l #1,-4(aS)
bra .S
*
*
.8:
clr.l -4(aS)
bra .S
* default: i=O;
.4:
contents of "in to dO
case 4: i--;
break;
sUb.l #3,dO
beq .6
sub.l #l,dO
beq .7
bra .8
.S:
First the compiler moves the value of the variable i to data register dO.
Each case portion is then translated, whereby a label is first placed
before each portion and then the individual statements are translated.
The use of the break statements at the close of each case portion
causes a branch to the instruction after the switch.
How is the s wit c h statement processed by the compiler? It first
causes the variable i to be loaded into data register dO and then the
branch to .4.
When the compiler reads case 3,the value 3 is subtracted from dO
and the result tested for O. A branch is made if true. In the second case
the value is tested against 4. But instead of loading the value of the
variable i back into dO and then subtracting 4, the compiler calculates
the difference between the first and second case (3-4). If this difference
is negative, then the difference is subtracted from dO (sub.l U,dO). If
the difference is positive, it is added (add.l fl,dO). Then you can
again test the data register against O.
If all of the case comparisons fail, the default statements are automatically executed (the default is optional, in which case the swi tch
statement is simply exited).
But naturally a swi tch statement is not always as ordered as was
shown above. It may contain more than two cases, and they do not
have to be ordered.
56
2.1 How
ABACUS
COMPILER
WORKS
switch (i)
(
case 1:
i++;
break;
case 0:
i--;
break;
case 6:
i=O;
break;
case 2:
i=O;
break;
We will ignore the fact that this isn't exceptionally elegant

programming-the compiler doesn't mind. What does it do? The
following assembly language code clarifies this:
move.l -4 (as) ,dO
bra .4
* contents of 'i' to dO
.6:
add.l #l,-4(aS)
bra .S
* case 1: i++;
* break;
.7:
sub.l #l,-4(aS)
bra .5
*
*
.8:
sub.l #l,-4(aS)
bra .S
* case 6:
* break;
.9:
sUb.l #l,-4(aS)
bra .S
*
*
case 2: i--;
break;
*
*
*
*
case 0:
case 1:
case 2:
default:
default:
default:
case 6:
.10: dc.w .7-.11-2

dc.w .6-.11-2
dc.w .9-.11-2
dc.w .S-.11-2
dc.w .S-.11-2
dc.w .S-.11-2
dc.w .6-.11-2
.4: cmp.l #7,dO
bcc .5
asl.l #1,dO
move.w .10(pc,dO.w),dO
.11:
jmp (pc,dO.w)
*
*
case 0: i--;
break;
i--;
(0)
(2)
(4)
(12)
* default
* dO * 2
.5:
This s wit c h statement is translated similar to the first. Only the
individual case selection is handled somewhat differently.
57
First the case blocks are translated in the order in which they appear
in the C program. They are then placed in order in the jump table which
contains the addresses (relative to the PC) of all of the case blocks.
The offsets are then sorted according to the order of the cases to which
they point and are stored.
Under normal circumstances you can test a maximum of 100 case
conditions. If you want to test more cases-which seems mther unlikely.
the -Y option allocates more space for additional case conditions.
- Y 1 0 0 reserves 100 entries for the cas e conditions. Each entry
consists of four bytes, whereby two bytes are used for the offset and the
other two bytes for the integer number for which the case is tested.
When selecting a case condition the value of the variable i (contained
in dO) is doubled (left-shifted) and used an index into the jump table.
Since this table consists only of offsets which are specified in 16-bit
words, i only has to be doubled instead of quadrupled, as would be
necessary if the table consisted of the absolute addresses of the case
blocks.
Since some cases are not tested (case 3:, case 4:, case 5:), the
jump table is filled at these locations with the address of the command
after the switch statement (label.S).
If the cases to be tested are far apart from each other, then they are not
selected with a jump table. This would not be an efficient use of
memory because the default label would appear many times in the jump
table. Therefore the fIrst method is used here-the variable to be tested
is placed in dO and the various case values are subtracted or added and
dO is tested against O.
Now we come to the simplest control structure: The if / e 1 s e

statement. These can be translated linearly, as they are written in C:
if (i == 1) i
= 0;
else
i++;
In assembly language this looks like:
.4:
.S:
cmp.l #1,-4(aS)
bne .4
clr.l -4 (as)
bra .5
add.l t1,-4(a5)
*
*
else i++;
==
no
* yes
* continue
The label poses the only problem when translating the if statement.
The compiler notes which label number it must use after the branch
command. for example. and which label must be placed before a command.
58
2.1 How
ABACUS
C COMPILER
WORKS
Another problem is the evaluation of mathematical expressions. What

does the compiler do with i = (j *5)+ 201 The compiler starts with the
deepest level of parentheses and works it way up.
To describe this process in detail would be too complex here. If you
aren't sure which operations have priority over others, you can add more
parentheses. This ensures the portability of your program, even if other
compilers translate mathematical expressions somewhat differently.
Another control statement is the goto statement, although few
structure-conscious C programmers use it:
goto Label;
Label:
The assembler source generates a bra or jmp command. The compiler

also must ensure that the label being jumped to is in the same function
as the goto statement
2.1.4
Function calls
In addition to using control statements and variable assignments, you
can also write your own functions or use existing ones. There are two
types of functions: those which return a value, and those which do not
Both types let you pass arguments to them. Arguments are passed
through the stack.
func(i,j,k)
int i;
long
j;
struct nonsense *k;
int 1;
1=0;
func (i,k,k)
In assembly language, this somewhat senseless recursive function looks
like this:
_func:
link as.#.2
movem.l .3- (sp)
move.114(a5),-(sp)
move.llO(a5),-(sp)
move. 1 8(a5),-(sp)
;storage for local variables

;registers on stack
;parameter k on stack
; for subsequent call
;j on stack
;i on stack
S9
jsr _func
lea lO(sp),sp
movem.l (sp)+,.3
unlnk as
rts
;call routine
;restore stack
;load registers again
;release local storage
;two bytes for int object
;register list
.2 equ -2
.3 reg
The following figure illustrates how the local variables are placed on
the stack:
Pointer
long
lO(aS)_ >1-----_
8 (a5) -
> ~'"""'"=="'-'''"II
lnt
link a5,i.2
as->-(sp);
a5 = sp;
sp->sp+#.2
(.2 equ -2)
~2(a5) ->~~~~
___I
sp _ >1-----_1
movem.l .6,-(sp)
Figure 2.1
You can see that the input arguments are accessed with positive
indexing on as. (e.g., 8(aS, while negative indexing accesses local
variables (-2(aS.
When the input arguments are accessed, the old value of register as.
placed on the stack by the link command, and the return address, placed
on the stack by any JSR, must be skipped. Beyond this, the input
parameters are simply accessed in reverse order on the stack.
60
2.1 How A C COMPILER
ABACUS
WORKS
The compiler doesn't check if the correct number of arguments were

passed to a function.
Let's assume that a function expects three long variables as input
arguments, but you pass this routine three int values instead. The
three int variables occupy only six stack bytes, while the routine
access~s 12 bytes {3* 4 (long.
If you now change these parameters (in the above example, k=O <=>
"clr.l14(a5)"), which is possible in C, you may end up changing a
return address because when the arguments were passed, only six bytes
were allocated on the stack instead of 12.
Correct Porameter Hondling
FolIe parameter Handl'ng (only Int',)
Pointer
14(aS)
long
lO(aS)
int
link as,t.2
link as,t.2
-2 (as) -2 (as) - >
Local
Variable
Local
Variable
Figure 2.2
For functions which return a value, you must specify the type of the
return value. This is given immediately before the function name. There
are also functions of type void, which means that they do not return a
value. If you do not explicitly set the return type of a function, the
compiler assumes that it is into
The type of the function's return value is stored in the symbol table
just like a variable type:
void Function ( , , )
Symbol table:
"Function"/code_function/code_void
61
If a function is in another module or in a library, the type can be

declared with extern like an external variable:
extern struct *MsgPort FindPort();
Symbol table:
"FindPort"/code_external_function/code_structure_type/
"MsgPort"
You can now be sure that the compiler will not "complain" when you
assign the return value of FindPortO, which is a pointer to a
MsgPort structure, to a pointer of type *MsgPort. Basically, the
objects of the pointers don't matter when doing pointer assignments.
Pointer variables are always four bytes long, so that no conversions
have to be made. There is a second type of pointer, however, the BCPL
or CPTR pointers. These are leftovers from the BCPL language. Their
contents are shifted right by two bits so that they can access only long
words. BCPL pointers occur only as input parameters, and then only
rarely I such as in certain I/O routines.
You can also use extern to assign different return values to functions
in other modules. Let's say that the routine atolO is defined in
module 2 and that it actually returns a long value. If you now define
in module 1 that this routine returns a value of int type (extern
int atolO), the compiler won't be any the wiser because the symbol
table for the second module is not available. Note that this declaration
must take place before the function is used. Even when you write your
own functions of type other than void, you must ensure that they are
defined (or declared before they are used) so that they are entered in the
symbol table.
For functions, the input argument types are irrelevant. There are compilers, however, which check the input parameters for "type purity" so
that errors like those described above don't occur. The Aztec compiler is
not one of these.
62
2.3 THE ASSEMBLER
ABACUS
2.2
The Assembler
After the compiler has created the assembly language source, the
assembler can begin its work. It must create the object file from the
assembly source. Its call is as follows:
as [>output_file) [options) program.asm
The object file consists of executable machine code whenever possible.

Only unresolved external references have to be resolved by the linker.
Let's look at the following assembler source:
dseg
ds 0
public Str
Str:
- dc.l .1+0
;allocate
cseg
dseg
.1
dc.b 72,97,108108,111,0
cseg
global i,2
;data segment
;reserve 0 bytes
;global variable
memory with address of the
; code segment
;data segment
; "Hello"
;code segment
public -main
main:
- link
a5,#.3
movem.l .4,-(sp)
add.l 1, str
jsr _exit
;start here
;str++;
.5
movem.l (sp)+,.4
unlk as
rts
.3 equ 0
.4 reg
public exit
public -:-begin
dseg
end
From this assembly language source the assembler creates relocatible

code. Relocatible means that the program is executable at any location
in memory. It is not restricted to a particular address.
But the assembly language source is not set up such that it creates
relocatible code. Such code uses PC-relative mode addressing (e.g., jmp
$1 o(pc. The assembler source contains only instructions with absolute addressing (e.g., jsr_exit).
63
2.
A MIGA C FOR ADVANCED PROGRAMMERS
During assembly, these absolute instructions are converted to relative

instructions. The directives public and global playa role in this.
They enter the specified labels in the symbol table which contains all of
the global labels in the module.
In the first pass the assembler searches the program for such labels. At
the same time it checks to see if a label in the symbol table corresponds
to something in the source file. This is the case with rna in, for
example. First this label is placed in the symbol table (public
_main) and then defined L main: ). global_i,2 places the label_i
in the symbol table. The memory storage for these variables is not
made available until the program is linked. The bss directive also
reserves storage for a variable. The label associated with the storage is
not placed in the symbol table, however. Therefore bss is well suited
for declaring static variables within functions.
When an instruction now accesses a memory location specified by a
label, the symbol table is searched to see if this label is defmed in the
same module. If this is the case, the offset from the current position to
the label can be calculated for the instruction.
When calculating the offset, it is important to know whether the label
describes a memory location in the code segment or the data segment.
From the first pass through the program the assembler knows how
many bytes the program contains. If an instruction accesses a label in
the data segment, you must take into account the fact that the data
segment is appended to the code segment.
The code and data segments are two distinct areas. If the directive dseg
stands before an instruction in the assemble language source, then all
subsequent data is written into the data segment until a cseg occurs,
marking the start of the code segment.
In the following listing you can clearly see that the code and data
segments are two separate areas. Whenever the directive dseg or cseg
appears, the address in the second column changes (the code segment!
data segment program counter).
Aztec 68000 Assembler 3.4a 1-25-87
1 0000:
dseg
2 0000:
ds a
3 0000:
public str
4 0000:
str:5 0000: xxxx xxxx
dc.l .1+0
6 0004:
cseg
7 0000:
dseg
8 0004:
.1
9 0004: 4861 6c6c 6fOO
dc.b 72,97,108,108, 111,0
10 OOOa:
cseg
11 0000:
global i,2
public -main
12 0000:
main:
13 0000:
- link a5,i.3
14 0000: 4e55 0000
15 0004:
movem.l .4,-(sp)
64
2.3 THE ASSEMBLER
ABACUS
16
17
18
19
20
21
22
23
24
25
26
27
0004:
0008:
OOOc:
OOOe:
OOOc:
OOOe:
0010:
0010:
0010:
OOOc:
OOOc:
OOOa:
52ad xxxx
4eba xxxx
add.l #1, str

jsr _exit-
.5
movem.l (sp)+,.4
unlk as
rts
.3 equ 0
.4 reg
public exit
public :begin
4e5d
4e75
0000 0000
0000
dseg
end
This listing is obtained by using the -1 option of the assembler. Here

you can see the assembler changing absolute addressed commands into
PC-relative instructions. The machine code for jmp exi t would be
$4eb9. In the listing (line 17), the code for jmp exit is $4eba. This
stands for j s r xxxx(pc). The offset of the exi t routine cannot
been determined yet because this routine is not defmed in this module.
The fIrst pass of the assembler creates this listing. This can be seen by
the fact that line 5 is full of x's. This happens because in the fIrst pass
it isn't known where the string resides in the data segment Similarly in
line 16 (add.1 #1, str). Since the start of the data segment has not
yet been determined. the offset to the str variable cannot be determined.
The object fIle which the compiler creates consists of executable
machine code whenever possible; a header which specifies the size of
the program; the number of labels in the symbol table, etc.; and the
symbol table itself:
II
a- .g:ra8a
$
$.1
$
$.'
$.8
link
add.!
jar
$.c
$.c
unlk
rt.
$04
..
.',
11,
-exit
.s
,. $0 (as)
Clta-aa"lIIa:Dt
,
fl.
fl.
$0.
$
$Ie
$Oe
dc.1
"Sello"
n.
...
lJab.~-tlbl.
Cod..-for-Coda-Se9llent
Acid.r.all 1117
"_i"
RESOLVED
Code-for-Cocle-SelJllleDt.
Ackireaa 0
I+-
"-mun"
RESOLVED
Code-for-Extern
Address 1111
"-exit"
UllRBSOLVED
code- for-Extern
Addrel!llil 1177
Figure 2.3
".bellin"
UNRESOLVED
6S
2.
Let's look at the listing and the assembly language source. When you
compare the two, it appears that the assembler is not assembling the
instructions movem.l .4-(sp) (line 15) and movem.l -(sp),.4 (line
19).
The assembler "optimized away" these instructions because they are

clearly senseless. The instructions are useful only when there are registers to save. This is only the case when register variables are used,
however.
The assembler also removes branches to the next command:
brs .7
.7 ..
This construction is also useless since the instruction at label .7 would

execute without the branch command. Also, all j s r instructions are
replaced with bsr, ifpossible, which executes faster.
Other assembler options exist in addition to the -n option:
Assembler
options
-0 filename
The object code is written to the specified file.
-Idirectory
This option tells the compiler which directory to

search for include files.
-L
This option produces a listing of the assembly

source. The listing is created in the first pass of the
assembler, in the file <filename>.lst.
-N
This option disables object code optimization.
-Snurnber
This option creates a squeeze table with numeric

entries. An entry is one byte long. The assembler
allocates a default of 1000 bytes (entries) if the
number argument is omitted. The squeeze table is
required for the optimization of the object code.
-v
This (Verbose) option prints memory statistics.
-ZAP
This option deletes the source file after assembly.
-c
This option creates "large code" (see Section 2.3).
-D
This option creates "large data."
-Ename[=value]
This option assigns the argument n a me to

value. This assignment is the equivalent of the
name EQU value. The val ue argument is
optional, and if not specified, the symbol is
assigned the value 1.
66
2.3
ABACUS
2.3
THE LINKER
The linker
The linker resolves unresolved references in the individual modules. To
do this, the object files (modules) are usually linked to the linker
libraries. The call is as follows:
In [>output_filel object_file.o [options]
Let's look at the last example program, which called the _ exi t
routine. This routine wasn't defined in the module in which it was
called, however. The _ exi t definition takes place in the linker library
c.lib. This must be linked to the object file which contains the
exit call.
The linker frrst reads the object file. It looks at the symbol table. From
the symbol table the compiler learns that _ exi t is an unresolved
external reference. The linker notes this in a table which lists all of the
unre-solved references. If one of these references is later resolved in a
module, then this is noted in the table.
Now the linker reads each module in order. Up to now, the commands
which called external routines, for example, were not complete. The hex
code for the PC-relative j s r instruction exists, but the offset of the
routine to jump to is unknown. These offsets can be calculated in a
second pass and then entered at the appropriate location.
Let's look at the following assembler source:
dseg
ds a
public _Str
Str:
- dc.l .1+0
;allocate
cseg
dseg
.1
dc.b 72,97,108108, Ill, a
cseg
global i,2
public :=main
main:
- link
as,#.3
movem.l .4,-(sp)
add.l I, str
jsr _exit
.5
movem.l (sp)+,.4
unlk as
rts
.3 equ
.4 reg
;data segment
;reserve a bytes
;global variable
memory with address of the string
; code segment
;data segment
; "Hello"
;code segment
;start here
;str++;
a
67
2.
public exit
public :-begin
dseg
end
In line 17 you see the instruction j s r _ exi t. The assembler generated

the hex code $4eba xxxx for this instruction. jsr exit actually
occupies four bytes. But the assembler uses only the first two bytesnamely those used for the code for the PC-relative j s r instruction
itself ($4eba). The remaining two bytes, which specify the offset for
this instruction, remain initialized. This initialization is handled by the
linker, which can calculate all of the necessary offsets in a second pass
and enter them in the appropriate locations.
Now we come to a question you are probably asking yourself: When
you link an object file with a linker library, are all of the functions of
the library linked in or just the functions actually used by the object
file? Here the Aztec linker is very economical. It appends to your program only the library routines actually used, when linking two object
files, it combines all of the functions.
But here again we have a question: Since you can link c.lib to a
module with In Ofile.o -lram:c as well as In Ofile.o
ram:c.lib, and since c.lib is just an object file which contains
many functions, how are only the needed routines from c.lib added to
the Ofile and not the entire c.lib?
The answer is that the linker libraries are not object files in the sense of
the files which are created by the assembler. The linker libraries consist
of a combination of several normal object files. These libraries are organized in a somewhat different manner. While the symbol table is placed
at the end of the file in normal object files, the table moves to the
beginning of a linker library. Then comes the executable machine code.
If the linker is to link a library to an object file, which is almost
always the case, it recognizes the library by the special code at the
beginning of the file. The linker then reads the library symbol table,
which it uses to immediately resolve external references in the object
file. It also notes which library routines are used.
When the linker has read the symbol table, it then comes to the
executable code of the library. It removes the functions that it noted and
appends them to the normal object flIes. In the second pass it calculates
all of the necessary offsets and sets them.
The linker is not only responsible for resolving unresolved external
references. It also links startup code to the object file. This is accomplished by the assembler directive public .begin..begin is stored
in the list of unresolved references in the first pass and then appended to
the program later, when linking with c.lib.
68
ABACUS
2.3 THE LINKER
In addition, the machine language instruction jmp .begin (in PCrelative form) is created for each module which contains the directive
public .begin. This is necessary only for the first object file to be
linked. This command must be the first one in the program. In the
other modules this command just takes up space (four bytes). Therefore
when you use multiple modules you should make sure that the label
.begin appears only in the first module. You should compile all other
modules with the +b option.
Now we come to the startup code. First it opens the DOS and
MathFFP libraries. Then the startup code determines whether the program was started from the Workbench or the OJ.
When starting from the Workbench, it must first wait for the startup
message. If this message passes arguments, they are interpreted as a
Lock structure and designate the Locked directory as the current
directory. The Tooltype window opens, if necessary, and defaults to
standard input/output.
When starting from the CLI, the startup code ensures that the program
gets the number of arguments and the address of the argument array.
These are made available by the OOS through the CIS (Command Input
String). The CLI's startup code prepares the parameters so that they can
be passed through main(argc, argyl.
After processing the parameters string, the main routine of the C
program is called. Somewhere in the startup code, which consists partIy

of machine language and partly of C code, the call main(argc,argv)
appears. Directly after this is the C call_exi t(O);. This _exi to
ensures that the OOS and MathFFP libraries are closed, that the
memory space allocated in connection with preparing the command
parameters is released and the command exitO is called. This
_ exi to is precisely the routine which your program can call to terminate the program prematurely.
But let's return to the linker libraries. Since the startup code must
always be linked to a C program, c.lib must always be part of your
program. There are other libraries, however, such as m.l ib. This
library must always be linked to an object file if floating-point
arithmetic is used. It isn't enough to simply link this library to the
object file. You must also compile the program with the +ff compiler
option. To use double precision floating point arithmetic, you must
compile the program with the +fi option and link with the mx.lib
library. When using the 68881 floating-point coprocessor you must use
mB.lib and compile the program with +fB.
But even c.lib and m.lib are not the only libraries in the compiler
package. If you have ever looked around on the disks, you would have
found a c32.lib, a cl.lib and a cl32.lib. The other libraries also
appears in these forms: ml.lib, ml32.lib, etc.
69
The linker libraries with a 1 in their names must be linked to the object
fIles when you want to use the "large data" and "large code" memory
model.
2.3.1
"Large data" and "small data"

Now we come to the differences between "large data" and "small data".
As the name implies, these two memory models refer to the data segment. In the "small data" model, the data segment can be only 64K in
size. This is because access to the data is performed with an address
register which points to the middle of the data area and through positive
and negative offsets, between 0 and 32768, can only access 64K bytes.
Here the PC-relative addressing with 16-bit offset comes into play.
With the "large data" model, data are accessed by their absolute
addresses. But since it is unknown from the beginning where the program resides in memory, all of the instructions which access absolute
memory locations in the data segment must be corrected. This is the
reason that a "large data" program takes longer before it starts to run.
Absolute address access is also slower than PC-relative access to
memory locations. The advantage of the "large model" is that the data
segment can theoretically occupy the entire free memory.
2.3.2
"Large code" and "small code"

The models "large code" and "small code" refer to the code segment of
the program. It doesn't involve the executable code so much as the data
which are also stored in the code segmenL
With the "large code" model data can reside in the entire code segment,
even scattered throughout memory. The "small code" model, on the
other hand, allows data to reside only in 32K of the program counter.
If the "small code" program is larger, an address table is created with
absolute addresses through which the outlying memory areas can be
accessed indirectly.
Also, a jump table is used for branches outside this 32K. Correcting
this jump and address table normally takes less time then correcting the
instructions in a "large code" program, which uses absolute addresses
exclusively. With the "small code" model an address register must be
set to the middle point of the data segment, while "large code" accesses
the data segment absolutely. The absolute 32-bit addressing is also
slower than the 16-bit offset addressing.
70
2.3
ABACUS
THE LINKER
If you want to use either "large code" or "large data," you must compile
and assemble the program with the +C or +D option. In both cases you
must link in a library with an 1 in the name (such as cl.lib or
c132.lib).
Now you may be wondering what the "32" means in names like
c32.lib and cl32.lib. This 32 stands for 32-bit integers. The int
variables used in these libraries are 32 bits in length instead of 16.
This is especially important for compatibility between Lattice and
Aztec programs. Lattice programs don't use 16-bit int variables. The
compiler extends variables and constants to 32 bits. This extension is
not performed by the Aztec compiler when using an int variable as a
function parameter, 16 bits are passed instead of 32. The Amiga library
routines require 32 bits, however, and the arguments may be interpreted
incorrectly, or the system may even crash.
The Aztec compiler can emulate the Lattice compiler when you extend
all int variables and constants to 32 bits. This is accomplished with
the +L compiler option.
The C support functions in c32.lib, cl32.lib, etc., were compiled
with this +L option and are useful when compiling Lattice source codes
under Aztec.
Here is a list of the linker options:

Linker
options
~O filename
The executable program is stored as filename. If

this option is omitted, the executable program
retains the object file's name without the extension
of. o.
-Lfilename
This option instructs the linker library to be linked

to the object file.
-F filename
The linker reads the arguments from the ASCII file

specified here. This is useful if you want to pass a
large number of arguments to the linker.
-T
This option stores the linker symbol table as an

ASCII file. The name of this ASCII file is the
object filename with an extension of . sym. This
ASCII file contains information about the file-in
which the segment code is located. This file
contains "hunks" of executable instructions, system
variables and program variables as well. These
hunks are enclosed in the labels Hx org and
_Hx_end, whereby x represents the gIven hunk
number. In programs which consist of only one
module, the program variables are stored between
71
2. C
CoMPILER OPERATIONS
H2 org and
H2 end. The program begins
at_HO_org. -
-w
This option stores information for the d b

debugger. This information includes the labels and
their offsets from the start of the program (see -T
above).
-v
This option prints hunk statistics.
+O[i]
This option instructs the linker to write the subsequent object module to the next or specified code
segmenL
+C[cdb]
This option instructs the linker to load the code

segment (+Cc), the data segment (+Cd), or both
segments (+Cb) into chip memory.
+F [cdb]
This option instructs the linker to load, the code

segment (+Fc), data segment (+Fd) or both (+Fb)
into fast memory.
+A
This option instructs the linker to begin each

object module on a longword address.
-c.
72
-Q
These options instruct the linker to store information for the source level debugger (not currently
implemented).
+Q
This option instructs the linker to suppress the

procedures of the object files being linked.
-M
This option prevents user-written linker library

functions from overwriting the linker library routine (see stack overflow).
+L
This option instructs the linker to assume that the

following object modules are "true" Amiga linker
libraries. Attempting to link c32.1ib under this
option was unsuccessful.
2.3 THE LINKER
ABACUS
+s
+ss
+sss
The +s option instructs the linker to write the

executable code of each object module and the
linker library routines used by the modules into
separate hunks. The +ss option instructs the linker
to write the code into one hunk, providing the code
does not exceed 8K in length (after 8K a new hunk
is used). The +sss option instructs the linker to
use a hunk for each module. You should note that
the linker libraries are combinations of multiple
modules and so each routine used is written in a
separate hunk. Without specifying +s, +ss and
+sss the entire code is written into one hunk.
73
2.
2.4
The debugger
Theoretically your C program should run correctly. But what do you do
if the program doesn't run, and you just can't explain why the program
doesn't do what you had wanted it to?
The debugger can help you find errors. It monitors executable programs
for runtime errors. There are two basic types of errors:
1.
Those which return incorrect results
2.
Those which cause a Guru Meditation
With the first type of error, you enter the debug mode of the included
debugger (bin directory of the third disk) with the al command. This
command waits until a program has been started from the Workbench or
the CLI.
Mter you have started the program the debugger responds immediately.
It attempts to load a symbol table (Loading syms ...). This symbol
table is similar to that which you examined for the compiler. Here the
table must be created by the linker (-w option), which contains all of
the labels used as well as their offsets from the start of the program.
Therefore in the development phase you should link your program with
-We The final and error-free version of the program should not be linked
with the -W option, especially if you intend to publish the program in
any form. This additional information is very useful for others who
want to know how your program works.
Mter the symbol table loads, the debugger displays all of the CPU
registers. In addition to the program counter (pc), status register (Sr)
and condition flags (x, n, Z, v, c), all of the data and address registers are
also displayed. In addition, the debugger disassembles the first instruction of the C program, e.g.:
' __HO_org jrnp .begin'
As mentioned previously, this instruction (jmp .begin) must be the

frrst one in every C program, since it handles such tasks as reading the
command parameters.
Now you can disassemble additional instructions of the program by

selecting u (unassemble). You may see instructions like main link
a5,4t-6, etc. You can thank the symbol table created by the linker and
loaded by the debugger that you see the word _ mai n instead of a more
or less meaningless number.
74
2.4
ABACUS
THE
DEBUGGER
But being able to see the program doesn't help much when looking for
errors. Therefore the debugger gives you something called single-step
mode. Each time you select s, the current or previously displayed
instruction executes. The registers are then redisplayed and the next
command is disassembled.
It's very tedious to have to step through the startup code at the beginning of the program. The startup code cannot be omitted, however,
since it is here that the command parameters are processed and important libraries are opened. Therefore you should set a breakpoint at the
routine _main (bs_main (breakpoint set. The _main routine is
called at the end of .begin.
After setting the breakpoint you can start the program by selecting g
(go). The program (startup code) executes until a breakpoint is reached.
Since this occurs at _main, you can proceed through the actual
program in the single-step mode, whereby the command parameters
fetched by the startup code are also taken into account.
If during single-stepping you come across a call to printf, for
example, you don't have to worry, you don't have to go step by step
through this long routine. The t (trace) command executes this routine
and returns to the next instruction in the program.
If you have a copy of the assembler listing augmented by the C
statements as comments (-T compiler option). you can go through the
program step by step, and when you discover an error, you know
exactly which C statement is at fault.
Now to debugging a program which gives you a Finish ALL disk
acti vi ty ... requester. One possibility would be to debug the
program in the manner just described until you can find the error. Or
you can debug the task in which the error occurs.
Imagine the following example: A program leads to a division by O.
This displays the requester just mentioned, and you can return to normal
operation only by resetting the computer.
But when you start the debugger (through a new eLI) or are in an
already running debugger, you may be able to save the system before a
total crash, and perhaps save important data (by jumping to a routine
which stores the data, for example).
To do this, you must use the ap command (debug post mortem). With
this command you can debug a task after the Finish... requester. You
are asked for the number of the task which you want to debug. It would
probably be useful to have previously used the at command to view
the current system tasks. There you might see a line like:
3:
(OOOOa708)
Wait
Initial eLI
<div>
75
2.
If you want to debug the task in which the program di v is running,

you must enter the number 3 when you are asked for the task to debug.
ap suggests which task to debug by noting which task just caused an
error. After entering the number the debugger goes to the location in the
selected task (program) in which the error occurred You can then debug
the program as before.
Two other commands can be useful when debugging: de and dg. dg
displays the contents of all global variables (display globals), while de
specifies at which address a given label is found (display code).
Back to task debugging. In addition to the a p method you can also
debug any task in the system, not just those which cause a serious error
(you can debug such tasks with a p). In addition there is the a s
command. Here too, you are asked for the task to be debugged and the
debugger tries to load the symbol table for the task. The program must
be on the disk in the current directory.
But the first disassembled line of the task is:
'fc08b8
rts'
This rts instruction doesn't belong to the current task itself. It

belongs to an essential part of the task handling mechanism. This
instruction is located in the section of the operating system which
ensures that only one task is running and all other tasks are passive
until one of these tasks is activated by a message from the keyboard,
mouse, disk, etc.
How do you access the task? You just have to enter the g (go)
command or the s (step) command, and then "click" the task, move or
change the size of a window managed by this task.
There are also tasks which wait for a keypress or something similar.
The important thing about task debugging is to make sure that a
"message" is sent to the task after s. You can execute the task without
hindrance from the debugger with ar (resume).
To get data about the current task, use the ai (information) command
to display information about the task or its task structure. In addition to
the status, you will also see the Trap Allocation and Trap Able mask,
whether the task is part of a process (Process Info) or a eLI structure.
The contents of the most important structure elements are displayed,
whereby the names of the elements can be shortened somewhat ("Rslt"
for "Result").
76
2.S
ABACUS
2.5
TRICKS AND TIPS
Tips and tricks

Now that we have explained how the compiler, assembler and linker are
used, how they work, and how to find errors, we want to give you a few
programming tips to get you on your way.
2.5.1
Accessing absolute memory locations

You have probably wondered at one time or another how to access a
specific memory location in C. Remember that pointers store the starting addresses of variables, structures, unions, bit fields, etc. If you can
increment pointers (actually the address to which a pointer points), why
can't you assign an absolute value to a pointer? The only thing to note
is the type of variable to which a pointer points. For example, if you
want to read a hardware register, the pointer you use should point to
type uword, since all hardware registers occupy two bytes. This allows
you to read the entire word
An example which can explain access to absolute memory locations is
accessing a elAA register, which occupies one byte. The address of this
register is $bfeOOl. With this register you can read the joystick fire
buttons and the left mouse button. Bit 7 (2"7 = Ox40) of this register
indicates whether or not the mouse button is being pressed. Note that
this register is active low, meaning that if bit 7 is set, then the mouse
button is not being pressed, while if it is cleared, the button is being
pressed.
/* mouse.c */
#define Register_Address OxbfeOOl;
char *CIAA_Regl;
/* access to CIAA register 1 */
main ()
{
ClAA_Regl = (char *)Register_Address;

printf (" <mouse button> running Waiting for left
mouse button !\n");
while ((*CIAA_Regl & Ox40) == Ox40);
printf ("Button pressed!! !\n");
77
2.
2.5.2
Dynamic arrays
The second programming example concerns the creation of dynamic
arrays. Let's say you want to store an arbitrary number of values. The
number of values to be stored is not detennined until the program is run
(by the user, for example).
Nonnally the user is given some restrictions on the use of the program.
Perhaps the maximum number of elements in a given array is limited
to 100. This is not particularly elegant, however. It would be better if
the maximum number of elements were limited only by available
memory. Here again pointers can help us.
First derme a pointer to the elements which make up the array:
int *array;
Then check to see how much memory is available with AvailMemO.

It depends on the application whether you want this memory in chip
memory (lower 512K) or fast memory.
size = AvailMem(MEMF_CHIP);
Now calculate how many array elements can be stored in this space:
NumElements = size/sizeof(*array);
You can now ask the user how many elements are used (elements).
Then you just have to allocate the memory:
array
AllocMem(sizeof(*array) * elements, MEMF_CHIP);
Now you can access the individual array elements with array [x] and
they are stored in the allocated memory.
78
2.S TRICKS AND TIPS
ABACUS
2.5.3
Function tables
Function tables are arrays which store the addresses of individual routines. Such an array is defined as follows:
void (*functions[4]) ();
This is an array of pointers to functions with the return value void.

After the definition, you just have to assign the address of the individual
functions to the array elements:
/*****************************************************/
/*
Table.c
*/
/*
(c) Bruno Jennrich
*/
/*
*/
/* This program shows you how to
*/
/* easily program jump tables.
*/
/*
*/
/* The functions are stored in an array
*/
/* which can be accessed with an index variable.
*/
/*****************************************************/
void (*functions [4]) (); /* Functions is a four-element */

/* array of pointers to functions */
/* The functions */
void FunctO ()
{
printf ("Function 0 !\n");

void Funct 1 ()
{

void Funct2 ()
void Funct3 ()
{
79
main (argc, argv)

int
argc; /* (argument count) */
char **argv; /* (argument values) */
{
int i;
functions[O] = FunctO;
functions [1] = Funct1;
functions [2] = Funct2;
functions [3] = Funct3i
i f (argc != 2)
{
printf (n Call: < TABLE Number_of_the_function >\n");

exit (0);
i = atoi (argv[l]);
i f i>-=O) && (i<=3
(*functions [i]) () I
else printf ("Number of the function must be between 0\
- and 3\nn);
80
3.
Intuition
and
C
3.
ABACUS
3.
INTUITION AND
Intuition and C
This chapter concerns programming the Amiga operating system,
especially system programming using Intuition. What do we mean by
system programming? Here are some examples:
How do I tell my Amiga operating system that I want to use a

certain graphic mode?
How can I tell the system that the user has made some inputs (with
the mouse or with the keyboard)?
These are two questions which come under the topic "system programming." There are certainly others, but this chapter answers these questions precisely.
Intuition is a type of communication (user interface) between you, the
user and the Amiga. By communication we mean any realtime exchange
of information. The keyboard and mouse are the most obvious input
devices-both are included with the computer, but there are certainly
other devices. Output occurs mainly to the video screen. There are also
printers and disks, but Intuition has no control over these devices.
Screens
What does Intuition offer? Let's look at what Intuition can do with
output. When booting with the Workbench disk you immediately see
two output displays offered by Intuition. The first and most important
is the screen, which forms the foundation of every output. An
important example is the Workbench screen, which appears when
booting. It accompanies the user through everything he does, and acts
as the basic output device for many programs.
Windows
Output rarely appears only on the screens, since they're difficult to

implement. To make things easier, you can use windows. the second
output level of Intuition.
Gadgets
Now we come to user input. Gadgets are the simplest input media.
They are found in both screens and windows. The simplest gadgets react
to a single mouse click. Naturally there are more complex types into
which you can enter text, but these involve keyboard access. First we'll
concentrate on mouse input, then look at keyboard input.
Menus were intended for elementary program-specific functions. They
are also controlled by the mouse, but the emphasis is on text selection,
while the gadgets are generally represented with graphics. There are,
however, exceptions to these rules.
83
3.
INTUITION AND
3.1
AMIGA
Windows
As we mentioned in the introduction to this chapter, windows are the
most important input/output medium.
We wanted to present a detailed description of a complete application

you could program in C, then continue to improve upon it throughout
this book. We decided on creating a text editor for C programs.
Therefore this section supplies you with information about
programming fundamental I/O, and selecting what to use from the
many options. You will also see how much Intuition offers to the
programmer.
Mter listing the various parameters available, we'll see how to configure one of the windows managed by your program. As a result you will
learn how to make use of the various Intuition commands.
In addition, we'll write a general window routine which you can use in
your own programs with only a few changes. You should prepare a disk
for this data, since you'll be using the modules created throughout the
course of this book.
3.1.1
Window parameters and selecting them

You need a number of parameters for creating a window. These parameters can control every conceivable aspect of the window. To lead into
this, remember the familiar properties found in the average Workbench
window:
External
features
Sizes and positions of every window can be adjusted. You can set the
following values: X position (upper left comer). Y position (upper left
comer), width and height.
H you want to change the size of a window, it is desirable to set

maximum and minimum window sizes. When working with the
Workbench you can generally resize the directory windows as much as
you want. However, requesters displayed on the WorkBench screen can
only be made smaller. As the programmer, you can set the maximum
and minimum sizes with MaxWidth. MaxHeight, MinWidth and
MinHeight.
In addition, you can set the colors used to draw the window border and
the gadgets found within the window. based upon the screen colors. The
84
ABACUS
3.1
WINDOWS
two colors are designated DetailPen and BlockPen. Each window

also has a title line that contains text (again, programmer controlled).
For those not satisfied with the system gadgets offered by and in part
managed by Intuition, you can add more gadgets to a window. This is
how the scroll bars in the directory windows are managed, for example,
because they are not defined as system gadgets. As a programmer you
can construct any gadget you can think of.
The last external property you can specify about a window you have
created is that system gadgets can be included or omitted, depending on
what makes sense for the given application.
Internal
features
Now that we've discussed the external parameters of windows, let's

look at those which you might not recognize by just looking at a
window.
One seldom used option is to redefine the appearance of the checkmark
used to check menu items. You've probably heard of programming
Intuition to manage a custom screen for you. Let's assume that you
have a custom screen and you want to put your new window in it A
flag tells the operating system that the window is to appear on your
screen and not on the Workbench screen. In addition, Intuition needs to
know on which custom screen the window should appear, since it is
possible to have more than one custom screen open at a time.
Therefore, if you want to open a window on a custom screen, designate
this with a flag and store the pointer to your screen in the window
structure.
Programinternal
features
The range of applications for which windows are intended is so broad

that more than one type is needed. Therefore, various methods have
been prepared from which programmers can choose one or more types.
Refreshing
window
displays
Window contents, which may include text as well as graphics, should

be preserved independent of other windows. This should even be the
case when other windows overlay it and cover parts of it If Intuition
were to simply store the hidden areas temporarily (a process called
buffering), it would involve a great deal of memory. In some cases
buffering may not be necessary. If you know what is in the window and
it is relatively easy to reconstruct, then you can do all this in the
program. The best known examples are the directory windows in the
Workbench. There the system simply redraws the icons after other windows cover them. Workbench restores the graphics without buffering
from Intuition. This takes up much less memory when multiple
windows overlap each other.
You must consider the uses for a window when developiong a program.
There are three basic application needs:
1. Programs where the contents of the windows is completely known
and whose data are stored in a different form. An example of this is
8S
3.
INTUITION AND
AMIGA
word processing, where the text appears as graphic infonnation on

the screen, and is also stored in the text buffer. For such applications it's best to perfonn the reconstruction yourself. This takes as
little memory as possible away from other programs running at the
same time.
2. Programs unable to reconstruct a damaged portion of a window in a
reasonable amount of time. Here Intuition handles the storage of
the overlapped areas to save work on the program's part. This problem often occurs with graphic programs, where the time spent
redrawing the image far outweighs the time involved in buffering
it. You must strike a careful balance between memory usage and
time. Memory usage should be carefully planned on any computer,
even in a computer with 1 meg of RAM.
3. Programs which require more graphic display than is actually
displayable. For this category a completely new method has been
developed. The program displays its graphic in a completely inde
pendent bit-map, and Intuition uses the desired portion from this
bit-map for the window. Therefore the graphic memory supplies
reference points for reconstructing the window.
Let's summarize the three modes in a table. The table contains the
names of the flags which must be set for the corresponding mode. Note
that each mode is mutually exclusive of the others.
Table 3.1:
Window
refresh
Borderless
windows
Refresh
Simple
Flag
Supported
Complete
SMART REFRESH
SUPER_BITMAP
SIMPLE REFRESH
Memory usage
None, since Intuition doesn't
bother with the window
Corresponds to hidden areas
The entire graphic in the
window is buffered
Your window doesn't always have to appear with the familiar border,
title line, gadgets and border lines.
Viewed at its lowest level, a window is just an area on the screen that
separates its contents from other areas present. The border which
Intuition draws merely acts as a visible indicator of the border. If all the
windows on the Workbench were drawn without this border, it would be
very confusing to the user.
However, borderless windows can serve a practical purpose in many
applications. You can prevent the border from being drawn when setting
up a new window by setting the BORDERLES S flag. Just be sure that a
borderless window won't confuse the user.
86
ABACUS
BACKDROP
windows
3.1 WINDOWS
The title of this paragraph refers to a window which no longer has all
the properties of a normal window. It will always be the window
farthest back, and can be overlapped by any other windows. The user
cannot bring this window to the foreground. This window has no front
or back gadgets-only the close gadget is allowed.
You can use this BACKDROP window type as the basis for any
application. This BACKDROP window can hold the background for an
animation program, or can display program status information, or even
hold gadgets for accessing program tools. The BACKDROP property
supports these applications, since a new window cannot be placed
behind it. Intuition manages the BACKDROP window, so further
programming is unnecessary. If you want to use this window type, you
must set the BACKDROP flag.
An interesting combination would be a BORDERLESS-BACKDROP
window, which would be very useful for the examples just mentioned.
But other types can be combined as well.
An extra
Earlier we described how to create a window which had no border. Here
window border we describe the opposite situation. You know that a normal window is
just a rectangular area with a visual border. A few problems result from
this: If you write a drawing program which uses a normal window, you
can easily overwrite the title bar, the gadgets or the border. The window
only prevents you from drawing outside of the window's border-the
rest of the window is open game.
GIMMEZERO
ZERO
How do you cure this problem? The simplest way; avoid drawing
outside the window's inner surface. But let's be realistic, users don't
blindly accept the universal programmer's instruction, "Don't do it."
The developers of Intuition created GIMMEZEROZERO mode to solve
this problem. This mode changes some window settings. The window
consists of two fields: One field contains all of the gadgets, the title bar
and the border. The other field is the surface in which the user can draw
without crossing over into gadgets or border art. The name
GIMMEZEROZERO comes from the fact that this inner window always
has the upper left comer coordinates 0,0, regardless of the size and location of the window.
This mode requires more memory and processing time. Also, the portion in which the drawing program can operate has different dimensions
from those specified in the window definition. This window type needs
two additional variables which you read to find out the size of the actual
drawing surface. These variables are called GZ Z He i g h t and
G Z Z wid t h . This area lies within the following values:
BorderLeft, BorderTop, BorderRight and BorderLeft.
Since the mouse coordinates of a window always refer to the entire
window area, a GIMMEZEROZERO window can cause certain problems.
Intuition also has relative mouse coordinates to the GIMMEZEROZERO
window: GZZMouseX and GZZMouseY.
87
3.
INTUITION AND
Scrolling data
AMIGA
Intuition offers the programmer the ability to manage a much larger

graphic surface than can be displayed on the screen. SUPER_BITMAP
mode displays a window type where you can display part of a bit-map
of almost any size in a window. If you want to see a different section,
the system displays that section. This is useful when scrolling
graphics.
If you want to use this window type, set the SUPER_BITMAP flag in a
program. It's possible to combine this with GIMMEZEROZERO or

BACKDROP.
System
gadgets
Gadgets are fields which can be activated by moving the mouse pointer
onto the gadget and clicking the left mouse button. Intuition offers four
gadgets as a default set. These four system gadgets support functions
which cannot normally be handled by the program. System gadgets also
have a variety of properties. Their positions and appearance are preset;
only the color can be changed. They can always be recognized by the
user in any program, and the user doesn't have to guess what they
mean. Let's look at each of them.
Close gadget
The close gadget closes a window. You are familiar with the procedure
from the Workbench: You click on the gadget in the upper left corner of
the window to close the window. The close gadget appears as a box
with a large point in its center.
Drag bar
The drag bar appears as three heavy lines in the title bar of a window.
You can use it to change the position of the window. Move the mouse
pointer onto the drag bar and press and hold the left mouse button. As
long as you hold down the mouse button the window moves with the
mouse pointer.
Front and
back gadgets
These gadgets always come as a pair. They change the priority of overlapping windows. The front and back gadgets are found in the upper
right comer. The light rectangular surface is the important part: It represents the current window's new position.
Sizing gadget
If a program allows you to change the size of a window. the window

has a sizing gadget in its lower right corner. It depicts two different size
rectangles. Click on it and press and hold the left mouse button. Now
moving the mouse changes the window's size. The new size can only
be set within the predefmed minimum and maximum sizes.
Since the sizing gadget enlarges the border of the window, the area in
which the Intuition gadgets are located. the programmer must also specify where this border is taken from the actual output window. The right
border or lower border are possibilities, depending on whether you want
to have more rows or more columns in the window. It is also possible
to reserve both areas for Intuition.
88
3.1
ABACUS
WINDOWS
It should be noted that Intuition receives a message from the close
gadget but that Intuition doesn't close that window. The reaction to the
close gadget depends on the program controlling the window. Just as
the program receives a message when the user wants to close the window, it also receives a message when the size of the window has been
changed, since this must be taken into account by the program.
3.1.2
How Intuition manages windows

You now know about the basic window parameters. Before you try out
these parameters, you should know the principles used to access
Intuition commands. You might also be interested in internal data
transmission in the Amiga.
3.1.2.1
Accessing the Intuition library

The entire Amiga operating system is made up of system routines
called libraries. A library consists of various routines, which accept
parameters and sometimes return values.
Not all of the libraries are available when the user turns on the
computer. If they are already loaded, a program must still indicate that it
wants to use a certain library, since access requires the use of a pointer.
This is an EXEC command (EXEC is the library which is active
immediately after powering up) which, like all EXEC commands, is
available to every program:
library = OpenLibrary(LibraryName, Version)
DO
Al
DO
You pass the library's function (in this case intuition.library)

and version number. Newer library versions may contain functions not
found in older libraries. You can prevent an older version of a library
from being loaded by specifying the version number.
We will use the basic commands, however, since there are already
windows in Intuition. The first lines look like this:
/* OpenIntuition.c, section 3.1.2.l.A */
struct IntuitionBase *IntuitionBase;
void
*OpenLibrary();
if (! (IntuitionBase = (struct IntuitionBase *)
OpenLibrary("intuition.library", OL)
89
3.
INTUITION AND
AMIGA
printf ("Intuition library not found!\n");

exit (FALSE);
}
How it works
The first structure definition introduces a pointer which points to the

Intuition function list. The compiler requires this pointer since all
Intuition access occurs relative to this pointer.
An additional untyped pointer also appears for the OpenLibrary
function. You may want to open more than the Intuition library, in
which case type conversion may cause problems.
The if test first calls the function itself. Then its return value (the
pointer to the Intuition functions) is tested for validity. This value must
be something other than null, in which case it represents the memory
address at which the library begins. A null value indicates an error. If
the if condition is true, the program displays a message in the
AmigaDOS window and exits.
Now you need to know how to conclude library access. When you
actually obtain access (and only then) you can simply close the library
again with the EXEC function CloseLibraryO:
CloseLibrary(LibraryPointer)
-414
Al
In your application this would appear as:

CloseLibrary(IntuitionBase);
3.1.2.2
The NewWindow structure

Now you know how to open and close a library. Let's look at windows
again. All of the properties discussed at the beginning of this chapter
can be specified when you open or change a window later. A structure is
used to initialize the values for Intuition.
The NewWindow structure listed below contains all of the values
required by Intuition to configure a new window. The structure looks
like this:
/* NewWindow structure *1
struct NewWindow
{
Oxoo
Ox02
Ox04
Ox06
90
00
02
04
06
SHORT
SHORT
SHORT
SHORT
LeftEdge;
Top Edge;
Width;
Height;
3.1 WINDOWS
ABACUS
Ox08
Ox09
OxOa
OxOe
Ox12
Ox16
Ox1a
Ox1e
Ox22
Ox26
Ox28
Ox2a
Ox2c
Ox2e
Ox30
08
09
10
14
18
22
26
30
34
38
40
42
44
46
48
UBYTE DetailPen;
UBYTE BlockPen;
ULONG IDCMPFlags;
ULONG Flags;
struct Gadget *FirstGadget;
struct Image *CheckMark;
UBYTE *Title;
struct Screen *Screen;
struct BitMap *BitMap;
SHORT MinWidth;
SHORT MinHeight;
USHORT MaxWidth;
USHORT MaxHeight;
USHORT Type;
As you know, a structure consists of a many different variables. You
may wonder why numbers precede each variable name in the variable
declaration. These numbers indicate the offset of each variable from the
start of' the structure. This can be helpful when working with the
debugger, or if you are using assembly language. The numbers do not
coincide with the actual C structure declaration.
The ftrst column contains the offsets in hexadecimal notation, and the
second in decimal notation. The last number indicates the total length
of the structure.
All library communication occurs in predefmed functions to which you
pass variables (e.g., OpenLibrary and CloseLibrary). Intuition,
like most other libraries, uses structures of its own for various management pwposes.
Creating a
NewWindow
structure
Here we want to explain the setup of the Newwindow structure. In

addition, we'll show you two functions which will help you in writing
your own programs later.
The ftrst function comprises everything which must be opened, initial..
ized, allocated or organized at the start of the program. This includes
memory allocation, opening libraries, clearing ftelds, etc.
The second function comprises everything which must be closed at the
end of the program. If you don't close a library after you fmish with it,
the memory allocated for managing it won't be returned to the system.
The solution to these programs looks quite simple. one small problem
needs a solution: Something may go wrong during initialization. In
this case you should display an error message and terminate the
program. You cannot forget to close everything which has already been
opened, however. And here is where the problem occurs.
91
3.
INTUITION AND
AMIGA
If you simply jump to the close routine, it might try to close things
which weren't open in the first place. This guarantees a system crash.
So, let's write a close function which examines the pointers obtained
during initialization. This function should only close open items, and
leave everything else alone. Let's look at the two functions:
/***************************************/
/*
*/
/* 3.1.2.2.B
Open_All.c
/*
*/
/* Function: Open everything we need

/*
/*
*/
===================================
/* Author:
/*
Date:
Comment:
----------
/* Wgb
16.10.1987
----------
Intuition only
/*
*/
*/
*/
*/
*/
*/
*/
/***************************************/
*OpenLibrary () ;
OpenLibrary (n intuition .library", OL)
{
printf ("Intuition library not found! \n");

Close_All () ;
exit(FALSE);
}
/***************************************/
/*
*/
/*
3.1.2.2.C
Close_All.c
/*
*/
/* Function: Close everything opened

/*
/*
*/
===================================
/* Author:
/*
/* Wgb
Date:
----------
16.10.1987
/*
Comment:
----------
Intuition only
*/
*/
*/
*/
*/
*/
*/
/**************************************/
if (IntuitionBase)
How it works
92
The first function should already look familiar to you. The contents of
the second function are almost silly, but the important part is the i f
test The program would crash without this test if the Intuition library
could not be opened for some reason.
3.1
ABACUS
WINDOWS
Both functions look quite simple at the moment, but the basic form
will continually expand as we continue. Enter them carefully and save
them both on your program disk and in a subdirectory created specifically for storing function blocks. Later, when either of them is required,
we will not present the entire listing, just references to the existing part
and the additions required.
Now you can start writing your first program to open and close a
simple window. Let's specify the properties for your window:
Window size
The position and size of the window don't matter too much for your
test. Just be sure that the width and height are large enough to fit
gadgets and the window border. We set ours at position 160, 50 with
the dimensions 320, 150. Write the values in a Newwindow variable
called FirstNewWindow:
FirstNewWindow.LeftEdge
FirstNewWindow.TopEdge
FirstNewWindow.Width
FirstNewWindow.Height
= 160;
= 50;
= 320;
= 150;
Since you don't want to omit the sizing gadget, set the screen dimensions to the maximum. Remember, the minimum shouldn't be too
small.
= 160;
= 50;
= 320;
150;
Since you don't know how to manage screens on your own yet, use the
Workbench screen. The window type is set and you won't have to fmd a
pointer:
FirstNewWindow.Type
FirstNewWindow.Screen
= WBENCHSCREEN;
=
NULL;
For now, we'll skip menus and custom gadgets:

FirstNewWindow.FirstGadget = NULL;
FirstNewWindow.CheckMark
= NULL;
Title bar
Now let's get to the graphic details. First you have the window title.
You can pass a pointer to a string ending with a null, or you can pass a
null pointer to indicate an untitled window. If no system gadget is
permitted, then no title bar is displayed.
When using the Aztec compiler a type cast should precede the string
to avoid a warning from the compiler.
FirstNewWindow.Title
= (UBYTE *)"System prograrrming test";
93
3.
INTUITION AND
AMIGA
Two interesting and seldom used parameters are the two pen colors.
Depending on the screen, they can accept values from 0 to 31, which
naturally depends on the number of bit-planes. A special value is -1
(Oxff), which uses the default screen color values. This assigns all of
the windows in the screen the same colors, allowing global color
changes as well. The example uses Workbench color values, but not the
default values. This makes the new window independent of the
superordinate screen's parameters, even if they currently contain the
same values.
FirstNewWindow.DetailPen
FirstNewWindow.BlockPen
= 0;
= 1;
You can only initialize two more parameters. These two flag parameters
present so many options that we must layout descriptions of each flag.
They can be divided into six basic groups: Gadget flags, sizing gadget
position, refresh flags, window type flags, mouse flags and miscellaneous flags.
Gadget flags
Sizing gadget
position
These flags must be set according to the system gadgets you want in
the window.
WINDOWDRAG
Adds a drag bar to the title bar.
WINDOWDEPTH
Adds front and back gadgets.
WINDOWCLOSE
Adds a close gadget readable by the program.
WINDOWS I ZING
Adds a sizing gadget for window size control.
If the sizing gadget is used, two additional flags must be integrated,

because all system gadgets are placed in the window borders. It should
be clear that the rust three are placed in the upper border. You have two
options for the sizing gadgets:
Refresh nags
SIZEBRIGHT
The sizing gadget goes in the right border.
SIZEBOTTOM
The sizing gadget goes in the bottom border.
Window contents can easily be destroyed if the window is overlapped by

another window. Intuition offers various ways of getting around this
problem:
SIMPLE REFRESH
94
This refresh status means that Intuition does not do

anything to restore the window. Any reconstruction
must be handled by the program itself.
ABACUS
3.1 WINDOWS
SMART REFRESH
Here Intuition buffers every hidden area of the window and then restores the area when made visible.
This causes problems only when resizing windows.
If you make the window smaller, the eliminated
portions are not buffered and not restored.
SUPER BITMAP This type of refresh uses the most memory. In
some cases you may need to store a large graphic
and display it in a window. This is both a window
type and a refresh mode, in that the window can
always be restored by the information contained in
the bit-map, and it also manages graphics larger
than the display window.
NOCAREREFRESH
The SMART REFRESH and SIMPLE REFRESH

window types require at least some program
cooperation. Intuition loses its way when the
window is resized with SMART REFRESH and for
SIMPLE REFRESH the overlaid area must be
redrawn. The program receives a message saying
that something must be done. The message port
and how it is used are explained later. If you don't
want to do anything, then there is no need to send
these messages. This flag tells Intuition that you
don't want this information.
Window type
flags
We have already discussed various window types. Here are the flag
names with brief comments:
This window always has lower priority than other
windows. Only one of these windows should exist
per screen.
BACKDROP
SUPER BITMAP Although SUPER BITMAP is also a refresh type,

it also applies as window type. There is only a
flag.
BORDERLESS
Intuition does not draw a window border.
GIMMEZEROZERO
Intuition keeps the border and all gadgets separate

from the window's contents.
Mouse flags
These two flags control the status of the mouse.

REMOUSE
If you set this flag you get continuous messages

about the mouse's position.
95
3. INTUITION AND C
RMBTRAP
AMIGA
All mouse signals from the right mouse button

convert into signals from the left button. This can be
used when there are no menus present.
Miscellaneous This last flag determines whether or not the window is active when
open.
ACTIVATE
If this flag is set, the window is active immediately

when it is opened. This is often important since input
is always directed to the active window. This can
cause problems if the user has already entered something and this flag opens a new window. Use such
windows with caution!
All of these flags are represented as set or cleared bits in the flag entry
of the NewWindow structure. When programming in C you can use
the symbolic names. But for assembly language programmers and those
who want more information, you need the hex values for the flags.
Here's a table of the flags and their values:
Table 3.2:
Window flags
FlaR Name
WINDOWS I ZING
WINDOWDRAG
WINDOWDEPTH
WINDOWCLOSE
Hex Value
OxOOOOOOO1L
OxOOOOOOO2L
OxOOOOOOO4L
OxOOOOOOO8L
GadRet ~roul>
System gadgets
SIZEBRIGHT
SIZEBBOTTOM
OxOOOOOO1OL
OxOOOOOO20L
Gadget position
NOCAREREFRESH
SIMPLE REFRESH
SMART REFRESH
SUPER- BITMAP
OxOOO20000L
OxOOOOOO40L
OxOOOOOOOOL
OxOOOOOO80L
Refresh types
BACKDROP
GIMMEZEROZERO
BORDERLESS
OxOOOOO1OOL
OxOOOOO400L
OxOOOOO800L
Window types
REPORTMOUSE
RMBTRAP
OxOOOOO200L
OxOOOlOOOOL
Mouse flags
ACTIVATE
OxOOOO1OOOL
Miscellaneous
This is the definition you will use in your Newwindow structure:

FirstNewWindow.Flags
= WINDOWDEPTH
I WINDOWSIZING
WINDOW DRAG I WINDOWCLOSE I
SMART_REFRESH;
Now that you have some knowledge of the simple flags, we'll spend a
little time looking at the IDCMP (Intuition Direct Communication
Message Port) flags. First we need a definition of the IDCMP.
96
3.1 WINDOWS
ABACUS
The IDCMP is a very complex data channel which carries infonnation

about all of the areas managed by Intuition. Here we'll just list them,
since the topic is so comprehensive that we dedicated an entire section
to it (Section 3.6). We include some infonnation about it here because
any interaction with Intuition requires the IDCMP.
You can get information about the following Intuition peripherals:
Windows, gadgets, menus, mouse, keyboard, disk, Preferences, clock.
But since you don't know anything about this yet, we'll use a little
trick. Instead of using a close gadget to close the window opened by the
program, we'll close it after a short time delay. This way you can
bypass using the IDCMP for now:
FirstNewWindow.IDCMPFlags
NULL;
This concludes the structure definition of FirstNewWindow. Take a

look at the program below. It consists of the two functions
Open AIIO and Close AIIO, the window structure, and a main
function:
Program 3.1:
First window
/***************************************/
/*
*/
/*
3.1.2.2.0
First_window.c
*/
/*
*/
1* Program: Window local definition
*/
*1
1*
==================================~
1*
*/
/* Author:
/*
1* Wgb
1*
DAte:
----------
Comment:
*/
----------
*/
*1
*/
/*
*1
/***************************************/
10/16/1987
first test
window
iinclude <exec/types.h>
#include <intuition/intuition.h>
struct Window
*FirstWindow;
struct NewWindow FirstNewWindow;
main 0
{
= 160;
=
50;
97
3.
INTUITION AND
FirstNewWindow.DetailPen
FirstNewWindow.BlockPen
FirstNewWindow.IDCMPFlags
FirstNewWindow.Flags
320;
150;
0;
1;
NULL;
WINDOWDEPTH I WINDOWSIZING
I WINDOWDRAG I WINDOWCLOSE
I SMART_REFRESH;
FirstNewWindow.FirstGadget= NULL;
FirstNewWindow.CheckMark = NULL;
FirstNewWindow.Title = (UBYTE *) "System test";
= NULL;
FirstNewWindow.BitMap
= NULL;
FirstNewWindow.MinWidth
100;
FirstNewWindow.MinHeight
50;
640;
FirstNewWindow.MaxWidth
200;
FirstNewWindow.MaxHeight
WBENCHSCREEN;
FirstNewWindow.Type
Delay(180L);
1***************************************
*
*
* Function: Open library and window
*
* =================================--= *
*
Author:
Date:
Comment:
---------- ----------
*
* Wgb
*
*
10/16/1987
*
*
*
*
*
*
**************************************1
void
*OpenLibrary();
struct Window *OpenWindow();
{
printf("Intuition library not found!\n");

Close_All () ;
exit(FALSE);
}
if (! (FirstWindow = (struct Window *)

OpenWindow(&FirstNewWindow)
{
98
3.1 WINDOWS
ABACUS
printf("Window will not open!\n");

Close_All () ;
exit(FALSE);
}
/***************************************
Function: Close everything opened

===================================
*
*
Author:
*
*
* Wgb
Date:
Comment:
10/16/1987
*
*
just Intuition
and window
*
*
*
*
*
*
*
*
*
**************************************/
if (FirstWindow)
if (IntuitionBase)
CloseWindow(FirstWindow);
C1oseLibrary(IntuitionBase);
How it works
The listing contains a few things that may be unfamiliar to you. First,
the Open AIIO function has been modified to open the window, and
Close_ A1IO has been correspondingly modified to close the window.
How does a window open once you define a NewWindow structure? It
is very important that the structure be defined before the mainO function, otherwise the structure values are not available to the other functions. The OpenWindow simply passes the address of the definition
structure (NewWindow). As the return value you get a pointer to the
window structure which was created based on your data. It also contains
additional management information. OpenLibrary must be checked
to ensure there were no errors (you can't assume that everything worked
properly with OpenwindowO). There are many things that could cause
an error in OpenWindowO, such as running out of memory. The
Newwindow structure can then be erased again. It is only needed for
initialization.
WindowPointer
DO
= OpenWindow(NewWindowStructure)
-204
AO
Closing the window has also been added to the Close_AIIO function.
The structure is similar to what you have seen already. The important
thing is that the window closes first, and then the Intuition library. If
these actions were reversed, the operating system would no longer be
able to find the ClosewindowO function and the program would
crash.
CloseWindow(WindowPointer)
-72
AO
99
3.
INTUITION AND
AMIGA
A time delay appears between Open-.A110 and Close-.A110 in the

main program with the DelayO function. This operating system puts
the task of your program into wait status until the specified time has
elapsed (time is specified in "ticks", 60 ticks per second), since this
program doesn't have the ability to read the close gadget in order to
close the window again. Instead of the DelayO function you could also
use a for loop to achieve the delay. But this would have slowed down
the multitasking even though you were only waiting.
Let's look at the start of the program and its comment blocks. Before
the mainO function you define two structures which you need for your
window. The first is the NewWindow structure under the name
FirstNewWindow. This allocates space for the information which
Intuition requires to set up your window. The second structure is a
pointer to a window structure. A Window structure contains much
more information than a Newwindow structure, and is linked to other
windows in the screen. The construction of the system which Intuition
sets up and manages looks like this:
"Ground zero" is a screen, generally the Workbench screen. This
Screen structure, explained in the next section, contains a pointer to
the first window in the screen. The Window structure itself has a
pointer to the following windows. If there are several screens, then the
frrst screen also stores a pointer to the next screen, which can contain
the same window chaining. Here is a figure to clarify the structure'
ScreenlWindOM-Linking
Figure 3.1
The structure definition described above must have access to the
operating system's Intuition functions in order for our new functions to
work. These can only be accessed once the library has been opened. You
must know the address of the library, stored in the pointer
*IntuitionBase.
Finally, we should say something about comments. Comments always

contain information about the author and the creation date. This is the
100
3.1 WINDOWS
ABACUS
date the first lines were written. A new date represents a fundamental
change in the routine. The purpose of the program or function is also
explained in the header.
3.1.2.3
The Window structure

The Window structure stores all of the infonnation needed for Intuition
to manage a window. Here it is:
struct Window
{
OxOO
Ox04
Ox06
Ox08
OxOa
OxOc
OxOe
OxlO
Ox12
Ox14
Ox16
Ox18
OxIc
Ox20
Ox24
Ox28
Ox2C
Ox2E
Ox32
Ox36
Ox37
Ox38
Ox39
Ox3A
Ox3E
Ox42
Ox46
Ox4A
Ox4E
Ox4F
Ox50
Ox51
Ox52
Ox56
Ox5A
Ox5E
Ox62
Ox63
Ox64
Ox68
00
04
06
08
10
12
14
16
18
20
22
24
28
32
36
40
44
46
50
54
55
56
57
58
62
66
70
74
78
79
80
81
82
86
90
94
98
99
100
104
struct Window *NextWindow;

SHORT LeftEdge;
SHORT TopEdge;
SHORT Width;
SHORT Height;
SHORT MouseY;
SHORT MouseX;
SHORT MinWidth;
SHORT MinHeight;
USHORT MaxWidth;
USHORT MaXHeight;
ULONG Flags;
struct Menu *MenuStrip;
UBYTE *Title;
struct Requester *FirstRequest;
struct Requester *DMRequest;
SHORT ReqCount;
struct Screen *WScreen;
struct RastPort *RPort;
BYTE BorderLeft;
BYTE BorderTop;
BYTE BorderRight;
BYTE BorderBottom;
struct RastPort *BorderRPort;
struct Window *Parent
struct Window *Descendant;
USHORT *Pointer;
BYTE PtrHeight;
BYTE PtrWidth;
BYTE XOffset;
BYTE YOffset;
ULONG IDCMPFlags;
struct MsgPort *UserPort;
struct MsgPort *WindowPort;
struct IntuiMessage *MessageKeYI
UBYTE DetailPen;
UBYTE BlockPen;
struct Image *CheckMark;
UBYTE *ScreenTitle;
101
3.
INTUITION AND
Ox6C
Ox6E
Ox70
Ox72
Ox74
Ox78
Ox7C
Ox80
Ox84
};
AMIGA
108
110
112
114
116
120
124
128
132
SHORT GZZMouseX;
SHORT GZZMouseY;
SHORT GZZWidth;
SHORT GZZHeight;
UBYTE *ExtData;
BYTE *UserData;
struct Layer *WLayer;
struct TextFont *IFont;
As with the NewWindow structure, the numbers preceding the structure

element names are the offsets from the start of the structure.
Let's go through the parameters one by one.
*NextWindow
A pointer to the next wi ndow structure in the

screen. This pointer maintains the window chaining
mentioned.
LeftEdge,TopEdge
Position of the window on the screen, as defined in
Newwindow.
Width, Height Dimensions of the window, as defined in
Newwindow.
MouseX,MouseY
Mouse coordinates, relative to the upper left comer

of the window.
MinWidth,MinHeight
Minimum values as defined in NewWindow.
MaxWidth,MaxHeight
Maximum values as defined in NewWindow.
Flags
Flag list, as defined in Newwindow.
*MenuStrip
Pointer to the window menu structure (see Section

3.8 for more information). No menus can be set
through Newwindow. This must be done with a
separate function.
*Title
Pointer to string containing the window title text.
*FirstRequester
Pointer to the first requester set in this window (see
Section 3.5).
*DMRequest
102
Pointer to double menu requester (see Section

3.5.4).
3.1
ABACUS
WINDOWS
ReqCount
Counter for the requesters opened in this window.
*WScreen
Pointer to screen in which this window opened.
*RPort
Pointer to the RastPort setup for this window.
BorderLeft, BorderTop, BorderRight, BorderBottom

Border size specifications.
*BorderRPort Pointer to the RastPort for the border (used for
GIMMEZEROZERO windows).
*Firstgadget Pointer to the fIrst gadget in a linked list of all the
gadgets in this window.
*Parent,*Descendent
Chaining previous and next window, to make
opening and closing easier.
*Pointer
Pointer to graphic for this window's mouse pointer

(mouse pointer can be redefIned for each window).
PtrHeight,PtrWidth
Mouse pointer size (X value cannot exceed 16).
XOffset,YOffset
Offsets specifying point in the pointer representing
the click point.
IDCMPFlags
The flags as defIned in Newwindow.
*UserPort
Message port for data communication.
*WindowPort
Message port for data communication.
*MessageKey
Message port for Intuition reports.
DetailPen,BlockPen
Pen colors as defIned in NewWindow.
*CheckMark
Pointer to the checkmark defIned for the menus.
* ScreenTi tle Pointer to the string for the screen title (can only
be set by the function SetWindowTitles()).
See explanation below.
GZZMouseX,GZZMOuseY
Mouse coordinates for a GIMMEZEROZERO
window (border is automatically taken into account
and subtracted from the normal value).
103
3.
INTUITION AND
GZZWidth,GZZHeight
Window size for a GIMMEZEROZERO window.
*ExtData
Pointer to external data structures (not used).
*UserData
Pointer to a data structure set by the programmer.
*WLayer
Pointer to the Layers structure of this window

(RPort->Layer also contains the same value).
*IFont
Pointer to the font used for Intuition text output in

this window.
The Window structure contains some pointers to very important

Intuition and graphic elements. The following figure should clarify the
structure:
Intuition-WindoH-Linking
I
I
FirstRequester
YMiR...ester
4j First6adget .
Figure 3.2
RastPort
BorderRastPort
All of these values allow easy access. If you want to know the width of
your GIMMEZEROZERO window, for example, you simply enter the
following in your source code:
width
= FirstWindow->GZZWidth;
You can also change a given value. Perhaps you want to use a different
graphic for the checkmark which appears next to selected menu items.
Just enter the following in your source code:
FirstWindow->CheckMark = NewCheckMark;
Later you'll see the importance of access to window parameters.

Let's return to the window program. You can see how simple programming in Intuition can be. That won't change in general, but typing
in all that initialization is still a lot of effort. You have to find a
104
3.1 WINDOWS
ABACUS
method to make structure definition easier. We wrote the following

definition:
1* 3.1.2.3.B.1.firstnew_struct *1
struct NewWindow FirstNewWindow =
{
160, 50,
320, 150,
0, I,
1*
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
NULL,
WINDOWDEPTH
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *)"System test",
NULL,
1* Screen
NULL,
1* BitMap
100, 50,
1* Min Width, Height
640, 200,
1* Max Width, Height
WBENCHSCREEN,
1* Type
.;
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*/
*1
This can be shortened to the bare essentials of information:

1* 3.1.2.3.B.2.short_firstnew *1
struct NewWindow FirstNewWindow
160, 50, 320, 150, 0, 1, NULL,

WINIXl'IDEPTH Iw:rNIXNISIZIN31 WINInIDRAG IWINIJC.W:IC6E ISMZIRT_REFRESH,
NULL, NULL, (UBYTE *)"System test",

NULL, NULL, 100, 50, 640, 200, WBENCHSCREEN
};
The advantage is clear: you have much less to type in because the
structure name appears only once. It's up to you whether or not you
want to include the comments.
The structure definition must precede the mainO function. This makes
it a global variable available to all functions. If you want to change a
few values you can do so by accessing each element through its name.
Note that when initializing structures in this manner the order dictates
what you are actually initializing. Changing things around can easily
lead to a crnsh.
105
3.
INTUITION AND
3.1.2.4
AMIGA
A summary of window functions

Now that we are done with initialization, we can begin to build on what
we have accomplished. You should use the first_window. c
program as the basis for the demonstrations which follow. Keep the
same window parameters you used there. It is easier if you use a global
structure definition as explained above.
You have already seen the first two Intuition functions related to
windows. For opening we used the openWindowQ function and for
closing we used CloseWindowQ.
In the programs which follow, make sure that all windows which are
opened are closed again. If a program ends without closing all of the
windows it opened, the pointers to the remaining windows are lost and
they remain lost until the computer is turned off or reset. A screen
containing such windows cannot be closed either.
Changing
window titles
Let's take a look at the rest of the window functions. The first is
SetWindowTitles(). This lets you change the title of a window
after it has been configured. To demonstrate this, insert the following
lines into the First_Window. c program:
1* 3.1.2.4.A Setwindowstitle.c 1*
/* add after Delay(180L) */
SetWindowTitles(FirstWindow,
"New window title", "The screen has a title now too!");
Delay (180L) ;
As you can probably tell from the text, the window is assigned a new
title. Then the screen gets a title. When you click on a window the
screen title line changes appearance. In most cases the program starts
from the Workbench, and the screen doesn't have a special title. The
same is true for the CLI. The text "Workbench Screen" is simply placed
in the title line. You as the programmer can choose a different title for
your active window. The format of this command looks like this:
SetWindowTitles(Window, WindowTitle, ScreenTitle);
-276
AO
Al
A2
window is the pointer to the structure of the window in question. The
titles are passed as pointers to strings ending in null.

Now that you have seen how it works, you should give some thought
to practical uses for SetWindowTi tlesO. Let's imagine a word
processor. Here you probably work on several documents in succession
or at the same time. It's vitally important that you know which document you're working on at the moment. You'd like the name of the
106
3.1
ABACUS
WINDOWS
document to appear as the title of the word processor window. In addition, most programmers also add the path. This lets you see the name
and path of the disk currently under access. The only way to give a title
to the screen is to use SetWindowTitlesO.
If you want to change one title parameter of SetWindowTitlesO
without changing the other parameters (e.g., if you want to change the
title of a window without changing the title of the screen), you can pass
-1 as the parameter instead of a pointer. This leaves the appropriate title
unchanged.
It is also possible to clear the title of either a window or the screen or
both. This is done by passing null to a string instead of a pointer.
Window
activation
The next window handling function first appeared in Version 1.2 of the
Intuition library. The ActivateWindowO function allows the
programmer to activate a window at any time during program
execution. The only argument passed to it is a pointer to the window
structure. Insert the following lines in the original version of the
Fir s t Win do w c program (not including the addition of
SetWindowTitlesO as entered above):
/* 3.1.2.4.B.activatewindow.c*/
/* place after Delay(180L) */
ActivateWindow(FirstWindow);
Delay(180L);
Compile the progam and then start it Click in a different screen or in

some other window as soon as the window appears, and watch what
happens. Now your window is deactivated and after a delay it
automatically reactivates. Here's the format for Acti vateWindow:
ActivateWindow(Window);
-450
AO
This function is important for input, which occurs only in the active
window. If a program checks to see if a window is active which is not
allowed to be active because no inputs are allowed, it can activate
another window. This prevents the user from entering information in
that window. See Section 3.7 (Reading the rDCMP flags) for a sample
application.
Moving
windows
It may also become necessary to change the position of a window.

Perhaps a new window must be opened at the same position as the old
one, but you want the old window to remain visible. You could leave
this up to the user, but it might get aggravating if it happened often. If
you as the programmer know when it is necessary to change the
position of a window, then you can use the MoveWindowO function.
The arguments you pass are the window pointer and two relative values
which specify the movement in the X and Y directions.
107
3.
INTUITION AND
AMIGA
But there is a danger associated with these delta (movement) values. If

you move the window even one pixel outside the screen, the system
crashes. Therefore you must check delta values. To do this we refer to
the information in the window structure, which contains the current
window position. Insert the following into the FirstWindow. c
program:
1* 3.1.2.4.C. movewindow.c *1
1* insert after line Delay(180L) *1
DeltaX = FirstWindow->LeftEdge;
DeltaY = FirstWindow->TopEdge;
MoveWindow(FirstWindow, -l*DeltaX, -l*DeltaY);
Delay(180L);
Before you save the source code for this program remember to declare
the two delta variables as short, to match the window structure.
Access to the Amiga library requires 32 bit integers so be sure to use
the +L option of the Aztec C compiler to compile this program so all
integers are 32 bits long. When the new program is then started, the
newly opened window moves to the upper left comer of the screen after
a short pause.
Here is the fonnat of the new function:
MoveWindow(MyWindow, DeltaX, DeltaY);
-168
AO
DO
D1
Sizing
windows
A function in a similar category is SizeWindowO. You may want to

change the size or shape of a window from within the program. in cases
when the user doesn't have the option of changing it himself. Good
programs often give the user a gadget to click on to either reduce or
enlarge the size of a window.
The first example appears below. A program at the end of this section
contains the second example. Add these lines to FirstWindow. c
program:
1* 3.1.2.4.D.sizewindow.c *1
1* insert after Delay(180L) *1
DeltaX = FirstWindow->LeftEdge - FirstWindow->MinWidth;
DeltaY = FirstWindow->TopEdge - FirstWindow->MinHeight;
SizeWindow(FirstWindow, -l*DeltaX, -l*DeltaY);
Delay(180L);
Remember to declare the variables as short. Be sure to use the +L

option of the Aztec C compiler to compile this program so all integers
are 32 bits long. This window type requires that you assign values to
the MinWidth and MinHeight variables. Otherwise problems can
arise when you reduce the window to the size of a single pixel. This
108
3.1
ABACUS
WINDOWS
error can occur only if you have not selected a sizing gadget, and did not
have to set the minimum and maximum sizes.
The format is very similar to that of the MoveWindowO function:
SizeWindow(MyWindow, Oe1taX, Oe1taY);
-288
AO
DO
01
Make sure you don't use any illegal values for the deltas. The window
may not be made smaller or larger than the screen allows.
Limiting
window size
WindowLimitsO allows you to set the maximum and minimum
sizes of a window after opening the window.

For example, if your program recognizes that it can no longer enlarge a
graphic window due to memory limitations, you can simply update the
maximum size values. The user can then only reduce the window's size.
The example below, which can again be inserted into
FirstWindow. c program, reads the old maximum values and then
changes the old maximum values to new maximums, after the initial
delay. The window can then be enlarged only up to the new limit
1* 3.1.2.4.E.windowlimits.c *1
MinWidth
MinHeight
= FirstWindow->MinWidth;
= FirstWindow->MinHeight;
NewMaxWidth = 200;
NewMaxHeight= 100;
Success = WindowLimits(FirstWindow, MinWidth, MinHeight,
NewMaxWidth, NewMaxHeight);
De1ay(180L);
Before you can try this out, you must declare the limit values as
USHORT and the result value (Success) as BOOL. Be sure to use the
+L option of the Aztec C compiler to compile this program. The
routine returns 1RUE if values were all within the allowed limits.
If the maximum values are reset and the window is already larger than
the new limits, the program returns FALSE as the result. Here is the
format of the function call:
Result
DO
WindowLimits(MyWindow, MinWidth, MinHeight,

-318
AO
DO
D1
MaxWidth, MaxHeight);
D2
D3
109
3.
INTUITION AND
AMIGA
A newly opened window is always placed in front of all others. The fact
Moving
that one window can cover all the others should be familiar to you from
windows to
front and back your daily use of Intuition. The programmer needs to know how to put
a window into the background when it is not currently needed, and then
returns it to the foreground. The user controls this procedure with the
front and back gadgets. There are two corresponding Intuition functions:
windowToBackO and windowToFrontO.
Both functions require just the pointer to the window structure as the
argument. Since you probably have the DOS window open when you
start all of these programs, you can see how these functions work. Add
the following lines to FirstWindow. c program:
/* 3.1.2.4.F.
windowt~back_front.c
*1

WindowToBack(FirstWindow);
Delay(180L);
WindowToFront(FirstWindow);
Delay(180L);
After a short delay the new window goes behind the large AmigaDOS
window then reappears. Here is the format:
WindowToBack(Window);
-306
AO
.
WindowToFront(Window);
-312
AO
3.1.3
Example window application programs

The first part of this section limited us to viewing structures and
window functions which support them. The following sections use
example programs to demonstrate how simple it is to use windows
under Intuition.
At the beginning we will introduce several general examples which can
be used in a variety of applications. Later you'll find window structures
and program fragments which were developed for the main project of
this book. Many are used in the C program editor that will be developed
later on in this book. If you are interested in this excellent editor, enter
the basic functions found in this section now. In the following
subsections about screens, output, gadgets, etc. you'll find extensions
to your basic window program together with descriptions. But you'll
need much more than the whole Intuition chapter to complete an editor.
The third portion of this section contains functions and routines which
deal with various aspects of text editing.
110
3.1
ABACUS
3.1.3.1
WINDOWS
All-purpose windows
Each listing here suits a particular range of applications. You can easily
adapt the programs to your own needs by making a few changes to the
arguments.
About graphic Let's assume that the screen to be drawn on has the appropriate color
settings. We'll use the Workbench colors and screen. Section 3.2 shows
programs
you how to open your own screens.
Since no operating system functions support drawing graphics on a
screen, you have to use windows. The window should resemble a screen
as much as possible: maximum size, completely blank and moveable
up and down.
Your window can use the full size normally occupied by the screen. For
the blank surface, enable the BORDERLESS flag to prevent drawn
borders. If the program opens other windows, you will have problems,
because these windows may be placed behind your drawing window.
You won't be able to access them because your window has no border
or gadgets. Omitting the front and back gadgets contradicts good
programming sense, nor would it be user-friendly. Simply enable the
BACKDROP flag to have a window with the properties of a screen.
Here's part of the listing:
/* 3.1.3.1.A. GraphicWindow1.c *1
{
0, 0,
640, 200,
0, 1,
NULL,
BACKDROP 1
BORDERLESS
SMART_REFRESH,
NULL,
NULL,
NULL,
NULL,
NULL,
0, 0,
0, 0,
WBENCHSCREEN,
};
1* LeftEdge, TopEdge
1* Width, Height
*1
*1
/* DetailPen, BlockPen *1
1* IDCMP Flags
*1
/* Flags
*1
1* First Gadget
/* CheckMark
/* Window Title
1* Screen
*1
*1
*1
*1
/* BitMap
/* Min Width, Height
/* Max Width, Height
*/
1* Type
*1
*1
*/
You won't find a complete listing printed here because it's unnecessary.
Simply use this structure in the First_Window. c program.
Let's look at the structure values: The position and dimensions correspond to the normal Workbench screen. The window contains no
111
3.
INTUITION AND
AMIGA
gadgets and is defined as BACKDROP and BORDERLESS. The refresh

mode is set to SMART REFRESH since SUPER BITMAP is needed
only when graphics areused which are larger than the dimensions of the
window. We did not use a title so the title bar won't appear. The result
is a completely blank screen which you can use as the drawing surface.
The only problem is the fact that the title bar of the screen may still be
visible. It normally overlaps any BACKDROP window, but you don't
want it there for a drawing program. You can use ShowTitleO,
which determines whether or not the title bar of the screen is displayed.
As arguments you need a pointer to the screen in question and the set
value. The following main function performs the task when you add
the two functions:
/******************************************/
/*
*/
/* 3.1.3.1.B. GraphicWindow2.c
*/
/*
*1
/* Program: Window for a graphics program */
1* ==========================~~=======~
*1
/*
*/
/* Author: Date:
Comments:
*/
/*
/* Wgb
1* JLD
----------
12/06/1987
01/06/1988
----------
BACKDROP
BORDERLESS
1*
*/
*1
*/
*1
/******************************************/
#include <exec/types.h>
struct Window
*FirstWindow;
struct IntuiMessage *message;
{
0, 0,
640, 200,
0, 1.
NULL,
/*
WINDOWCLOSE
BACKDROP I
BORDERLESS I
SMART_REFRESH,
NULL,
NULL,
NULL,
NULL,
NULL,
0, 0,
0, 0,
112
*/
/* Width, Height
*/
/* DetailPen, BlockPen */
/* IDCMP Flags
*/
1* Flags
*/
1* First Gadget
/* CheckMark
/*Window Title*/
1* Screen
1* BitMap
*/
*/
*/
*/
*/
*/
3.1
ABACUS
WBENCHSCREEN,
};
/* Type
WINDOWS
*/
main ()
{
ShowTitle(FirstWindow->WScreen, FALSE);
Delay(180L);
exit (TRUE) ;
}
/***************************************
*
*
===================================
* Author:
*
*
*
Date:
Comment:
10/16/1987
Wgb
*
*
*
*
*
*
**************************************/
void
*OpenLibrary();
{
printf ("Intuition library not found! \n");

Close_All () ;
exit(FALSE);
}
if (! (FirstWindow
= (struct Window *)

Close_All () ;
exit (FALSE) ;
}
113
3.
INTUITION AND
AMIGA
/***************************************
*
*
* Function: Close everything opened *
* =================================== *
*
*
Comment:
* Author: Date:
*
*
*
---------- ---------10/16/1987 just Intuition *
* Wgb
and window
*
*
*
*
**************************************/
if (First Window)
if (IntuitionBase)
The next example illustrates the GIMMEZEROZERO window type. The

program first opens a simple window. This window closes when you
click on the close gadget (you'll finally get to see how to test gadgets).
/***************************************
* Program: Window GIMMEZEROZERO.C Test*
* =================================== *
*
*
* Author: Date:
Comments:
*
*
---------- ---------*
* Wgb
10/16//1987 close with
*
01/09/1989 a Gadget
*
* JLD
*
*
***************************************/
struct Window
*FirstWindow;
{
160, 50,
320, 150,
0, 1,
CLOSEWINDOW,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
114
/*
/*
/*
/*
/*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
*/
*/
*/
*1
*/
3.1
ABACUS
GIMMEZEROZERO I
SMART_REFRESH.
NULL.
1*
NULL.,
1*
(UBYTE *)"GIMMEZEROZERO
NULL.
1*
NULL.
1*
100. 50,
640. 200,
WBENCHSCREEN.
};
First Gadget
CheckMark
Test".
Screen
BitMap
1* Min Width. Height
1* Max Width. Height
1* Type
WINDOWS
*1
*1
*1
*1
*1
*1
*1
main ()
(
ULONG MessageClass;
USHORT code;
struct Message *GetMsg();
FOREVER
{
if (message = (struct IntuiMessage *)

GetMsg(FirstWindow->UserPort
{
MessageClass = message->Class;
code = message->Code;
ReplyMsg(message);
switch (MessageClass)
{
case CLOSEWINDOW : Close_AII();

exit (TRUE);
break;
}
}
1***************************************
*
*
* Function: Library and Window open *
* =================================== *
*
* Author:
*
*
*
wgb
Date:
Comments:
---------- ----------
10/16//19B9
*
*
*
*
*
***************************************1
115
3.
INTUITION AND
AMIGA
void
*OpenLibrary();
OpenLibrary("intuition.library". OL)
{
printf("Intuition Library not found!\n");

Close_All ();
exit(FALSE);
}

{
printf(JlWindow will not open!\n");

Close_All () ;
exit(FALSE);
}
/***************************************
* Function: Close everthing opened
* =================================== *
*
* Author: Date:
*
---------10/16/1987
* Wgb
*
*
Comments:
----------
for Intuition
and Window '\
*
*
*
*
*
***************************************/
if (FirstWind6w)
if (IntuitionBase)
Before you compile and run the program, lets talk a bit about reading
the system gadgets. In order to get information about the gadget status
you need a Me s sag e structure, defined at the beginning of this
program. In addition you need a function to get messages for you. The
GetMsgO function returns a pointer to a message data structure. You
then divide this message into MessageClass, the origin of the
message, and Code. This Code is included in the message, but you
have no use for it yet. In a switch statement you then test to see if the
message is of type CLOSEWINDOW (a symbol defined in one of the
include files) and if so, the window is closed.
This method can be used for much more, but this will be handled in
several sections: Section 3.4 (Gadgets), Section 3.7 (Reading the
IDCMP flags) and Section 3.8.3 (Reading the menus).
116
3.1
ABACUS
WINDOWS
Let's expand the first program by having it display the coordinates of

the mouse in the DOS window from which you started the program. To
read the mouse coordinates you must define two more variables:
SHORT Mx,My;
Both are assigned values in the FOREVER loop before the if test
Mx = FirstWindow->MouseX;
My = FirstWindow->MouseY;
And are printed with:

printf("X: %d Y: %d\n", Mx, My);
After you have run the first version of the program, make the following
changes: Add the GIMMEZEROZERO flag to the NewWindow structure. Then declare the variables Gx and Gy as short and assign them
values after Mx and My as follows:
Gx = FirstWindow->GZZMouseX;
Gy = FirstWindow->GZZMouseY;
The printf line must also be expanded:

printf("X: %d Y: %d ; Gx: %d Gy: %d\n", Mx, My, Gx, Gy);
Now when you run the program you will see the difference between the
normal and GIMMEZEROZERO windows.
Note the amount of free memory space given in the Workbench display
when neither of the programs are running. Then start the first program
and write down the memory display value (you will see this when you
click somewhere on the Workbench surface with the left mouse button).
Then end the first program by clicking on the close gadgeL After starting the second program you wi1l notice that it requires more memory
than the first.
3.1.3.2
Program routines for text editing

Let's create the windows we could use in the text editor we will create.
First we have to consider what we need to assemble a project like a text
editor.
You'll need at least three windows for the editor: One for text output,
one for file selection and one for requesters and messages.
Editor window The first window allows text output and editing. We'll call this window
the editor window.
117
3.
INTUITION AND
File selector
The second window will be configured as a fIle selector box. This type
of box lists all filenames in a directory, and lets you select a fIle using
the mouse. Actually this involves a requester, but we will not be discussing gadgets and requesters until later. It is only important that you
have the file available so that you can use it.
Message
window
We will use the last window as a simple window for messages or

requesters. Here again it is important only to have an appropriate file
ready.
What properties should the editor window have? It should use a window
the size of the entire screen in order to display as much text as possible.
Thus, the size of the Workbench screen gives us the maximum window
values. You can use any values you like for the minimum, as long as
the sizing gadget is still accessible. Finally, we're interested in the
various window flags. You should use all the gadgets geared for
maximum user-friendliness. We recommend SMART REFRESH mode,
since SUPER BITMAP is overkill, and SIMPLE REFRESH will only
provide extrawork for the program. Here is the window structure for
our editor:
Structure 3.1: 1* 3.1. 3 .2. A. editor window *1

Editor window struct NewWindow EditorWindow
0, 0,
640, 200,
0, 1,
1*
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
NULL,
WINDOWDEPTH
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *)"System programming Editor",
NULL,
/* Screen
NULL,
1* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
/* Type
);
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
The same parameters set for the editor window apply to the fIle selector
box, except that the file selector box doesn't take up the entire screen.
We only include the WINDOWDEPTH and WINDOWDRAG gadgets
because WINDOWCLOSE is replaced by another gadget, and sizing
would complicate the whole thing even more. Here is this structure:
118
3.1 WINDOWS
ABACUS
Structure 3.2:
File selector
box
/* 3.1.3.2.B.filewindow */
struct NewWindow FileWindow =
{
180, 25,
/* LeftEdge, TopEdge
250, 150,
/* Width, Height
0, 1,
/* DetailPen, BlockPen
NULL,
/* IDCMP Flags
WINDOWDEPTH I
/* Flags
WINDOWDRAG I
SMART_REFRESH,
NULL,
/* First Gadget
NULL,
/* CheckMark
(UBYTE *)"File-Select-Box",
NULL,
/* Screen
NULL,
/* BitMap
0, 0,
0, 0,
WBENCHSCREEN,
/* Type
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
};
The last window which you will always need for the text editor is the
message window. It mainly displays error messages, so it is somewhat
smaller than the file selector box. In addition, it doesn't contain any
system gadgets since it is read and then confirmed, not moved or
resized. The window structure is rather simple:
Structure 3.3:
Message
window
/* 3.1.3.2.C.msgwindow */
struct NewWindow MessageWindow =
{
250, 100,
140, 80,
0, 1,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
0, 0,
0, 0,
WBENCHSCREEN,
};
/* Left Edge, TopEdge
/*
/*
/*
/*
/*
/*
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
First Gadget
CheckMark
/*
/*
/*
/*
/*
Screen
BitMap
Min Width, Height
Max Width, Height
Type
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
119
3.
INTUITION AND
3.1.3.3
AMIGA
Window for a new CL I

We still need one more window for an editor program.
Our planning begins with writing a small editor which can be used in
the eLI instead of the old ED program. Naturally. you need a window
for this. What follows is the main function with the window structure:
Program 3.5:
NeweLl:
1****************************************1
1*
*1
1* 3.1.3.3.A.
CLled.c
*1
1*
*1
*1
/* Program: A new CLI Ed
/* ==================================== *1
1*
*1
Comments:
*1
1* Author: Date:
1*
*1
/* Wgb
10/20111987 just the window
*/
01/0911989 for testing

1* JLD
*1
1*
*1
1****************************************1
#inc1ude <exec/types.h>
#inc1ude <intuition/intuition.h>
struct Window
*FirstWindow;
struct NewWindow ConsoleWindow =
{
0, 0,
640, 101,
2, 3,
1* Width, Height
CLOSEWINDOW,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
ACTIVATE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *)"Wgb Prod. presents BECKERshell",
NULL,
1* Screen
NULL,
1* BitMap
120
*/
*/
/* DetailPen, BlockPen *1
1* IDCMP Flags
*/
1* Flags
*/
*1
*1
*1
*/
3.1 WINDOWS
ABACUS
640, 50,
640, 200,
WBENCHSCREEN,
};

1* Type
*/
*/
*/
main 0
(
ULONG MessageClass;
USHORT code;
FOREVER
{

(
ReplyMsg(message);
{
case CLOSEWINDOW : Close_All();

exit(TRUE);
break;
1***************************************
* Function: Open everything
*
*
* =============%=:=================== *
*
*
* Author:
* Wgb
Date:
Comments:
10/20/1987
* No parameters
*
*
*
*
*
*
***************************************/
struct Library *OpenLibrary();

if (!(IntuitionBase = (struct IntuitionBase *)
OpenLibrary ("intuition library", OL)
121
3. INTUITION AND C
AMIGA

Close_All ();
exit(FALSE);
}

OpenWindow(&ConsoleWindow)
{
printf ("Window can't be opened! \n");

Close_All () ;
exit (FALSE) ;
}
/***************************************
*
*
*
*
*
*
*
*
*
Function: Close everything
*
*
=================================== *
Author:
Wgb
Date:
Comments:
10/20/1987
No parameters
*
*
*
*
*
*
***************************************/
if (FirstWindow)
if (IntuitionBase)
122
3.2
ABACUS
3.2
SCREEN FUNDAMENTALS
Screen Fundamentals
The last section described many parameters which affect the appearance
of windows. Screens play an even more important role in graphic
display. Each window parameter depends on the screen to some extent.
Window color, resolution and position can be affected by the screen, and
base their definitions on the context of the screen.
What purpose does the screen serve in Intuition? The screen contains
the essential display parameters. It acts as the background for all visual
output But output has different uses, and there are many different basic
assumptions. Let's take two really crude examples:
1.)
Word processing's main use for screen output results in text

display. As you can see from any book you only need two colors
for this (one for text and one for background). It is important that
the characters are displayed in high resolution for easy reading.
2.)
Multi-color graphics need many colors, as the name suggests.

The resolution is important, but some modes don't place as
much importance on resolution (such as HAM).
These two examples make it clear that different basic properties have to
be available to program. The Amiga goes one step farther here. It
allows the programmer to display several different modes using screens,
where only one might normally be displayed.
Intuition allows any number of these screens with any number of
configuring modes. In principle the procedure is similar to that used for
windows. The following pages explain this in more detail.
3.2.1
Creating custom screens

We use a different method for analyzing screen attributes than we used
for analyzing windows. You may have become familiar with several
types of windows from working with the Workbench. Screens in general are more limited in scope, so the Amiga user may assume that the
Workbench screen is all there is. Not so.
As with all operating system routines, structures are used to pass the
required information. Therefore, a NewScreen structure exists for
creating new screens, similar to the Newwindow structure for creating
windows.
123
3.
INTUITION AND
3.2.1.1
The
AMIGA
NewScreen
structure
Lets take a look at this NewScreen structure and its individual parameters. Here it is:
1* 3.2.1.1.A.newsereen_struet *1
struet NewSereen
{
OxOO
Ox02
Ox04
Ox06
Ox08
OxOa
OxOO
OxOe
OxOe
Ox10
Ox14
Ox18
Ox1e
Ox20
};
00
02
04
06
08
10
11
12
14
16
20
24
28
32
SHORT LeftEdge;
SHORT TopEdge;
SHORT Width;
SHORT Height;
SHORT Depth;
UBYTE DetailPen;
UBYTE BloekPen;
USHORT ViewModes;
USHORT Type;
struet TextAttr *Font;
UBYTE *DefaultTitle;
struet Gadget *Gadgets;
struet BitMap *CustomBitMap;
The fIrst four arguments have similar meanings as in the Newwindow

structure. LeftEdge and TopEdge specify the position of the screen
on the monitor. The current versions of Intuition only allow you to
change the vertical position, however. LeftEdge is included for
compatibility with future versions.
Width and Height let you specify the screen width and height in
pixels. You may change the resolution size within limits. For example.
you can create a screen only 100 pixels wide but 300 pixels high in
non-interlace mode. The maximum values for any screen are 72Ox400
pixels (720X456 PAL).
The Depth setting of a screen appears as a new argument Here you set
the number of bit-planes allocated for a screen. This determines both the
number of colors that can be displayed at once and the amount of
memory required. We are interested mainly in the number of colors,
calculated according to the formula Colors = 2"Depth.
Let's look at the parameters DetailPen and BlockPen. These
determined the colors which Intuition used to draw windows. Carried
over to the screen, they set the colors used by Intuition when drawing
the title bart the front gadget and back gadget.
It should be noted that only colors allowed by the Depth can be seL
For example, if you set a Depth of three bit-planes, then the
maximum number of colors allowed is eight (numbered zero to seven).
124
3.2 SCREEN FUNDAMENTALS
ABACUS
Therefore, seven is the highest value for DetailPen or BlockPen.

The same is true for windows, which are completely dependent of the
screen settings.
Mter the pen colors comes a flag by the name of ViewModes in the
structure. This is the most important setting-resolution.
NewScreen
As we mentioned before, different types of programs have different res0lution requirements. Intuition has a maximum resolution of 320 pixels
in the X direction in normal cases. This means that in the rectangular
space of your screen there are 320 horizontal points. These pixels are
relatively wide and can easily be seen by the eye. In higher resolution
the pixel width splits in half, and you have 640 horizontal points.
It's a similar situation in the vertical direction. Normally this is divided

into 200 rows in the NTSC version of the Amiga (256 in the PAL
version). Here too, you can double the number of lines, but it's
accomplished in a different manner. The electron beam of the cathode
ray tube writes two sets of information to the screen in alternation, each
set shifting slightly vertically from the other.
Here is a summary of all the possible screen settings:
Table 3.3:
ViewModes
Flat name
OxOOOO8000L
OxOOOOOOO4L
Description
Screen with double X resolution
Screen with double Y resolution
HAM
OxOOOOO800L
EXTRA_HALFBRITE OxOOOOOO80L
4096 color mode

64 color mode, 5 bit-planes
HIRES
LACE
Hex value
PFBA
OxOOOOOO40L
DUALPF
SPRITES
OxOOOOO400L
OxOOOO4000L
VP_HIDE
OxOOOO2000L
GENLOCK AUDIO
GENLOCK VIDEO
OxOOOOOlOOL
OxOOOOOOO2L
Dual playfield mode

Sprites can be used
Combines video picture into the

graphics through hardware
Before discussing the next value in the structure we would like to

clarify the relationship between Depth and ViewModes. Since the
video chip must manage a large amount of memory for your screens, it
can become overloaded. You shouldn't be surprised if the video chip
can't display more than 16 colors in 640 pixel resolution. With 320
pixels it can display the maximum of 32 colors. If you want to use
more colors, you must use the HAM mode instead of the bit-map
method.
125
3.
INTUITION AND
Screen type
The next structure value sets the screen type. There are only two types:
the Workbench screen, opened by the system, and the CustomScreen
(user screen).
Associated with these two flags are three others which let you enter
special features. Intuition must organize memory for the bit-maps. You
can do this yourself, however, and set it with another parameter. To
inform Intuition that you have already allocated graphic memory, you
must set the third flag.
The fourth flag enables you to open the new screen behind all others.
This lets you draw a graphic while the screen is hidden without having
to fIrst open the screen and then move it to the background.
The fIfth flag disables drawing the system gadgets and the title bar. In
the program you must then ensure that the right mouse button cannot
be used. This can be done by setting a flag. Otherwise the title bar with
the menus would be drawn, but since deleting is disabled, it would not
be removed.
Here is a summary of the two screen types and the fIve flags just
described:
Table 3.4:
Screen type
Hex value
Description
FlaK name
WBENCHSCREEN OxOOOOOOOlL Screen type
CUSTOMSCREEN OxOOOOOOOFL
CUSTOMBITMAP
SCREENBEHIND
SCREENQUIET
SHOWTITLE
BEEPING
OxOOOOOO40L Custom bit-map

OxOOOOOO80L Screen opens in the background
OxOOOOOlOOL No gadgets
OxOOOOOOlOL (These last two flags set by
Intuition itself)
OxOOOOOO20L Screen flashes
Just like each window, a screen has a title bar containing a text string.
You can set this title through a window. But if no windows are active,
the title bar must remain empty.
Text insertion The next two parameters allow basic text to be displayed. First, you can
use Font to point to a TextAttr structure, from which you
determine the font used for the screen title and all other Intuition text
displayed in the screen. If you want to use the font which the user
selected with Preferences, use the value null (0) instead of a
pointer here.
The second value is simply a pointer to a string. If you use null here,
no title appears in the title bar.
All we can say about the pointer to the gadget structure is that it was
included in the NewScreen structure for upward compatibility. Unfor-
126
3.2
ABACUS
SCREEN FUNDAMENTALS
tunately, you can't add custom gadgets to a screen. You should always
initialize this value as null.
The last pointer normally points to a Bi tMap structure initialized by
the program itself. If you don't use this function, you should use the
value null here.
3.2.1.2
The first screen listing

Before we go through the NewScreen structure in detail as we did for
the NewWindow structure, you should take a look at the following
listing:
Program 3.5:
Window on
custom screen
1***************************************1
1*
*1
1* 3.2.1.2.A customscreen_1.c
*1
1*
*1
1* Program: Window on Custom Screen
*1
1* =================================== *1
1*
*1
Comments:
1* Author: Date:
*1
1*
*1
10/16/1987 simple Screen
1* Wgb
*/
*/
1*
*/
1*
/***************************************/
finclude <exec/types.h>
struct
struct
struct
struct
IntuitionBase
Screen
Window
IntuiMessage
*IntuitionBase;
*FirstScreen;
*FirstWindow;
*message;
struct NewScreen FirstNewScreen =

0, 0,
640, 200,
1,
0, 1,
HIRES,
CUSTOMSCREEN,
NULL,
(UBYTE *)"Screen Test",
NULL,
NULL,
/* Width, Height
/* Depth
1* DetailPen, BlockPen
/* ViewModes
/* Type
/* Font
*/
*/
*/
*/
*/
/* Gadgets
/* CustomBitMap
*/
*/
*1
*/
127
3.
INTUITION AND
};
1*
1*
1*
1*
1*
160, 50,
320, 150,
0" I,
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
CLOSEWINDOW,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *) "Test Custom-Screen",
NULL,
1* Screen *1
NULL,
1* BitMap
100, 50,
640, 200,
CUSTOMSCREEN,
1* Type
};
main ()
(
ULONG MessageClass;
USHORT code;
FOREVER
{

{
ReplyMsg(message);
{
case CLOSEWINDOW : Close_All()t

exit (TRUE) ;
break;
128
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
3.2
ABACUS
SCREEN FUNDAMENTALS
/***********************************************
* Function: Open library, screen and window
* =========================================== *
*
*
Comments:
* Author: Date:
*
---------- ---------*
*
10/16/1987
* Wgb
*
*
*
*
*
***********************************************/
Open_All ()
(
void
*OpenLibrary();
struct Screen *OpenScreen();
OpenLibrary(lintuition.library", OL)
(
printf("Intuition library cannot be found!\n");

Close_All () ;
exit(FALSE);
}
if (! (FirstScreen = (struct Screen *)

OpenScreen(&FirstNewScreen)
{
pr int f (" Screen cannot be opened! \n ") ;

Close_All () ;
exit(FALSE);
}
FirstNewWindow.Screen = FirstScreen;
(
printf("Window cannot be opened!\n");

Close_All () ;
exit(FALSE);
}
/***************************************
* Function: Close everything opened
* =================================== *
*
*
Comments:
* Author: Date:
*
*
*
---------- ---------10/16/1987 Window, screen *
* Wgb
and intuition
*
*
*
*
***************************************/
129
3.
INTUITION AND
AMIGA
Close_All ()
{
if (FirstWindow)
if (FirstScreen)
if (IntuitionBase)
CloseScreen(FirstScreen);
How it works
You should be familiar with the structure of this listing. After the general pointer and structure definitions comes the main function which
calls the OpenAllO function, then branches to a loop which waits
until you click on the close gadget The CloseAllO function ends the
program. Let's look at the four main sections:
Among the pointers we find one for screen structure. It places the
window later. First you need the NewScreen structure, which opens a
simple screen which has the same properties as the Workbench screen,
with the exception that you request only two colors (i.e., one bitplane).
You should recognize the ma~n program since we explained how 10
read the close gadget at the end of the last section.
The Open AllO function includes a screen opening routine. The order
of openingl.s as follows: Library (no library, no openScreenO function), screen. window. A pointer to the custom screen appears in the
window.
The Close AII0 function closes the window ftrst, then the screen,
then the Iibffiry. Everything occurs in the reverse of the order of the
opening process.
If you like, you can change one or two of the arguments of the
NewWindow or NewScreen structure. Don't make too many changes.
at once though, since it's easy to overlook an error.
3.2.1.3
The Screen structure

After this programming interlude, let's return to the Screen structure
set up by Intuition. It contains many values. pointers and flags. Here is
the structure, complete with offsets:
/* 3.2.1.3.A. screenstructure *1
struct Screen
{
OxOO
Ox04
OxOS
OxOA
130
00
04
08
10
struct Screen *NextScreen;

struct Window *FirstWindow;
SHORT LeftEdge;
SHORT TopEdge;
ABACUS
OxOC
OxOE
Ox10
Ox12
Ox14
Ox16
Ox1A
Ox1E
Ox1F
Ox20
Ox21
Ox22
Ox23
Ox24
Ox25
Ox26
Ox27
Ox2B
12
14
16
18
20
22
26
30
31
32
33
34
35
36
37
38
39
43
SHORT Width;
SHORT Height;
SHORT MouseY;
SHORT MouseX;
USHORT Flags;
UBYTE *Title;
UBYTE *DefaultTitle;
BYTE BarHeight;
BYTE BarVBorder;
BYTE BarHBorder;
BYTE MenuVBorder;
BYTE MenuHBorder;
BYTE WBorTop;
BYTE WBorLeft;
BYTE WBorRight;
BYTE WBorBottom;
struct TextAttr *Font;
struct ViewPort ViewPort;
OxOO
Ox04
Ox08
OxOC
Ox10
Ox14
Ox18
Ox1A
Ox1C
Ox1E
Ox20
Ox22
Ox23
Ox24
Ox28
00
04
08
12
16
20
24
26
28
30
32
34
35
36
40
struct ViewPort *Next;

struct ColorMap *ColorMap;
struct CopList *Dsplns;
struct CopList *Sprlns;
struct CopList *Clrlns;
struct UCopList *UCoplns;
SHORT DWidth;
SHORT DHeight;
SHORT DxOffset;
SHORT DyOffset;
UWORD Modes;
UBYTE SpritePriorities;
UBYTE reserved;
struct Raslnfo *Raslnfo;
};
Ox53
83
struct RastPort RastPort;
OxOO
Ox04
Ox08
OxOC
Ox10
Ox14
Ox18
Ox19
Ox1A
Ox1B
Ox1C
OxlD
Ox1E
Ox1F
Ox20
Ox22
Ox24
Ox26
Ox28
00
04
08
12
16
20
24
25
26
27
28
29
30
31
32
34
36
38
40
struct Layer *Layer;

struct BitMap *BitMap;
USHORT *Areaptrn;
struct TmpRas *TmpRas;
struct Arealnfo *Arealnfo;
struct Gelslnfo *Gelslnfo;
UBYTE Mask;
BYTE FgPen;
BYTE BgPen;
BYTE AOlPen;
BYTE DrawMode;
BYTE AreaPtSz;
BYTE linpatcnt;
BYTE dummy;
USHORT Flags;
USHORT LinePtrn;
SHORT cp_x;
SHORT cp_y;
UBYTE minterms[8];
131
3.
INTUITION AND
AMIGA
Ox30
Ox32
Ox34
Ox38
Ox39
Ox3A
Ox3C
Ox3E
Ox40
Ox42
Ox46
Ox4E
Ox5C
Ox64
48
50
52
56
57
58
60
62
64
66
70
78
92
100
SHORT PenWidth;
SHORT PenHeight;
struct TextFont *Font;
UBYTE AlgoStyle;
UBYTE TxFlags;
UWORD TxHeight;
UWORD TxWidth;
UWORD TxBaseline;
WORD TXSpacing;
APTR *RP_User;
ULONG longreserved[2];
UWORD wordreserved[7];
UBYTE reserved[8];
};
OxB7 183
struct BitMap BitMap;
OxOO
Ox02
Ox04
Ox05
Ox06
Ox08
Ox28
00
02
04
05
06
08
40
UWORD BytesPerRow;
UWORD Rows;
UBYTE Flags;
UBYTE Depth;
UWORD pad;
PLANEPTR Planes[8);
};
OxDF 223
struct Layer_Info LayerInfo;
OxOO
Ox04
Ox08
OxOC
Ox18
Ox3C
Ox4A
Ox4E
Ox50
Ox51
Ox52
Ox54
Ox58
Ox5C
00
04
08
12
24
60
74
78
80
81
82
84
88
92
struct Layer *top_layer;

struct Layer *check_lp;
struct Layer *obs;
struct MinList FreeClipRects;
struct SignalSemaphore Lock;
struct List gs_Head;
LONG longreserved;
UWORD Flags;
BYTE fatten_count;
BYTE LockLayersCount;
UWORD LayerInfo_extra_size;
WORD *blitbuff;
struct LayerInfo_extra *LayerInfo_extra;
};
Ox13B
Ox13F
Ox140
Ox141
Ox143
Ox147
Ox14B
Ox14F
315
319
320
321
323
327
331
335

UBYTE Detai1Pen
UBYTE BlockPen;
USHORT SaveColorO;
struct Layer *BarLayer;
UBYTE *ExtData;
UBYTE *UserData;
};
*NextScreen Like the windows, all open screens are linked together. The link pointer resides in this variable.
132
3.2
ABACUS
SCREEN FUNDAMENTALS
*FirstWindow
Each screen contains the pointer of the fIrst window

opened. The windows are then linked together.
LeftEdge, TopEdge, width and Height
These four values were explained in conjunction with

the NewScreen structure above.
MouseX,MouseY
As with the windows, these variables contain the

mouse coordinates relative to your screen.
Flags
This value is initialized as in the NewScreen structure, with two additional flags:
SHOWTITLE
This flag is set when the title bar is displayed with

ShowTitleO
BEEPING
This flag is set when the screen flashes with

DisplayBeepO.
*Ti tIe
As with the windows, this is a pointer to the title

text string.
*DefaultTitle
This pointer comes into play when a window is

opened without a specifIc title. Intuition then uses the
DefaultTitle.
The following nine variables of the Screen structure apply to all

windows in the screen as well as the screen itself:
BarHeight
Title bar height, measured in screen lines.
BarVBorder
Vertical border width.
BarHBorder
Horizontal border width.
MenuVBorder Vertical menu border width.

MenuHBorder Horizontal menu border width.
WBorTop
Top window border width.
WBorLeft
Left window border width.
WBorRight
Right window border width.
WBorBottom
Bottom window border width.
*Font
Pointer to the default font to be used by Intuition.

Initialized by the NewScreen structure.
133
3.
INTmTION AND
AMIGA
ViewPort
This structure contains infonnation about the display

needed for the Copper, the video chip.
RastPort
This structure manages the screen and contains all

infonnation required for drawing on the screen.
Bi tMap
This structure contains more precise infonnation

about the bit-map. It also contains pointers to each
bit-map.
Layerlnfo
Layers are the basis of the window management. This

infonnation structure is the beginning.
*FirstGadget
Pseudo-pointer to the screen custom gadgets (not

implemented).
DetailPen,BlockPen
The two drawing pens, as explained for NewScreen

and Newwindow.
3.2.1.4
SaveColorO
Since the background color changes when a screen

flashes, this location acts as a buffer for color O.
*BarLayer
This pointer points to the layer (graphic storage area)

in which the title bar of the screen is stored.
*ExtData
Pointer to external data which can be added.
*UserData
Pointer to data managed by the user.
The screen functions

As you learned in the example programs, there are also open and close
functions for screens, called OpenScreenO and a CloseScreenO.
The arguments are also similar: the pointer to the NewScreen structure for the fonner, and the structure pointer returned by Intuition for
the latter.
As you may have guessed, the user has quite a number of screen functions available for him/her. Let's take a look at them.
For instance, Intuition functions support clicking on the front and back
gadgets. With ScreenToFrontO and ScreenToBackO you can
move any screen to which you have the pointer to the foreground or
background. To demonstrate this, remove the gadget test from the
screen example program listed above. The main function should then
look like this:
134
ABACUS
main ()
{
Delay(180L);
Close_All () ;
exit (TRUE) ;
)
The DelayQ line must also be inserted so that you notice that
something is happening when the new screen opens. Add all of the
extensions after the Delay(180L); line unless told otherwise. First an
example of ScreenToBackO and ScreenToFrontO:
ScreenToBack(FirstScreen);
Delay(180L);
ScreenToFront(FirstScreen);
This very simple example moves the screen to the background and the
foreground after a short delay. but it shows you how to use the
functions. Here is the general syntax with the arguments, registers and
function offset
ScreenToBack(Screen);
-246
aO
ScreenToFront(Screen);
-252
AD
The same possibilities which can be used for a custom screen can also
be used for the Workbench screen, of course. Normally the problem is
that you don't have a pointer to the Workbench screen. Another
function does this without using pointers:
WBenchToBack();
-336
WBenchToFront();
-342
Insert these three lines:

WBenchToFront();
Delay(180L);
WBenchToBack();
You may be familiar with the following situation in BASIC: You try
to move out of the text area with the cursor in the editor and the screen
flashes and beeps. Intuition sends this flash as a warning. The following routine produces five warning signals in a row:
135
3. INTUITION AND C
for (i=l; i<6; i++)

{
for (j=O; j<10000; j++);

DisplayBeep(FirstScreen);
}
To start the new program, fIrst declare the two loop variables i and j
as into The program then flashes the new custom screen fIve times.
If you want to flash all of the screens (e.g., if you don't know what
screen is currently in the foreground), then null can be used as the
pointer to tell Intuition that it should flash all of the screens:
DisplayBeep(NULL);
Here is the format of the function call:

DisplayBeep(Screen);
-96
No
WorkBench
AO
Many applications may run out of graphic memory very quickly. This
cannot be corrected by expanding the memory because the chip RAM
area cannot be expanded beyond 512K.
The Amiga developers decided that a program would not need the Workbench screen if it opened a screen of its own. In these cases you can
close the Workbench screen. The only condition: No other programs
can be running which display data on a window in the Workbench
screen. If this condition is met, Intuition lets you close the Workbench
window with CloseWorkBenchO.
How can you start a program without using the Workbench? Unfortunately you cannot start your program from the CLI window because the
CLI is an independent program which outputs to the Workbench screen.
Therefore we will give the program an icon so that it can be called by
double-clicking on the icon.
Add the following lines to CustomScreenl. c program and save it
program, and remember
to link it with startup.o. (You may have to compile startup.h).
At the end of the Open_AllO function after opening the window add:
to disk as CloseBench . c. Compile the new
CloseWorkBench();
At the start of the Close_AllO function add:

OpenWorkBench();
Next copy the CLI. INFO icon to CloseBench. INFO. Close all
programs that you started from the workbench, the CLI is a running
program, and start the Close Bench program. When you pull the
screen down you will notice that there is no longer a Workbench screen
behind it.
136
3.2
ABACUS
Moving
Screens
SCREEN FUNDAMENTALS
Each screen whose title bar is not hidden by a BACKDROP window can
be moved vertically by the user. As with windows, an Intuition
function allows screens to be moved. This is no problem with
MovescreenO and the corresponding delta values.
We have prepared two lines which move the screen back up if you
move it down. Insert the following lines in CustomScreenl. c
program in the FOREVER loop before the message test:
if (FirstScreen->TopEdge != 0)
MoveScreen(FirstScreen, 01, -11);
The general format of MoveScreenO is:

MoveScreen(Screen, DeltaX, DeltaY);
-162
AO
DO
Dl
Since the ShowTitleO function was explained in the window section,

just two functions remain which deal with screens. The first is
GetScreenDataO, which returns a summary of screen data.
All of the values could be read through a simple structure access, but
then you don't get a summary at a particular time because all of the
parameters could change as you read them. Therefore you set up a buffer
for this function in which the data are stored. Then you call the function
and all of the arguments transfer to your buffer. This is especially
useful if you want information about the Workbench screen. Since you
can specify the type of the window with a flag, it is easy to set the
Workbench type. This allows you to easily acquire normally inaccessible data. Since an example is rather difficult to formulate, here is the
syntax:
Success
DO
= GetScreenData(Buffer,
-426
AO
Size, Type, Screen);

DO
D1
Al
MakeScreenO lets you manipulate a screen managed by Intuition.

After calling MakeScreenO and passing it a screen pointer, it waits
until Intuition View is no longer needed, performs its task using
MakeVPortO and releases Intuition View.
Here's the format of MakeScreenO:

MakeScreen(Screen);
-378
AO
These are all the Intuition functions which control screens. We hope the
examples helped you understand the functions.
137
3.
INTUITION AND
3.2.2
AMIGA
Examples of using screens

In conclusion we would like to present some short programs and
extended the program sections which you can use in your existing
programs. You'll find that some of these program excerpts are intended
for the "big editor project" later on in this book.
Unfortunately, we don't have all that many examples. This is because
screens are required only for special applications.
3.2.2.1
General examples
In this section we present two screen examples in the subject of graphic
programs. These examples may give you some ideas for your own
programming.
First is a super-high-resolution screen with just one bit-plane. Superhigh-resolution means a resolution of 640x400 pixels on an NTSC
monitor (or 640x512 on a PAL monitor). We chose just one bit-plane
because we use enough memory in the higher resolution. You must
remember that you only have 512K of graphic memory available
(CHIP_MEMORY).
Structure 3.3:
Super screen
1* 3.2.2.l.A. superscreen struct *1

struct NewScreen SuperScreen
0, 0,
640, 400,
1,
0, 1,
HIRES I
LACE,
CUSTOMSCREEN
SCREENQUIET,
NULL,
NULL,
NULL,
NULL,
};
1*
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
Depth
DetailPen, BlockPen
ViewModes
*1
*1
*1
*1
*1
1* Type
*1
1* Font
*1
1* Gadgets
/* CustomBitMap
*1
*1
This mode is suited for digitized pictures and for super hi-res graphics.
Here is the complete example program:
138
3.2
ABACUS
SCREEN FUNDAMENTALS
/***************************************
* Program: Superscreen.c
*
* =================================== *
*
*
Comments:
* Author: Date:
*
*
*
---------- ---------10/16/87
one Screen
* Wgb
*
*
*
*
*
***************************************/
struct
struct
struct
struct
IntuitionBase
Screen
Window
IntuiMessage
*IntuitionBase;
*FirstScreen;
*FirstWindow;
*message;
struct NewScreen Super Screen
0, 0,
640, 400,
1,
0, 1,
HIRES
LACE,
CUSTOMSCREEN
SCREENQUIET,
NULL,
NULL,
NULL,
NULL,
};
/*
/*
/*
/*
/*
LeftEdge, TopEdge
Width, Height
Depth
DetailPen, BlockPen
ViewModes
*/
*/
*/
*/
*/
/* Type
*/
/* Font
*/
/* Gadgets
/* CustomBitMap
*/
*/
160, 50,
320, 150,
/* Width, Height
0, 1,
CLOSEWINDOW,
/* IDCMP Flags
/* Flags
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
/* First Gadget
NULL,
NULL,
/* CheckMark
(UBYTE *)"Test Custom-Screen",
NULL,
/* Screen
/* BitMap
NULL,
*/
*/
*/
*/
*/
*/
*/
*/
*/
139
3.
INTUITION AND
AMIGA
100, 50,
640, 150,
CUSTOMSCREEN,
};

/* Type
*/
*/
*/
main ()
{
Open_All () ;
Delay(180L);
Close_All 0 ;
exit (TRUE) ;
}
/***********************************************
* Function: Library, Screen and Window open
=========~============================~===
**
*
***********************************************1
Open_All 0
{
void
*OpenLibrary();
{
printf ("Intuition Library not found! \n");

Close_All 0 ;
exit (FALSE) ;
)
if (! (First Screen = (struct Screen *)

OpenScreen(&SuperScreen)
{
printf("Screen not opened, no page!\n");

Close_All();
exit (FALSE) ;
}
FirstScreen;

{

Close_All ();
exit(FALSE);
}
140
3.2
ABACUS
SCREEN FUNDAMENTALS
1***************************************
* Function: Close everything opened *

* =================================== *
*
*
***************************************1
Close_All ()
{
if (FirstWindow)
if (FirstScreen)
if (IntuitionBase)
Partial screens In the second example we want to show you that you don't have to use
the full resolution. An interlace screen can also be initialized with 200
lines. It then takes up only half the monitor. It should be mentioned
that any other screen used along with an interlace screen also flickers.
Structure 3.4:
Utility screen
1* 3.2.2.1.B. screen util *1

struct NewScreen Utility =
1*
1*
1*
1*
HIRES,
1*
CUSTOMSCREEN,
1*
NULL,
1*
(UBYTE *) "Utility-Screen",
NULL,
1*
NULL,
1*
0, 150,
640, 50,
2,
2, 1,
LeftEdge, TopEdge
Width, Height
Depth
DetailPen, BlockPen
ViewModes
Type
Font
*1
*1
*1
*1
*1
*1
*1
Gadgets
CustomBitMap
*1
*1
};
The structure presented here opens a screen in the lower 50 lines of the
Workbench screen. You might want to open a BORDERLESS window
here in which you could place gadgets for calling up various utility
functions. Be sure to adjust your window size so it fits on the screen.
3.2.2.2
Program sections for the text editor

You may want to use your C source code editor while the compiler is
running, or you may want it removed completely from the screen if, for
example, there are too many windows open on the Workbench and the
processing time takes too long. In that case, it's a good idea for a new
program to open a new screen. Then you can exit the program and work
on the Workbench undisturbed.
The structure presented here is configured especially for text editing. It
uses only one bit-plane to use as little memory as possible. This is
141
3.
INTUITION AND
AMIGA
C FOR
ADVANCED PROGRAMMERS
enough, since text doesn't rely heavily on colors. High-resolution

makes the text easily readable.
Structure 3.5:
Editor screen
/* 3.2.2.2.A. edscreen */
struct NewScreen EditorScreen =
0, 0,
640, 200,
/* Width, Height
1,
/* Depth
1, 0,
HIRES,
/* ViewModes
CUSTOMSCREEN,
/* Type
NULL,
/* Font
(UBYTE *)"Abacus C Editor Screen",
NULL,
/* Gadgets
NULL,
/* CustomBitMap
};
142
*/
*/
*/
*/
*/
*/
*/
*/
*/
3.3
ABACUS
3.3
OUTPUT
Output
After having completed the frrst two sections of this chapter, you now
know almost every capability of Intuition screens and windows; how
they open and close; and how other functions interact with them.
But what use is a screen or window that appears on the screen and does
nothing? Except, perhaps, letting the user move it up or down.
At the beginning of this book we mentioned that windows are the basis
for all input and output. The screens simply exist to subdivide areas and
specify default properties of the windows. You know how to set up
screens and windows, but how do you control input and output? This
section discusses the subject of output.
The Amiga offers two types of output. The frrst type comes from the
output functions found in graphics.library. These are the basis
for all of the screen output functions. The second type is supported by
intuition.library. It offers only three output functions, but they
are flexible enough that all other elements can be constructed from
them. All user elements such as gadgets, requesters, and menus use the
three graphic functions supplied by Intuition.
You see how important it is to explore these functions. Later you will
see how easy it is to construct the most complex objects using the
building block system of Intuition.
In addition, you'll get to use the windows for a real application.
3.3.1
Text output
We'll go through the individual output options step by step. Let's look
at text output frrst, since it's the most important form of communication with the user.
When starting out, it is important to know the parameters your text can
have. They are easy to understand.
143
3.
INTUITION AND
3.3.1.1
AMIGA
Position, color, character mode

About the position: You must determine the exact pixel at which your
text is written relative to your window. In addition, you can set the
color used for the background and foreground (the text lines
themselves). Here you access a certain number of color pens which the
screen offers you. Pen number -1 (OxFP) plays a special role. This
number represents the window's default color.
As you are probably aware, the operating system has various modes
available for producing OutpuL

JAMl
The first mode takes into account what is drawn but not the
background. This displays the text lines only, without disturbing the
background. This mode is called JAMl because color 1 is "jammed"
into the graphic.
JAM2
In the second mode Intuition displays both the characters and the back-
ground. This mode is called JAM2 because both colors are drawn.
INVERSEVID
The third mode is similar to the second except that colors are
exchanged. You need INVERSEVID for the default colors, however,
which are designated with -1 (OxFF) and cannot be simply exchanged.
COMPLEMEN'r
Mode number four, COMPLEMENT, is similar to one except that the

background is drawn correspondingly. Set points are erased and erased
points are seL
Here are the four modes:
Table 3.5:
DrawModes
Drawin! mode
JAMl
JAM2
INVERSEVID
COMPLEMENT
Description
Only pen 1 used
Both pens used
Like JAM2, with pens exchanged
Like JAMl set points cleared,-cleared set
With the last text parameter you can specify a character font as a
TextAttr structure.
3.3.1.2
The :IntuiText structure

As usual,the Intuition parameters are stored in a structure, which looks
like this:
144
ABACUS
3.3 OUTPUT
1* 3.3.1.2.A. intuitext_struct *1
struct IntuiText
{
OxOO
Ox01
Ox02
Ox03
OxOs
Ox07
OxOB
OxOF
Ox13
};
00
01
02
03
05
07
11
15
19
OBYTE FrontPen;
OBYTE BackPen;
OBYTE DrawMode;
SHORT LeftEdge;
SHORT TopEdge;
struct TextAttr *ITextFont;
OBYTE *IText;
struct IntuiText *NextText;
The last two elements of the structure need some explaination. IText
represents a pointer to a string terminated by null. With the pointer
Next Text, the system makes it possible to link several such
Int ui Text structures together into a chain. The advantage of this lies
in the call to produce the output. This capability means that you aren't
limited to a single line or color setting per call. This allows maximum
flexibility with just one function call.
3.3.1.3
Fonts and type styles

Any of the fonts on the Workbench disk can be used, but Intuition
offers another option.
A default font can be set for any window. If you want to use this font
(generally set in Preferences), then you just use null instead of a
pointer.
You can also access a custom font through the diskfont.library.
You should usually use the two ROM fonts, however, to retain
readability and consistency. These fonts are always accessible and have
uniform dimensions.
To select the two ROM fonts you fIrst need the TextAttr structure,
which defInes both the font and its style. ThiS structure looks like this:
1* 3.3.1.3.A. textattributes *1
struct TextAttr
{
OxOO
Ox04
Ox06
Ox07
Ox08
00
04
06
07
08
STRPTR ta_Name;
OWORD ta_YSize;
OBYTE ta_style;
OBYTE ta_Flags;
);
145
3.
INTUITION AND
AMIGA
t a _ Narne contains a pointer to a string which states the font name. If

you want to use the ROM fonts, insert topaz.font. Make sure that
the font name is entirely in lowercase letters or the operating system
will not recognize it. If you have opened other fonts with the
diskfont.library, you can access them as well.
With the next value you set the height of the font in pixels. This
determines the size of the font and also the number of characters per
line. ROM fonts have two sizes:
Table 3.6:
Standard fonts
Characters
Font size
TOPAZ SIXTY
Sixty
Eighty
TOPAZ EIGHTY
Height
At 640 pixels = 9 pixels
At 640 pixels = 8 pixels.
The next element in the TextAttr structure, ta Style, allows you

to set something called the soft-style. This lets you change the appearance of a normal font by applying certain functions to it. Using this
you can implement underlining, italics, boldface and extended characters. The following flags are available and can be freely combined:
Table 3.7:
Type styles
Flag name
FS NORMAL
FSF ITALIC
FSF- BOLD
FSF UNDERLINED
FSF EXTENDED
Hex value
OxOO
Ox04
Ox02
OxOl
Ox08
Meaning
No effect
Italic
Boldface
Underline
Extended characters
In the last value of the structure you use flags to specify additional
information about your character set. These flags are called font preference flags. We are interested in just two of these flags-the two which
indicate from where Intuition should get the font. You can use ROM
fonts or disk fonts. Since topaz.font is in ROM, you set the corresponding flag:
Table 3.8:
Font type
Flag name
FPF ROMFONT
FPF DISKFONT
FPF REVPATH
FPF_TALLDOT
FPF WIDEDOT
FPF PROPORTIONAL
FPF DESIGNED
FPF REMOVED
Hex value
OxOl
Ox02
Ox04
Ox08
OxlO
Ox20
Ox40
Ox80
Meaning
Get font from ROM
Get font from disk
To use topaz.font with a line length of 80 characters and normal

appearance, the structure would look like this:
/*3.3.1.3.B. test font */
struct TextAttr TestFont
{
(STRPTR) "topaz. font" ,
146
3.3 OUTPUT
ABACUS
TOPAZ_EIGHTY,
FS_NORMAL,
FPF_ROMFONT
};
3.3.1.4
PrintIText
Now that you've seen the individual parameters, it's time to work with
the IntuiText structure.
The P r i n tIT ext function is required for displaying the
IntuiText. In addition, you need an argument from the window in
which the output appears. Your argument is the RastPort, without
this it is impossible to perform any output.
You can do this with the following definition and assignment:
struct RastPort *MyWindowRastPort;
MyWindowRastPort = MyWindow->RPort;
Now you need an appropriate IntuiText structure to start output.

We've written one for you:
1* 3.3.1.4.A.firsttext *1
struct IntuiText FirstText
1*
1*
1*
1*
FrontPen, BackPen
DrawMode
LeftEdge, TopEdge
15, 0,
Font (Standard)
NULL,
(UBYTE *) "System programming on the Amiga!",
1* NextText
NULL
1, 0,
JAM2,
*1
*1
*1
*1
*1
};
Insert this definition in the Customscreenl . c program. Remember

that this definition must precede the mainO function.
RastPort can be determined in the mainO function after
Open_ AllO fmishes with the assignment statement given above.
You are almost ready to call PrintITextO. All you need yet are
coordinates.
147
3.
INTUITION AND
You may wonder why you need them. Take a look at the following
graphic:
B 59 199
PrintIText()Offset-Defto
9~iiii~==========:3~.
Offset of window from function (S9l, Sal)
50
190
Itt:"'"
Hello. this Is a test.
T
Relative position of structure
(2B, 20)
Figure 3.3
When you call the PrintITextO function, you determine the point
relative to the RastPort to which all other parameters in the structures refer. This point becomes the origin. The position specified in the
IntuiText structure then relates to this. Insert the following line in
your Customscreenl. c program:
PrintIText(MyWindowRastPort, FirstText, 10L, 20L);
Note:
to
Be sure to specify the offsets of Print ITex as long values or the

Aztec C compiler may have problems. Here is the complete listing:
1***************************************
*
*
* ===================z=z=======z===== *
*
*
**
Program: PrintIText.c
Comments:
* Author: Date:
*
*
*
10/16/1981 PrintIText Test *
* Wgb
Window
*
*
*
*
***************************************/
--------- ----------
iinc1ude <exec/types.h>
iinclude <intuition/intuition.h>
struct Window
*FirstWindow;
148
ABACUS
3.3
160, 50,
320, 150,
/*
/*
/*
/*
/*
0, 1,
LeftEdge, Top Edge

Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
NULL,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
/* First Gadget
NULL,
/* CheckMark
(UBYTE *)"System programimg test",
NULL,
/* Screen
NULL,
/* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
/* Typ
OUTPUT
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
};

{
/* FrontPen, BackPen
*/
JAM2,
/* DrawMode
*/
15, 0,
*/
NULL,
/* Font (Standard)
*/
(UBYTE *) "System programing on the Amiga!",
NULL
/* NextText
*/
};
1, 0,
main ()
{
struct RastPort *MyWindowsRastPort;
MyWindowsRastPort
FirstWindow->RPort;
PrintIText(MyWindowsRastPort, &FirstText, 10L, 20L);

Delay(180L);
Close_All();
}
/***************************************
* Function: Library and Window open
===================================
*
*
***************************************/
149
3. INTUITION AND C
AMIGA
void
*OpenLibrary();
Open Library ("intuition.library", OL)
{

Close_All () ;
exit(FALSE);
}

{
printf("window will not open!\n");

Close_All () ;
exit(FALSE);
}
/***************************************
*
*
*
Function: Close All

=====~=======~~===================
*
*
*
*
***************************************/
if (FirstWindow)
if (IntuitionBase)
PrintIText format:
PrintIText(RastPort, IText, LeftOffset, TopOffset);
-216
AO
Al
DO
Dl
Linked text
To round everything out, here is an example of a linked IntuiText

structure, which you can print by calling PrintIText as above. The
system takes care of whether or not multiple strings exist, and displays
everything correctly. Note that when defining these structures in this
manner the last text in the list must be defined first.
struct IntuiText FourthText
1, 0,
JAM2,
15, 20,
150
/* FrontPen, BackPen
/* DrawMode
*/
*/
*/
ABACUS
3.3
NULL,
1* Font (Standard)
(UBYTE *) "This text is an example ",
NULL
1* NextText
};
OUTPUT
*1
*1
struct IntuiText ThirdText =

{
0, 1,
1* FrontPen, BackPen
JAM2,
1* DrawMode
15, 30,
NULL,
1* Font (Standard)
(UBYTE *) "of linking ",
&FourthText
1* NextText
};
struct IntuiText SecondText
*1
*1
*1
*1
*1
2, 0,
1* FrontPen, BackPen *1
JAM2,
1* DrawMode
*1
15, 40,
1* LeftEdge, TopEdge *1
NULL,
1* Font (Standard)
*1
(UBYTE *) "multiple strings ",
&ThirdText
1* NextText
*1
};
{
3, 0,
JAM2,
1* DrawMode
15, 50,
NULL,
1* Font (Standard)
(UBYTE *) "with IntuiText.",
&SecondText
1* NextText
};
*1
*1
*1
*1
*1
IntulText structure
UBYIE:
Function front Pen
UBYIE:
BatkPen
UBlYE:
DrlNltode
SHORT:
LeftEdae
SHIRl:
T~Edge
srtutt Textftttr:
*ITextAttr
UBYTE:
*!Text
strutt IntuiText r*IIextText
UBYIE:
ri frontPen
UBYlE:
BltkPen
UBT'tE:
DrlNltode
SHORT:
LeftElli
SHIIlI:
TI!PEdge
srtUtt TextRttr:
*ITextAttr
UBYlE:
*!Text
strutt IntulText t*IIextText
UBYIE:
FrontPen
UBYlE:
BatkPen
UBlYE:
DrlNltode
SHORT:
LeftEdge
SHIIlI:
lopEdge
srtutt TextAttr:
*!TextAttr
UBYIE:
*IText
strutt Intullext
*IIextText
r+
Figure 3.4
151
3. INTUITION AND C
3.3.2
Drawing lines
The second output element which Intuition supports is line output.
You may be familiar with the importance of these lines if you have
ever seeri a Workbench requester. Each gadget has a double border surrounding it. The scroll bars in the directory windows are also framed
with a box made up of these lines.
First we'll examine the parameters in the Border structure so that you
can draw simple lines.
3.3.2.1
Color, position, etc.

As with the IntuiText structure, the Border structure can be used
to establish the starting position of the first line to be drawn. As you
saw with PrintITextO, an offset is used as the relative origin for
the position specified in the function call.
FrontPen and BackPen can also be set, although BackPen is not
used (a line really doesn't have a defined background). For this same
reason you can't use all four modes provided for DrawMode. Only
JAMl and INVERSVID make sense because the other two refer to the
second color (again, an unused color in this case).
3.3.2.2
The Border structure

Let's look at the Border structure before going on to the parameters:
/* 3.3.2.2.A. borderstruct */
struct Border
{
OxOO
Ox02
Ox04
OxOS
Ox06
Ox07
Ox08
Oxoc
Ox10
};
152
00
02
04
05
06
07
08
12
16
SHORT LeftEdge
SHORT TopEdge;
UBYTE FrontPen
UBYTE BackPen;
UBYTE DrawMode;
BYTE Count;
SHORT *XY;
struct Border *NextBorder;
ABACUS
3.3
OUTPUT
Unlike the IntuiText structure, no definition of a font exists for the

border. But you need a value which tells the types of comer points your
sequence oflines will have.
Correspondingly you also have a pointer to tables of coordinate pairs
instead of a string. Count specifies the number of pairs in this table.
As with texts, multiple border structures can be linked together to
perform tasks like change line colors.
3.3.2.3
Coordinate table for line drawing

The number of coordinate pairs in the table is stored in the structure in
the variable Count. You have to supply a pointer to the first element.
A coordinate table consists of short value pairs; one value for the X
position and one for the Y position. These values are viewed as offsets
from LeftEdge and TopEdge, which you initialized in the Border
structure.
To complicate matters, we should mention that these coordinates are
just offsets to the coordinates of the DrawBorderO function. The
following graphic makes this clear:
DrawBorderC)-Offset-DeftD
8
58
188
158
2BO
Off.et of window from function (Sal, Sal)
5B
(1BO, 60)
,j.
(23B, 68)
J,
188
Relative position of structure (58, 18)
Figure 3.5
A coordinate table could look like this:
SHORT TestValues[]
{
0, 0,
50, 0,
50, 12,
153
3. INTUITION AND C
AMIGA
0, 12,
0, 0
};
This table describes a rectangular box with a width of 50 pixels and a

height of 12 pixels. Note that you need five values for a box with four
comers: The start value and the four destination values for the lines.
3.3.2.4
DrawBorder function
Now that we've taken stock of the situation, let's take advantage of
Intuition's command for drawing borders. To do this you need a completely defined Border structure and a coordinate table:
/* 3.3.2.4.A. testborderstruct *1
struct Border TestBorder =
{
50, 20,
2, 0,
JAM1,
5,
&TestValues,
NULL
};
Let's use the table printed above as the coordinates. In addition, you
need the RastPort and the offsets for the function:
DrawBorder(MyWindowRastPort, TestBorder, 10L, 10L);
Add the Border structure, the coordinate table, the RastPort and
the function call to CustomScrteen C program. After starting the
modified program, a small box appears in the window. Here is the
complete program:
1***************************************
*
*
*
*
Program: DrawBorder.c
===================~===-=======~===
* Author: Date:
*
* Wgb
10/16/1987
----------
*
*
Comments:
---------Border box
in Window
..*
*
....*
..
*
*
***************************************1
154
3.3 OUTPUT
ABACUS
struct Window
*FirstWindow;
160, 50,
320, 150,
0, 1,
NULL,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
/*
/*
/*
/*
/*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
1* First Gadget
/* CheckMark
(UBYTE *)"System programing test",
NULL,
/* Screen
/* BitMap
NULL,
100, 50,
640, 200,
WBENCHSCREEN,
/* Typ
};
NULL,
NULL,
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
SHORT TestValue[]
{
0,
0,
50, 0,
50, 12,
0, 12,
0, 0
};
struct Border TestBorder

{
50, 20,
2, 0,
JAM1,
5,
TestValue,
NULL
};
main 0
{
lSS
3.
INTUITION AND
AMIGA
MyWindowsRastPort = FirstWindow->RPort;
DrawBorder(MyWindowsRastPort, &TestBorder, lOL, lOLl;
Delay(lBOL);
1***************************************
*
*
* =================================== *
***************************************1
void
*OpenLibrary();
struct Window *Openwindow();
OpenLibrary (nintuition .libraryn, OL)
{
printf(nlntuition Library not found!\nn);

Close_All();
exit (FALSE) ;
}

{
printf(nWindow will not open!\nn);

Close_All () ;
exit(FALSE);
}
1***************************************
*
*
*
* =================================== *
*
*
* Function: Close all
***************************************/
if (FirstWindow)
if (IntuitionBase)
}
156
3.3
ABACUS
OUTPUT
Here is the format of the new function:

DrawBorder (RastPort, Border, 'LeftOffset, RightOffset);
-108
AO
A1
DO
D1
Finally, here is an example which combines several Border structures. This is how programmers create multicolor boxes around system
requesters.
/* 3.3.2.4.B.multiborders */
SHORT WhiteValues[]
{
0, 0,
50, 0,
50,12,
0,12,
0, 0
};
SHORT RedValues[]
{
0, 0,
54, 0,
54,16,
0,16,
0, 0
};
struct Border WhiteBorder =

{
20, 20,
1, 0,
JAM1,
5,
&WhiteValues,
NULL
};
struct Border RedBorder

{
18, 18,
2, 0,
JAM1,
5,
&RedValues,
&WhiteBorder
};
DrawBorder(MyWindowRastPort, &RedBorder, 10L, 10L);
157
3.
INTUITION AND
Border structure
SHORT:
leftEdge
Fu nctlon SHDRT:
TllpEdae
UBYTE:
FrontPen
UBYTE:
BatkPen
UBTYE:
Drawllode
BYTE:
Count
SHORT:
*xy
strutt Border:
*HextBorder
SHDRT:
rt leftEdge
f-
SHORT:
TopEdge
UBYTE:
FrontPen
UBVTE:
BatkPen
UBTYE:
DrilMl10de
BYTE:
Count
SHORT:
*xy
strutt Border:
*HextBorder
r-t
r-
SHORT:
leftEdge
SHII!T:
TopEdge
UBYTE:
FrontPen
UBYTE:
BatkPen
UBTYE:
Drawllode
BYTE:
Count
SHORT:
*xy
strutt Border:
*HextBorder
r+
r-
Figure 3.6
3.3.3
Graphic output
Now let's look at Intuition's most important output function: graphics.
Look at the Workbench with its icons. Each icon and system gadget is
a small graphic image.
3.3.3.1
Size, position, color, etc.

Like text and lines, each graphic also has a certain position in the
window, which, as you know, acts as a relative offset to the coordinates
in the function call.
In addition to the position specification, two values determine the size
of the graphic. Each graphic can be viewed as a rectangular surface, for
which you specify the width and height of the surface.
You may already be aware that each graphic on the Amiga is divided
into several bit-planes. The bit-planes are basically individual sets of
bits in memory, each controlling color output. One bitplane provides
two colors, two bit-planes provide four colors, three provide eight
colors, etc. We must specify the number of bit-planes used by our
graphics so that in can be managed accordingly.
158
ABACUS
3.3.3.2
3.3
OUTPUT
The Image structure

/* 3.3.3.2.A. imagestructure */
struct Image
{
OxOO
Ox02
Ox04
Ox06
Ox08
OxOA
OxOE
OxOF
Ox10
Ox14
};
00
02
04
06
08
10
14
15
16
20
SHORT LeftEdge;
SHORT TopEdge;
SHORT Width;
SHORT Height;
SHORT Depth;
USHORT *ImageData;
UBYTE PlanePick
UBYTE PlaneOnOff;
struct Image *NextImage;
The fIrst five values have been explained. Let's look at the remaining
four.
*ImageData
This pointer points to one or more data fIelds containing graphic data.
*NextImage
This pointer lets you connect multiple Image structures. This allows you to easily generate complex
graphics.
The Intuition DrawImageO function writes the graphic data into the
bit-planes of our window or screen.
The P l.aneP ick flag tells the function which planes of your graphic
to transfer to those of the window. Each bit represents a plane:
Table 3.9:
P l.aneP ick,
Pl.aneOnO
PlanePick value (binarv)

00000000
00000001
00000010
00000100
00000011
00000101
00000111
Planes used
None
Plane 0
Plane 1
Plane 2
etc.
Plane 0 and 1
PianeOand2
Plane 0, 1 and 2
etc.
This table should make it easy to determine which bit-planes of the

graphic are transferred to the window bit-planes. What can this method
be used for? It can be used to tum off certain bit-planes of a graphic.
This is often used to display a graphic in different colors.
159
3.
INTUITION AND
AMIGA
The other flag, P laneOnOff, makes use of the unused bit-planes.

With the same values listed in the table above you can easily specify
the bit-planes which are f1lled with ones. This gives you an easy way to
fill a graphic with a color.
3.3.3.3
The graphic data in the Image structure

Let's take a look at the structure of graphic data. You can make graphic
image development easier by drawing each item out on graph paper
first. Each square represents a pixel. You must do this for all bit-planes,
then you can calculate the values for your data.
First divide the rectangle into columns of four pixels each. You can
view these as binary numbers, whereby each set pixel stands for a one,
and calculate the value accordingly. Do this for each column in every
line and write down the hexadecimal digits from left to right.
Here is a sample grid:
For your own applications
HH+++rHH++~~++rH~++Hrl++~
HH+++rHH++~~++rH~++Hrl++~
HH++~HH++~~++~~++HH++~
~++rH~++Hrl+++rHH++~HH++H
H4++~~.++HH++~HH++~H4++~
H4++~~.++HH++~HH++~H4++~
H4++~rlH,++HH++~HH++~Hrl++H
H4++hHHHf++Hrl++~HH++~Hrl++H
~++hHHH++Hrl+++rHH++~Hrl++H
Figure 3.7
HH++~HH++~~++~~++HH++~
HH+++HHH++~~++~~++HH++~
HH++~HH++~~++~~++HH++~
HH++~HH++~~++MH~++HH++~
HH++~HH++~~++~~++HH++~
~~~uu. .~~~~~. .~~~
Bltplane 1
ax
ax
ex
Ox
ex
ex
ex
ex
ex
ex
ex
ex
ex
ex
ax
ax
, ax
, ex
,,Ox
ex
,ex
,ex
,ex
,ex
,ex
,ex
,ex
,aX
, e~
,Bx
,ax
, ax
Sltplant 2
ax
ax
ax
Ox
ex
ax
ax
ax
ex
a~
Bx
ex
ax
ex
ax
ax
, ax
,ax
,ax
, ax
, Ox
, ex
,ax
, ax
I ax
,ex
,ex
, ax
,ex
I ax
,ax
,ax
If your graphic consists of multiple bit-planes, you must list the data
for the first bit-plane, then the data for the second, etc.
3.3.3.4
The
Drawlmaqe
function
Now we corne to some visible examples. Let's look at the first Image
structure:
160
ABACUS
3.3
OUTPUT
/* 3.3.3.4.A.imageexample *1
struct ImageExample =
{
20, 10,
16, 16,
2,
&ExampleData[O],
3, 0,
NULL
};
This is a graphic representation of the Image structure chaining:

Imaga data exampla
n-r~"TT"n-r~"TT"ITTT'1
4++++++-H-++t-I-tt++-l
-+.++++++-H-++l-H+++-I
.w-H++Hf++t-I
~H++Hf++t-I
-++H-++HH++-I
-++t-H++t++++IH++t1f+tffi-t+tt-I
-+++-HH+++-I+I
.w-H-++HH++-I
-+.++++++-H-+-I+H+++-I
4++++++-H-+++I-tt++-l
H+H+++1H+H++f-If+tH
H-HH+H+H+ttt+H-tH
Bltplana 1
Bltplana 2
BxD7F8,
BxD7FB,
BdE1E,
BxlE1E,
Dx78BD,
Bx78BD,
Dx7BoB,
Bx7BBD,
Bx7B8D,
Bx7BBDi
OdElE,
OdElE,
oxB7FB,
Bxo7FB,
oxBDBo,
oxBDBB,
BxBBBB,
BxDBBo,
BxBBBB,
Bx8B8B,
BxDDBB,
Bx8888,
BxD808,
oxBOBB,
BxB8BB,
8xBD88,
oxBOBO,
oxB8Bo,
BxBBDB
BxBDBB
BxBDDB
BxBBDo
BxFEDo
BxFBDo
Bx8DDo
Ox888o
BxFBBo
BxFE80
odDBo
ox8DBo
oxB8Bo
Bxooo8
OxooBo
OxoOOD
OxoOBD
BxBDBB
8xFEOB
8xFBOD
Dx8000
8x8888
OxFBOD
BxFEBB
OxBOB8
8xOB88
BxDD80
oxoOB8
Figure 3.8
The data field for the example looks like this (remember when inserting
this code in CustomScreenl.c that it must be defined~ the
Image structure):
1* 3.3.3.4.B. sampledata */
USHORT ExampleData[] =
{
/* first BitPlane *1
Ox07F8, OxOOOO,
Ox07FO, OxOOOO,
Ox1E1E, OxOOOO,
Ox1E1E, OxOOOO,
Ox7800, OxFEOO,
Ox7800, OxF800,
Ox7800, OxOOOO,
Ox7800, OxOOOO,
Ox7800, OxF800,
Ox7800, OxFEOO,
Ox1E1E, OxOOOO,
Ox1E1E, OxOOOO,
Ox07F8, OxOOOO,
Ox07F8, OxOOOO,
OxOOOO, OxOOOO,
161
3. INTUITION AND C
OxOOOO, OxOOOO,
/* second BitPlane */
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxFEOO,
OxOOOO, OxF800,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxF800,
OxOOOO, OxFEOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO
};
Finally, you have to find the RastPort of the window and call the
DrawlmageO function:
Drawlmage(MyWindowRastPort, Example, 10L, 10L);
The format:
Drawlmage(RastPort, Image, LeftOffset, TopOffset);
-114
AO
A1
DO
D1
Remember that your graphics chip can access only the chip memory
(the lower 512K). There are two ways to ensure that graphic data are in
this memory area. A linker option (+c in Aztec) transfers all data to the
chip memory. This is probably the simplest method. But it can also be
done within a program. To do this you must allocate space in chip
memory and then copy the graphic data into it. Then you can initialize
the graphic and display it. Here is the complete Drawlmage. c
program, remember to use the +c option in the Aztec linker to force
the program into chip memory.
/***************************************
*
*
* Program: Drawlmage.c
*
* =================================== *
*
*
Comments:
* Author: Date:
*
---------*
*
10/16/1987 Use +C linker
* Wgb
*
option (Aztec) *
*
*
*
***************************************/
----------
162
3.3
ABACUS
OUTPUT
struct Window
*FirstWindow;
{
160, 50,
320, 150,
0, 1,
/*
/*
/*
/*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
/* Flags
NULL,
WINDOWDEPTH
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
/* First Gadget
NULL,
NULL,
/* CheckMark
(UBYTE *)"System programming test",
/* Screen
NULL,
NULL,
/* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
/* Typ
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
};
USHORT ExampleData[]
{
/* first BitPlane */
Ox07F8, OxOOOO,
Ox07FO, OxOOOO,
Ox1E1E, OxOOOO,
Ox1E1E, OxOOOO,
Ox7800, OxFEOO,
Ox7800, OxF800,
Ox7800, OxOOOO,
Ox7800, OxOOOO,
Ox7800, OxF800,
Ox7800, OxFEOO,
Ox1E1E, OxOOOO,
Ox1E1E, OxOOOO,
Ox07F8, OxOOOO,
Ox07F8, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
/* second BitPlane */
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxOOOO,
OxOOOO, OxFEOO,
OxOOOO, OxF800,
163
3.
INTUITION AND
AMIGA
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
};
OxOOOO,
OxOOOO,
OxFBOO,
OxFEOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO,
OxOOOO
struct Image Example

{
20, 10,
32, 16,
2,
&ExampleData[O],
3, 0,
NULL
};
main ()
{
MyWindowsRastPort = FirstWindow->RPort;
DrawImage(MyWindowsRastPort, &Example, 10L, lOLl;
Delay(lBOL);
1***************************************
*
*
* =================================== *
*
*
***************************************1
void
*OpenLibrary();
OpenLibrary ("intuition. library" , OL)
{
164
3.3 OUTPUT
ABACUS

Close_All () ;
exit(FALSE);
}

{
printf(IIWindow will not open!\n");

Close_All () ;
exit(FALSE);
}
/***************************************

*
* =================================== *
***************************************/
i f (FirstWindow)
if (IntuitionBase)
3.3.4
Examples of graphic applications

We want to show you a few applications which you may find of
interest
You can see how easy it is to initialize the structure values. The large
number of parameters allows you to configure graphics just the way
you want them. This can be a problem in itself. If you have a number
of identical strings, for example, it gets annoying if you have to
constantly put the same values into the IntuiText structures. In this
section we'll present a function which will help you out
In addition, this section includes some small graphics, texts and lines
which you will use in defining the gadgets described in Section 3.4.
We're expanding the "toolkit" needed to write our C source code editor,
a little at a time.
165
3.
INTUITION AND
3.3.4.1
AMIGA
Easy text definition

Often you'll find that you have to use IntuiText structures to define
multiple texts. Since you don't always choose a new character color and
you usually limit yourselves to one font, defining twenty almost identical structures can be aggravating.
Here is a solution: First you define a default structure in which you
enter all the parameters, but not the pointer to the string. Here is an
example:
Structure 3.6:
De f a ul. t Text
/*3.3.4.1.A.textdefinition */
struct TextAttr DefaultFont
{
(STRPTR) "topaz.font",
TOPAZ_EIGHTY,
FS_NORMAL,
FPF_ROMFONT
};
struct IntuiText DefaultText {
1, 0,
JAM2,
1, 1,
&DefaultFont,
NULL,
NULL
};
/*
/*
/*
/*
/*
/*
FrontPen, BackPen
DrawMode
LeftEdge, TopEdge
Font
Textpointer
NextText
*/
*/
*/
*/
*/
*/
The TextAttr structure is also initialized here for correct font usage.
For the strings which are later inserted into this structure, you define a
pointer array of type char so that you can access each string easily:
/*3.3.4.1.B.stringaccess *1
char *Textarray[] =
{
"First text string",

"here is the second",
"and another",
"and here we have the last and the longest!"
};
Now you just need an array of IntuiText structures into which you
will copy the default structure values and then add the string pointer.
struct IntuiText AllTexts[4];
Let's look at the subroutine which does all this:
166
3.3
ABACUS
Function 3.3:
Make_Text
OUTPUT
/*3.3.4.1.C.Make_Text function */
Make_Text (Number, Text, Struktur)
int
Number;
char
*Text[];
struct IntuiText Struktur[];
int i;
for (i=O; i<Num,ber; i++)
{
Struktur[i]
= DefaultText;
Struktur[i] .IText
(UBYTE *) Text[i];
Struktur[i] .NextText = &Struktur[i+1];
}
Struktur[Number-1].NextText = NULL;
}
How it works
The subroutine does its work in a small loop. Here the structure defined
as an array element is first assigned the default structure values. Then
you add the pointer to the first string and join this first IntuiText
structure to the second. The loop then executes again for the next value.
To prevent the last structure from pointing to a nonexistent structure,
the extra pointer must be deleted again.
If you don't want to link the texts together, you can omit the two lines
responsible for this. Here is a complete listing of a demo program

using the Make_Text () function:
/***************************************
* Program: Make_Text.c
*
* =================================--= *
*
*
* Author: Date:
Comments:
*
*
*
10/16/1987 Make text
* Wgb
*
01/16/1988 Structures
* JLD
*
*
*
***************************************/
---------- ----------
struct Window
*FirstWindow;
160, 50,
400, 150,
/* Width, Height
*/
*/
167
3.
INTUITION AND
AMIGA
0, 1,
NULL,
1* IDCMP Flags
WINDOWDEPTH I
1* Flags
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
NULL,
1* Screen
NULL,
1* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
1* Typ
};
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
struct TextAttr DefaultFont

{
(STRPTR) "topaz. font" ,

TOPAZ_EIGHTY,
FS_NORMAL,
FPF_ROMFONT
};
struct IntuiText DefaultText
1, 0,
JAM2,
1, 1,
&DefaultFont,
NULL,
NULL
};
1*
1*
1*
1*
1*
1*
FrontPen, BackPen
DrawMode
LeftEdge, Top Edge
Font
Textpointer
NextText
char *Textfield[]
{
"First Text text",

"here is the second",
"and one more",
"here is the last and the longest text!"
};
struct IntuiText AllText[4];
main ()
{

Open_All () ;
Make_Text(4, Textfield, AllText);
MyWindowsRastPort
168
= FirstWindow->RPort;
*1
*1
*1
*1
*1
*1
3.3
ABACUS
OUTPUT
PrintIText(MyWindowsRastPort, &AIIText[O], 10L, lOLl;

Delay(180L);
/***************************************
*
*
*

=============~=====================
*
*
***************************************/
void
*OpenLibrary();
{

Close_All 0 ;
exit(FALSE);
}

{
print f ("Window will not open! \n") ;

Close_All 0 ;
exit(FALSE);
}
/***************************************
===================================
*
*
*
***************************************/
if (FirstWindow)
if (IntuitionBase)
Make_Text(Number, Text, Struktur)
169
3. INTUITION AND C
AMIGA
int
Number;
char
*Text[];
struct IntuiText Struktur[];
int i;
for(i=O; i<Nurnber; i++)
{
Struktur[i]
Struktur[i] .IText
Struktur[i] .TopEdge
Struktur[i] .NextText
Defau1tText;
= (UBYTE *) Text[i];
= 20+i*10;
= &Struktur[i+1];
Struktur[Number-1].NextText
= NULL;
3.3.4.2
Borders for gadgets

This subsection prepares you for the next section, where we'll look at
gadgets. These gadgets use all three output elements which you've seen
so far. A box often delimits the clickable area of a gadget, so we will
develop a Border structure here which you can use later.
Let's take single and double boxes placed around gadgets as the first
example. Double boxes are used in DataRetrieve and BeckerText.
for example, to designate which is the default gadget.
Here is the single border (for eight letters):
/* 3.3.4.2.A. sing1eborder
SHORT Sing1eVa1ues[)
{
0, 0,
66, 0,
66,10,
0,10,
0, 0
};
170
*1
ABACUS
3.3
OUTPUT
struct Border SingleBorder =

{
0, 0,
1, 0,
JAM1,
5,
&singleValues,
NULL
};
and the double border;

/*3.3.4.2.B.doubleborder*/
SHORT DoubleValues[]
{
0, 0,
70, 0,
70,14,
0,14,
0, 0
};
struct Border DoubleBorder =

{
-2, -2,
2, 0,
JAM1,
5,
&DoubleValues,
NULL
};
The DoubleBorder call can be easily appended with the pointer:

SingleBorder.NexBorder = &DoubleBorder;
Next we offer a solution for the problem of defIning a gadget border for
text whose length is not known until the program is running. A
Border structure and value array must be defined preceding the
main () function. This would look like this:
/*3.3.4.2.C.borderlong*/
SHORT Values []
{
0, 0,
999, 0,
999,10,
0,10,
0, 0
};
struct Border Bord =

{
0, 0,
1, 0,
171
3.
INTUITION AND
AMIGA
JAM1,
5,
Values,
NULL
};
In addition, you need an IntuiText structure with everything but the
pointer to the text:

/*3.3.4.2.D.generic_intuitext */
struct IntuiText Text =
{
1, 0,
JAM2,
2, 2,
NULL,
(UBYTE *) "Test",
NULL
};
/*
/*
/*
/*
/*
/*
FrontPen, BackPen
DrawMode
LeftEdge, TopEdge
Font (Standard)
Textpointer
NextText
*/
*/
*/
*/
*/
*/
Both Text and Bard are written to RastPart at the same offsets:
DrawBorder(RastPort, Bord, 10L, 30L);
PrintIText(RastPort, Text, 10L, 30L);
Before this can be done, you have to call a function which calculates the
correct border values for the text:
/*3.3.4.2.E.Calc*/
Calc_Border (Text)
struct IntuiText
Text;
int Width
0;
Width = IntuiTextLenght(Text);
Values [2]
Values [4]
Width+4;
Width+4;
The only new thing is the IntuiTextLengthO function, which

returns the text width in pixels, while accounting for the font size.
Once you have this value, you simply insert it in the array. Here is a
complete example program:
/***************************************
* Program: Calc_Border.c
*
*
*
*
*
172
============================~=~====
Author:
Wgb
Date:
10/16/1987
Comments:
for testing
*
*
*
*
*
*
*
3.3
ABACUS
only
OUTPUT
*
*
***************************************1
struct Window
*FirstWindow;
160, 50,
320, 150,
0, 1,
1*
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
NULL,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
1* First Gadget
NULL,
1* CheckMark
NULL,
NULL,
1* Screen
1* BitMap
NULL,
100, 50,
640, 200,
WBENCHSCREEN,
1* Typ
*1
*1
*1
*1
*/
*1
*1
*1
*1
*1
*1
*1
};
SHORT Values [ 1
{
0, 0,
999, 0,
999,10,
0,10,
0, 0
};
struct Border Bord

{
0, 0,
1, 0,
JAMl,
5,
Values,
NULL
};
struct IntuiText Text =
{
1, 0,
*1
173
3. INTUITION AND C
AMIGA
JAM2,
2, 2,
NULL,
(UBYTE *) "Test",
NULL
};
1*
1*
1*
1*
1*
DrawMode
LeftEdge, TopEdge
Font (Standard)
Textpointer
NextText
*1
*1
*1
*1
*1
main ()
{
struct RastPort *RP;
RP
= FirstWindow->RPort;
Calc_Border(&Text);
DrawBorder(RP, &Bord, 10L, 30L)1
PrintIText(RP, &Text, 10L, 30L);
Delay(180L);
Close_All 0 ;
}
1********************************
* Function: Open All
*
*
*
********************************1
void Open_All ()
{
void
*OpenLibrary()1
if(! (IntuitionBase = (struct IntuitionBase *)
(

Close_All 0;
exit(FALSE);
}
if (!(FirstWindow = (struct Window *)

{
printf("Window will not openl\nn);

Close_All 0;
exit (FALSE) ;
}
174
ABACUS
3.3
OUTPUT
1***************************************
*
*
* Function: Close All
***************************************/
void Close_AII()
{
if (FirstWindow)
if (IntuitionBase)
Calc_Border (Text)
struct IntuiText
*Text;
int Width
Width
0:
0:
0;
IntuiTextLength(Text);
Values[2J
Values[4J
0:
0:
Width+4;
Width+4;
3.3.4,,3
Symbols say more than words

As an extension to this section on images, we want to introduce a
graphic which we could use later in our editor as a gadget icon.
It is somewhat annoying for the user if he always has to drag and resize
the window to make it fill the screen just to click a gadget. It's just as
annoying to reduce it in size.
Since you can do all this with window functions, you can easily perform this action using a gadget. Now you need an icon for the gadget.
The icon represents a small window and a large window, with a line
connecting them to suggest the change in size.
175
3. INTUITION AND C
AMIGA
eLi-wide Image data

Bltplane 1
-t+H-t+H-t+H-t+H-t+++++1-H BxffBB, BxBOBS
Ox8100,
Ox!Uff,
-t-t-+HH-tlI-H-t-t-+H+t-+-H-t-t-+H Bx8CBl,
-t-t-+t-~-t+t-H++++-ir-t+++-H OxffBl,
..-+-H--..+++-H4-+++-HH-+++1-4 OxlOCl,
H-t___++-I+t-~-t+-a-I+++-H-+++-H+++-H"'" DxlB21,
H-t-e-++-I+t-++-I-t+.-...t++++-I-+++-H+++-H,.... OxlBBl,
-+++-H+++-H-+++-H~ BxlOOl,
-+++++++-1-++++-11-++-1-4 Bxlfff,
H-t-t+++-I++++-I-t++-H+++-H-+++-H+++-H,.... BxBBBO,
H-t-t+++-I++++-I-t++-H+++-H-+++-H+++-H,.... BxODBD,
t-H-H-t-H-H-t-H-t+t-H-H-t-H-t+++-I++++1H BxBDDO I
H+++-t+tt-H-++++-t-++H-t+Hf-t-t++!-++~ BxODBO I
H+++-t+tt-H-++++-t-++H-t+Hf-t-t++!-++~ BxDDDD,
L..U..........I..I...I.~L..U....L..L..Ju...L."-U...I..I..U-I...I...I..L...I..,I...L...I..J BxBBBB I
-t+t+++t-i-t+t++-H-H
-t+t+++t-i-t+t++-H-H
Figure 3.9
Resize Gadget
OxDDOO
DxODDO
DxDDDB
BxDBOB
OxOBOB
OxBBOB
OxBBDB
BxOOOB
OxBBOB
OxDBBB
DxBBBB
BxBDDO
BxDOOB
OxDBOO
OxDDBS
It shouldn't be a problem for you to build an Image structure from the

picture and the grnphic data.
176
3.4 GADGETS
ABACUS
3.4
Gadgets
Application programs need the input of general information, like text or
data in order to operate properly. In addition, they need instructions to
guide them in their work.
As an experienced Amiga owner you know that gadgets provide the
easiest way to send information to the computer. These are rectangular
selectable fields displayed in windows or screens. Programs react when
you click the mouse button with the pointer on a gadget
The Amiga can support more than just simple gadgets which just pass
on a selection. Some gadgets can be turned off and on, some can be
used for text input, and others can specify positions. These different
gadget types can be further divided into subgroups. Using gadgets, you
have many different ways to get information from the user.
You wouldn't be able to recognize these gadgets if they were only
placed on the screen, Intuition allows the user to "decorate" these
gadgets with borders, graphics and text. This is why the display
techniques explained in the previous section are so important
3.4.1
Different gadgets, different applications

Intuition offers two primitive gadget groups. The first group contains
the system gadgets. Intuition offers direct support to access these
gadges, and predefines their appearance and opemting mode. See Section
3.1 for more information. The second group is intended for the
progmmmer and supports custom gadgets. You can select from three
different gadget types which have completely different properties.
Boolean
gadgets
Boolean gadgets are intended for very simple commands and data which
consist of just a boolean value (true or false). They offer two modes:
The first mode is called "hit select." This means that Intuition sends a
report when you click the gadget with the left mouse button (called the
select button in this case). The appearance of the gadget changes while
the mouse selects it. If you move the mouse pointer away or release the
select button, the appeamnce reverts to its original form. This mode can
be used to issue very simple, specific commands.
177
3.
INTUITION AND
AMIGA
With the second mode, "toggle select," you turn the gadget on with the
ftrst click. It then changes its appearance as if the mouse button were
being held on it. A second click changes the gadget to its original state.
This method is particularly good for turning choices on and off.
Proportional
gadgets
If you want to represent areas or positions, this is the gadget type for
you. With it you can display specific proportions in one or two
dimensions and let the user of your program change them.
These gadgets consist of a rectangular box containing an object. We're
using a generic term here because you can define any graphic you like
as the object. For example, in Preferences the screen position object
represents the edge of the Workbench screen, which you can set using a
proportional gadget. In this case it is two-dimensional. An example of a
one-dimensional proportional gadget is the scroll bar in a directory
window. The bar, your "object," represents the size of the window
display and can be moved back and forth.
For each proportional gadget you can defme the size of the scroll box
(container), the object used (knob), and the size of the object.
String gadgets The string gadget allows simple text input. Like a proportional gadget,
there is a "container" into which you enter a single line of text. The
length is irrelevant: Intuition scrolls the text as necessary.
Editing is made easier by additional functions like Undo.
Now you know all of the gadget types and their basic properties. Let's
look at the details of each type.
3.4.1.1
Boolean gadgets
One of the most common gadget types is the boolean gadget. Boolean
gadgets appear three times in the system gadgets: Close gadget, front
gadget and back gadget. Boolean gadgets represent the basic gadget type,
which is why we will discuss them in such detail.
Each gadget can be assigned a position, height and width. Let's take a
look at the structure before we go into the other values:
1* 3.4.1.1.A.booleanstruct *1
struct Gadget
{
OxOO
Ox04
Ox06
Ox08
OxOA
178
00
04
06
08
10
struct Gadget *NextGadget;

SHORT LeftEdge;
SHORT TopEdge;
SHORT Width;
SHORT Height;
3.4
ABACUS
OxOC
OxOE
Oxl0
Ox12
Ox16
Ox1A
Ox1E
Ox22
Ox26
Ox28
Ox2C
};
12
14
16
18
22
26
30
34
38
40
44
GADGETS
USHORT Flags;
USHORT Activation;
USHORT GadgetType;
APTR GadgetRender;
APTR SelectRender;
struct IntuiText *GadgetText;
LONG MutualExclude;
APTR Speciallnfo;
USHORT GadgetID;
APTR UserData;
Just like all of the other Intuition structures you have looked at, gadgets
can be linked together. This is the purpose of the first pointer. After the
values for the left and top edges, height, and width you find a flag
which performs several tasks.
Flags first show you the settings for the gadget graphic. You can
specify how the appearance of the gadget is changed when it is clicked.
There are four GADGHIGHBITS available for this:
GADGHCOMP
All points in the click area are complemented

(reversed).
GADGHBOX
A box is drawn around the click area.
GADGHlMAGE
A different Image or Border is displayed.
GAGDHNONE
No change takes place. The flag also contains the

position specifications for the gadget:
GRELBOTTOM
When you set this flag, the TopEdge variable of the

structure describes an offset relative to the bottom of
the window. If this flag isn't set, then TopEdge
contains a relative specification from the upper border
of the window.
GRELRIGHT
When you set this flag, the LeftEdge variable of

the structure describes an offset relative to the right
border of the window. If this flag isn't set, then
LeftEdge is relative to the left border.
The size of the gadget can also be set in relationship to the window:
GRELWIDTH
If you set this flag, the value of the Width variable

is subtracted from the window width and the result
then represents the actual width of the gadget. If you
do not set this flag, the width represents an absolute
value.
179
3.
INTUITION AND
GRELHEIGHT
AMIGA
If you set this flag, the value of the Height variable
is subtracted from the window height and the result

represents the actual height of the gadget If you clear
the flag, the value is interpreted as absolute.
There are three flags for gadget status:
GADGIMAGE
If you set this flag, the ImageRender variable is

assumed to point to an Image instead of a Border.
The meaning also extends to the SelectRender
variable.
GADGDISABLED
This flag is set (it can be set only at the beginning

and thereafter it can only be read) when the gadget is
unselectable.
SELECTED
This flag is set (it can be set only at the beginning

and thereafter it can only be read) when the gadget is
selected.
Here are the flags together with their hex values:

Table 3.10:
Gadget nags
Fla2
GADGHIGHBITS
GADGHCOMP
GADGHIMAGE
GADGHBOX
GADGHNONE
Hex value
OxOOO3L
OxOOOOL
OxOOO2L
OxOOOlL
OxOOO3L
GRELBOTTOM
GRELRIGHT
GRELWIDTH
GRELHEIGHT
OxOOO8L
OxOOlOL
OxOO20L
OxOO40L
GADGlMAGE
OxOOO4L
SELECTED
OxOO80L
GADGDISABLED OxOlOOL
Now we come to the next value of the Gadget structure. ThIS value is
a set of flags which contain information about how the gadget is acti
vated and whether it is located in the window border.
Let's first look at the flags which have to do with the activation status:
TOGGLESELECT
This flag defines a certain type of boolean gadget.

Thereafter the gadget is treated as a switch: On the
first click it turns on and it remains in this state until
clicked again.
180
ABACUS
3.4
GADGETS
GADGIMMEDIATE
With simple boolean gadgets the programmer can

choose between two select modes: If this mode is
chosen, the program receives a message as soon as
the gadget is clicked.
RELVERIFY
If this flag is set, Intuition waits with the message
until the user also releases the button in the click

area. This allows the user to change his mind after
having pressed the mouse button.
There are still more variations for the position of the gadget These refer
more to the management of the gadget position than the position of the
gadget proper. You know that the border of a GIMMEZEROZERO window is handled specially so that the graphic in the window cannot be
disturbed.
You can also place your custom gadgets in the window border. To do
this, set the flags which follow.
These flags determine the gadget's location:
RI GHTBORDER
The gadget is placed in the right window border.
LEFTBORDER
The gadget is placed in the left border of the window.
TOPBORDER
The gadget is placed in the top border of the window.
BOTTOMBORDER
The gadget is placed in the bottom border of the

window.
Width of window border
Top border
Gadgets in Uindow
....................................
..................... ..
.............................. .. .........................
".11
f
I
t 1,1
' "
I '"
I , . , , . , I " , . , . "
I I
I I I . _
I I
I I . . . . . . . . . . . . . . . . . . . . . . . I I I I I
II I.' I I I I 1""
I . I I ' . , . I I .- I , , I I
I I I
........................................... " ............ .

, I "
" , , , I I
I I I I I I I I I . I "
Left
border
I I
I I I I I I I I I "
I I
...........................................................
I I I I I I,' I "
Figure 3.10
L I:[]~ ~~L: . ._.~i:' :h:::I;~~:::i::~::::a:::~:::t::~:':i:':~:':'~::::~:::~::d:::~:::~::t::I~~~.

T
T:':..
Bottom border
Right border
181
3. INTUITION AND C
AMIGA
The following table lists all of the activation flags:

Table 3.11:
Activation
flags
Activation f1al!
TOGGLES ELECT
RELVERIFY
GADGIMMEDIATE
Hex value
Ox01OOL
OxOOO1L
OxOOO2L
Comment
Switch gadget
RIGHTBORDER
LEFTBORDER
TOP BORDER
BOTTOMBORDER
OxOO1OL
OxOO20L
OxOO40L
OxOO80L
Position in GZZ border
STRINGCENTER
STRINGRIGHT
LONGINT
ALTKEYMAP
BOOLEXTEND
ENDGADGET
FOLLOWMOUSE
Ox0200L
Ox0400L
Ox0800L
OxlOOOL
Ox2000L
OxOOO4L
OxOOO8L
Reserved for string gadgets

For integer gadgets
Custom keyboard table
For requester
ForIDCMP
Let's return to the Gadget structure. We now come to the last flag
value which has to do with the type of gadget. Gadget Type also
contains more information:
BOOLGADGET
A boolean gadget
PROPGADGET
A proportional gadget (more on this later)
STRGADGET
A string gadget with which simple text input can be

performed. For an integer gadget, set the LONGINT
flag in the Flags variable of the structure.
These three flags determine the gadget type. With two more flags,
which are set by the system, you can tell where the gadget carne from
and where it lies.
SCRGADGET
specifies that the gadget belongs to a screen and
SYSGADGET
identifies the system gadgets in a list
!fa gadget is placed in the borderofa GZZ window, the GZZGADGET

flag must also be set
182
3.4 GADGETS
ABACUS
Table 3.12
Gadget types
Gadget Type
BOOLGADGET
GADGETOOO2
PROPGADGET
STRGAD GET
Hex value
OxOOO1L
OxOOO2L
OxOOO3L
OxOOO4L
SYSGADGET
SCRGADGET
GZZGADGET
GIMMEZEROZERO
REQGADGET
Ox8000L
Ox4000L
Ox2000L
OxlOOOL
Comment
Unused
System gadget
Screen gadget
Window gadget
Reauester midget
Finally we come to the visible characteristics of the gadgets. We will

concern ourselves with the graphic that normally associates with
gadgets. Height, width and position are used to set the position and
dimensions of the click field. Now it is time to make these visible to
the user as well. To do this use a Border or Image structure. A
pointer to one of these two structure types is stored in the
GadgetRender variable. This graphic is then drawn when the gadget
is displayed.
If you have also specified that a different appearance is desired when the
gadget is clicked, then another pointer to an Image or Border structure must be in the SelectRender variable.
You can't just use a Border structure one time and an Image structure the next. As you can see above, a flag exists for each mode. This
means that the GAGDIMAGE flag must be set for an Image structure
and cleared for a Border structure.
GADGIMAGE also affects SelectRender, but it doesn't say that a
pointer has to be stored there. Intuition looks for a structure pointer
there only if the GADGIMAGE flag is set.
Intuition also offers a pointer to an IntuiText structure so that you

can also put a label on your gadget. This text has nothing to do with
selecting the gadget.
The last four values of the structure are for more complex applications.
Since you are primarily interested in practical applications, lets ignore
them for the moment. We will come back to them later in our
discussion.
The first
gadget
Now let's tum to programming. We have prepared the following gadget

structure for the first example:
/* 3.4.1.1.B.boolgadget */
struct Gadget BoolGadget =
{
NULL,
10, 40,
60, 20,
/* NextGadget
/* Width, Height
*/
*/
*/
183
3.
INTUITION AND
AMIGA
GADGHBOX,
RELVERIFY,
BOOLGADGET,
(APTR)&GadgetBorder,
NULL,
&GadgetText,
NULL,
NULL.
1,
NULL,
};
Structure
description
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
Flags
Activation
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
Special Info
GadgetID
UserData
*/
*1
*1
*1
*1
*1
*1
*1
*1
*1
This defines a gadget of type boolean, which has a rectangular click

field in the upper left comer of the window. When the gadget is
selected, a border is drawn around the field. The gadget doesn't activate
until the selection is verified. The gadget has a border and a label to
make it recognizable and to explain its purpose to the user.
Something new is the value GadgetID, where you can enter a value
from 0 to 65535. This is used to distinguish between gadgets. All you
need to add are the Border Structure, the coordinate table and the texL
Here they are:
/* 3.4.1.1.C.gadgettext */
struct IntuiText GadgetText
{
2, 0, JAM2, 4, 1, NULL, (UBYTE *)"Gadget" ,NULL,

};
SHORT GadgetPairs[}
{
0, 0, 61, 0, 61, 21, 0, 21, 0, O.

};
struct Border GadgetBorder
-1, -I, I, O. JAMl, 5, GadgetPairs, NULL,

};
Insert the new program fragments into First window. c program

from Section 3.1. Remember that the IntuIText and Border
Structures must be defined preceding the Gadget structure. Finally, add
a pointer to the window structure which points to a user gadgeL The
entire window structure then looks like this:
/* 3.4.1.1.D.firstnewwindow *1
160, 50,
320, 150,
184
/* Width, Height
*1
*1
0, 1,
CLOSE WINDOW
GADGETUP,
WINDOWDEPTH
1* IDCMP Flags
*1
/* Flags
*1
3.4 GADGETS
ABACUS
WINDOWSIZING I
WINDOWDRAG I WINDOWCLOSE
SMART_REFRESH,
&BooIGadget,
/* First Gadget
NULL,
/* CheckMark
(UBYTE *) "Gadget Programming Test",
NULL,
/* Screen
NULL,
/* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
1* Type
};
*/
*1
*/
*/
*/
*1
*1
Take a look at the new IDCMP flag, which sends you messages whenever a gadget is selected.
With this addition you also have to extend the testing of the IDCMP
flag for this new case. You can get messages from a self-defined user
gadget as well as from the close gadget
1*3.4.1.1.E.othergadget_loop */
FOREVER
{
if message = (struct IntuiMessage *)

GetMsg(FirstWindow->UserPort == NULL)
(
Wait(lL
continue;
FirstWindow->UserPort->mp_SigBit);
ReplyMsg(message);
{
case GADGETUP
nr = nr + 1;
printf("Gadget activated %d times!\n", nr);
break;
case CLOSEWINDOW
Close_AII();
exit(TRUE);
break;
The test loop here offers another important advantage over the previous
one: The Wai function places the task in the wait state until a signal
occurs. Therefore the loop doesn't use any processor time when it has
nothing to do.
to
The test for the new gadget is easy. The swi tchO statement tests to
see if a user-defined gadget has been selected. Since you only used one,
you don't have to sort things out any further, and output a counter
185
3. INTUITION AND C
AMIGA
which indicates how many times the gadget has been clicked. It's
important not to forget to declare this variable.
How it works
After the program starts, a window containing a boolean gadget appears

on the Workbench screen. The program receives a message whenever
you click on this gadget with the mouse button. Since you have set
RELVERIFY, the button must also be released while the pointer is
within the click field. A message with the number of clicks appears in
the DOS window. You can end the program by clicking the close
gadget.
Boolean gadgets always work well for simple processes. They allow the
user to easily issue or confirm a command, or terminate an action. Here
is a complete example listing for a boolean gadget. Remember to
specify the 32-bit version of the libraries if you use the +L option of
the Aztec compiler.
/**************************************
*
*
*
* ================================== *
* Gadgets: Boolean-Gadget.c
*
*
Comments:
* Author: Date:
*
*
*
10/16/1987 only for
* Wgb
*
testing
*
*
*
*
**************************************/
---------- ----------
struct Window
*FirstWindow;
SHORT GadgetPairs[]
0, 0, 61, 0, 61, 21, 0, 21, 0, 0,

);
-1, -1, 1, 0, JAM1, 5, GadgetPairs, NULL,

);
struct IntuiText GadgetText =

2, 0, JAM2, 4, 7, NULL, (UBYTE *) "Gadget" , NULL,
};
186
3.4 GADGETS
ABACUS
struct Gadget BoolGadget ;

{
NULL,
10, 40,
60, 20,
GADGHBOX,
RELVERIFY,
BOOLGADGET,
NULL,
&GadgetText,
NULL,
NULL,
1,
NULL,
);
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
Activation
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
/* SpecialInfo
1* GadgetID
1* UserData
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
struct NewWindow FirstNewWindow ;

{
160, 50,
320, 150,
0, 1,
CLOSEWINDOW
GADGETUP,
WINDOWDEPTH
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
1*
1*
1*
1*
1* Flags
*1
&BoolGadget,
NULL,
1* First Gadget
1* CheckMark
*1
*1
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
*1
*1
*1
*1
. *1
*1
NULL,
NULL,
1* Screen
1* BitMap
100, 50,
640, 200,

*1
WBENCHSCREEN,
};
1* Type
*1
*/
main ()
{
ULONG MessageClass;
USHORT code, nr ; 0;
187
3.
INTUITION AND
AMIGA
RefreshGadgets(&BoolGadget, FirstWindow, NULL);

FOREVER
{

{
Wait(lL
continue;
ReplyMsg(message);
{
case GADGETUP
: nr += 1 ;
printf("Gadget activated tu times!\n", nr);
break;
case CLOSEWINDOW
Close_All 0;
exit (TRUE);
break;
Open_All ()
{
void *OpenLibrary();
printf (IIIntuition Library not found!\n");
Close_All () ;
exit(FALSE);
}
if (! (FirstWindow = (struct Window

*)OpenWindow(&FirstNewWindow)
(

Close_All();
exit(FALSE);
}
Close_All ()
{
if (First Window)
if (IntuitionBase)
}
188
CloseLibrary(IntuitionBase),
3.4
ABACUS
3.4.1.2
GADGETS
All other gadget types are extensions of the boolean gadget, including
the proportional gadget. This gadget is used for the Workbench scroll
bars. This is useful when the material to be displayed is larger than the
actual display area. Then a scroll bar indicates the relative size of the
display area and allows the user to view a different portion of the whole.
Preferences uses a two-dimensional proportional gadget to allow you to
set the position of the Workbench screen. This is another application of
this gadget type.
For the fIrst proportional gadget, let's take a simple basic gadget structure with a long vertical click field:
1* 3.4.1.2.A.propgadget */struct Gadget PropGadget =
{
1* NextGadget
NULL,
100, 20,
20, 60,
GADGHCOMP,
RELVERIFY,
PROPGADGET,
(APTR) Buffer,
NULL,
NULL,
NULL,
&ExampleProp,
/* Width, Height
1* Flags
/* Activation
/* Gadget Type
/* GadgetRender
/* Select Render
/* GadgetText
/* MutualExclude
1* Speciallnfo
1* GadgetID
1* UserData
2,
NULL,
*/
*/
*1
*1
*/
*/
*/
*1
*/
*/
*/
*/
*1
};
As you can see, this structure is a little short on arguments. But you
don't need many: You don't need a text in this example, and a border
around the container is drawn automatically for proportional gadgets.
We have, however, defIned the value SpecialInfo. This pointer is
used whenever a boolean gadget isn't enough. It then points to an
extension structure. In this case you need a P ropInfo structure,
which looks as follows:
/* 3.4.1.2.B.propinfostruct*/
struct Proplnfo
{
OxOO
Ox02
Ox04
Ox06
Ox08
OxOA
OxOC
00
02
04
06
08
10
12
USHORT
USHORT
USHORT
USHORT
USHORT
USHORT
USHORT
Flags;
HorizPot;
VertPot;
HorizBody;
VertBody;
CWidth;
CHeight;
189
3. INTUITION AND C
OxOE
OxlO
Ox12
Ox14
Ox16
};
AMIGA
14
16
18
20
22
USHORT
USHORT
USHORT
USHORT
HPotRes;
VPotRes;
LeftBorder;
TopBorder;
So you see, you need more data for a proportional gadget. These contain
the dimensions of the container and the knob. The knob is the actual
graphic that is moved by the user. The container is the actual area in
which the knob can be moved. Here are the descriptions of each
parameter:
Flags
You need a few additional values for proportional

gadgets. You can select from among the following
arguments:
AUTOKNOB
This flag indicates that you have not defined a custom

graphic for the knob and that Intuition should use a
rectangle as default. Intuition creates this rectangle,
and automatically adapts it to any situation. The size
of the knob can be set with the next two structure
values.
FREEHORIZ
If you set this flag, the knob can be moved horizontally. The knob can be moved in both directions if
you also set the FREEVERT flag.

FREEVERT
If you set this flag, the knob can be moved vertically.

The knob can be moved in both directions if you also
set the FREEHORI z flag.
KNOBHIT
You cannot set this flag yourself. Intuition sets this

flag if the knob was just clicked with the mouse.
PROPBORDERLESS
Normally Intuition draws a border around the gadget

box. Setting this flag suppresses the border.
190
HorizPot
Specifies the horizontal position of the knob.
VertPot
Specifies the vertical position of the knob.
HoriZBody
Here you can set the horizontal increment of knob

movement. This is the value which the knob moves
when you click in the container, but not on the knob
itself.
Vert Body
Here you can set the vertical increment of knob

movement. This is the value which the knob moves
when you click in the container, but not on the knob
itself.
3.4 GADGETS
ABACUS
None of the following values need be set by the programmer. They are
calculated by Intuition and can be read after the gadget has been initialized.
cwidth
Container width
CHeight
Container height
HPotRes
Horizontal knob resolution
VPotRes
Vertical knob resolution
LeftBorder
Current position of the left border of the container
TopBorder
Current position of the top border of the container
The example code used to access the Proplnfo structure would look
something like this:
1*3.4.1.2.C.exampleprop*1
struct Proplnfo ExampleProp
AUTOKNOB I FREEVERT,
Ox8000,
Ox8000,
Ox0800,
Ox0800,
0,
0,
0,
0,
0,
0
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
Flags
HorizPot
VertPot
HorizBody
VertBody
CWidth
CHeight
HPotRes
VPotRes
LeftBorder
TopBorder
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
};
Now you can insert the structure definitions in your program. Remember that the Proplnfo structure must be defmed first so that the value
of &ExampleProp is known to the gadget structure. The same
applies to the window structure. in which the address of the gadget
structure is set in place.
Even after you've done all this. you're still not finished. With a
proportional gadget you get a message when the gadget has been
released (GADGET UP) instead of when you ''press it" (GADGETDOWN).
You have to specify this in your loop:
Program 3.3:
Test loop for
proportional
gadget
1*3.4.1.2.D.prop forever *1
FOREVER
{
if message - (struct IntuiMessage *)

{
191
3.
INTUITION AND
Wait(lL
continue;
ReplyMsg(message);
(
case GADGETUP
printf("Position: u%\n",
ExampleProp.VertPot);
break;
case CLOSEWINDOW
Close_All () ;
exit(TRUE);
break;
case GADGETDOWN
nr += 1;
printf ("Gadget activated for
the %nr time!\n", nr);
break;
Here is a complete program listing for creating a proportional gadget:

/***************************************
*
*
* Program: Proportional-Gadget.c
*
* =================================== *
*
* Author:
*
* Wgb
*
*
Date:
Comments:
----------
----------
10/16/1987
for testing
only
*
*
*
*
*
***************************************/
struct Window
*FirstWindow;
WORD Buffer[4];
struct Proplnfo ExampleProp
{
AUTOKNOB I FREEVERT, /*
Ox8000,
/*
Ox8000,
/*
Ox0800,
/*
Ox0800,
/*
192
Flags
HorizPot
VertPot
HorizBody
Vert Body
*/
*/
*/
*/
*/
3.4
ABACUS
0,
0,
0,
0,
0,
0
1*
1*
1*
1*
1*
1*
CWidth
CHeight
HPotRes
VPotRes
LeftBorder
TopBorder
GADGETS
*1
*1
*1
*1
*1
*1
};
struct Gadget PropGadget
1* NextGadget
*1
1* Width, Height
*1
GADGHCOMP,
1* Flags
*1
RELVERIFY,
1* Activation
*1
PROPGADGET,
1* Gadget Type
*1
(APTR) Buffer,
1* GadgetRender
*1
NULL,
1* Select Render
*1
NULL,
1* GadgetText
*1
NULL,
1* MutualExclude
*1
(APTR) &ExampleProp, 1* SpecialInfo
*1
2,
1* GadgetID
*1
NULL,
1* UserData
*1
NULL,
100, 20,
20, 60,
};

160, 50,
320, 150,
0, 1,
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
rDCMP Flags
CLOSEWINDOW
GADGETUP,
WINDCWDEPTH 1
1* Flags
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
&PropGadget,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *) "Gadget Programmng Test",
NULL,
1* Screen
NULL,
1* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
1* Type
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
};
main ()
{
ULONG MessageClass;
USHORT code, nr ~ 0;
193
3.
INTUITION AND
AMIGA
FOREVER
{
if message ~ (struct IntuiMessage *)

GetMsg(FirstWindow->UserPort ~= NULL)
{
Wait(1L
continue;
ReplyMsg(message);
{
case GADGET UP
ExampleProp.VertPot);
printf("Position: %u\n",
break;
case CLOSEWINDOW
Close_All ();
exit (TRUE) ;
break;
case GADGETDOWN
nr +~ 1;
printf ("Gadget activated %d.
times!\n", nr);
break;
1***************************************
***************************************1
void
*OpenLibrary();
if (! (IntuitionBase~ (struct IntuitionBase *)
{

Close_All();
exit (FALSE);
)

194
3.4 GADGETS
ABACUS

Close_All () ;
exit(FALSE);
}
1***************************************
*
*
*
*
*
Function: Close all
***************************************1
if (FirstWindow)
if (IntuitionBase)
3.4.1.3
The string gadget

String gadgets, like proportional gadgets, are an extension of boolean
gadgets. As you might imagine, you insert a Stringlnfo structure
instead of a Proplnfo structure. This is about the same length as the
Proplnfo structure, and contains all of the extra infonnation that a
string gadget needs.
You have no doubt used a string gadget before. Even the Workbench
offers them. When you change the name of a disk or fIle you select the
Rename option in the Workbench menu. A small window appears into
which you can enter the new name. This is a string gadget In this case,
the window is the same size as the gadget, but this is not important
You have to set up a structure for a string gadget just like any other
gadget. You must define a horizontal click field, since text input is
limited to a single line.
1* 3.4.1.3.A. stringgadget */
struct Gadget StringGadget
{
NULL,
50, 40,
120, 10,
GADGHCOMP,
RELVERIFY t
STRINGCENTER,
STRGADGET,
/* NextGadget
1* Width, Height
1* Flags
1* Activation
*1
*1
*1
*1
*1
1* Gadget Type
*1
195
3.
INTUITION AND
NULL,
&GadgetText,
NULL,
(APTR)&Stringlnfo,
3,
NULL,
/*
/*
/*
/*
/*
/*
/*
GadgetRender
Select Render
GadgetText
MutualExclude
SpecialInfo
GadgetID
UserData
*/
*/
*/
*/
*/
*/
*/
};
Some variables have additional or different values for a string gadget

The ftrst is the Flags flag. It now describes the appearance of the
cursor and not the entire click field.
Four more flags can be used in string gadget activation:
STRINGCENTER
This flag centers the text in the gadget when set. The
default setting (unset STRINGCENTER flag) leftjustifies the text in the gadget.
STRINGRIGHT This flag right-justifies the text in the gadget when
set. The default setting (unset STRINGRIGHT flag)
left-justifies the text in the gadget

LONGINT
This flag indicates an integer gadget (lets the user

enter numbers only).
ALTKEYMAP
Setting this flag defines the value AltKeyMap in

the Stringlnfo structure, thus allowing access to
an alternate keyboard driver.
After you specify the gadget type with STRGADGET, you must not
forget to identify the click field with a graphic. Borders are drawn
automatically only for proportional gadgets. We used a simple Border
structure here:
/*3.4.1.3.B.gadgetpairs*/
SHORT GadgetPairs[) =
{
0, 0, 122, 0, 122, 12, 0, 12, 0, 0,

};
-2, -2, I, 0, JAMl, 5, GadgetPairs, NULL,

};
The gadget text is not required, but you should include it so that the
user knows what to enter:
196
3.4 GADGETS
ABACUS
/*3.4.1.3.C/gadgettext*/
{
2,0, JAM2, -40, 1, NULL, (UBYTE *)"Text" ,NULL,

};
The last additional information worth mentioning is the pointer to the

Stringlnfo structure. This new structure contains the following
parameters:
/* 3.4.1.3.D.stringinfostruct*/
struct Stringlnfo
{
OxOO
Ox04
Ox08
OxOA
OxOC
OxOE
Ox10
Ox12
Ox14
Ox16
Ox18
Ox1C
Ox20
Ox24
00
04
08
10
12
14
16
18
20
22
24
28
32
36
UBYTE *Buffer;
UBYTE *UndoBuffer;
SHORT BufferPos;
SHORT MaxChars;
SHORT DispPos;
SHORT UndoPos;
SHORT NumChars;
SHORT DispCount;
SHORT CLeft;
SHORT CTop;
struct Layer *LayerPtr;
LONG Longlnt;
struct KeyMap *AltKeyMap;
};
The first two values are the most important. They point to a buffer in
which the strings are stored. The first buffer contains the text which
was actually entered. The second buffer is used for the Undo editing
feature. The user just has to press a key to restore the text.
You can quickly create such a buffer. You simply define an array with
the corresponding number of characters. It looks like this:
#define STRINGSIZE 80
unsigned char StringBuffer[STRINGSIZE] = "Hello Amiga!";
unsigned char UndoBuffer [STRINGSIZE];
You don't have to provide an undo buffer, but then the user can't use
this function.
You can use BufferPos to initialize the cursor at a specific location.
This is used when text is already in the buffer before the call, as in the
example above.
Since Intuition does not know how large the text buffer is, you have to
tell it in MaxChars how many characters can be entered. Note that the
terminating null byte must be included in the string length.
197
3. INTUITION AND C
AMIGA
The last value which can be set by the programmer is DispPos. The
number which Intuition finds there designates the first character which
is displayed in the gadget field.
All other values are handled by Intuition itself. You can read them for
information.
UndoPos
Cursor position in the undo buffer.
NumChars
Number of characters currently in the buffer.
DispCount
Number of characters which can be displayed in the

gadget field.
CLeft
The offset of the container from the left border of the

window.
CTop
The offset of the container from the upper border.
LayerPtr
A pointer to the layer in which the gadget is located.
LongInt
After pressing <Return>, this variable contains the

number entered for an integer gadget for later reading
and processing.
Al tKeyMap
As mentioned above, you can place a pointer to a

custom keyboard table here which is then used for
input.
Let's look at some practical applications now. Insert the gadget,

StringInfo, and other structures into your standard program 3.1.
The defmitions look like this:
/* 3.4.1.3.E.stringadditions */
unsigned char StringBuffer[STRINGSIZE] = "Hello Amiga!";
struct Stringlnfo Stringlnfo
&StringBuffer[O],
&UndoBuffer[Oj,
0,
STRINGSIZE,
0,
0,
0,
0,
0, 0,
NULL,
0,
NULL,
};
198
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
Buffer
Undo Buffer
Buffer Position
MaxChars
Display Position
Undo Position
NumChars
Display Counter
CLeft, CTop
LayerPtr
Longlnt
AltKeyMap
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
3.4
ABACUS
SHORT GadgetPairs[]
GADGETS
0, 0, 122, 0, 122, 12, 0, 12, 0, 0,

};

};
2,0, JAM2, -40,1, NULL, (UBYTE *)"Text" ,NULL,

);

NULL,
50, 40,
120, 10,
GADGHCOMP,
RELVERIFY I
STRINGCENTER,
STRGADGET,
NULL,
&Gadget Text,
NULL,
(APTR)&Stringlnfo,
1,
NULL,
);
/* NextGadget
/* Left Edge , TopEdge
/* Width, Height
/* Flags
/* Activation
*/
*/
/*
/*
/*
/*
/*
/*
*/
*/
*/
*/
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
Speciallnfo
/* GadgetID
/* UserData
*/
*/
*/
*/
*/
*/
*/
160, 50,
320, 150,
0, 1,
/*
/*
/*
/*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
CLOSEWINDOW
GADGETUP,
WINDOWDEPTH
/* Flags
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
/* First Gadget
&stringGadget ,
NULL,
/* CheckMark
(UBYTE *)"String Gadget Test",
NULL,
/* Screen
NULL,
/* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
/* Type
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
};
199
3.
INTUITION AND
AMIGA
For the program use a slightly modified test loop:

3.4.1.3.F.testloop *1
(
printf ("The string is: %s\n",
case GADGETUP
&St~ingBuffer[O]);
break;
case CLOSEWINDOW
How it works
Close_All();
exit (TRUE) ;
break;
A string gadget appears in the window. Clicking on the gadget allows

you to edit the text in it When you press the <Return> key. the
program printsilie string entered in the CLI window.
The program receives a GADGETUP message each time the user presses
<Return> while in a string gadget. This can be used as a message that
the input is finished. Here is a complete example program using a
string gadget:
1**************************************
*
*
*
* ================================== *
* Program: String-Gadget.c
*
*
Comments:
* Author: Date:
*
*
---------- ---------*
10/16/1987 for testing
* Wgb
*
only
*
*
*
*
**************************************1
tinclude <exec/types.h>
tinclude <intuition/intuition.h>
struct Window
*FirstWindow;
unsigned char StringBuffer[STRINGSIZE]
struct Stringlnfo Stringlnfo
"Hello Amiga";
StringBuffer,
UndoBuffer,
0,
STRINGSIZE,
0,
200
1*
1*
1*
1*
1*
Buffer
Undo Buffer
Buffer Position
MaxChars
Display Positoin
*1
*1
*1
*1
*1
3.4 GADGETS
ABACUS
0,
0,
0,
0, 0,
NULL,
0,
NULL,
};
1*
1*
1*
1*
1*
1*
1*
Undo Position
NumChars
Display Counter
CLeft, CTop
LayerPtr
LongInt
AltKeyMap
*1
*1
*1
*1
*1
*1
*1
SHORT GadgetPairs[] =
{
0, 0, 122, 0, 122, 12, 0, 12, 0, 0,

};
struct Border GadgetBorder =

{
-2, -2, 1, 0,JAM1, 5, GadgetPairs, NULL,

};
2, 0, JAM2, -40, 1, NULL. (UBYTE *)"Text" ,NULL,

};

{
NULL,
50, 40,
120, 10,
GADGHCOMP,
RELVERIFY I
STRINGCENTER,
STRGADGET,
NULL,
&GadgetText,
NULL,
(APTR)&StringInfo,
1,
NULL,
1*
1*
1*
1*
1*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
Activation
*1
*1
1*
1*
1*
1*
1*
1*
1*
1*
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
Special Info
GadgetID
UserData
*1
*1
*1
*1
*1
*1
*/
*1
*1
*/
*/
};
*1
160, 50,
320, 150,
0, 1,
CLOSEWINDOW
GADGETUP I
GADGETDOWN,
/* Width, Height
/* IDCMP Flags
*1
*1
WINDOWDEPTH
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
/* Flags
*1
*J
201
3.
INTUITION AND
/* First Gadget
/* CheckMark
&StringGadget,
NULL,
*/
*/

NULL,
NULL,
/* Screen
/* BitMap
*/
*/
100, 50,
640, 200,

*/
*/
WBENCHSCREEN,
/* Type
*/
);
main ()
(
ULONG MessageClass;
USHORT code;
RefreshGadgets(&StringGadget, FirstWindow, NULL);

FOREVER
{

(
Wait(lL
continue;
Rep1yMsg(message);
{
case GADGET UP
%s\n",&StringBuffer[O);
case CLOSEWINDOW
printf("The string is:

Close_All () ;
exit(TRUE);
break;
Open_All ()
{
202
3.4
ABACUS
GADGETS

printf (IIIntuition Library not found! \n");
Close_All () ;
exit(FALSE);
}

{
printf(IIWindow will not open!\n");

Close_All () ;
exit (FALSE) ;
}
Closeyll ()
{
if (FirstWindow)
if (IntuitionBase)
CloseLibrary (IntuitionBase) ;
3.4.2
Gadgets in action
Now we will present an example of multiple applications of the three
gadgets just described.
Gadgets are user friendly because of their graphic nature. But string and
proportional gadgets are not without problems. That is why we want to
employ a few examples that can be built into other programs, but have
been designed specifically with our editor in mind.
Until now the gadgets were always connected with the Newwindow
structure, and the program simply displayed them. This must be
changed, because not all gadgets are needed when you first open a
window. Gadgets are usually inserted temporarily into a window then
removed from the window after the user response has been determined.
The Intuition gadget functions are used to control the display of the
gadgets.
The largest use of gadgets are in requesters. They are collections of
multiple gadgets. Because we need many gadgets, we want to give you
some examples that will be completed in the chapter on requesters.
203
3. INTUITION AND C
3.4.2.1
AMIGA
Gadget functions
First we want to concern ourselves with the problem mentioned above,
that not all gadgets are always wanted when opening a window. In
many programs, having a gadget inserted in a window can be useful.
Intuition has many functions that can do this.
Let's imagine that we've opened a nonnal window. Then, during
program execution, a gadget is placed in the window structure and then
appears on the screen.
We described the operation in single steps because we must proceed in
the same single steps when programming. As a base program you will
use the First_window. c program from section 3.1, including the
global NewWindow structure. Define a toggle select boolean gadget
with the following gadget structure, but don't place it in the
Newwindow structure:
1*3.4.2.1.A.togglegadget *1
SHORT GadgetPairs[]
{
0, 0, 61, 0, 61, 21, 0, 21, 0, 0

};
-1, -1, 1, 0, JAMl, 5, GadgetPairs, NULL

};
struct IntuiText ToggleText
3, 0, JAM2, 4, 7, NULL, (UBYTE *)"Toggle" ,NULL

};
struct Gadget ToggleGadget
NULL,
120, 40,
60, 20,
GADGHBOX,
RELVERIFY
TOGGLESELECT,
BOOLGADGET,
NULL,
&ToggleText,
NULL,
NULL,
1,
NULL
};
204
1*
1*
1*
1*
1*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
Activation
*1
*1
*1
*1
*1
1*
1*
1*
1*
1*
1*
1*
1*
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
Speciallnfo
GadgetID
UserData
*1
*1
*1
*1
*1
*1
*1
*1
3.4
ABACUS
GADGETS
A toggle gadget goes into the "selected" status when clicked on, and
when they are clicked on again they are "unselected."
We place this gadget in the window using an Intuition function. It is
called AddGadgetO and has the following syntax:
RealPosition
DO
AddGadget(Window, Gadget, Position);

-42
AO
A1
DO
We fIrst pass the window in which the gadget should be inserted along
with the address of the gadget itself. Next we test for the position that it
should take in the list As a result we get the current position.
For our toggle gadget the command looks like this:
Pos
= AddGadget(FirstWindow,
ToggleGadget, -1L);
The -1 states that the gadget should be inserted as the last thing in the
list.
With this function call the gadget takes control in the window list, but
it cannot yet be seen by the user. It still needs an additional function
call, called automatically when a new window is set up. We call this
function using its name:
RefreshGadgets(Gadgets, Window, Requester);
-222
AO
A1
A2
This function shows all of the given gadgets in the window list. Not
everything new is displayed with this; we test with gadgets to see
which gadget the routine should begin with. That is why we placed our
new gadget at the end of the list. We enter its number, and this is
displayed.
We insert RefreshGadgetO following the AddGadgetO statement:
RefreshGadgets(Pos, FirstWindow, NULL);
This makes the gadget visible. Now let's look at the toggle gadget
check:
1* 3.4.2.1.B.toggle_forever *1
FOREVER
{

(
Wait(1L
continue;
205
3.
INTUITION AND
AMIGA
GadgetPtr = (struct Gadget *) message->IAddress;
SelectMode= GadgetPtr->Flags & SELECTED;
ReplyMsg(message);
{
case GADGET UP
i f (SelectMode)
{
printf (ltActivated
It);
else
{
printf(ltDeactivated\n lt );
}
break;
case CLOSEWINDOW
Close_All();
exit (TRUE) ;
break;
The test to see if a gadget was pressed is done with a case statement.
The gadget status is also analyzed to see if it is "selected" or
"unselected". A simple AND comparison does this. The program also
displays the status of the gadget in the eLi window. Here is the
complete program listing:
/**************************************
* Program: Toggle-gadget.c
*
* ================================== *
*
*
Comments:
* Author: Date:
*
*
---------- ---------*
*
* Wgb
*
only
*
*
*
**************************************1
struct Window
*FirstWindow;
SHORT GadgetPairs[]
{
0, 0, 61, 0, 61, 21, 0, 21, 0, 0,
206
3.4 GADGETS
ABACUS
};

};
struct IntuiText GadgetText =

{
2, 0, JAM2, 4, 7, NULL, (UBYTE *)"Gadget" ,NULL,

};
struct Gadget BoolGadget
NULL,
10, 40,
60, 20,
GADGHCOMP,
TOGGLESELECT,
BOOLGADGET,
NULL,
&GadgetText,
NULL,
NULL,
1,
NULL,
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
Activation
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
Special Info
GadgetID
UserData
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
);

{
160, 50,
320, 150,
0, 1,
CLOSEWINDOW
GADGETUP,
/*
/*
/*
/*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
*/
*/
*/
*/
WINDOWDEPTH
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
/* Flags
*/
&BoolGadget,
NULL,
/* First Gadget
/* CheckMark
*/
*/
(UBYTE *)"System programmng test",

NULL,
NULL,
/* Screen
/* BitMap
*/
*/
100, 50,
640, 200,

*/
*/
WBENCHSCREEN,
};
/* Type
*/
207
3.
INTUITION AND
AMIGA
main ()
{
ULONG MessageClass;
USHORT code;
RefreshGadgets(&BoolGadget. FirstWindow. NULL):

FOREVER
{

{
Wait(lL
continue;
Select Mode = GadgetPtr->Flags & Selected;
ReplyMsg(message);
{
case GADGETUP
if (SelectMode)
{
printf(nActivated\n");
}
else
printf(nDeactivated\n A ) ;
break;
case CLOSEWINDOW
Close_All();
exit(TRUE);
break;
Open_All ()
{
OpenLibrary(nintuition.library". OL)))
208
3.4
ABACUS
GADGETS
printf (ttIntuition Library not found! \ntt);

Close_All () ;
exit(FALSE);
}

(
printf(ttWindow will not open!\n");

Close_All () ;
exit(FALSE);
}
Close_All ()
{
if (FirstWindow)
if (IntuitionBase)
At the moment this new toggle gadget has no further function. What
can turn this gadget on and off? Another gadget, of course. We'll add a
new gadget and let it tum the toggle gadget on and off. This way we
learn about two new functions.
As the second gadget we choose a completely normal boolean gadget to
minimize programming hassles. Here's the structure:
/* 3.4.2.1.C secondgadget */
SHORT SelectPairs[]
{
0, 0, 61, 0, 61, 21, 0, 21, 0, 0,

};
struct Border SelectBorder
-1, -1, 1, 0, JAMl, 5, SelectPairs, NULL,

};
struct IntuiText Select Text
1, 0, JAM2, 4, 7, NULL, (UBYTE *)"Select" ,NULL,

};
struct Gadget Select Gadget
{
NULL,
10, 40,
60, 20,
GADGHCOMP
GADGDISABLED,
RELVERIFY,
BOOLGADGET,
(APTR)&SelectBorder,
NULL,
/* NextGadget
/* Width, Height
/* Flags
*/
*/
*/
*/
/*
/*
/*
/*
*/
*/
*/
*/
Activation
Gadget Type
GadgetRender
Select Render
209
3.
INTUITION AND
AMIGA
&Se1ectText,
NULL,
NULL,
2,
NULL,
};
*!
/* GadgetText
/* Mutua1Exc1ude
/* SpecialInfo
*/
*/
/* GadgetID
/* UserData
*/
*/
(If you want you can use another gadget instead)

Structure
description
As usual, we have supplied you with the border and the IntuiText
structure. We now turn our attention to the Select gadget, which is
the same as a normal boolean gadget with the exception that it is
"disabled." It is set in this condition using the GADGDISABLED flag;
the user cannot choose it. The entire click area is displayed in a graphic
status called "ghosted."
This second gadget must also be added to the window. We can construct
a second AddGadgetO function. Intuition supplies another function in
case more gadgets must be added. The entire list of gadgets can be added
to the window with AddGListO. Set a pointer to the second gadget in
the ftrst gadget. Now, you can insert both gadgets into the window list
using the Rea 1 P 0 s i t ion 0 function. The format for
RealPositionO:
Rea1Position = AddGList(Window, Gadget, Position, Numgad,
DO
-438
AO
Al
DO
D1
Requester);
D2
The ftrst three arguments may be familiar to you from the simple functions. By using Numgad you can state how many gadgets the list
contains, and Requester must point to the requester when you need
one. In this case, though, the value is stored with null.
The same problem that we had with AddGadgetO is also encountered
with RefreshGadgetsO. We ftnd the first gadget's position with
AddGListO. We state this with the Refresh function. Only two
gadgets should be newly displayed. For this reason there is also a list
refresh command:
RefreshGList(Gadgets, Window, Requester, Numgad);
-432
AO
Al
A2
DO .
Similar to the AddGList function, Numgad specifies the number of

gadgets a Refresh should execute.
Now we turn back to our program. It should return the boolean gadget
as free because the toggle gadget was activated and returned. To do this
we must be able to differentiate between the self deft ned gadgets.
Gadget ID helps us here. The number entered there makes it possible
for us to identify each gadget. This is why you choose the different
numbers for each gadget in the program.
210
3.4
ABACUS
GADGETS
Consequently the check must now be changed. You should familiarize

yourself with the two functions that you can use to influence the status
of a gadget. They are:
OffGadget (gadget, Window, Requester);
-174
AO
A1
A2
OnGadget(gadget, Window, Requester);

-186
AO
A1
A2
Both functions need the same arguments: a pointer to the gadget whose
status should be changed, the pointer to the window in which the gadget
is found and a pointer to a requester, if it is presenL The last variable
does not concern us at the moment.
Change the check so that when the SELECT flag is positive the
program calls OnGadgetO, and when the flag is negative, the program
calls OffGadgetO. Every time you pass a pointer to the SELECT
gadget:
switch (GadgetNr)
(
case 2
printf (lI s e1ected! \n") ;

break;
case 1
if (selectmode)
(
printf("activated ");
OnGadget(&SelectGadget, FirstWindow,NULL);
RefreshGadgets(&SelectGadget, FirstWindow,NULL);
}
else
(
printfC"deactivated\n");
OffGadget (&SelectGadget, FirstWindow,NULL);
}
break;
}
break;
Insert this instead of the if check by the case GADGETUP. In addition you must calculate the ushort variable GadgetNr before this
point:
GadgetNr
How it works
= GadgetPrt->GadgetID;
When you have added all of the new sections to the first program, you
should see the following when the program is run:
A window with two gadgets appear. The right gadget turns the left
gadget on and off. The second gadget fIrst appears as "ghosted." When
you click on the right gadget, the type of the other gadget should be
readable, not ghosted. Now you can select this gadget as well. The
program shows its reactions using text in the CLI window. The text
211
3.
INTUITION AND
AMIGA
"activated" and "deactivated" appear for the toggle gadget. When you can
click on the left gadget, the program sends the text "selected!"
Here is the complete listing of the program:
/********~*****************************
*
*
*
* ================================== *
* Program: OnOff-gaggets.c
*
*
Comments:
* Author: Date:
*
*
*
---------- ---------10/16/1988 for testing
*
* Wgb
only
*
*
*
*
**************************************/
struct Window
*FirstWindowi
struct IntuiMessage *messagei
SHORT SelectPairs[]
{
0, 0, 61, 0, 61, 21, 0, 21, 0, 0,

}i
struct Border Select Border =

-1, -1, 1, 0, JAM1, 5, SelectPairs, NULL,
}i
struct IntuiText Select Text
1, 0, JAM2, 4, 7, NULL, (UBYTE *) "Select" ,NULL,

}i
struct Gadget SelectGadget
{
NULL,
10, 40,
60, 20,
GADGHCOMP,
RELVERIFY,
BOOLGADGET,
NULL,
&SelectText,
NULL,
NULL,
212
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
NextGadget
LeftEdge, Top Edge
Width, Height
Flags
Activation
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
SpecialInfo
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
3.4
ABACUS
2,
NULL,
};
1* GadgetID
1* UserData
GADGETS
*1
*1
SHORT GadgetPairs[]
(
0, 0, 61, 0, 61, 21, 0, 21, 0, 0

};

(
-1, -1, 1, 0, JAM1, 5, GadgetPairs, NULL

};
struct IntuiText BoolText
3, 0, JAM2, 4, 7, NULL, (UBYTE *)"BoolGad" ,NULL

};
struct Gadget BoolGadget =
(
&SelectGadget,
120, 40,
60, 20,
GADGHBOX,
RELVERIFY
TOGGLESELECT,
BOOLGADGET,
NULL,
&BoolText,
NULL,
NULL,
1,
NULL
};
1*
1*
1*
1*
1*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
Activation
*1
*1
*1
*1
*1
1*
1*
1*
1*
1*
1*
1*
1*
Gadget Type
GadgetRender
Select Render
Gadget Text
MutualExclude
SpecialInfo
GadgetID
UserData
*1
*1
*1
*1
*1
*1
*1
*1

(
*1
*1
*1
*1
160, 50,
320, 150,
0, 1,
CLOSEWINDOW
GADGETUP I
GADGETDOWN,
1*
1*
1*
1*
WINDOWDEPTH
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
1* Flags
*1
&BoolGadget,
NULL,
1* First Gadget
1* CheckMark
*1
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
*/
213
3.
INTUITION AND
AMIGA
NULL,
NULL,
1* Screen
1* BitMap
*1
lOa, 50,
640, 200,

*1
*1
WBENCHSCREEN
};
1* Type
*1
*/
main 0
(
ULONG MessageClass;
USHORT code, selectmode, GadgetNr;
struct Gadget *Gadgetptr;
RefreshGadgets(&BoolGadget, FirstWindow, NULL);

FOREVER
{

{
Wait (1L
continue;
MessageClass = message->Classi
GadgetNr = GadgetPtr->GadgetID;
selectmode= GadgetPtr->Flags & SELECTED;
ReplyMsg(message);
switch (MessageClassl
{
case GADGET UP
case GADGETDOWN
switch (GadgetNr)
{
case 2
printf ("selected! \n") ;

break;
case 1
if (selectmode)
(
printf("activated H);
OnGadget(&SelectGadget,
FirstWindow,NULL);
214
3.4
ABACUS
GADGETS
RefreshGadgets(&SelectGadget, FirstWindow,NULL);
}
else
{
printf ("deactivated\n") ;
Of f Gadget (&SelectGadget,
FirstWindow,NULL);
break;
break;
case CLOSEWINDOW
Close_All();
exit(TRUE);
break;
Open_All ()
{
{
printf ("Intuition Library not found! \n") ;

Close_All () ;
exit(FALSE);
}

{
printf (IIWindow will not open! \n") ;

Close_All () ;
exit(FALSE);
}
Close_All ()
{
if (FirstWindow)
if (IntuitionBase)
Now that we've finished with boolean gadgets, let's describe the other
two types. First we should turn our attention to the proportional
gadgets, for which there are two special functions. This example will
supply a graphic to the knob. Here is the pertinent Image structure
with the graphic data:
215
3.
INTUITION AND
AMIGA
USHORT KnobGraphic[]
{
OxCCCC,
Ox6666,
Ox3333,
Ox9999,
OxCCCC,
Ox6666,
Ox3333,
Ox9999,
OxCCCC,
Ox6666,
Ox3333,
Ox9999,
OxCCCC,
Ox6666,
Ox3333,
Ox9999
};
struct Image Knob
0, 0,
16, 16,
1,
&KnobGraphic[O],
1, 0,
NULL
};
The gadget structure with the PropInfo belongs to this gadget:

struct PropInfo SliderProp
{
FREEVERT,
Ox8000,
Ox8000,
Ox0800,
Ox1000,
0,
0,
0,
0,
0,
0
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
Flags
HorizPot
VertPot
HorizBody
Vert Body
CWidth
CHeight
HPotRes
VPotRes
LeftBorder
TopBorder
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
};
struct Gadget Slider
{
NULL,
260, 40,
24, 120,
GADGHNONE
GADGIMAGE,
RELVERIFY,
PROPGADGET,
216
/*
/*
/*
/*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
/* Activation
/* Gadget Type
*/
*/
*/
*/
*/
*/
3.4
ABACUS
(APTR)&Knob,
NULL,
NULL,
NULL,
(APTR)&SliderProp,
1,
NULL,
};
Note:
/*
/*
/*
/*
/*
/*
/*
GadgetRender
Select Render
GadgetText
MutualExclude
SpecialInfo
GadgetID
UserData
GADGETS
*/
*/
*/
*/
*/
*/
*/
The graphic data of the mover must be stored in chip memory or it will
not be displayed, use the +c option of the Aztec compiler to place this
data in chip memory.
We should read the mover's position. Insert these two lines in the
gadget check:
case GADGETUP :printf("Mover position: %x/n",
moveprop.VertPot);
break;
You can check and evaluate the slider's position in the program using
these methods. Here is a complete example program:
/***************************************
*
*
*
* =============================;===== *
* Program: Prop-Image-Gadget.c
*
*
Comments:
*
* Author: Date:
*
---------- ---------*
* Wgb
*
only, place in *
*
chip ram
*
*
***************************************/
struct
struct
struct
struct
IntuitionBase
Screen
Window
IntuiMessage
*IntuitionBase;
*FirstScreen;
*FirstWindow;
*message;
USHORT KnobGraphic[)
{
OxCCCC,
Ox6666,
Ox3333,
Ox9999,
OxCCCC,
Ox6666,
Ox3333,
217
3.
INTUITION AND
AMIGA
Ox9999,
OxCCCC,
Ox6666,
Ox3333,
Ox9999,
OxCCCC,
Ox6666,
Ox3333,
Ox9999
};
struct Image Knob
0, 0,
16, 16,
1,
&KnobGraphic[O],
1, 0,
NULL
};
{
FREEVERT,
OxBOOO,
OxBOOO,
OxOBOO,
Ox1000,
0,
0,
0,
0,
0,
0
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
Flags
HorizPot
VertPot
HorizBody
VertBody
CWidth
CHeight
HPotRes
VPotRes
LeftBorder
TopBorder
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
};

{
NULL,
260, 40,
24, 120,
GADGHNONE
GADGIMAGE,
RELVERIFY,
PROPGADGET,
(APTR)&Knob,
NULL,
NULL,
NULL,
(APTR)&SliderProp,
1,
NULL,
};
218
1*
1*
1*
1*
NextGadget
LeftEdge, Top Edge
Width, Height
Flags
Activation
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
SpecialInfo
GadgetID
/* UserData
1*
1*
1*
1*
1*
1*
1*
1*
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
3.4
ABACUS
GADGETS

{
160, 0,
320, 200,
/* Width, Height
0, 1,
CLOSEWINDOW
/* IDCMP Flags
GADGETUP,
WINDOWDEPTH
/* Flags
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
&Slider,
/* First Gadget
NULL,
/* CheckMark
(UBYTE *)"Proportional-Gadget",
NULL,
/* Screen
NULL,
/* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
/* Type
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
};
main ()
{
ULONG MessageClass;
USHORT code;
FOREVER
{

(
ReplyMsg(message);
(

exit(TRUE);
break;
case GADGETUP :printf("MOVer position: %x/n",
moveprop.VertPot);
break;
219
3. INTUITION AND C
AMIGA
Open_All 0
{
void
*OpenLibrary();
Open Library (rrintuition. libraryn, OL)
(
printf(nlntuition Library not found!\n");

Close_All () ;
exit(FALSE);
)

*'
printf("Window will not open!\nrr);

Close_All () ;
exit(FALSE);
}
Close_All ()
{
i f (FirstWindow)
if (IntuitionBase)
There are situations in which it is important to change the position of

the slider during the program (e.g., size or something similar). We
would like to describe one method which can be used with any gadgeL
Then we will explain two functions that are used only for proportional
gadgets.
General tip for all gadgets: To change any setting on a gadget, you
must fIrst remove this gadget from the window list. When this is not
done, errors occur. RemoveGadgetO is a function which permits a
gadget to be removed from the list.
Position - RemoveGadget(Window. Gadget);
eDO
-228
AD
Al
Now you can make changes to the Gadget structure, the Proplnfo
structure and all associated structures. Then it can be added back to the
list with AddGadgetO. To conclude you must update the screen using
RefreshGadgetO
In other cases where entire portions of gadgets are removed, changed and
then reentered in the list, there is an additional function:
220
3.4
ABACUS
GADGETS
Position = RemoveGList(Window, Gadget, Numgad);

OeO
-444
AO
Al
DO
Here you give the number of gadgets that should be removed. The
opposite, AddGListO, has already been discussed. Don't forget to
update the graphic with RefreshGListO.
The second way, only for proportional gadgets, saves the time of
deleting and reinserting every gadget from the list The Modi fyP ropO
function lets you change all possible gadget settings in the Proplnfo
structure, after describing the gadget. The gadget is then automatically
redisplayed:
ModifyProp(Gadget, Window, Requester, Flags, HorizPot,
-156
AO
A1
A2
00
01
VertPot, HorizBody, VertBody);

02
03
04
Unfortunately this function has a disadvantage: In most applications it

is often necesary to change the settings using ModifyPropO.
However, when RefreshGadgetsO performs it function, all the
gadgets following in the list are also "refreshed." This can be annoying,
therefore NewModifyPropO was added to the Intuition functions.
NewModifyProp(Gadget, Window, Requester, Flags, HorizPot,
-156
AO
A1
A2
DO
01
VertPot, HorizBody, VertBody, Numgad);

02
03
04
05
The new function serves exactly the same purpose as ModifyProp.

You can use NewModifyProp () to select how many gadgets in the
list should be refreshed. We recommend 1 as a good value, unless you
have other gadgets which relate to the change and should be changed
accordingly.
File Selector
We would now like to present the structure for a file selector box, used
to select a single file name from a list of many file names. This is used
to load a file into an application program. Examples of these are
loading pictures into a drawing program or loading a text file into an
editor.
If you thought that you could print out the names with the
Int ui text structure, in conjunction with the Gadget structure,
you're wrong. The programming can get rather complicated when a file
list is present that contains more names that can fit in the window. The
text must be scrolled to allow the user to see all of the files.
It would be best to create a routine to display the list of names, beginning with a certain one, in a predetermined order. The number of names
that are to be displayed is the number of gadgets that have no graphics
associated with them. They are only known by their click area and the
221
3.
INTUITION AND
AMIGA
text displayed in them. This cannot be more than one line, 8/9 points,
and must be within the maximum number of characters of a filename.
Because these are all boolean gadgets, we can use the simple gadget
structure:
1*3.4.2.1.D.filegadget_struct*1
struct Gadget FileGadget =
{
NULL,
80, 0,
200, 8,
GADGHCOMP,
RELVERIFY,
BOOLGADGET
REQGADGET,
NULL,
NULL,
NULL,
NULL,
NULL,
0,
NULL,
1*
1*
1*
1*
1*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
Activation
*1
*1
*1
*1
*1
1*
1*
1*
1*
1*
1*
1*
1*
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
Speciallnfo
GadgetID
UserData
*1
*1
*1
*1
*1
*1
*1
*1
};
Because of the large width of your gadget we would like to ask that you
now load the First_window. c program from Section 3.1 again and
change the window structure there so that the window has a LeftEdge
of 120 and Width of 360. We want to begin developing a file select
box, and we need a wide enough window to do so. Look at the design:
File select box used as a wIndow
Load/save file
Next dIrectory
level up
String gadgets
Figure 3.11
222
812345678'812345678
812345678'B12345678
812345678'812345678
812345678'01Z345678
I. OK 1
_ 812345678'012345678
II":=:iCancel 8123456m012345678
~
812345678'812345678
81Z345678'812145678
J.CHDIRII
.
. 912345678'812345678
812345678'812345678
912345678'812345678
812345678'812345678
812345678'012345678
Proportional
gadget
Path: _ _ _ _ __
File::------
Boolean
gadgets
3.4
ABACUS
GADGETS
You also need a large field for the filenames, a scroll bar (which you
programmed above), two string gadgets and three boolean gadgets. After
these digressions for an overview of your project, we turn back to the
many gadgets for selecting a name.
With the gadget structure defined first we don't get very far. You need
something more, defining every structure individually is too hard.
Remember the example of the IntuiText structure. There we
reproduced a structure quite simply. That would be handy to have here:
Preceding the main function we define this gadget structure and a field
with unfilled structures.
struct Gadget Files[15];
We need a function in the program that copies the general structure and
at the same time corrects the Y position.
/*3.4.2.1.E.makefl1es*/
Make_Fl1es(Structur)
struct Gadget Struktur[];
{
int i;
for (i=O; 1<15; i++)
{
Struktur[i]
Struktur[i] .GadgetID
Struktur[i] .NextGadget
= FileGadget;
= 30
=
=
+ i
i + 1;
* 8;
&Struktur[i+l];
Struktur[14].NextGadget
NULL;
How it works
The function puts the general structure in the structure field and corrects
the Y value (TopEdge). In addition it links the entire list of 15
gadgets together so that you can move them into the window with a
simple call (AddGList (. Furthermore, every gadget receives an ID
beginning with 1. This lets you establish later which of the user gadgets were clicked on. A Refresh is unnecessary in this case because
no graphic is used.
Now insert the above defined proportional gadget into the program.
Change the variable LeftEdge to a value of 300. The ID of these
gadgets must also receive a value greater than 15 so that it is differentiated from the others. We recommend 20, so there is space for enlarging.
You are now missing the two strings. Because there are only two, you
can define them normally through the structure. Here we encounter the
wonderful condition that we only need one Undo buffer because only
one gadget can be active at a time.
223
3. INTUITION AND C
AMIGA
/* 3.4.2.1.F.buffers*/
unsigned char PathBuffer[512] = "df1:";
unsigned char FileBuffer[31]
I
unsigned char UndoBuffer [512];
.....
struct Stringlnfo Pathlnfo

{
&PathBuffer[O],
&UndoBuffer[O],
5,
511,
0,
0,
0,
0,
0, 0,
NULL,
0,
NULL,
},
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
Buffer
Undo Buffer
Buffer Position
MaxChars
Display Position
Undo Position
NumChars
Display Counter
CLeft, CTop
Layerptr
Longlnt
AltKeyMap
*/
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
/*
Buffer
Undo Buffer
Buffer position
MaxChars
Display Position
Undo Position
NumChars
Display Counter
CLeft, CTop
LayerPtr
Longlnt
AltKeyMap
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*/
*/
*/
*/
*/
*1
*1
*1
*/
*1
*1
struct Stringlnfo Filelnfo

&FileBuffer [0],
&UndoBuffer[O],
0,
31,
0,
0,
0,
0,
0, 0,
NULL,
0,
NULL,
};
SHORT StringPairs[]
{
0, 0, 248, 0
},
struct Border StringBorder

{
0, 9, 1, 0, JAM1, 2, StringPairs, NULL,

}
struct IntuiText PathText
1, 0, JAM2, -58, 1, NULL, (UBYTE *)"Path:" ,NULL,

};
struct IntuiText FileText =

{
1, 0, JAM2, -58, 1, NULL, (UBYTE *) "File :" , NULL,

};
224
3.4
ABACUS
struct Gadget Path
GADGETS
NULL,
80, 150,
250, 8,
GADGHCOMP,
RELVERIFY,
STRGADGET,
(APTR)&StringBorder,
NULL,
&PAthText,
NULL,
(APTR) &Pathlnfo,
21,
NULL,
/* NextGadget
*/
*/
/* Width, Height
*/
/* Flags
*/
/* Activation
*/
/* Gadget Type
*/
/* GadgetRender
*/
/* Select Render
*/
/* GadgetText
*/
/* MutualExclude
*/
/* SpecialInfo
*/
/* GadgetID
*/
/* UserData
*/
};
struct Gadget File

{
&Path,
80, 165,
250, 8,
GADGHCOMP,
RELVERIFY,
STRGADGET,
(APTR)&StringBorder,
NULL,
&FileText,
NULL,
(APTR)&Filelnfo,
22,
NULL,
};
/* NextGadget
*/
*/
/* Width, Height
*/
/* Flags
*/
/* Activation
*/
/* Gadget Type
*/
/* GadgetRender
*/
*/
/* Select Render
/* GadgetText
*/
/* MutualExclude
*/
*/
/* SpecialInfo
/* GadgetID
*/
/* UserData
*/
You need boolean gadgets as the last gadgets for your file select box.
You simply need one border structure, three IntuiText structures,
and three gadget structures:
/*3.4.2.1.G.finalgadgets*/
SHORT SelectPairs[]
{
0, 0, 61, 0, 61, 21, 0, 21, 0, 0,

l;
struct Border Select Border =
-1, -1, 1, 0, JAM1, 2, StringPairs, NULL,
};
struct IntuiText OKText
1, 0, JAM2, 22, 5, NULL, (UBYTE *)"

};
OK
II
,NULL,
225
3. INTUITION AND C
AMIGA
struct IntuiText CancelText
1, 0, JAM2, 2, 5, NULL, (UBYTE

};
struct IntuiText ReturnText
*)
"Cancel" ,NULL,
*)
"CHDIR" ,NULL,
1, 0, JAM2, 2, 5, NULL,
};
(UBYTE
struct Gadget OK
NULL,
10, 40,
60, 20,
GADGHCOMP,
RELVERIFY,
BOOLGADGET,
NULL,
&OKText,
NULL,
NULL,
23,
NULL,
};
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
1*
NextGadget
LeftEdge, TopEdg~
Width, Height
Flags
Activation
Gadget Type
GadgetRender
Select Render
GadgetText
MutualExclude
SpecialInfo
GadgetID
UserData
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
Activation
Gadget Type
GadgetRender
Select Render
/* GadgetText
1* MutualExclude
1* SpecialInfo
1* GadgetID
1* UserData
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
struct Gadget Cancel =

&OK,
10, 65,
60, 20,
GADGHCOMP,
RELVERIFY,
BOOLGADGET,
NULL,
&CancelText,
NULL,
NULL,
24,
NULL,
};
struct Gadget Return
*/
*1
&Cancel,
10, 90,
60, 20,
GADGHCOMP,
RELVERIFY,
BOOLGADGET,
NULL,
&ReturnText,
NULL,
226
1*
1*
1*
1*
1*
1*
1*
1*
1* NextGadget
*!
1* Width, Height
*1
1* Flags
*/
1* Activation
*1
1* Gadget Type
*1
1* GadgetRender
*1
1* Select Render
*1
1* GadgetText
*1
1* MutualExclude
*1
3.4
ABACUS
NULL,
25,
NULL,
GADGETS
/* SpecialInfo
/* GadgetID
/* UserData
*/
*/
*/
};
Now you have three linked gadget lists: the file gadgets without graphics, the two strings gadgets, and the boolean gadgets. You also have a
proportional gadget. Now construct all of the gadget definitions in the
standard program and think about the window measurements. Then add
the following after the open_AllO function call in the main
function:
/* 3.4.2.1.H.makingmorefiles */
Make_Files(Files);
FilePos = AddGList(FirstWindow, Files, -lL, 15, NULL);
Strgpos = AddGList(FirstWindow, File, -lL, 2, NULL);
SlctPos = AddGList(FirstWindow, Return, -lL, 3, NULL);
PropPos = AddGadget(FirstWindow, Slider, -lL);
RefreshGList(File, FirstWindow, NULL, 2);
RefreshGList(Return, FirstWindow, NULL, 3);
RefreshGList(Slider, FirstWindow, NULL, 1);
We don't want to give the gadget check an evaluation. Here we insert an

output of the GadgetID in case it is handled as being self defined. The
important program sections are found in the example numbered
3.4.2.1.c.1. It is missing the line:
printf("Gadget number: %d\n", GadgetNr);
A complete program of this file selector is presented in Section 3.5.4.
3.4.2.2
A new sizing gadget

Now we finally come to a custom sizing gadget. When you click on
this gadget, and the window is not at the maximum size that it can
assume on the screen, it assumes full size. When it reaches the
maximum size on the screen, clicking it again brings the size to its
minimum. Be careful that the MinWidth and MinHeight values are
initialized.
/*3.4.2.2.A.sizing*/
struct Gadget Sizing =
{
NULL,
-10, 15,
8, 10,
GADGHCOMP
GRELRIGHT,
/*
/*
/*
/*
NextGadget
LeftEdge, TopEdge
Width, Height
Flags
*/
*/
*/
*/
227
3. INTUITION AND C
AMIGA
RELVERIFY I
RIGHTBORDER,
BOOLGADGET,
(APTR)&SizingImage,
NULL,
NULL,
NULL,
NULL,
0,
NULL,
};
/* Activation
*1
/* Gadget Type
1* GadgetRender
/* Select Render
/* GadgetText
/* MutualExclude
1* SpecialInfo
1* GadgetID
/* UserData
*/
*/
*/
*/
*/
*/
*/
*/
As the Image structure, we ask you to use the structure that has been
implemented at the end of the preceding section. The check routine
should have the following modifications:
/*3.4.2.2.b.modifications */
Sizing()
{
SHORT ScreenWidth, ScreenHeight;

ScreenWidth = Window->WScreen->Width;
ScreenHeitgh= Window->WScreen->Height;
NULL & Window->TopEdge == NULL)
ScreenWidth &
Window->Height
ScreenHeight)
SizeWindow(Window, -l*(Window->Width Window->MinWidth),
if (Window->LeftEdge
if (Window->Width
==
==
-l*(Window->Height Window->MinHeight);
MoveWindow(Window, -l*Window->LeftEdge,
-l*Window->TopEdge);
SizeWindow(Window, ScreenWidth - Window->Width,
ScreenHeight - Window->Height);
228
3.5 REQUESTERS
ABACUS
3.5
Requesters
Requesters are an extension of gadgets. Gadgets are used by programs to
request infonnation and user selections during program execution.
Sometimes a program must request information or user input
immediately during its execution. The computer must then ask the user
for the information needed, before the program can execute further.
A requester appears to request this information. A requester is a box
containing at least one gadget to aid user response. Text can be added to
the gadgets to clarify the problem or question. If the programmer
prefers, he can construct a requester entirely of graphics.
System
requester
The simplest example is the system requester. They appear when a

certain disk is required by the operating system. This requester allows
you to either Retry or Cancel the operation.
We want to place the fIle select box window we created in the previous
section in a requester. This requester will ask whether the user wants to
load or save a fIle. The user needs information such as the fIlename and
the directory in which it can/should be found.
We mentioned that requesters were supported by Intuition. You can also
draw your own fairly easily.
3.5.1
Automatic requesters
Requesters are a collection of many gadgets. You have already seen that
gadget programming is simple but requires many parameters.
Intuition offers a simple solution to this problem: Use the automatic
requester. Automatic requesters consist of two gadgets and one
prompting text. This simple requester doesn't require any complex
programming. The greatest advantage to the automatic requester is it
doesn't need a structure of its own. You simply give the function the
pointer to the lnt ui Text structure and you get a value back when the
user responds to the requester.
You need a pointer to a window to work with your automatic requester.
This must not be initialized.
229
3. INTUITION AND
AMIGA
Next three IntuiText structures can be given. The ftrst represents a

general text used for an explanation of the problem (Le., an
explaination or question). The other two texts act as gadget texts, as
you saw from the system requesters. The left comer states the positive
text, listing a consent or confirmation such as Retry. The right comer
states the negative text, listing denial, rejection or cancellation, such as
Cancel
We are still missing two flags. The ftrst represents the positive gadget
You can set this flag at zero because clicking on the gadget activates the
positive condition. The second flag does the same job as the first,
except it represents the negative gadget. These last two values are
known. Because Intuition doesn't know the measurements of the text
and the gadgets, you can set the requester's size at this point. Intuition
calculates the border sizes for the gadget text-so you don't have to.
Here's the general format of our new function with all of the arguments
in one short description:
Response = AutoRequest(Window, BodyText, PositiveText,
DO
-348
AD
Al
A2
NegativeText, PositiveFlags. NegativeFlags,
A3
DO
Dl
Width, Height);
D2
D3
Arguments
Window
Pointer to a Window structure.
BodyText
IntuiText general text structure.
PositiveText
IntuiText for Yes gadget.
Negati veText
Intui Text for No gadget
PositiveFlags
IDCMP flag for Yes gadget.
Negati veFlags
IDCMP Flag for No gadget.
Width
Width of auto requester window.
Height
Height of auto requester window.
The requester example does not need the First_window. c program.

You simply initialize the Int ui Text structures and assign all of the
values of the function. You get the response of the user by checking the
positive gadget. The following program reads it and displays a
corresponding text in the CLI window.
230
3.5 REQUESTERS
ABACUS
/**********************************
*
* ;===;==;====================== *
*
* Program: AutoRequester.c
*
* Author:
*
*
*
*
*
Wgb
Date:
Comments:
---------- ---------12/12/1987
*
*
*
*
**********************************/
finclude <intuition/intuition.h>
struct Window
*Window = NULL;
struct IntuiText Body
{
0, 1,
JAM2,
10, 10,
NULL,
(UBYTE *)"00 you wish to program requesters?".
NULL
};
struct IntuiText YES
{
2, 3,
JAM2,
5, 3,
NULL,
(UBYTE *)" I do I!!",
NULL
};
struct IntuiText NO
{
2, 3,
JAM2,
5, 3,
NULL,
(UBYTE *)" I don't
NULL
"
I" I
);
main ()
{
BOOL Answer;
231
3.
INTUITION AND
AMIGA
Answer = AutoRequest(Window, &Body, &YES, &NO, OL, OL,

320L, 60L);
if (Answer == TRUE)
printf("That is the truth!\n");
else
printf("The truth is always the best.\n");
*OpenLibrary ();
{

Close_All () ;
exit(FALSE);
}
Close_All ()
{
if (IntuitionBase)
How it works
The special feature of this program is that we can display a system

requester even if we don't have a reference window. This is important,
for example, for the request to release memory when it is impossible to
open the output window. To do this we set the window pointer to null.
This is not possible with every requester function.
responds to clicking a gadget, pressing a key
combination or inserting a disk. Pressing right <Amiga><V> selects
the Retry gadget, while pressing right <Amiga><B> selects the Cancel
AutoRequest
gadget.
The auto requester window has one small disadvantage We can't defme
any features of the requester window, and the AutoRequest assigns
the text, gadget graphics and dimensions.
232
3.5 REQUESTERS
ABACUS
3.5.2
System requesters
The use and programming of Auto-requesters is very simple. One
drawback: There is no possibility for outside manipulation from the
programmer.
The AutoRequest () function calls BuildSysRequest () which
does allow modifications. The following stipulations must be met to
work with BuildSysRequest () :
First, a window must be opened that should state the text size. Then we
assign texts for both gadgets. In addition, we need the IDCMP flag for
the requester window and the size of the requester.
The routine constructs a requester based on these arguments, in the
same style as AutoRequest: Body text in the top section of the
requester, gadgets in the lower left and right comers stating positive and
negative responses. A communication channel returns the pointer to the
requester window, either based on the pointer we assigned, or through
parameters given by the Workbench.
We must write the checking routine for the new requester. We can rely
on all IDCMP options for this routine. All input is permissible-it
doesn't matter whether the input comes from the mouse or the keyboard.
When the routine has problems opening the requester, this requester
changes into a recoverable Alert, just like an AutoRequest. Then
BuildSysRequest () returns a true value to a window instead of a
pointer. This tells us which of the two answers were selected.
Here's the format of this requester making function:
Reqwindow = BuiIdSysRequest(Window, BodyText,
DO
-360
AO
Al
PositiveText, NegativeText, IDCMPFIags, Width, Height);
A2
A3
DO
D2
D3
struct
struct
struct
struct
struct
ULONG
SHORT
Window
*ReqWindow;
*Window;
Window
IntuiText
*BodyText;
IntuiText
*PositiveText;
IntuiText
*NegativeText;
IDCMPFIags;
Width, Height;
For your check routine you'll need some information about the gadgets.
The following gadget flags should be set for each gadget:
BOOLGADGET, RELVERIFY, REQGADGET and TOGGLESELECT
233
3.
INTUITION AND
The gadgets should look the same, just to keep things consistent. Use
the AUTO flags in the <intuition/intuition.h> include
fIle:
AUTODRAWMODE
Value Description
JAM2 Background and foreground color
AUTOFRONTPEN
AUTOBACKPEN
OL
lL
Character color 1
Character color 2
AUTOLEFTEDGE
AUTOTOPEDGE
6L
3L
Positions of the IntuiTexts
Auto-Name
AUTOITEXTFONT NULL
Character sets for texts: Workbench-Zs
AUTONEXTTEXT
Pointer to further texts
NULL
All flags can be used in the IntuiText structures so that a uniform

user interface is possible. You should inform the user if he must keep
watch for other conventions.
With this description it should be possible for you to open a system
requester. The only thing missing is the reaction when closing the
requester. This is needed to free up all structures and memory that were
used. This task is taken care of by the FreeSysRequest ()
function. It uses the pointer returned by the BuildSysRequest ()
routine. If only a true value is given, this function may not be called.
3.5.3
Making our own requester

Sometimes the auto requester just isn't enough. In many cases you need
more than a yes/no answer. Enter the custom requester.
The system needs so many statements for a custom requester that it
pays to set up a structure. This structure contains all of the information
about the type, extent and appearance of the requester. By creating the
structure we add the advantage of flexibility. We must program our own
checking routines because the system no longer has access to its
defaults for requesters, as it does for AutoRequest ()
Think of a custom requester as a window into which we place multiple
gadgets. At least one gadget tells the program that input has ended.
Then we can analyze the data. This opens up custom requester design.
234
3.5 REQUESTERS
ABACUS
3.5.3.1
Requester structure
We need a requester structure for each requester. They contain information similar to that found in the window structure:
struct Requester
{
OxOO
Ox04
Ox06
Ox08
OxOA
OxOC
OxOE
OxlO
Ox14
Ox18
OxlC
OxlE
OxlF
Ox23
Ox43
Ox47
Ox4B
Ox6F
.;
00
04
06
08
10
12
14
16
20
24
28
30
31
35
67
71
75
struct Requester *OlderRequest;

SHORT LeftEdge;
SHORT TopEdge;
SHORT Width;
SHORT Height;
SHORT RelLeft;
SHORT RelTop;
struct Gadget *ReqGadget;
struct Border *ReqBorder;
struct IntuiText *ReqText;
USHORT Flags;
UBYTE BackFill;
struct Layer *ReqLayer;
UBYTE ReqPadl[32];
struct BitMap *ImageBMap;
struct Window *RWindow;
UBYTE ReqPad2[36];
III
The variable OlderRequester needs only to be initialized, then it

can only be used by Intuition for management purposes.
LeftEdge, TopEdge, width and Height state the position and
size of the requester in the window.
Re ILeft and ReI Top serve a very useful function. Set the flag
POINTREL in the flags value. Then the requester position is given
relative to the mouse pointer position instead of at the LeftEdge and
TopEdge positions. This way you can place a requester so that a
certain gadget lies beneath the mouse pointer.

The next three pointers in the structure let you use Gadgets,
Border and Texts. A requester cannot exist without gadgets.
Border lets you can give the requester an attractive border, and
IntuiText specifies the explanatory text
The most important value of the structure, flags, is a flag. You can
set the flag POINTREL from here so that the requester appears relative
to the mouse pointer position. A second flag named PREDRAWN indicates that all Border, Intui Text and Image structures should be
ignored. The ImageBMap takes over the entire graphic. Intuition
searches for an initialized bit-map and fIlls it with graphics. Notice that
235
3.
INTUITION AND
AMIGA
this new graphic also corresponds to the click area of the gadgets,
because these no longer have their own graphic.
Intuition also sets its own flags in this value. The REQOFFWINDOW
flag indicates that the requester is found outside of the bordering
window, and REQACTIVE indicates when the requester is active. The
flag SYSREQUEST is set when it is handled as a system requester (see
AutoRequest and SystemRequest).
The next variable in the structure is BackFill. Here you find the
number of the drawing pen that should fill in the background of the
requester.
ReqLayer is filled from Intuition and contains a pointer to the layer
that contains the requester.

ImageBMap is a pointer to a custom bit-map.
RWindow is a variable managed from Intuition. It points to the
window where the requester appears.
3.5.3.2
Establishing a requester structure

Because of the difficulty in collecting the gadgets for a requester, you
can add the example below to your own programs. First we place the
file selector box in a requester. This makes the program handling
simpler and more structured. You could use a file select box for our
"big" project (the editor).
/* 3.S.3.2.a. setreqstructure */
struct Requester FileSelectBox;
InitRequest(&FileSelectBox);
FileSelectBox.LeftEdge =
FileSelectBox.TopEdge =
FileSelectBox.Width
=
FileSelectBox.Height
=
FileSelectBox.ReqGadget=
Structure
description
236
2;
10;
360;
200;
&Return;
The measurements of the requester fit those of the window previously

used for the file select box. The Ini tRequest () function set all
arguments to null. Next we expand values. The gadget value points to
the first gadget and moves to the NextGadgetPointer of each
structure. That's why you must link the four gadget blocks. Because the
REQGADGET flag must be set in each gadget of a requester, the
example below takes you through all four gadget blocks:
3.5
ABACUS
REQUESTERS
/*3.5.3.2.b. reqgadgetset */
unsigned char PathBuffer[512] = "dfl:";
nn.,
(
OxCCCC,
OxCCCC,
OXCCCC,
OXCCCC,
Ox6666,
Ox6666,
Ox6666,
Ox6666,
Ox3333,
Ox3333,
Ox3333,
Ox3333,
Ox9999,
Ox9999,
Ox9999,
Ox9999
};
struct Image Knob
0, 0, 16, 16, 1, &KnobGraphic[O], 1, 0, NULL

};

(
FREEVERT,
Ox8000,
Ox8000,
Ox0800,
OxlOOO,
0,
0,
0,
0,
0,
};
struct StringInfo PathInfo
&PathBuffer[O], &UndoBuffer[O],
5, 511, 0, 0, 0, 0, 0, 0, NULL, 0, NULL
};
struct StringInfo FileInfo
&FileBuffer[O], &UndoBuffer[O],
0, 31, 0, 0, 0, 0, 0, 0, NULL, 0, NULL
};
SHORT StringPairs[]
(
0, 0, 248, 0
};
SHORT SelectPairs[)
{
0, 0, 61, 0, 61, 21, 0, 21, 0, 0,

};
0, 9, 1, 0, JAM1, 2, StringPairs, NULL,

};
struct Border Select Border
237
3.
INTUITION AND
AMIGA
-1, -1, 1, 0, JAM1, 5, SelectPairs, NULL.

};
1, 0, JAM2, -58, 1, NULL, (UBYTE *) "Path:" ,NULL,

};
struct IntuiText FileText
1, 0, JAM2, -58, 1, NULL, (UBYTE *) "File:" ,NULL,

};
1, 0, JAM2, 22, 5, NULL, (UBYTE *)"OK" ,NULL.

};
struct IntuiText CancelText
1, 0, JAM2, 2, 5, NULL, (UBYTE *)"Cancel" ,NULL,

};
1, 0, JAM2, 2, 5, NULL, (UBYTE *)"Return!" ,NULL.

};
struct Gadget FileGadget
{
NULL,
80, 0, 240, 8,
GADGHCOMP, RELVERIFY, BOOLGADGET I REQGADGET,
NULL, NULL, NULL,
NULL, NULL, 0, NULL
};
struct Gadget Slider {
&Files [0),
322, 30, 24, 120,
GADGHNONE I GADGIMAGE, RELVERIFY, PROP GADGET I
REQGADGET,
(APTR)&Knob, NULL, NULL,
NULL, (APTR)&SliderProp, 1, NULL
};
struct Gadget Path
&Slider,
80, 150, 250, 8,
GADGHCOMP, RELVERIFY, STRGADGET I REQGADGET.
(APTR)&StringBorder, NULL, &PathText.
NULL, (APTR)&Pathlnfo. 21. NULL
238
3.5
ABACUS
REQUESTERS
}; /*165 pa1*/
struct Gadget File =
{
&Path,
80, 165, 250, 8,
GADGHCOMP, RELVERIFY, STRGADGET I REQGADGET,
(APTR)&StringBorder, NULL, &FileText,
NULL, (APTR)&FileInfo, 22, NULL
}; /*185pal*/
struct Gadget OK
&File,
10, 40, 60, 20,
GADGHCOMP, RELVERIFY I ENDGADGET , BOOLGADGET
REQGADGET,
(APTR)&SelectBorder, NULL, &OKText,
NULL, NULL, 23, NULL
);
struct Gadget Cancel

&OK,
10, 65, 60, 20,
REQGADGET,
(APTR)&SelectBorder, NULL, &CancelText,
);
&Cancel,
10, 90, 60, 20,
(APTR)&SelectBorder, NULL, &ReturnText,
};
3.5.3.3
The requester functions

We would now like to show you how to develop a program that uses
the gadgets described above. This program makes it possible to use the
file select box in real-world applications.
We need our First_Window program from Section 3.1. Define the
NewWindow structure as printed, as well as the requester structure.
Before this, we had to insert the gadgets with your "appendix." Now the
program waits for a signal to activate a requester. We activate it. You
239
3. INTUITION AND C
AMIGA
can insert a condition for it later, like a menu item selection, for
example.
1* 3.5.3.3.a.requester additions *1
{
1* LeftEdge, ~opEdge *1
1* Width, Height
*1
1* DetailPen, B+ockPen *1
1* IDCMP Flags
*1
160, 0,
370, 200,
0, 1, ,
CLOSEWINDOW
GADGETUP I
REQCLEAR,
WINDOWDEPTH
1* Flags
WINDOWSIZING ,
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *)"Requester Test",
NULL,
1* Screen
1* BitMap
NULL,
100, 50,
640, 200,
WBENCHSCREEN,
1* Type
*1
*1
*1
*1
*1
*1
*1
*1
);
We use the simple Intuition function Request () to call the requester.

It only needs two arguments. Here is the general format for our
program:
Success - Request(Requester, Window);
DO
-240
AO
A1
This call blocks the transfer of input. We only get signals from the
requester. It is best if you write a subroutine for handling the requester
input. We form the program very distinctly by calling the routine after
initialization:
FileSelectBox_Request();
The routine itself tests the data transfer as before. We must pay special
attention to the REQCLEAR flag. It indicates that the user has ended
input and would like to return to normal program execution. The signal
is released from all gadgets that set the ENDGADGET flag when
activated. These are the boolean gadgets Cancel and Ok in our case:
240
3.5
ABACUS
REQUESTERS
1***************************************
*
*
* Function: Request Check
*
*
*
*
*
*
Author: Date:
Comments:
---------- ----------
Wgb
12/28/1987
*
*
*
***************************************1
FileSelectBox_Request()
{
BOOL Waiton = TRUE;

ULONG MessageC1ass;
struct Gadget
*GadgetPtr;
USHORT GadgetID;
while (Waiton)
{

GetMsg(FirstWindow->UserPort ~~ NULL)
(
Wait(lL FirstWindow->UserPort->mp_SigBit);
printf(nSIGNAL!\nn);
continue;
)
GadgetPtr ~ (struct Gadget *) message->IAddress;
GadgetID s GadgetPtr->GadgetID;
ReplyMsg(message);
if (MessageClass
==
REQCLEAR)
Waiton s FALSE;
printf(nEnd Requester\nn);
continue;
if (MessageClass
switch (GadgetID)
GADGETUP)
case SLIDER
printf(nSlider\nn);
break;
case PATH
printf(npath\n");
break;
case FILE
printf("File\n");
break;
case RETURN
printf("Return!\n R ) ;
241
3.
INTUITION AND
AMIGA
break;
if (GadgetID <= 16 && 1 <= GadgetID)

{
/* Table output */
printf ("Gadget-Nr. %d\n", GadgetID);
}
3.5.4
Requesters and accessing the user

Double menu requesters allow more service features. With this requester
type the user can call a requester whenever he wants to use its
functions. The call occurs by double-clicking the right mouse button
(also called the menu button).
A DMRequest is released every time the user double-clicks the right
mouse button. This requester can be activated for a window with
SetDMRequest () . When the requester appears, the program receives
a report over the IDCMP that the first requester is active. It can then
branch off to a special check loop. Here the handling is the same as for
any other requester. When the program doesn't have DMRequest
capability, everything returns to normal with ClearDMRequest () .
A program is especially user friendly when it offers a DMRequest.
For example, the file select box could be made from multiple
DMRequesters. Then the routine checks if the file should be saved,
loaded or deleted, instead of letting the user select a filename then
performing the function. Then each disk operation can be performed
with a double-click on the right mouse key.
The requester handling for the programmer remains unchanged. The user
can work with the requester in the method described above, and then read
the data. The major difference occurs in the call. While conventional
requesters execute from a function call in the program, this requester is
only defined and can be called anytime. That's why it is important to
check the rDCMP flags. They can test the first encounter with the
requester because the previous routines bypassed the
ClearDMRequest().
Here is the complete file requester program:
242
3.5 REQUESTERS
ABACUS
1***************************************
*
*
3.5
*
Program: Custom Requester
*
****************************************
* Author:
* Wgb
Date:
*
*
*
*
*
Comments
12-26-88
* Compile and link options
* ------------------------
* cc CustomRequest (-n for SourceDB)

* In CustomRequest -lc +Cd (-g)
*
*
****************************************/
struct
struct
struct
struct
IntuitionBase
Screen
Window
IntuiMessage
*IntuitionBase;
*FirstScreen;
*FirstWindow;
*message;
unsigned char PathBuffer[512] = "dfl:";

"";
{
OxCCCC,
OxCCCC,
Oxcccc,
OxCCCC,
};
Ox6666,
Ox6666,
Ox6666,
Ox6666,
struct Image Knob
Ox3333,
Ox3333,
Ox3333,
Ox3333,
Ox9999,
Ox9999,
Ox9999,
Ox9999
0, 0, 16, 16, 1, &KnobGraphic[O], 1, 0, NULL

};
{
FREEVERT,
Ox8000,
Ox8000,
Oxl000,
Ox0800,
0,
0,
0,
0,
0,
};
struct StringInfo PathInfo
&PathBuffer[O], &UndoBuffer[O],
243
3. INTUITION AND C
5, 511, 0, 0, 0, 0, 0, 0, NULL, 0, NULL

};
struct Stringlnfo Filelnfo =
&FileBuffer[O], &UndoBuffer[O],
0, 31, 0, 0, 0, 0, 0, 0, NULL, 0, NULL
};
SHORT StringPairs[]
{
0, 0, 248, 0
};
SHORT SelectPairs[]
{
0, 0, 61, 0, 61, 21, 0, 21, 0, 0,

};
0, 9, 1, 0, JAM1, 2, StringPairs, NULL.

};
struct Border Select Border
-1, -1, 1, 0, JAM1, 5, SelectPairs, NULL.

};
1, 0, JAM2, -58, 1, NULL, (UBYTE *)"Path:" ,NULL,

};
struct IntuiText FileText
1, 0, JAM2, -58, 1, NULL, (UBYTE *)"File:" ,NULL,

};
1, 0, JAM2, 22, 5, NULL, (UBYTE *)"OK" ,NULL,

};
struct IntuiText Cancel Text
1, 0, JAM2, 2, 5, NULL, (UBYTE *) "Cancel" ,NULL,

};
1, O. JAM2, 2. 5, NULL, (UBYTE *)"CHDIR!" ,NULL.

};
struct Gadget FileGadget
244
3.5 REQUESTERS
ABACUS
NULL,
80, 0, 240, 8,
NULL, NULL, NULL,
NULL, NULL, 0, NULL
};

{
&Files[O],
322, 30, 24, 120,
GADGHNONE I GADGlMAGE, RELVERIFY, PROP GADGET I
REQGADGET,
(APTR)&Knob, NULL, NULL,
NULL, (APTR)&SliderProp, 1, NULL
}
struct Gadget Path

{
&Slider,
80, 150, 250, 8,
(APTR)&StringBorder, NULL, &PathText,
NULL, (APTR)&Pathlnfo, 21, NULL
}; /*165 pal*/
struct Gadget File
{
&Path,
80, 165, 250, 8,
(APTR)&StringBorder, NULL, &FileText,
NULL, (APTR)&Filelnfo, 22, NULL
}; /*185pal*/
struct Gadget OK
&File,
10, 40, 60, 20,
REQGADGET,
(APTR)&SelectBorder, NULL, &OKText,
};
struct Gadget Cancel
&OK,
10, 65, 60, 20,
REQGADGET,
(APTR)&SelectEorder, NULL, &CancelText,
};
245
3. INTUITION AND C
AMIGA

{
&Cancel,
10, 90, 60, 20,
(APTR)&SelectBorder, NULL, &ReturnText,
}i
struct Requester FileSelectBox;
{
160, 0,
370, 200,
0, 1,
*/
/* Width, Height
*/
/* IDCMP Flags
*/
CLOSEWINDOW
GADGETUP I
REQCLEAR,
WINDOWDEPTH
/* Flags
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
/* First Gadget
NULL,
/* CheckMark
(UBYTE *) "Requester Test" ,
NULL,
/* Screen
NULL,
/* BitMap
100, 50,
640, 200,
WBENCHSCREEN,
/* Type
};
#define
#define
#define
#define
SLIDER 1
PATH 21
FILE
22
RETURN 25
main ()
{
ULONG MessageClass;
USHORT code;
Make_Files(Files);
InitRequester(&FileSelectBox);
FileSelectBox.LeftEdge
2;
FileSelectBox.TopEdge = 10;
FileSelectBox.Width
= 360;
246
*/
*/
*/
*/
*/
*/
*/
*/
3.5 REQUESTERS
ABACUS
FileSelectBox.Height
200;
FileSelectBox.ReqGadget z &Return;
Request(&FileSelectBox, FirstWindow);
FileSelectBox_Request();
EndRequest(&FileSelectBox, FirstWindow);
/*
Close_All();
exit (TRUE) ;
*/
FOREVER
{
if (message = (struct IntuitMessage *)

GetMsg(FirstWindow->UserPort)
{
ReplyMsg(message);
{

exit (TRUE) ;
break;
/***************************************
*
*
Author:
Date:
comment:
----------
*
*
*
*
*
*
---------10/16/1987
* Wgb
*
*
*
*
*
**************************************/
void
*OpenLibrary();
OpenLibrary ("intuition .library", OL)))
{

Close_All () ;
exit (FALSE) ;
}
247
3.
INTUITION AND
AMIGA
C FOR

{

Close_All();
'exit (FALSE) ;
}
/***************************************
*
*
*
*
*
Function: Close everything opened

===================================
Author:
* Wgb
*
*
Date:
10/16/1987
Comment:
*
*
*
*
just Intuition
and window
*
*
*
*
**************************************/
if (FirstWindow)
if (IntuitionBase)
/****************************************
* Function:
*
*
Gadget duplication
*****************************************/
Make_Files (Struktur)
struct Gadget Struktur[];
int i;
for (i=O; i<15; i++)
{
Struktur til
Struktur[i] .GadgetID
Struktur[i].NextGadget
FileGadget;
30 +i * 8;
i + 1;
=&Struktur [i+11;
Struktur[14].NextGadget = NULL;
/***************************************
*
*
*
248
Function: Request Check

===================================
*
*
*
3.5
ABACUS
REQUESTERS
*
*
Comments:
* Author: Date:
*
---------- ---------*
*
12/28/1987
* Wgb
*
*
*
*
*
***************************************/
FileSelectBox_Request()
{
BOOL Waiton = TRUE;

ULONG MessageClass;
struct Gadget
*GadgetPtr;
USHORT GadgetID;
while (Waiton)
{

GetMsg(FirstWindow->UserPort)) == NULL)
{
Wait(1L FirstWindow->UserPort->mp_SigBit);
printf("SIGNAL!\n");
continue;
)
GadgetID = GadgetPtr->GadgetID;
ReplyMsg(message);
if (MessageClass == REQCLEAR)
{
Waiton = FALSE;
printf(IIEnd Requester\n");
continue;
}
if (MessageClass
switch (GadgetID)
GADGETUP)
case SLIDER
printf ("Slider\n") ;
break;
case PATH
printf ("Path\n");
break;
case FILE
printf ("File\n");
break;
case RETURN
printf("Return!\n");
break;
249
3.
INTUITION AND
AMIGA
if (GadgetID <= 16 && 1 <= GadgetID)

{
/* Table output */
printf("Gadget-Nr. %d\n", GadgetID);
}
250
3.6
ABACUS
3.6
ALERTS
Alerts
Alerts are relatives of the requester family. Alerts receive top priority
from the system, but they're at the bottom of the user-friendliness list
Both requesters and alerts help get a decision from the user. Let's look
at the differences between alerts and requesters:
Requesters are highly developed windows that collect data: Alerts

are screen areas that display the last information possible before
the system crashes.
You can make requesters in any size, position, and form: Alerts
shove all screens aside, or append themselves to the current
display. In addition, Alerts can only be displayed on half of the
screen at full screen width.
Requesters have many input options and support all gadget

types: Alerts allow only two different inputs, accessed through
the mouse button.
Requesters only pause the task from which you call them: Alerts
halt the entire operating system.
You now know the differences between the two communication

channels. Alerts have very special uses in our programs.
Intuition manages Guru meditations as well as an AutoRequest in
an alert when not enough memory exists. Alerts need very little
memory for using the Amiga's graphic capabilities.
Note:
Alerts pack a certain amount of shpck value. Is the problem so serious

that you want an alert to appear? Or is it something recoverable, that
would be a lot friendlier if you used a requester instead? Alerts upset
most users, whether the message is serious or not. Therefore, always
think carefully about the proper programming of requesters and alerts, if
only out of courtesy to the user.
251
3.
INTUITION AND
3.6.1
AMIGA
Applications for alerts

The first important alert application is Insufficient memory.
This can cause a library or screen to fail to open. We gave
Open All () and Close All () a text using printf () during
the teSt phase of our functions. But the user may not always open the
CLI window and start the program using it If this "message channel" is
closed, we can report this with an alert
This can be used in many cases if, for example an interface card must be
present to operate our program, or perhaps executable memory cannot
handle the program size.
The above mentioned reasons could generate an alert if you don't think
that a simple AutoRequester will be enough. The alert stops the
execution of the entire system and waits for you to react. This alert
places all other processes in a lower priority.
3.6.2
Creating and configuring an alert

The DisplayAlert () function needs three arguments to construct
an alert: Alert type, alert height and alert text.
Alert type
We first assign the number which Intuition reads and handles as the
alert type. You differentiate between two recognizable types by their
most significant byte of the long variable. A RECOVERY_ALERT
(which returns to the program after the alert display) has a label of 00,
and a DEADEND_ALERT (displays the alert then resets the machine)
has a label of 80.
Alert height
We give the function the height of our alert in screen lines. The alert
then pushes all screen lines below it to fit on the top of the screen.
Alert text
The pointer to the alert text points to one string that states the desired
information. The alert text consists of many alert strings concluded by
continuation bytes. Continuation bytes are either end of line characters
or end of paragraph characters. At the beginning each alert string has a
position statement in pixels, followed by the text in normal string
format. Then we find a byte that reads the next string, if it is unequal to
null. The alert text closes when done loading the strings.
Unfortunately, Intuition supports little of the complications of alert
management. You won't find the structure in the include fIle.
252
3.6
ABACUS
ALERTS
First we establish a structure for the alert strings:

/* 3.6.2.a. alertmessage_struct */
struct AlertMessage
{
OxOO
Ox02
Ox03
Ox53
Ox54
J;
00
02
03
83
84
SHORT Left Edge;

BYTE TopEdge:
char AlertText[50];
BYTE Flag;
The biggest problem encountered when defining the structure is the

length of the text. We must accept a default value for this; you can
change it to anything. Be careful that the alert appears in top a z
sixty font, and that the text line is no longer than 60 characters.
We can now lump our alert strings together into one unit. For this we
establish an array in this structure:
1* 3.6. 2.b. firstalert *1
#define NOEND OxFF
jdefine END OxDD
struct AlertMessage UserAlert[]
(
50,
50,
50,
50,
24,
34,
44,
54,
n
n
n
n
This is the first alert I have written!!", NOEND,

------------------------------------", NOEND,
", NOEND,
<Left Button> CANCEL <Right Button> RETRY", END
);
We defined two lables to signal the continuation bytes. NOEND indicates that another text follows, and END indicates the end of the text.
After defining our first alert structure, we can call it with the
DisplayAlertO function. This function has the following format:
Resply
DO
- DisplayAlert(AlertNumber, String, Height);

-90
DO
AO
A1
We can insert the following known values into our text program:
Reply
DisplayAlert(RECOVERY_ALERT, &UserAlert, SOL);
Here is the entire source code for an alert:

1***************************************
*
* Program: First Alert
* ===================================
Comments:
* Author: Date:
*
*
*
Wgb
12/29/1987
First Alert
*
*
*
*
*
*
*
***************************************/
253
3.
INTUITION AND
AMIGA
struct AlertMessage
SHORT LeftEdge;
BYTE TopEdge;
char AlertText[50];
BYTE Flag;
};
#define NOEND OxFF

#define END
OxOO
struct AlertMessage UserAlert[]
50, 24, "
.. , NOEND,
50, 34, It
.. , NOEND,
50, 44, n
", NOEND,
50, 54, It
RETRY", END
};
This is the first alert I have written!!!
----------------------------------------<Left Button> CANCEL
<Right Button>
main ()
{
BOOL Reply;
Reply
= DisplayAlert(RECOVERY_ALERT,
&UserAlert, 80L);
if (Reply == TRUE)
printf ("Left mouse button was pressed!\n");
else
printf("Right mouse button was pressed!\n");
1***************************************
* Function: Open all
*
* =================================== *
*
* Author: Date:
Comments:
*
...
for Intuition
Wgb
10/16/1987
...*
*
***************************************1
254
'3.6 ALERTS
ABACUS
Open_All ()
{
void
*OpenLibrary () ;

Open Library ("intuition library" , OL)
{

Close_All () ;
exit (FALSE) ;
}
1***************************************
*
*
* =================================== *
*
*
* Author: Date:
Comments:
*
* -----*
10/16/1987 for Intuition
* Wgb
*
*
*
*
* Function: Cloae all
***************************************/
Close_All ()
{
if (IntuitionBase)
How it works
The program first prepares all of the necessary structures for the
function call. Both include files are included because without their
defmitions, the alert is not possible.
If you define the alert string with all of the coordinates and constants,
the main program can call the Open All () function to ensure access
Intuition. Then DisplayAlert() starts and processes the return
value. Everything can be closed again with Close_All ( ) .
to
Note:
To examine the workings of a DEADEND_ALERT, you should change

the alert type. Remember that in a DEAD END ALERT, control passes
on to the reset routine.
-
3.6.3
Alerts for program projects

An alert built into a program means work for the programmer that
could have been spared. The aim in programming is to keep the process
as efficient as possible.
255
3.
INTUITION AND
It's important to install additional alerts at the end of all work in the
program. These alerts usually replace error messages given in the eLI
window during the development phases.
We recommend the following system for appending alerts: Your
program is ready and you've already determined all points in the
program at which an alert can replace a simple error message. Now you
can develop the text to be displayed in the proper case.
For this we recommend a general alert design, and maybe a multi-alert.
Multi-alerts insert just a specification on the margin. The alert texts can
be stored in a file accessed from the structure definition. This file is
included last:
#include <exec/devices.h>
#include <graphics/gfxbase.h>
#includes "Disk:Directory/Alert"
In the Alert file we find, for example, the following error messages
for a word processor:
/*******************************************1
1*
*/
1* 3.6.3.a. alert include
*1
1*
*1
1* include : Alert Structure wI Definition */
/*
=========~%==~====~============~=~=~~~~
/*
/* Author:
Date:
comments:
*1
*/
*/
/*
---------- ---------*1
/* Wgb
29.12.1987 for word processor */
1*
*1
1*******************************************1
struct AlertMessage
{
SHORT LeftEdge;
BYTE TopEdge;
char AlertText[50};
BYTE Flag;
};
#define NOEND OxFF

#define END
OxOO
256
--
--
------------------------------------------------------------
3.6
ABACUS
ALERTS
struct AlertMessage OutMem[] =

{
50,
50,
50,
50,
50,
24, "
34, "
44, n
54, II
64, II
There is not enough memory

Please free up some memory!
Start the program"again!
Mouse key to continue
};
" , NOEND,
, NOEND,
II
II
NOEND,
" , NOEND,
" , END
#define OutMemSize 80L

struct AlertMessage NoLibrary[]
50,
50,
50,
50,
50,
50,
24, "
34, II
44, II
54, "
64, II
74, II
Graphics.library cannot be opened

The program is stopped!
Make sure that this library is
on the Workbench!
II
, NOEND,
,
",
,
,
"
NOEND,
NOEND,
NOEND,
NOEND,
END
",
,
",
",
",
NOEND,
NOEND,
NOEND,
NOEND,
END
II
II
II
};
#define NoLibraySize 90L

struct AlertMessage Ende[] =
{
50,
50,
50,
50,
50,
Important functions cannot be

reached!
Crash! Crash!
24, "
34, n
44, "
54, II
64, II
};
II
#define EndeSize 80L
For the DisplayAlert () call we included the height needed for each
alert in a *define. The program text contains the following lines:
if (Window
==
NULL)
printf ("Unfortunately no window can be opened! \nn);

Close_All();
exit (FALSE) ;
}
Replace it with this:

if (Window
== NULL)
DisplayAlert(RECOVERY_ALERT, &NoWindow, NoWindowSize);

Close_All () ;
exit (FALSE) ;
}
257
3.
INTUITION AND
3.6.4
AMIGA
Understanding the Guru Meditation

Even if you've never programmed any alerts, you probably have had an
acquaintance with Guru Meditations. Generally you assume the message
on the top line is true because most folks can't interpret the funny
numbers underneath. This is a misunderstanding: Gurus report to the
user why the operating system must be reset.
Gurus and why This is especially important if you do your own programming. If you
know why your own program runs into errors in the test phase, you can
usually find the source of the errors. The only problem with the Guru
Meditation is that it always seems to display the same text, and a
number which doesn't give a clue as to the problem. This is something
we can help change. The following tables are used to decipher any Guru
Meditation.
Guru numbers
The Guru number consists of multiple segments. We can break them

down into a general format:
Guru Meditation : TT55FFGGG.AAAAAAAA
The letters have the following meanings:

Letters in code
TT
55
FF
GGGG
AAAAAAAA
Description
AlERT_TYPE
System class
Error class
Exact description
Address of error solving task
Just a word about the Guru codes listed below: We write out the entire
term along with the number code. If an alternate exists, we include the
word OR.
There are only two possibilities for TT. You read about these values
under the label ALERT TYPE. It can be either a DEADEND ALERT
(80) or a RECOVERY_ALERT (00) (only the latter type returns-data).
10000000 I RECOVERY
ALERT
Next we will discuss the system class. Here we find the

SUBSYSTEM_CODE label, where the error occurs.
SUBSYSlfEM_CODE
258
100000000 I 68000
3.6 ALERTS
ABACUS
LIBRARIES
01000000
02000000
03000000
04000000
05000000
060000OO
07000000
08000000
09000000
OAOOOOOO
DEVICES
1()()()()()OO
11000000
12000000
13000000
14000000
15000000
exec.library
.graj)hics.library
l~rs.lilLrary
intuition.library
math.library
clist.library
dos.library
ram.library
icon.lib@!Y
expansion.library
audio.device
console.device
gameport.device
k~board.device
trnckdisk.device
timer.device
RESOURCES
22000000 Misc
MIse
Be sure that the error message is at the beginning of the subsystem list
Here the processor automatically gets the words. When this occurs, it
acts as a processor trap. The rest of the locations of the Guru
Meditation have other meanings:
TRAP CODES
00000OO2 Data or address bus error with measures

00000003 Addressing error (uneven addres~
00000OO4 Illegal instruction
00000005 Division Qy zero
00000OO6 eRR Instruction
00000007 TRAPV Instruction
00000008 Privilege violation
00000OO9 Single-step mode
OOOOOOOA Line A emulator (opoode 1010)
OOOO)OOB Line F emulator~ 11111
If an error occurred after the subsystem list, then the ERROR_CLASS
code brings us closer to the cause of the error:
259
3.
INTUITION AND
ERROR CLASS
AMIGA
00010000
00020000
00030000
00040000
00050000
000600OO
00070000
Not enough memory_

Lilm!Jy cannot be constructed
Library cannot be opened
Device cannot be ~ned
No reaction from the hardware
I/O error
I/O not present
The last four numbers of the error code give us still more detailed
information. These are specified in the subsystems:
exec.l.ibrary
01000000
81000002
81000003
81000004
81000005
81000006
81()()()()()7
81000008
81000009
8100000A
checksum error excepting CPU

checksum error at the starting address
checksum of a library
not enough memory for library
error is memory list entry
not enollgh memQIY for interrupt
pointer error
error in Sema12hore
memQ!Y area was freed ~ for the second time
pointer error with exception
graphics.library
82010001
82010002
82010003
82010004
82010005
82010006
82010007
82010008
82010009
8201000A
8201000B
82010030
82011234
not enough memory for copper list

not enough memory for copper instruction list
the c~ list is full
division error in the copper list
not enough memory for the copper list head
not enough memory for "long frame"
not enough memory for "short frame"
not enough memory for the fill routine
not enough memQ!Y for the text routine
not enough memory for the blitter bit-ma~
incorrect memory area
error during the establishment of a view port
GfxNoLCM
(no intermediate memory availabld
l.ayers.l.ibrary
260
103000001 I not enough memory for Layers
3.6 ALERTS
ABACUS
intuition.library
84000000
04000001
84010002
04010003
04010004
84010005
84000006
84010007
84010008
84000009
8401000A
840 1OOOB
8400000c
8400000D
8400000E
8400000F
gadg_et ~ is unknown
type error with AN .,gadget
not enou"Eh memQ!Y for buildinga~rt
not enough memory for a menu
not enough memory for a submenu
not enou.zh memQ!Y for the menu list
incorrect position of the menu list
not enough memQ!Y forQpenScreel!Q
not enough memory for building a RastPort
unknown or erroneous SCREEN TYPE
not enough memory for p;adp;et
not enough memory for window
bad r~ister status when intuition.li~ ~n
incorrect message over the IDCMP
not enough memory for the message stack
not enoygh memQ!Y for the console. device
dos.library
07000001
07000002
07000003
07000004
07000005
07000006
07000007
07000008
07000009
0700000A
0700000B
0700000c
not enoygh memory at start up_

the task was not ended
QPkt error is encountered
dataj!acket was not e~ed
free pointer is inaccessible
erroneous data of a disk block
the bit-ml:lJlis destrQYed
key was already freed UP
the check sum is full of errors
disk error
key is outside of the allowable area
incorrect over write
ram.library
108000001 1 erroneous entry in the management list
expansion~ibrary
1OAOOOOO 1 1 error in the expansion
hardware
trackdisk.device
timer.device
15000001 error with access of the device

15000002 error with time coordination:
network fluctuation
disk.resource
bootstrap
130000001 1 error when analyzing the boot data
261
3. INTUITION AND C
3.7
Checking the IDCMP

A program really "comes to life" when it interacts with its user. We're
going to write a small demonstration program that asks for user input
The big question: How to get information from the user?
The answer: The Amiga has many input and output devices for data.
These include the audio.device, console.device.
timer.device, trackdisk.device, input.device, keyboard.device, gameport.device, narrator.device,
serial.device, parallel.device, printer.device, and
the clipboard.device. Some of these devices only provide input,
others only provide output, while some can perform both tasks. For
example, we can only transmit data over the audio. device.
narrator.device and printer.device. We can only receive
data from the timer.device, keyboard.device and
gameport . device. Data can be transferred in both directions with
the remaining devices. That's good communication.
All of these input and output devices have a disadvantage. When we
want to send and receive multiple data of different types, we need more
than one device to perform the task. Many devices have certain advantages, while others have strong disadvantages. The end result is that the
program opens a number of devices, but the programmer may lose sight
of what's open and what isn't
For this reason the Amiga's developers came up with the InCMP
(Intuition Direct Communications Message Port). This is a
combination of i npu t .dev ice, gamepo rt .dev ice.
timer.device, trackdisk.device and console.device.
with interfacing through Intuition.
The IDCMP is for data processing, filtering and preparation. Through
the assigned window you can specify which data you want to receive
and which data it should convert into the proper format. Or maybe you
would prefer to leave the data unchanged? This is also possible with the
IDCMP. If you would like to have more flexibility, then you should
avoid the graphics.library. Intuition is usually enough to handle
this. The IDCMP receives all input, and the graphic structures make
output possible.
Let's take a closer look at the IDCMP and its versatile input capabilities.
262
ABACUS
3.7.1
3.7 CHECKING THE IDCMP
Configuration and receiving

The IDCMP is not constantly on standby, ready for the programmer's
call. Intuition must know that a possibility exists for data reception.
First we need a window. Without a window, the Amiga can't receive
data through Intuition. Remember the Newwindow structure. The
IDCMPFlags entry lists the interface to IDCMP. When a flag indicates that we are interested in the information, Intuition establishes a
port for this window. We must first know which flags can be set. These
are divided into six groups.
The ftrst group lists the most important flags. They occupy the message port window:
SIZEVERIFY
When this flag is set the program receives a message

if the user changes the size of the window. This can
be important if the program is drawing and this work
must stop before the size can be changed. Thus the
window size cannot be changed before verification.
NEWSIZE

of a size change in the window. The value can be read
from the window structure.
REFRESHWINDOW

of whether the window contents must be redrawn.
Used on SIMPLE REFRESH and SMART REFRESH
window types. SMART_REFRESH wiiidows only
need the refresh flag when they are supplied with a
sizing gadget.
ACTIVEWINDOW

when the window is active.
INACTIVEWINDOW

when another window is activate.
The second group concentrates on the problem of gadget manipulation.
We learned about these flags earlier with our gadget check:
GADGETDOWN

when a gadget of type GADGIMMEDIATE is clicked
on.
263
3.
INTUITION AND
AMIGA
GADGETUP

when a gadget of type RELVERIFY is clicked on.
CLOSEWINDOW

when the close gadget is activated by the user.
Intuition does not close the window automatically.
The third group consists of expanded flags to control the gadgets.

REQSET
When this flag is set the program gets a message

when the user clicks on the fIrst requester in the
window.
REQCLEAR

when the user clicks on the last requester in the
window.
REQVERIFY

when Intuition tries to open the fIrst requester. Intuition then waits for a response to the message. You
can finish all output before the requester appears. If
further requesters open in the window. Intuition does
not give a message. During requester handling the
output remains visible.
The fourth group handles menu reading. Menu programming will be

described in the next section.
MENUPICK

when the user presses the right mouse (menu) button.
MENUVERIFY

when someone tries to activate the menus. As long as
no response occurs (e.g., with Reply ( ) ) the menus
cannot be presented.
The flags of the fIfth group check mouse status:

MOUSEBUTTONS

whether a mouse button is pressed or not. Remember
that no messages are sent when the left button is
pressed while the pointer is over a gadget. or when
the right button is used to manipulate the menus.
264
MOUSEMOVE

when the user moves the mouse. The REPORTMOUSE flag must also be set (or the FOLLOWMOUSE
flag. when a gadget is involved).
DELTAMOVE

when the user moves the mouse. Unlike the above
ABACUS
flag, the position returned is the difference from the

last position.
In the sixth group we fmd the unassigned flags.
INTUITICKS

when a tenth of a second elapses (in conjunction with
the timer.device). You only get further TICK
reports if the last one got a response.
NEWPREFS

when Preferences values change. This is important,
for example, if the program abandons the color
settings or the printer setting. Every window in
which this flag is set receives a message. A program
that changes the Preferences data checks to see if it
wants to inform the others.
DISKINSERTED
When this flag is set. the program gets a message

when a disk is inserted into the drive. Every window
in which this flag is set gets a message (in conjunction with the trackdisk.device).
DISKREMOVED When this flag is set the program gets a message
when a disk is removed from a drive. Every window

in which this flag is set gets this message.
RAWKEY

when a key on the keyboard is pressed. The keyboard
code is received in "raw" form, which means that code
doesn't get converted according to the keyboard table
(in conjunction with the input.device).
VANILLAKEY

when a key is pressed. The keyboard code is processed
and converted according to the keyboard table (in conjunction with the console.device).
265
3.
INTUITION AND
AMIGA
Here are all of the flags and their hex values:

Flag name
SIZEVERIFY
NEWSIZE
REFRESHWINDOW
AC'TIVEWINDOW
INACTIVEWINDOW
Hex value
OxOOOOOOOlL
OxOOOOOOO2L
OxOOOOOOO4L
OxOOO40000L
OxOOO80000L
Remarks
Window flags
GADGETDOWN
GADGETUP
CLOSEWINDOW
Ox00000020L
Ox00000040L
OxOOOOO200L
Gadget flags
REQSET
REQCLEAR
REQVERIFY
OxOOOOOO80L
OxOOOOlOOOL
OxOOOOO800L
Requester flags
MENUPICK
MENUVERIFY
OxOOOOOlOOL
OxOOOO2000L
Menu flags
MOUSEBUTTONS
MOUSEMOVE
DELTAMOVE
OxOOOOOOO8L
OxOOOOOOlOL
OxOO 1OOOOOL
Mouse flags
INTUITICKS
OxOO400000L
timer.device
NEWPREFS
OxOOOO4000L
Preferences change
DISKINSERTED
DISKREMOVED
OxOOOO8000L
OxOOOlOOOOL
trackdisk.device
RAWKEY
VANILLAKEY
OxOOOOO400L
OxOO200000L
input.device
console.device
WBENCHME SSAGE
LONELYMESSAGE
OxOOO20000L
Ox80000000L
Workbench message
No Intuition/IDCMP message
There are a number of ways to tell the system which region you decided
should receive the messages and information. The fIrst and the simplest:
Insert the corresponding flags in the Newwindow structure. Intuition
establishes the IDCMP; you need only worry about the check. The
second way: Use the ModifyIDCMP () function to display a second
window for information display, without establishing a port. Simply
give the window pointer and all of the flags you want set. This either
establishes a new port or closes an existing one. The command has the
following format:
ModifyIDCMP(Window, IDCMPF1ags);
-150
266
AO
DO
ABACUS
The third possibility converts the previous flags to the new flags'
values.
The functions that influence the InCMP, such as REPORTMOUSE () ,
are also important for the second method. Here the same thing happens,
but in a single flag.
3.7.1.1
The IntuiMessage structure

Now that you have some background on the situations for supplying
the InCMP messages, it's time to look at exactly what gets reported in
transmission.
We find in the window structure a pointer to the Intuition structure named MessageKey at byte 90. We also find another pointer at
byte 82 which points to the MessagePorts named UserPort and
WindowPort. You can process the input in conjunction with these
three message ports.
First let's look at the IntuiMessage structure:
1* 3.7.l.l.a.intuimessage_struct *1
struct IntuiMessage
(
Oxoo
Ox14
Ox18
OxlA
OxIC
Ox20
Ox22
Ox24
Ox28
Ox2C
Ox30
Ox34
00 struct Message ExecMessage;

OxOO 00 struct Node mn_Node;
OxOO 00 struct Node *In_Succ;
Ox04 04 struct Node *In_Pred;
Ox08 08 UBYTE In_Type;
Ox09 09 BYTE In_Pri;
OxOA 10 char *In_Name;
OxOE 14
OxOE 14 struct MsgPort *mn_ReplyPort;
Ox12 18 UWORD mn_Length;
Ox14 20
20 ULONG Class;
24 USHORT Code;
26 USHORT Qualifier;
28 APTR IAddress;
32 SHORT MouseX;
34 SHORT MouseY;
36 ULONG Seconds;
40 ULONG Micros;
44 struct Window *IDCMPWindow;
48 struct IntuiMessage *SpecialLink;
52
};
267
3.
INTUITION AND
Structure
description
AMIGA
The structure begins with a message structure. This is needed for the
data transfer of messages from Exec. This is interesting and that additional information can contain important data for other purposes. The
other information depends on the flag.
Variables
Exec needs ExecMessage so it can transfer the message into the

system.
Clas s is a ulong variable that marks the type of the data sent These
bits match the window IDCMP variables, and may be necessary to
identify the message.
Code contains data which depends on the released flag. The value
always has different meanings.
Qualifier is a copy of the ie_Qualifier variables transferred
from input. device.
MouseX and MouseY contain the mouse coordinates of the window up
to
the time of the message transmission.
Seconds, Micros duplicate the Amiga's system clock of the Amiga

up to the time of the message transmission.
IAddress is a pointer to the structure where you find the report. That
can, for example, be a gadget structure from which the impulse was
released.
IDCMpwindow is a pointer to the window from which the message
stemmed.
SpecialLink is a variable that is used from the system.
As you see, the IntuiMessage structure holds all of the data that
can be important for further processing. Here we find the basic data of
Intuition, whose evaluation is dependent on the impulse received.
3.7.1.2
MsgPort structures
'The MsgPort structure named UserPort signals the receipt of the
message. UserPort doesn't lie within the IntuiMessage structure. By setting a bit in the variable mp SigBi t we can wait for a
message. The structure contains the following elements:
268
ABACUS
/* 3.7.1.2.a. msgport struct */

struct MsgPort
{
OxOO
00
OxOO
Ox04
Ox08
Ox09
OxOA
OxOE
OxOE 14
OxOF 15
Ox10 16
Ox14 20
Ox22 34
struct Node mp_Node;

00 struct Node *In_Succ;
04 struct Node *In_Pred;
08 UBYTE In_Type;
09 BYTE In_Pri;
10 char *In_Name;
14
UBYTE mp_FIags;
UBYTE mp_SigBit;
struct Task *mp_SigTask;
struct List mp_MsgList;
};
Structure
description
The structure begins with a node structure which helps in recognizing
the port.
mp_Flags labels the type of message:
PA SIGNAL
PA SOFTINT
PA IGNORE
is set when a message is expected

releases a software interrupt
labels message that should be ignored
mp SigBi t is the variable that relates to us. The set bit indicates the
arrival of a task signal for us.

mp_SigTask is a pointer to the task that sent the signal.
mp_ MsgLi st heads the list of all of the messages sent to this port
You may be confused after reading all this. considering all we want to
do is check the IDCMP. The IDCMP'S complexity stems from its close
relationship with the Amiga's internal signal system. All we need to
understand for now is the mp SigBi t of the IDCMP. We wait for a
message through the signal forthis bit. What can be found in this bit is
immaterial.
3.7.1.3
Waiting for a message

Intuition establishes a message port and an IntuiMessage structure
for us. Our program should wait until it receives a message that it can
read and evaluate. The best route would be to use the Exec function
GetMsg () to see if UserPort contains a message:
269
3. INTUITION AND C
AMIGA
Message
= GetMsg(Window->UserPort);
We receive one of two results: Either GetMsg () returns a pointer to

the message structure (in this case, an IntuiMessage structure), or a
null pointer (i.e., no message present).
You can repeat this check using a loop until a message is present. That
loop looks like this:
1*3.7.1.3.a.msgwait
FOREVER
*1

{
1* Evaluation *1
This message check is very compatible with our multitasking computer. Although it has nothing to do yet, our program constantly
strives to read a message.
It would be better if the task were in "quiet" status until a message
actually is sent. For this there is an Exec function named Wait ().
wai t () lets you inform the system to not process the task until a
message for it appears. The new version of the FOREVER loop looks
like this:
1* 3.7.1.3.b.msgquiet */
FOREVER
{

{
Wait(lL
continue;
/* Evaluation */
First the loop inspects the UserPort for a message. If no message

exists, the system waits until one is received. Without a message,
program execution continues at the beginning of the FOREVER loop. If
a message does exist, or if the system returns to the Wa i t status, it
reads the message. When in Wait status, the loop begins where the
received message is read in (this is true in every case).
270
ABACUS
3.7.2
Reading the message

The message reading depends completely on the type of message
received. We need to read gadgets, menus, requesters, the keyboard and
the mouse. This strict subdivision increases the size and complexity of
the programming, but also increases the fmal program's flexibility. The
C language supports this flexibility, that makes programming the
checks to a minimum, providing you understand the programming
involved.
When receipt of the message is confIrmed, the program goes to the preevaluation process. We must fIrst establish the category into which the
message falls (Le., which flag value the Cl ass variable of the
IntuiMessage structure receives). For this examination we get the
standard values from the structure:
MessageClass z Message->Class;
MessageCode = MessageO>Code;
We can then determine from MessageClass which IDCMP flag has

released the message. In this sense we must continue the inspection. Do
not forget to reply to the message so the message system can continue
operation. No further messages can be received without a reply:
ReplyMsg(Message);
We now create checks for individual flag groups, with which we can
manage the data flow.
3.7.2.1
Gadgets
Once you set the pointer to the IntuiMessage structure, we can
begin with the reading. Let's look at how you program this reading to
accept gadget data. You have already read about the development of
gadget checking.
We fIrst test if it is handled as the gadget IDCMP flag:
if (MessageClass & (GADGETDOWN I GADGETUP
{
1* Gadget check *1
Then we provide the pointer to the Gadget structure:
271
3. INTUITION AND C
struct Gadget *GadgetPrt;

GadgetPtr = (struct Gadget *) Message->IAddress;
For the difference between multiple gadgets we need the GadgetId:

USHORT GadgetID;
GadgetID = GadgetPrt->GadgetID;
Note:
We recommend successive numbers replaced by a fdefine at the

beginning of the program, to allow easier recognition of gadgets.
Mter we determine the Gadget ID we can release the individual
actions. The source code makes more sense if you place longer reactions
in a function:
1* 3.7.2.1.a.gadcheck *1
switch (GadgetID)
{
case TEXTGADGET
Text_Evaluate();
break;
case SLIDER
New_Value 0 ;
break;
case
All further information can come through the pointer to the Gadget
structure. The mouse coordinates can be found in the IntuiMessage
structure-the gadgets are no longer stored here. The mouse coordinate
check is very simple:
MouseX
MouseY
3.7.2.2
=
=
(SHORT) Message->MouseX;
(SHORT) Message->MouseY;
Window messages
Window flag reading controls more than gadgets. Window flags also
return additional status information.
Two possibilities for this check: the first is direct insertion into the
system. We first select all window flags and read them further:
1* 3.7.2.2.a.windowmsg *1
if (MessageClass & (NEWSIZE I REFRESHWINDOW I ACTIVATEWINDOW
INACTIVATEWINDOW
WindowPtr = (struct Window *) Message->IAddress;
{
272
ABACUS
case NEWSIZE
: TWidth = WindowPrt->Width:
THeight = WindowPrt->Height;
break:
case REFRESHWINDOW
break;
case ACTlVATEWINDOW
break;
case INACTIVATEWINDOW: ActivateWindow(Windowptrl:
break;
Bear in mind that this method cannot check the SIZEVERIFY because
the exact inspection was answered with ReplyMsg () through the
message. Here a simple flag check before each reply will be enough.
You can simply jump to the ReplyMsg () function.
The last line is quite clever. These lines constantly keep the window in
our program active. As soon as the user clicks on another window, the
program reactivates our program window.
The Workbench also uses this method. When you select the Rename
item from the Workbench menu, a window containing a string gadget
opens. The window is the same size as the string gadget, and has no
system graphics. Click anywhere on the Workbench and the Rename
window remains active.
3.7.2.3
Requesters and execution

Requesters can normally only be exited from the program, and do not
usually obstruct program execution. It gives the user some comfort that
a certain function can be called at any time through DMRequest.
However, that leaves interruption time undetermined. This is why
requester flags exist.
When you place REQSET and REQCLEAR in the rDCMP of the
window, you always receive a message, if the first requester is activated
(our DMRequest) and when the last requester disappears (also our
DMRequest).
Because the End flag receives priority over the requester routine, we
only need to worry about REQSET. One program line circumvents this
problem:
if (MessageClass & REQSET) DMRequest_Evaluation();
273
3. INTUITION AND C
AMIGA
You can test to see if the process released or if the important work must
still be executed using REQVERIFY. REQVERIFY works just like
SIZEVERIFY. When the flag is set, the program frrst receives this
message before Intuition opens the requester. This can frrst be processed
after the message is answered with ReplyMsg ().
3.7.2.4
Menu flag analysis

Menus are the most important input option offered to the user by Intuition (next to gadgets). When the user presses and releases the right
mouse button, the program gets a message over the IDCMP if the
MENUPICK flag is set. We can then search in the Code array of the
IntuiMessage structure for the menu item number. The menu
points are not managed by numbers: An internal system makes the
identification process as easy as possible. The system allows us to
program submenus only to a certain extent for this reason.
Now we should look at the example menu's coding. The two bytes can
be "divided" as follows:
UUUOO MMMIMMM NNNNN
Here we see two bytes within three bit groups. The first group
(NNNNN) describes the menu number. We must separate this from the
others to get the number. The work executes the definition
MENUNUM (Code), found in the <intuition/intuition .h>
incl ude file.
The second group (MMMMMM) represents the number of the menu item.
It allows 31 menu titles and 63 menu items. To determine the number
of menu items available, use the definition ITEMNUM (Code).
The third group (uuuuu) contains the number of the submenu item if
one is present. The system allows 31 possible values here. These
values can be reached through the definition SUBITEM(Code).
Definitions makes the menu checking very easy. In our check loop for
all of the IDCMP messages, we must frrst determine whether a menu
message exists. Then we can branch off to a function that can make a
further analysis:
if (MessageClass & MENUPICK) Menu_Evaluation(Code);
The function calculates and utilizes the values one after another:
274
ABACUS
3.7 CHECKING TilE IDCMP
/* 3.7.2.4.a.menufigure */
Menu_Analysis (Menunumber)
USHORT Menunumber;
(
OSHORT Menu, MenuItem, Sub Item;

Menu
= MENUNUM(Menunumber);
MenuItem = ITEMNUM(Menunumber);
SubItem = SUBITEM(Menunumber);
switch (Menu)
(
case FILEMENU
FileAnalysis(MenuItem, SubItem);
break;
case WORKMENU
WorkAnalysis(MenuItem);
break;
FileAnalysis(MenuItem, SubItem)
USHORT MenuItem, SubItem;
(
$witch(Menultem)
-{
case LOAD
Load() ;
break;
case SAVE
Save ();
break;
case DELETE
switch (Subltem)
(
case FILE : DelFile () ;

break;
case TEXT
DelText ();
break;
break;
WorkAnalysis(MenuItem)
USHORT MenuItem;
(
switch (Menultem)
(
case MARK
Mark ();
break;
case EDIT
Edit ();
break;
case SEARCH
Search();
break;
275
3.
INTUITION AND
Instead of numbers we used expressions for the menus, menu items and
submenu items. This makes the program text easier to understand. Here
are the defines for the above routines:
1* 3.7.2.4.b.menu names *1
1* Menu names */
#define FILEMENO
#define WORKMENU
/* 1. Menu
FILEMENU */
#define LOAD
#define SAVE
#define DELETE
1* Sub menu DELETE */

#define FILE
#define TEXT
1* 2. Menu WORKMENO */
#define MARK
#define EDIT
#define SEARCH
3.7.2.5
Mouse reading
The mouse serves a number of purposes, including gadget selection and
menu item selection. The programmer can also monitor the mouse's
activities. The IDCMP directly reads the mouse. By setting flags we
indicate whether we would like to know that the user pressed or released
mouse buttons, or even that the user moved the mouse. Everything can
be received through the IDCMP.
In a drawing program it can be important to know when the mouse
button was pressed (e.g., so a point can be placed at that location).
There are two mouse buttons, and the system can check for either
button pressed or released. Accordingly there are four possible results:
SELECTDOWN,SELECTUP
Left mouse button (select) flags.
MENUDOWN,MENUUP
Right mouse button (menu) flags.
Be careful that the left mouse button only sends signals if it is not
needed for gadget work (gadget operation has a high priority). The same
goes for the right mouse button: The right mouse button registers only
if the flag RMBTRAP is set.
276
ABACUS
The check looks very simple in the program. We assume that the program needs both mouse buttons and manages no windows. We set the
IDCMP flags MOUSEBUTTONS and RMBTRAP in the window. When
we get a message, the Code array of the IntuiMessage structure
indicates the type of mouse button:
MousX = (SHORT) Message->MouseX;
MousY = (SHORT) Message->MouseYi
if (MessageClass & MOUSEBUTTONS) MausAbfrage(Code, MousX,
MousY) ;
These lines branch to a subroutine where further readings take place:

MousX
MousY
=
=
(SHORT) Message->MouseX;
(SHORT) Message->MouseY;
i f (MessageClass
MousY) ;
&
MOUSEBUTTONS) MouseRead(Code, MousX,
This line branches to a subroutine where further readings take place:

1* 3.7.2.S.a.readmouse *1
MouseRead(Flag, MousX, MousY)
USHORT Flag;
SHORT MousX, MousY;
{
switch (Flag)
{
case SELECTDOWN
Begin_Line(MousX, MousY);
break;
case SELECTUP
End_Line (MousX, MousY);

break;
case MENU DOWN
Begin_Erase(MousX, MousY);
break;
case MENUUP
End_Erase (MousX, MousY);

break;
The example function receives the mouse status along with the coordinates as the assignment value. Depending on which mouse status has
priority, either the lines function of the program executes or ends or the
delete function executes. We then assign the frrst or second comer pixel.
The next important point of the mouse check is the position. In the
first example we used this in part because the mouse position is
supplied to every IntuiMessage relative to the window. When we
want to program a penn anent coordinate display, we cannot wait for the
user to press a key to see the coordinates. He must always have the
277
3.
INTUITION AND
AMIGA
current condition. Intuition offers a message type for this. Simply set
the IDCMP flag REPORTMOUSE, and you receive each movemenL
Note:
We would like to warn you at this time! Each small movement of the
mouse, even one pixel, releases a message. You should also consider
that many steps are needed for the exact positioning.
We recommend a check for mouse movement:
MousX = (SHORT) Message->MouseX;
MousY = (SHORT) Message->MouseY;
The subroutine can then handle analysis. With the above mentioned
methods you get the absolute value for every mouse movement, relative
to the IDCMP window. It is easier in many applications if you know
the difference from the last position. For this change we must set the
DELTAMOVE flag.
3.7.2.6
Recognizing keyboard input

Keyboard input can only be recognized when either the VANILLAKEY
or RAWKEY flag is set. Either flag sets the keyboard mode we want
"see." Only one mode per window is allowed.
RAWKEY is different from VANILLAKEY in that we receive untranslated codes from RAWKEY (i.e., each key sends a number as the message).
When VANILLAKEY is active, the keyboard number is translated

according to the keyboard table. This gives us the keyboard table codes.
Remember this when programming, since the computer's keyboard
configuration does change from country to country. In America we use
an American keyboard arrangement, in Germany a German
configuration, in France a French configuration, etc.
278
ABACUS
3.7 CUECKING TilE IDCMP
Amiga 1000 keyboard

,
..
,
.-
3D
3.
3F
20
2.
2F
10
1F
OF
3C
.3
Amiga 500/2000 keyboard
J!.L
I
1B
3D
3.
20
2'
, HollIe
,
3A
01
PtUp
. .
3F
.A
, En. ,
10
OF
Figure 3.12
Raw Keyboard
codes
1.
Ina
oc
2F
J PtDn
1F
D., ,:43
3C
Note that the numeric keypad has its own numbering system. When
you want to program a check, you must search the Class array for the
RAWKEY flag, and in the Code array for the key number. The
Qualifier array returns a status message concerning special keys
like <Shift>, <Alt> and <etr}>.
279
3.
INTUITION AND
4.MIGA C FOR ADVANCED PROGRAMMERS
You can identify the key pressed with the following table:
QUALIFIER TYPES
IEQUALIFIER_LSHIFT
IEQUALIFIER_RSHIFT
IEQUALIFIER_CAPSLOCK
Hex value
OxOOO1L
OxOOO2L
OxOOO4L
IEQUALIFIER_CONTROL
OxOOO8L
IEQUALIFIER_LALT
IEQUALIFIER_RALT
OxOO1OL
OxOO20L
IEQUALIFIER_LCOMMAND
IEQUALIFIER_RCOMMAND
OxOO40L
OxOO80L
IEQUALIFIER_NUMERICPAD
Ox01OOL
IEQUALIFIER_REPEAT
Ox0200L
IEQUALIFIER_INTERRUPT
Ox0400L
IEQUALIFIER_MULTIBROADCAST
Ox0800L
IEQUALIFIER_MIDBUTTON
IEQUALIFIER_RBUTTON
IEQUALIFIER_LEFTBUTTON
OxlOOOL
Ox2000L
Ox4000L
IEQUALIFIER RELATIVEMOUSE
Ox8000L
The input.device driver allows division in the key groups, so you

can recognize the left <Shift>, right <Shift> and <Caps Lock> keys.
You can also find out whether information came from the numeric
keypad, and if the key found is in repeat mode. You can check the
mouse buttons for this data.
If you prefer to work with codes processed according to the keyboard
table, then Class returns a message of type VANILLAKEY. Under
Code we find the translated character code. This option is the simplesL
It doesn't matter which method you use in the program:
if (MessageClass & RAWKEY) RAW_Evaluation(Code, Qualifier);
if (MessageClass & VANILLAKEY) VANILLA_Evaluationchar)Code):
Various differences appear in reading the code received. While

VANI LLAKEY can simply work with the characters, RAWKEY must
first consult the keyboard table.
VANILLA_Analysis (ZChar)
char ZChar;
{
.280
ABACUS
Print(ZChar);
InputTextBuffer(ZChar);
RAW_Analysis (Code, Qualifier);

USHORT Code, Qualifier;
{
/* 1. Conversion according to the table *1

ZChar= KeyTab[Qualifier] [Code];
1* or 2. Conversion with individual differences *1

switch (Qualifier)
(
case IEQUALIFIER_LSHIFT
switch (Code)
{
1* Evaluate left Shift key *1

break;
case IEQUALIFIER_RSHIFT
switch (Code)
(
1* Evaluate right Shift key *1

break;
switch (Code)
{
1* Evaluate without Shift key *1
3.7.2.7
The disk drive

Let's look back at the file select box. This should display the directory
of the inserted disk. What happens if you remove the disk from the
drive and insert another disk? Nothing! The same view with the same
files appears. That's wrong-we should be able to read the new disk's
directory.
Two disk flags solve this problem. DISKREMOVED tells us when a
disk is removed, allowing the file table to be deleted. DISKINSERTED
starts the read operation that analyzes the new directory.
281
3.
INTUITION AND
AMIGA
/* 3.7.2.7.a.DISKRMV/INS */
if (MessageClass & DISKREMOVED)
{
/* Clear table */
if (MessageClass & DISKINSERTED)

{
/* Read new table */
282
3.8
ABACUS
3.8
MENUS
Menus
Has this ever happened to you in any computer language? You write a
program which does everything for the user except make coffee in the
morning. You've written subroutines to handle every detail, every error,
every contingency. The user has more functions in this program than he
knows what to do with. But you don't know how to present all these
functions in a fann that the user can easily learn and remember.
We could use the keyooard exclusively, then draw a keyboard overlay to
help the user remember. That's kind of silly, since the Amiga has
Intuition. Maybe we can code user access to functions and commands
through gadgets. That's better, but since we still need room for an input
and output window pair, where do we put the gadgets?
It doesn't matter to the Amiga what you use. Gadgets are good for
limited access, but menus allow the most user-friendly environment for
accessing program commands and functions.
The menus offered by Intuition let us divide all of the functions of our
program into groups. Most of these functions go under menu titles
named F i 1 e and Edi t. Each title lists menu items with names
applicable to the titles. The user selects a function by pressing and
holding the right mouse button, moving the mouse pointer to the menu
title, moving the pointer down to the desired item and releasing the
right mouse button. The Workbench menus were probably the first
menus you ever saw on the Amiga.
We'll write program code in this section that creates multiple menus
with multiple items. Some of the menu items even have submenu
items. These menus make use of all of the menu options.
3.8.1
General construction of a menu

It is important to know which structure the menus have, so that we can
learn our limitations in menu creation. Let's look at the Workbench
menus as our fIrst examples.
When the user presses the right mouse button the title bar changes to
the menu strip, which displays three names. These are the names of
each menu title. The strip itself uses the font used by the current screen.
We can program the title text, similar to setting the exact pixel in the
X direction.
283
3. INTUITION AND C
Now we consider the fIrst menu: The Workbench menu has six functions, none of which can be selected. All of the text appears in ghost
print. If you click on an icon then select the Workbench menu, many
menu items appear in readable print (you can select one of these).
As the second example we11 use an imaginary drawing program for the
menu strip. This menu uses graphic elements-menus can consist of

more than lines of text We can use graphics overall. In addition, we
aren't limited to one line of text/graphic menu item data-we can
determine how many texts or graphics per line appear per menu item.
Be careful not to overlap the individual items!
3.8.2
Making a menu
You now have an idea of what menus can do, thanks to the two examples listed above. Your fIrst goal is to create your own menu strip. We
would like to open a new screen with some additional arguments. We
need a window, without which a menu strip cannot exist since the two
are interrelated.
The following listing accesses a standard program for screens and
contains additional font settings:
/***************************************
*
*
*
* =================================== *
*
*
* Author: Date:
Comments:
*
* -----*
* Program : Baseprogram.c
* Wgb
*
2/ 3/1988
Empty Menu-Strip*
*
*
***************************************/
struct
struct
struct
struct
IntuitionBase
Screen
Window
IntuiMessage
*IntuitionBase;
*FirstScreen;
*FirstWindow;
*message;
struct TextAttr ScreenFont

{
(STRPTR) "topaz. font".

TOPAZ_SIXTY,
284
3.8
ABACUS
MENUS
FS_NORMAL,
FPF_ROMFONT
};

{
0, 0,
640, 200,
3,
0, 1,
/*
/*
/*
/*
HIRES,
/*
CUSTOMSCREEN,
/*
&ScreenFont,
/*
NULL,
/*
NULL,
/*
);
LeftEdge, TopEdge
Width, Height
Depth
DetailPen, BlockPen
ViewModes
Type
Font
*/
*/
*/
*/
*/
*/
*/
Gadgets
CustomBitMap
*/
*/

/*
/*
/*
/*
/*
0, 0,
640, 100,
0, 1,
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
CLOSEWINDOW,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
/* First Gadget
NULL,
/* CheckMark
(UBYTE *)"Test Menu-Strip" ,
NULL,
/* Screen (kommt noch)
NULL,
/* BitMap
100, 50,
640, 200,
CUSTOMSCREEN,
/* Type
};
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
*/
main ()
(
ULONG MessageClass;
USHORT code;
struct Message *GetMsq();
FOREVER
{

(
Wait(lL
continue;
285
3.
INTUITION AND
AMIGA
ReplyMsg(message);
{
case CLOSEWINDOW
Close_All () ;
exit (TRUE) ;
break;
/***********************************************
* =========================================== *
*
*
* Author: Date:
* Wgb
10/16/1987
Comments:
functions!
*
*
*
*
*
*
*
***********************************************1
Open_All ()
{
void
*OpenLibrary();
OpenLibrary(lintuition.library", OL)
{

Close_All () ;
exit(FALSE);
}
if (!(FirstScreen = (struct Screen *)

{
printf(nNo screen page!\n");

Close_All ();
exit(FALSE);
}
FlrstNewWindow.Screen
= FirstScreen;

Openwindow(&FirstNewWindow)
{

Close_All () ;
exit(FALSE);
286
3.8 MENUS
ABACUS
1***************************************
*
*
*
* =============================-===== *
*
*
* Author: Date:
Comments:
*
*
*
---------- ---------16.10.19S7 Window, Screen *
* Wgb
and Intuition
*
*
*
***************************************1
Close_All ()
{
if (First Window)
if (First Screen)
if (IntuitionBase)
CloseScreen(Firstscreen);
3.8.2.1
Filling the menu strip

This test program will help us construct the menu strip. For this we
need a structure for each menu. In addition to the menu titles, the menu
structure contains a flag, coordinate statements and two left pointers.
1* 3.S.2.1.a.menustruct *1
struct Menu
{
OxOO
Ox04
Ox06
OxOS
OxOA
OxOC
OxOE
OxOF
Ox13
Ox15
Ox17
Ox19
Ox1B
00
04
06
OS
10
12
14
15
19
21
23
25
struct Menu *NextMenu;

SHORT Left Edge;
SHORT TopEdge;
SHORT Width;
SHORT Height;
USHORT Flags;
BYTE *MenuName;
struct Menultem *Firstltem;
SHORT JazzX;
SHORT JazzY;
SHORT BeatX;
SHORT BeatY;
);
287
3. INTUITION AND C
AMIGA
Structure
description
*NextMenu
The menu structure contains a pointer to the result

structure, like so many Intuition structures. This
pointer helps in the chaining of multiple menus.
LeftEdge, width
These values describe the position and width of the

select box. This is the region in which the mouse
pointer must lie until the menu becomes active.
TopEdge,Height
These values aren't currently implemented. Later versions may allow graphics in the menu strip.
Flags
This flag determines whether or not the entire menu

should be selectable. If not, all of the menu items
appear in ghost print. To allow the choice, the
MEN U E NAB LED flag is set. Intuition sets the
MIDRAWN flag if the menu is chosen by the user.
*MenuName
This pointer points to the menu titles. You may only

enter a string here, which appears in the default font
of the screen.
*Menultem
Here we find the chaining together with another

structure. It contains all of the information for the
individual menu items.
All other variables are used from Intuition and do not need
initialization.
We'll create three menus in our test program to start. The first menu,
called Files, should support all disk functions. It can be used in any
program because disk access is important in most programs. The second
menu will be called Style and the third menu, Graphics.
Here are the three structures. All you need to do is type them in and
append them to the test program, before the main () function.
/* 3.8.2.1.b. menu_titles */
struct Menu Graphics =
{
NULL,
220, 0,
90, 10,
MENUENABLED,
(BYTE *) "Graphics",
NULL
};
struct Menu Style
{
288
3.8
ABACUS
MENUS
&Graphics.
100. 0,
70. 10,
MENUENABLED.
(BYTE *) "Style",
NULL
};
struct Menu File
&Style.
10, 0,
60, 10,
MENUENABLED,
(BYTE *) "File",
NULL
};
Now insert two new commands in our All () function. First we

must place this menu strip in the window. The Intuition function
SetMenuStrip () perfonns this task. It has the following fonnat:
Success
DO
= SetMenuStrip(Window,
-264
AO
Menu);
Al
The Window lists the window and its address. The menu strip is
already active. You can work with it as long as the window is active.
Our 0 pen _A 11 () function needs the following after the
OpenWindow () function:
SetMenuStrip(FirstWindow, &File);
We have not paid attention to the return value Success because it

always reads true.
The Close_All () function needs:
if (FirstWindow) ClearMenuStrip(FirstWindow);
The general fonnat for ClearMenuStrip () is:

ClearMenuStrip(window);
-54
AD
Remember that you must remove the menu strip before closing the
window, otherwise it could lead to a system crash. In addition, you
must remove the menu strip from the window before any menu change.
Mter the change it can again be added with SetMenuStrip ().
Program
description
Mter starting the program a screen with a window appears. As long as

the window is inactive, you can't reach the menu strip using the right
mouse button. When you activate the window, the menu strip also
becomes visible. You'll see three menu titles, without items (we'll fix
that). Click on the window's close gadget to end the problem.
289
3.
INTUITION AND
3.8.2.2
The MenuJ:tem structure

Now we'll complete the first menu's design. We'll use text because
they're easiest to manipulate. We need an Intui Text structure for
each text
/* 3.S.2.2.a. intuitext */
struct TextAttr Font =
{
(STRPTR) "topaz. font",

TOPAZ_SIXTY,
FS_NORMAL,
FPF_ROMFONT
I;
struct IntuiText LoadText
4, I, JAM2, I, 1, &Font, (UBYTE *)"Load", NULL

);
struct IntuiText SaveText =
4, 1, JAM2, I, I, &Font, (UBYTE *)"Save". NULL
);
struct IntuiText DeleteText

{
4, 1, JAM2, 1, 1. &Font, (UBYTE *)"Delete", NULL

};
struct IntuiText ProgramendText
4, 1, JAM2, I, 1, &Font, (UBYTE *)"Program End", NULL

};
We now have the menu texts but need a structure for each menu item.
The Menultem structure reserves all of the important data for each
menu item.
/* 3.S.2.2.h. menuitem */
struct MenuItem
{
OxOO
Ox04
Ox06
OxOS
OxOA
OxOC
OxOE
Ox12
Ox16
Ox1A
290
00
04
06
OS
10
12
14
IS
22
26
struct MenuItem *Nextltem;

SHORT LeftEdge
SHORT TopEdge;
SHORT Width;
SHORT Height;
USHORT Flags;
LONG MutualExc1ude;
APTR ItemFill;
APTR SelectFill;
BYTE Command;
3.8
ABACUS
Ox1B
Ox1F
Ox21
);
27
31
33
MENUS
struct Menultem *Subltem;

USHORT NextSelect;
Structure
description
*NextItem
This pointer points to the next menu item. We need

it to manage more than one item per menu. All of the
menu items are related to this pointer.
LeftEdge,TopEdge
These specify the position of the activation area called

the select box. The coordinates are relative to
LeftEdge and TopEdge of the menu structure.
Width,Height
These specify the measurements of the select box of

the menu items.
MutualExclude
Through the bits of these variables we can see if other .

menu items should be deactivated by choosing this
menu item.
ItemFill
This pointer either points to an Image structure or

to an IntuiText structure. You label this with the
ITEMTEXT flag (see below for a description of the
flags in this structure). .
SelectFill
The second pointer contains a pointer to one of the

two structures. This occurs only if one of the menu
items is activated and the HIGHlMAGE flag is also
set (see below for a description of the flags in this
structure).
Command
If you define a command under Flags, the ASCn
code of the character of the menu item that should be

activated is found here (see below for a description of
the flags in this structure).
*SubItem
As you may already know, it is possible to insert the

menus into another area. This pointer points to the
first submenu item of this menu item.
NextSelect
A variable that can be filled after choosing the menu

item from Intuition. Then another menu item is
chosen and here is its number.
Flags
Here are the flags which specify the appearance of the

menu items:
291
3.
INTUITION AND
Flags
AMIGA
Normally Intuition assumes a pointer to the Image structure under

ItemFill and SelectFill. By setting the flag ITEMTEXT you
set an IntuiText structure.
The HIGHFLAGS flag configure the highlighting of the menu item
currently under the mouse pointer. When you set HIGHCOMP, the
select box changes to inverse video. When you set HIGHBOX, the
select box appears with a surrounding box. HIGHIMAGE specifies the
display of a new text or graphic. The last flag, HIGHNONE, tells
Intuition to not do anything.
The ITEMENABLED flag indicates whether a specific menu item should
be selectable or not. Set ITEMENABLED if you want the menu item

selectable.
Command keys (keyboard abbreviations) can be assigned to any menu
item. Set the COMMSEQ flag to assign the item to a key combination.
Intuition also allows the selection and unselection of certain menu
items, using the CHECKIT flag. The menu item then stands in either
an active or inactive state.
Should a selectable/unselectable menu item be un selected when the
menu initializes, then you must set the CHECKED flag.
Intuition has reserved two additional flags in the variable Flags.
Specifying a submenu item of a menu item sets the I SDRAWN flag.
The HIGHITEM flag is set when the mouse pointer activates a menu
item.
Now, let's start programming our first menu items. You have hopefully read the IntuiText structure data above and typed these
structures in. We have four MenuItem structures to add to our
program:
1* 3.8.2.2.c.menuitems *1
struct Menultem Programend =
{
NULL,
1, 40,
120, 10,
ITEMTEXT I ITEMENABLED I HIGHCOMP,

0,
(APTR)&ProgramendText,
NULL,
0,
NULL,
);
struct Menultem Delete =,

{
&Programend.
292
3.8 MENUS
ABACUS
1, 26,
120, 10,
0,
(APTR) &DeleteText,
NULL,
0,
NULL,
o
};
struct Menultem Save

{
&Delete,
1, 14,
120, 10,
0,
(APTR)&SaveText,
NULL,
0,
NULL,
o
};
struct Menultem Load

{
&Save,
1, 2,
120, 10,
0,
(APTR) &LoadText ,
NULL,
0,
NULL,
o
};
Here is the complete program for our fIrst menu:

/***************************************
*
*
*
Program: Demonstration Menu-Strip

==================================~
*
*
*
*
*
Comments:
* Author: Date:
*
*
---------- ---------*
* Wgb
2/ 3/1988 Menu-Strip
*
one menu built *
*
*
*
***************************************/
293
3.
INTUITION AND
struct
struct
struct
struct
IntuitionBase
Screen
Window
IntuiMessage
*IntuitionBase;
*FirstScreen;
*FirstWindow;
*message;

{
TOPAZ_SIXTY,
FS_NORMAL,
FPF_ROMFONT
);

{
0, 0,
640, 200,
3,
0, 1,
1*
1*
1*
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
Depth
DetailPen, BlockPen
ViewModes
Type
Font
HIRES,
CUSTOMSCREEN,
&Font,
(UBYTE *) "Screen Test",
NULL,
1* Gadgets
NULL,
1* CustomBitMap
*1
*1
*1
*1
*1
*/
*1
*1
*1
};
0, 0,
640, 100,
0, 1,
1*
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
rDCMP Flags
Flags
CLOSEWINDOW,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *)"Test Menu-Strip",
NULL,
1* Screen (kommt noch)
NULL,
1* BitMap
100, 50,
640, 200,
CUSTOMSCREEN,
1* Type
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
);
struct IntuiText LoadText
4, 1, JAM2, 1, 1, &Font, (UBYTE *)"Load", NULL

};
294
3.8 MENUS
ABACUS
struct IntuiText SaveText
4, 1, JAM2, 1, 1, &Font, (UBYTE *)"Save", NULL

};

(
4, 1, JAM2, 1, 1, &Font, (UBYTE *)"Delete", NULL

);
4, 1, JAM2, 1, 1, &Font, (UBYTE *)"Program end", NULL

);
struct Menultem Programend
NULL,
1, 40,
120, 10,
0,
NULL,
0,
NULL,
);
struct Menultem Delete

{
&Programend,
1, 26,
120, 10,
ITEMTEXT J ITEMENABLED I HIGHCOMP,
0,
(APTR) &DeleteText,
NULL,
0,
NULL,
};
&Delete,
1, 14,
120, 10,
0,
(APTR)&SaveText,
NULL,
0,
NULL,
};
295
3.
INTUITION AND
AMIGA

&Save,
1, 2,
120, 10,

0,'
(APTR)&LoadText,
NULL,
0,
NULL,
};
struct Menu Graphic

NULL,
220, 0,
90, 10,
MENUENABLED,
NULL
};
struct Menu Style

{
&Graphic,
100, 0,
70, 10,
MENUENABLED,
(BYTE *) "Style",
NULL
};
struct Menu File

{
&Style,
10, 0,
60, 10,
MENUENABLED,
(BYTE *) "File",
&Load
};
main ()
(
ULONG MessageClass;
USHORT code;
296
3.8 MENUS
ABACUS
FOREVER
{

GetMsg(FirstWindow->OserPort == NOLL)
{
Wait(1L FirstWindow->OserPort->mp_SigBit);
continue;
}
ReplyMsg(message);
{
case CLOSEWINDOW
Close_All () ;
exit(TROE);
break;
/***********************************************
* =========================================== *
*
*
* Author: Date:
Comments:
*
*
*
* Wgb
16.10.1987
functioning!
*
*
*
*
*
***********************************************/
Open_All ()
{
void
*OpenLibrary();
{

Close_All () ;
exit (FALSE) ;
}

{
print f ("No screen page! \n") ;

Close_All ();
exit(FALSE);
}
297
3. INTUITION AND C
FirstNewWindow.Screen = FirstScreen;
if (!(FirstWindow = (struct Window *)
{

Close_All () ;
exit (FALSE) ;
}
/***************************************

*
* =================================== *
* Author: Date:
*
---------10/16/1987
* Wgb
*
Comments:
----------
Window, Screen
and Intuition
*
*
*
*
*
***************************************1
Close_All ()
{
if
if
if
if
(FirstWindow)
(FirstWindow)
(FirstScreen)
(IntuitionBase)
C1earMenuStrip(FirstWindow);
3.8.2.3
Adding extras to menu design

Most people would be satisfied with the above structures, definitions,
texts and settings. Not us. We currently have two item less menus, and
one menu without any submenu items. Pretty boring stuff. It needs a
little showmanship, as well as some user conveniences. Let's go
through the data step by step, and make full use of Intuition.
First is convenience (which is actually bottom priority in this
instance). Imagine yourself in the following situation: You are the long
time user of a program and know all its possibilities, but must always
take the mouse in your hand to go through the menus to select one
item.
298
3.8 MENUS
ABACUS
This mouse-to-keyboard-to-mouse switching is a pain. It would be

much easier if you could call the menu item from the keyboard as well
as from the menu. You can do it: Just set the COMMSEQ flag in the
Flags array of the MenuItem structure, and to give the Command
argument the ASCII value of the key. The <Amiga> key and our character appear to the right of the menu item. Widen the menu item box
by at least four characters (40 pixels): One character for space, two
characters for the A. character, and one character for the command key
itself. The command key is always called by pressing the right
<Amiga> key, then the key.
Widen all of the item boxes by 40 pixels and set the COMMSEQ flags in
the Flag arrays in each structure (don't forget to include a letter for
each command key). We recommend the following: "L" for Load, "S"
for Save, "D" for Delete. and "E" for end program. Now look at the
MenuItem structures:
/* 3.8.2.3.a.menuitem*/
struct Menultem Programend
NULL,
1, 40,
170, la,
ITEMTEXT
ITEMENABLED
HIGHCOMP
COMMSEQ,
HIGHCOMP
COMMSEQ,
HIGHCOMP
COMMSEQ,
0,
NULL,
Ox45,
NULL,
a
};
&Programend,
1, 26,
170, 10,
ITEMTEXT
ITEMENABLED
0,
(APTR)&De1eteText,
NULL,
Ox44,
NULL,
a
};
&Delete,
1, 14,
170, 10,
ITEMTEXT
ITEMENABLED
0,
(APTR)&SaveText,
299
3.
INTUITION AND
AMIGA
NULL,
Ox53,
NULL,
o
};
&Save,
1, 2,
170, 10,
ITEMTEXT I ITEMENABLED I HIGHCOMP I COMMSEQ,
0,
(APTR)&LoadText,
NULL,
Ox4C,
NULL,
o
Note:
Remember that Intuition differentiates between upper case and lower

case letters. For example, you can use right <Amiga><S> for one
command key, and <Shift> right <Amiga><S> for a different command
key.
Highlighting
The menu items alone aren't enough. A menu item is displayed in

inverse video to indicate it's selected, but we can add still more "bells
en' whistles." For example, we can program the menu to change the
menu item text when an item is selected. Take the Delete item: When
the item changes to inverse video, we can make the text change to
something like "REALLY?" or "Are you sure?", to remind the user of
what he's selecting. The HIGHFLAGS make this text swap possible.
For this we need a new IntuiText structure.We'll also change the
color at the same time:
/* 3.8.2.3.b.swaptext*/
{
5, 1, JAM2, 1, 1, &Font, (UBYTE *) "Delete ", NULL

};
struct IntuiText DeleteTextII
5, 1, JAM2, 1, 1, &Font, (UBYTE *) "REALLY?", NULL

};
You can change the text position or the font. Many options are
possible with this code.
We must remember to change the MenuItem structure. For one, the
HIGHIMAGE flag must replace the HIGHCOMP flag, and then the
SelectFill pointer must be supplied with the pointer to our text:
300
3.8 MENUS
ABACUS
1*3.8.2.3.c.highimage_struct*/
struct Menultem Delete =
{
&Programend.
1, 26.
170. 10.
ITEMTEXTI ITEMENABLED I HIGHIMAGE I COMMSEQ.
O.
(APTR)&DeleteText.
(APTR)&DeleteTextII.
Ox44.
NULL.
};
We kept the letter command sequence because it is always practical.

Compile, link and test the menu choice once.
For graphic menu display, we suggest you look to the other two
HIGHFLAG modes for help (HIGHNONE and HIGHBOX). The
HIGHBOX flag draws a box around the selected menu item. Perhaps you
want some menu items to nQ1 have boxes drawn around them. Use the
HIGHNONE flag. For example, if there are many functions in a menu,
it would be good if these were also arranged so that some can be
accessed in certain cases. If you don't want this, you can simply mark a
graphic or text menu item with a line and not have the program read it.
HIGHNONE actually helps you avoid any program abuse from the user.
For example, why enable the Save item at the start of an editor
program, if there's nothing in memory to save yet? We can enable or
disable the Save item as needed using the ITEMENABLE flag. Here's
the "new and improved" Save structure:
/*3.8.2.3.d.newsave_struct*/
struct Menultem Save =
{
&Delete,
1. 14.
170, 10,
ITEMTEXT I HIGHCOMP I COMMSEQ.

O.
(APTR)&SaveText.
NULL.
Ox53.
NULL.
};
301
3.
INTUITION AND
3.8.2.4
Submenus
We're not done yet-we need to go deeper yet in the menu strip. The
submenus are next. These menus appear when we select a menu item
which requires additional parameters or information.
How do we access a submenu? The Menultem structure performs this
task (see Section 3.8.2.2). We find a pointer to a Subltem therethat's our entrance to the submenu. If you insert the pointer to a new
Menultem structure, that acts as the beginning of the new submenu.
These methods give you the option of expanding the File menu. With
a File menu in a word processor, you may want to enter the ftle type
of the text you want loaded/saved. Enter the submenu. For our example
you can choose from three ftletypes: Text (resident text format used by
the word processor), ASCII (unformatted text file to allow transfer to
any computer), and IFF (Interchange File Format, Electronic Arts
universal ftle exchange format).
/*3.8.2.4.a.text_stuff*/
struct IntuiText TextText =
{
3, 1, JAM2, 1, 1, &Font, (UBYTE *) "Text", NULL

};
struct IntuiText ASCIIText
3, 1, JAM2, 1, 1, &Font, (UBYTE *) "ASCII", NULL

};
struct IntuiText IFFText =
{
3, 1, JAM2, 1, 1, &Font, (UBYTE *) "IFF", NULL

};
struct MenuItem IFF =
{
NULL,
140, 22,
100, 10,
0,
(APTR)&IFFText,
NULL,
Ox49,
NULL,
o
};
struct MenuItem ASCII

{
302
3.8
ABACUS
MENUS
&IFF,
140, 12,
100, 10,

0,
(APTR)&ASCIIText,
NULL,
Ox41,
NULL,
o
};
struct Menultem Text

{
&ASCII,
140, 2,
100, 10,

0,
(APTR) &TextText,
NULL,
Ox54,
NULL,
o
};
&Save,
1, 2,
170, 10,

0,
(APTR)&LoadText,
NULL,
0,
&Text,
o
};
The same submenu must be expanded with the Save function. You
must expand the Me nul t em structure with the pointer to the
submenus and remove the old command sequence:
1*3.8.2.4.b.bigsave_struct*1
struct Menultem Save =
&Delete,
1, 14,
170, 10,
ITEMTEXT I HIGHCOMP I COMMSEQ,

0,
(APTR)&SaveText,
NULL,
0,
303
3.
INTUITION AND
AMIGA
&SaverText,
o
};
Unfortunately, the Subltems must also be redefined because the old

command sequences cannot be used twice:
1*3.8.2.4.c.IFFsave_struct*1
struct MenuItem Save IFF =
{
NULL,
140, 22,
100, 10,
ITEMTEXT I ITEMENABLED I HIGHCOMP I COMMSEQ.
0,
(APTR)&IFFText,
NULL,
Ox46,
NULL,
};
struct MenuItem SaveASCII

{
&SaveIFF,
140, 12,
100, 10,
0,
(APTR)&ASCIIText,
NULL,
Ox43,
NULL,
o
};
struct MenuItem SaverText

{
&SaveASCII.
140, 2,
100, 10,
ITEMTEXT I ITEMENABLED I HIGHCOMP I COMMSEQ.
0,
(APTR)&TextText.
NULL,
Ox51,
NULL,
We can use the IntuiText structures a second time without thinking

about it.
304
3.8 MENUS
ABACUS
3.8.2.5
Style menu tutorial

You now have the essentials for programming your own menus. All we
need to do now is the menu descriptions. You can then define the
structures as you wish, and link them together.
The menu should consist of six text types. Each item should be accessible through a command key. The following texts types are standard:
Normal, Italics, Bold, Inverse, 80 characters and 60
characters. Each menu item must be defmed as an IntuiText
structure with its own TextAttr structure. In the TextAttr structure you give the respective text types. Now the user has a simple
choice and can immediately select a text style.
When the mouse is activate, the user should be able to see which text
types are currently enabled. The checkmark will serve our purposes. Set
the CHECKIT flag in the Flags variable. The checkmark character
must precede the text in the IntuiText structure (two spaces preceding the text should be enough). We also recommend that the
MENUTOGGLE flag be used to enable and disable text styles.
Next we must turn toward Mut ualExcl ude. This flag lets us change
checkmarked menu items to unchecked items, and unchecked items to
checkmarked items. Confused? It's actually very simple and very useful.
Look at a concrete example: if the user selects the Normal text type,
this disables all other text types. We accomplish this by using
MutualExclude. The same goes for the two text sizes. Only one
can be used at a time; the other must be deactivated.
MutualExclude works as follows: We have a long value for which
each bit represents a menu item. All of the menu items marked with a
set bit should be deactivated when a particular menu item is selected.
Bit 0 corresponds to the first menu item, bit 1 to the second menu
item, etc.
3.8.2.6
Graphic menus
After adding type style choices to the Style menu, you have the
option of adding graphics and color choices to the Graphic menu.
This menu goes well in a drawing or paint program.
The menu consists of two menu items. The first, Pen, has a submenu
with two more subitems: A fine point brush graphic and a thick point
brush graphic. The second menu item, Palette, has eight submenu
305
3.
INTUITION AND
AMIGA
items which specify the drawing color. This can be useful for both the
drawing program and the word processor alike.
Here are the Menultem structures for the Pen menu:
/*3.8.2.6.a.brushmenu_struct*/
struct MenuItem PenII =
{
NULL,
90, 2,
10, 10,
ITEMENABLED I HIGHCOMP,
0,
(APTR)&PenIIImage,
NULL,
0,
NULL,
};
struct MenuItem PenI

{
&PenII,
102, 2,
10, 10,
0,
(APTR)&PenIImage,
NULL,
0,
NULL,
};
And the MenuItem structure for the Graphic menu:

/*3.8.2.6.b.brushmenuitem */
struct IntuiText PenText =
{
0, 1, JAM2, 1, 1, &Font, (UBYTE *) "Brush", NULL

};
struct MenuItem Pen

{
&Colour,
2, 2,
100, 10,
ITEMTEXT t ITEMENABLED I HIGHCOMP,
0,
(APTR)&PenText,
NULL,
0,
&PenI,
a
};
306
3.8 MENUS
ABACUS
Here are the two Image structures and their values:

/*3.8.2.6.c.imagedata*/
USHORT PenIData[]
{
OxOO,
OxOl,
OxOO,
OxOO,
OxOO,
OxFO,
OxOO,
OxOO,
};
USHORT PenIIData[]
{
OxOO,
OxOl,
OxOl,
OxOO,
OxOO,
OxFO,
OxFO,
OxOO,
};
struct Image PenIImage

{
1, 1,
8, 8,
1,
&PenIData[O],
2, 0,
NULL
};
struct Image PenIIImage
1, 1,
8, 8,
2,
&PenIIData[O],
2, 0,
NULL
};
The data for the color choice submenu follow. We define a MenuItem
structure for each menu item. The submenu represents every color
through In t ui Text structure. First, the main menu:
/*3.8.2.6.d.colortext*/
struct IntuiText ColornText
{
0, 1, JAM2, 1, 1, &Font, (UBYTE *) "Colors", NULL

};
struct MenuItem Colorn

{
NULL,
2, 14,
100, 10,
307
3.
INTUITION AND
0,
(APTR)&ColornText,
NULL,
0,
&ColorO,
o
);
For the colors we take IntuiText structures which contain blank

spaces only. The background color changes so that color bars appear
instead of spaces. We show only the frrst three structures; the others are
easy to construct from there:
/*3.8.2.6.e.colOtext*/
struct IntuiText ColOText
{
0, 0, JAM2, 1, 1, &Font, (UBYTE *)"
.. , NULL
};
struct IntuiText Coil Text
0, 1, JAM2, 1, 1, &Font, (UBYTE *)"

};
struct IntuiText Co12Text
.. , NOLL
0, 2, JAM2, 1, 1, &Font, (OBYTE *)A
A, NULL
);
Defme the MenuItem submenu structure based on the same principles:
/*3.8.2.6.f.color2*/
struct Menultem Color2
{
&Color3,
90, 26,
50, 10,
ITEMTEXT I ITEMENABLED I HIGHBOX,
0,
(APTR)&Co12Text,
NULL,
0,
NULL,
o
);

&Color2,
90, 14,
50, 10,
308
3.8
ABACUS
MENUS
0,
(APTR) &CollText,
NULL,
0,
NULL,
o
};
struct MenuItem Col orO

{
&Colorl,
90, 2,
50, 10,

0,
(APTR)&Co10Text,
NULL,
0,
NULL,
};
Here is an example program:

/***************************************
* Program: Demonstration Menu-Strip
* =================================== *
*
*
Comments:
*
* Author: Date:
*
---------- ---------*
* Wgb
*
*
2/ 3/1988
Menu-Strip
*
*
***************************************/
struct
struct
struct
struct
IntuitionBase
Screen
Window
IntuiMessage
*IntuitionBase;
*FirstScreen;
*FirstWindow;
*message;

{
TOPAZ_SIXTY,
FS_NORMAL,
FPF_ROMFONT
};
struct NewScreen FirstNewScreen
309
3.
INTUITION AND
0, 0,
640, 200"
/*
/*
3,
/*
0, 1,
/*
HIRES,
/*
CUSTOMSCREEN.
/*
&Font,
/*
NULL,
/*
NULL,
/*
};
LeftEdge, TopEdge
Width, Height
Depth
DetailPen, BlockPen
ViewModes
Type
Font
*/
*/
*/
*/
*/
*/
*/
Gadgets
CustomBitMap
*1
*1

0, 0,
640, 100.
0, 1,
*/
/*
/*
/*
/*
*1
*1
*1
Width, Height
DetailPen, BlockPen
IDCMP Flags
Flags
CLOSEWINDOW,
WINDOWDEPTH I
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
SMART_REFRESH.
NULL,
/* First Gadget
NULL,
/* CheckMark
(UBYTE *)"Test Menu-Strip" ,
NULL,
1* Screen (kommt noch)
NULL,
/* BitMap
100, 50,
640, 200,
CUSTOMSCREEN,
/* Type
}I
USHORT PenIData[) =
{
OxOO,
OxOl,
OxOO,
OxOO,
OxOO,
OxFO,
OxOO.
OxOO.
};
USHORT PenIIData[]
{
OxOO,
Ox01,
OxOl,
OxOO,
OxOO,
OxFO,
OxFO,
OxOO,
};
struct Image Penllmage

{
1, 1,
8, 8,
1,
310
*/
*1
*1
*1
*1
*/
*1
*1
3.8 MENUS
ABACUS
&PenIData[O] ,
2, 0,
NULL
};
struct Image PenIIImage
{
1, 1,
8, 8,
2,
&PenIIData[O) ,
2, 0,
NULL
};
struct IntuiText ColorText

{
0, 1, JAM2, 1. 1, &Font, (UBYTE *) "Color", NULL

};
struct IntuiText PenText
0, 1, JAM2, 1, 1, &Font, (UBYTE *) "Pen", NULL

};
struct IntuiText LoadText ='

{
4. 1, JAM2, 1, 1, &Font, (UBYTE *)"Load", NULL

};
struct IntuiText Save Text
4, 1, JAM2, 1. 1, &Font, (UBYTE *) "Save", NULL

};
{
5, 1, JAM2, 1, 1, &Font, (UBYTE *)"Delete ", NULL

};
struct IntuiText DeleteTextII
5, 1, JAM2, 1, 1, &Font, (UBYTE *)"REALLY?", NULL

};
4, 1, JAM2, 1. 1, &Font, (UBYTE *)"Program end", NULL

};
struct IntuiText TextText
3, 1, JAM2, 1, :.., &Font, (UBYTE *)"Text", NULL
311
3. INTUITION AND C
AMIGA
};
struct IntuiText ASCIIText
3, 1, JAM2, 1, 1, &Font, (UBYTE *)"ASCII", NULL

};
struct IntuiText IFFText =

{
3, 1, JAM2, 1, 1, &Font, (OBYTE *) "IFF", NOLL

};
0, 0, JAM2, 1, 1, &Font, (UBYTE *)"

};
",
NULL
0, 1, JAM2, 1, 1, &Font, (UBYTE *)"

};
" , NULL
struct IntuiText Co12Text =

{
0, 2, JAM2, 1, 1, &Font, (UBYTE *)"

};
" , NOLL

0, 3, JAM2, 1, 1, &Font, (UBYTE *)"
};
" , NULL
0, 4, JAM2, 1, 1, &Font, (UBYTE *)"

};
" , NULL
0, 5, JAM2, 1, 1, &Font, (UBYTE *)"

};
" , NOLL

{
0, 6, JAM2, 1, 1, &Font, (UBYTE *)"

};
" , NOLL

0, 7, JAM2, 1, 1, &Font,
(OBYTE *)"
" , NOLL
(OBYTE *)"
" , NOLL
};

0, 8, JAM2, 1, 1, &Font,
};
312
3.8
ABACUS
MENUS
struct MenuItem PenII

NULL,
90, 2,
10, 10,
0,
(APTR)&PenIIlmage,
NULL,
0,
NULL,
o
};
struct Menultem PenI

{
&PenII,
102, 2,
10, 10,
0,
(APTR)&Penllmage,
NULL,
0,
NULL,
o
};

{
NULL,
90, 98,
50, 10,

0,
(APTR)&CoI8Text,
NULL,
0,
NULL,
o
};
struct Menultem Color?

{
&Color8,
90, 86,
50, 10,

0,
(APTR)&Col?Text,
NULL,
0,
NULL,
o
};
313
3.
INTUITION AND

{
&Color7,
90, 74,
50, 10,
0,
(APTR)&Co16Text,
NULL,
0,
NULL,
o
);
struct Menultem ColorS

{
&Color6.
90, 62,
50, 10,
0,
(APTR)&Co15Text,
NULL,
0,
NULL,
o
);
struct Menultem Color4 {
&ColorS,
90, 50,
50, 10,
0,
(APTR)&Co14Text,
NULL,
0,
NULL,
};

{
&Color4,
90, 38,
50, 10,
0,
(APTR)&Co13Text,
NULL,
0,
NULL,
};
314
3.8 MENUS
ABACUS

{
&Color3,
90, 26,
50, 10,
0,
(APTR)&Co12Text,
NULL,
0,
NULL,
};
{
&Color2,
90, 14,
50, 10,
0,
(APTR)&CollText,
NULL,
0,
NULL,
t;
struct Menultem Col orO
{
&Color1,
90, 2,
50, 10,
0,
(APTR)&ColOText,
NULL,
0,
NULL,
);
struct Menultem Color

{
NULL,
2, 14,
100, 10,
0,
(APTR)&ColorText,
NULL,
0,
&ColorO,
o
};
315
3.
INTUITION AND
struct MenuItem Pen

&Color,
2, 2,
100, 10,

0,
(APTR)&PenText,
NULL,
0,
&PenI,
a
};
struct MenuItem IFF

{
NULL,
140, 22,
100, 10,

0,
(APTR)&IFFText,
NULL,
Ox49,
NULL,
};
struct MenuItem ASCII

&IFF,
140, 12,
100, 10,

0,
(APTR)&ASCIIText,
NULL,
Ox41,
NULL,
o
};
struct MenuItem Text

{
&ASCII,
140, 2,
100, 10,

0,
(APTR)&TextText,
NULL,
Ox54,
NULL,
316
3.8
ABACUS
MENUS
};
struct MenuItem Save IFF =

{
NULL,
140, 22,
100, 10,
ITEMTEXT
ITEMENABLED
HIGHCOMP
COMMSEQ,
HIGHCOMP
COMMSEQ,
HIGHCOMP
COMMSEQ,
COMMSEQ,
0,
(APTR) &IFFText,
NULL,
Ox46,
NULL,
};
struct MenuItem SaveASCII
&SaveIFF,
140, 12,
100, 10,
ITEMTEXT
ITEMENABLED
0,
(APTR)&ASCIIText,
NULL,
Ox43,
NULL,
};
struct MenuItem SaverText

{
&SaveASCII,
140, 2,
100, 10,
ITEMTEXT I ITEMENABLED
0,
(APTR)&Te~tText,
NULL,
Ox51,
NULL,
};
struct MenuItem Programend

{
NULL,
1, 40,
170, 10,
ITEMTEXT
ITEMENABLED
HIGHCOMP
0,
NULL,
Ox45,
NULL,
317
3.
INTUITION AND
o
};
{
&Programend,
1, 26,
170, 10,
ITEMTEXT I ITEMENABLED I HIGHIMAGE I COMMSEQ,
0,
(APTR}&DeleteText,
(APTR)&DeleteTextII,
Ox44,
NULL,
};

{
&Delete,
1, 14,
170, 10,
ITEMTEXT I HIGHCOMP I COMMSEQ,
0,
(APTR}&SaveText,
NULL,
Ox53,
&SaverText ,
};
&Save,
1, 2,
170, 10,
0,
(APTR) &LoadText ,
NULL,
0,
&Text,
o
};
struct Menu Graphic

{
NULL,
220, 0,
90, 10,
MENUENABLED,
&Pen
};
318
ABACUS
3.8 MENUS
struct Menu Design

{
&Graphic,
100, 0,
70, 10,
MENUENABLED,
(BYTE *) "Design",
NULL
};
struct Menu File

&Design,
10, 0,
60, 10,
MENUENABLED,
(BYTE *) "File",
&Load
};
main ()
{
ULONG MessageClass;
USHORT code;
FOREVER
{

{
Wait(lL
continue;
ReplyMsg(message);
{
case CLOSEWINDOW
Close_All () ;
exit (TRUE) ;
break;
/***********************************************
319
3.
INTUITION AND
AMIGA
* =========================================== *
*
*
Author: Date:
Comments:
Wgb
functioning!
---------- ----------
16.10.1987
*
*
*
*
*
*
*
*
***********************************************1
Open_All 0
(
void
*OpenLibrary () ;
OpenLibrary(nintuition.libraryn, OL)
{
printf (nIntuition Library not found! \nn);

Close_All () ;
exit (FALSE) ;
}

{
printf(nNo screen page!\nn);

Close_All ();
exit (FALSE) ;
}
FirstScreen;

(

Close_All () ;
exit (FALSE) ;
}
1***************************************
*
*
*
* =================================== *
*
*
* Author: Date:
Comments:
*
*
---------- ---------*
* Wgb
10/16/1987 Window, Screen *
and Intuition
*
*
*
*
320
3.8 MENUS
ABACUS
***************************************/
Close_All ()
{
if
if
if
if
(FirstWindow)
(FirstWindow)
(FirstScreen)
(IntuitionBase)
ClearMenuStrip(FirstWindow);
3.8.3
Reading the menu strip

The menu strip mentioned in this section represents a universal menu
type because it consists of more than one menu, one menu item and
more than one submenu item. In addition, we find menu items that
release actions applicable to certain items. For this reason the menu
strip is suitable for a general test You can use the following functions
in each of your programs, because menu reading is especially flexible.
As for every IDCM access, we need the proper parameters. For this we
set the MENUPICK flag in the Window structure IDCMP variable.
Next we test for a menu message in the loop:
/*3.8.3.a.menureadforever*/
FOREVER
{

(
Wait(lL
continue;
Code = message->Code;
ReplyMsg(message);
(
case CLOSEWINDOW
Close_AII();
exit (TRUE) ;
break;
case MENUPICK
Menu_Analysis(Code);
break;
321
3.
INTUITION AND
AMIGA
When we get a message or type MENUPICK, we branch to the

Menu Analysis function, where we can examine one point after
another.
/*3.B.3.b.menu analysis*/
Menu_Analysis (Menunumber)
USHORT Menunumber;
(
USHORT Menu, Menultem, Subltem, NextMenu;

struct Menultem *MenuPoint;
Menu
= MENUNUM(Menunumber);
Menultem = ITEMNUM{Menunumber);
Subltem = SUBNUM{Menunumber);
MenuPoint
ItemAddress(&File, Menunumber);
switch (Menu)
(
case 0 : /* File - Menu */

break;
case 1 : /* Style - Menu */
break;
case 2 : /* Graphic - Menu */
break;
NextMenu = MenuPoint->NextSelect;
if (NextMenu) Menu_Analysis(NextMenu);
Here we choose the respective menus. The program calculates the

address of the Men u I t em structure with the function
ItemAddress (). We can read more data out. This uses the new
function if the value Next Select is read from a new menu number.
This routine also supports the choosing of multiple menu items.
Now we will go through the reading process, starting with the File
menu. Let's layout four menu items:
/*3.B.3.c.menuitems*/
switch (Menultem)
{
case 0 : /* Load */
break;
case 1 : /* Save */
322
ABACUS
3.8 MENUS
break;
case 2 : /* Delete */
Delete();
break;
case 3 : /* Program end */
Ende = FALSE;
break;
The last two items finish a process. Only the first two must be read for
now:
/*3.8.3.d.load_save*/
case 0 : /* Load */
switch (SubItem)
{
case 0 : /* Text */
Load(TEXT);
break;
case 1 : /* ASCII */
Load (ASCII) ;
break;
case 2 : /* IFF */
Load(IFF);
break;
break;
case 1 : /* Save */
switch (Subltem)
{
case 0 : /* Text */
Save (TEXT) ;
break;
case 1 : /* ASCII */
Save (ASCII) ;
break;
case 2 : /* IFF */
Save (IFF) ;
break;
break;
All submenu accesses call the same function. They pass parameters (in
this case, self-defined flags). TEXT tells the Load () or Save ()
structures that it is waiting for the command sequences needed to write
out/load normal text. ASCII would strip out any formatting and IFF
would save/load data in Interchange File Format.
The second menu requires an alternate method of reading. We have no
submenus, only a main menu where many features interact with one
323
3.
INTUITION AND
AMIGA
another. Only the text width is incompatible. Assuming that the program uses one variable for the text type and one for character width, the
following function makes sense:
/*3.S.3.e.textstyle*/
switch (MenuItem)
(
case 0 : /* Normal */
TextStyle = NULL;
break:
case 1 : /* Italics */
TextStyle ~= ITALIC:
break:
case 2 : /* Bold */
TextStyle ~= BOLD
break:
case 3 : /* Inverse */
TextStyle ~= INVERS;
break;
case 4 : /* SO Characters */
TextFont = SOL;
break;
case 5 : /* 60 Characters *1
TextFont = 60L;
break:
The third menu lets us return to simpler reading. The Pen item is fIrst;
the PenNr variable contains the number of the selected pen. The
variable ColorNr contains the color number after selecting that color:
/*3.S.3.f.brush_color*/
switch (Menultem)
{
case 0 : 1* Brush *1
PenNr = Sub Item + 1;
break;
case 1 : 1* Color *1
ColNr = SubItem;
break;
324
3.8 MENUS
ABACUS
3.8.4
Working with source code utilities

What do you know about source code utilities? These are what we call
programs that help the programmer in developing program source code.
Many source codes execute complex and large tasks-tasks that are
theoretially hard to process because they lean heavily on graphics. The
data entry itself isn't difficult, but involve lots of writing and
development time. A source code utility decreases the amount of
writing time. It provides the necessary definitions in a clear, easy-tounderstand format, which the developer may change to suit his/her own
needs.
You'll find such utilities very helpful when programming Intuition.
You've seen how many parameters are necessary just for a simple
window. This grows in complexity with every gadget (by one, two and
even three structures). You remember how much typing you had to do
to program in the requesters.
It's gotten to the point where writing such large structures yourself
stops being feasible. Since a program needs menus only when you have
multiple functions available, we need at least two structures per menu.
Some programs may feature over 50 menu items. For example,
BeckerText from Abacus has five menus with 42 menu items and 65
submenu items. For this you would have to define 84 Menultem
structures and 130 Intui Text structures. We can decrease this work
using a source code utility program.
PowerWindows is one such program that relieves a programmer of a
lot of work. It allows the programmer to easily create windows, menu
strips and gadgets-the three most important elements of Intuition. It
will write a commented source code to disk for the programmer. If you
are going to be using C on the Amiga a lot, PowerWindow is a good
investment
Here is an example source text generated by PowerWindow:
struct IntuiText IText1 - {
3, 7, JAM2,
1* front and back text pens and drawmode *1
20,1,
1* XY origin relative to container TopLeft *1
NULL,
1* font pointer or NULL for defaults *1
"
.,
1* pointer to text *1
NULL
/* next IntuiText structure *1
I;
struct MenuItem SubItem8 - {
NULL,
34,62,
1* next SubItem structure *1

1* XY of Item hitbox relative to
TopLeft of parent hitbox *1
68,10,
1* hit box width and height *1
CHECKIT+ITEMTEXT+ITEMENABLED+HIGHBOX,
1* Item flags *1
0,
1* each bit mutually-excludes a same-level Item *1
(APTR)&IText1, 1* Item render (IntuiText or Image or NULL) *1
NULL,
1* Select render *1
325
3.
INTUITION AND
AMIGA
/* alternate command-key */
/* no SubItem list for SubItems */
NULL,
NULL,
OxFFFF
1* filled in by Intuition for drag selections */
I;
struct IntuiText IText2 = {
3,6, JAM2,
/* front and back text pens and drawmode
/* XY origin relative to container TopLeft
20,1,
NULL,
1* font pointer or NULL for defaults
1* pointer to text
1* next IntuiText structure
NULL
.,
*1
*1
*1
*1
*1
I:
struct MenuItem SubItem7 = {
&SubItem8,
34,52,
/* next SubItem structure *1

1* XY of Item hitbox relative
to TopLeft of parent hitbox *1
68,10,
1* Item flags *1
0,
/* each bit mutually-excludes a same-level Item *1
(APTR)&IText2, 1* Item render (IntuiText or Image or NULL) */
NULL,
1* Select render *1
NULL,
1* alternate command-key *1
NULL,
/* no SubItem list for SubItems *1
OxFFFF
1* filled in by Intuition for drag selections *1
I;
struct IntuiText IText3 = I
3,5, JAM2,
1* front and back text pens and drawmode
20,1,
NULL,
II ,
/* pointer to text
NULL
*/
*1
*1
*/
*/
I;
struct MenuItem SubItem6
&SubItem7,
34,42,
/* next SubItem structure */
/* XY of Item hitbox relative

to TopLeft of parent hitbox */
68,10,
1* hit box width and height
1* Item flags
0,
1* each bit mutually-excludes a same-level Item
(APTR) &IText3 , 1* Item render (IntuiText or Image or NULL)
NULL,
1* Select render
NULL,
1* alternate command-key
NULL,
1* no SubItem list for SubItems
OxFFFF
1* filled in by Intuition for drag selections
*1
*1
*1
*1
*1
*1
*1
*1

3,4, JAM2,
20,1,
1* XY origin relative to container TopLeft
NULL,
"
1* pointer to text
NULL
*1
*1
*1
*1
*1
I;
I;

&SubItem6,
34,32,

to TopLeft of parent hitbox

68,10,
1* Item flags
a,
(APTR)&IText4, /* Item render (IntuiText or Image or NULL)
NULL,
1* Select render
NULL,
/* alternate command-key
NULL,
OxFFFF
I;
326
*1
*1
*1
*1
*1
*1
*1
*1
*/
ABACUS
3.8 MENUS

3,3, JAM2,
20,1,
NULL,
/* font pointer or NULL for defaults
1* pointer to text
",
NULL
/* next IntuiText structure
Ji
*1
*/
*/
*1
*/

&SubItem5,
34,22,
1* next SubItem structure * /

1* XY of Item hit box relative
68,10,
/* Item flags *1
0,
1* each bit mutually-excludes a same-level Item */
NULL,
/* Select render *1
NULL,
NULL,
1* no SubItem list for SubItems */
OxFFFF
Ji
3,2,JAM2,
20,1,
NULL,
1* pointer to text
",
NULL
I;
&SubItem4,
34,12,
*/
*/
*/
*/
*/

to TOpLeft of parent hitbox *1
68,10,
1* hit box width and height */
/* Item flags */
0,
NULL,
1* Select render */
NULL,
NULL,
1* no SubItem list for SubItems */
OxFFFF
Ii

3,1, JAM2,
20,1,
NULL,
M
/* pointer to text
NULL
I;
*/
*1
*1
*1
*1
struct MenuItem SubItem2 = I

&SubItern3,
34,2,
1* next SubItem structure */

68,10,
1* Item flags */
0,
/* each bit mutually-excludes a same-level Item */
(APTR)&IText7, /* Item render (IntuiText or Image or NULL) */
NULL,
1* Select render */
NULL,
1* alternate command-key */
NULL,
1* no SubItem list for SubItems *1
OxFFFF
I;
3, 0, JAM2,
/* front and back text pens and drawmode *1
20,1,
327
3.
INTUITION AND
NULL,
NULL

1* pointer to teKt *1
1* neKt IntuiTeKt structure *1
I;
struct Menultem Sublteml - t
&Subltem2,
34,-8,
1* neKt Subltem structure *1

1* XY of Item hitboK relative
68,10,
CHECKIT+ITEMTEXT+ITEMENABLED+HIGHBOX+CHECKED, 1* Item flags *1
0,
NULL,
1* Select render *I
NULL,
NULL,
1* no Subltem list for Subltems *1
OxFFFF
I;
3, 1, COMPLEMENT,
1* front and back teKt pens and drawmode
1,1,
NULL,
"Color",
1* pointer to text
NULL
I;
*1
*1
*1
*1
*1
struct Menultem Menultem2 - t

NULL,
0,11,
1* next Menultem structure *1

1* XY of Item hitboK relative to
TopLeft of parent hitboK *1
49,10,
/* hit box width and height *1
ITEMTEXT+ITEMENABLED+HIGHCOMP,
1* Item flags *1
0,
NULL,
1* Select render *1
NULL,
&SubIteml.
1* SubItem list *1
OKFFFF
Ii
struct IntuiText IText10 - f
3, 1, COMPLEMENT,
1* front and back teKt pens and drawmode
1,1,
NULL,
"!!!",
1* pointer to teKt
NULL
1* neKt IntuiTeKt structure
I;
*1
*1
*1
*1
*1
struct Menultem Subltem10 - I

NULL,
59,0,
1* next Subltem structure *1

25,10,
1* Item flags *1
0,
(APTR)&IText10, 1* Item render (IntuiTeKt or Image or NULL) *1
NULL,
1* Select render *1
NULL,
NULL,
1* no Subltem list for Subltems *1
QKFFFF
I;
3, 1, COMPLEMENT,
1,1,
NULL,
"111",
1* pointer to text
NULL
I;
328
*1
*1
*1
*1
*1
3.8
ABACUS
MENUS

&SubItem10,
34,0,

25,10,
/* hit box width and height */
/* Item flags *1
0,
(APTR)&IText11 1* Item render (IntuiText or Image or NULL) */
NULL,
/* Select render *1
NULL,
NULL,
/* no SubItem list for SubItems */
OxFFFF
/* filled in by Intuition for drag selections */
};

3, 1, COMPLEMENT,
1,1,
NULL,
"Pen",
/* pointer to text */
NULL
*1
*/
*1
*/
};
struct MenuItem MenuItem1 - {

&MenuItem2,
0,0,
1* next MenuItem structure *1

49,10,
1* Item flags *1
0,
NULL,
/* Select render */
NULL,
&SubItem9,
/* SubItem list *1
OxFFFF
};
struct Menu Menu3 NULL,

135,0,
66,0,
MENUENABLED,
"Graphic",
&MenuItem1
1* next Menu structure */

/* XY origin of Menu hit box
relative to screen TopLeft */
/* Menu hit box width and height */
/* Menu flags */
1* text of Menu name */

/* MenuItem linked list pointer *1
};

3, 1, COMPLEMENT,
20,1,
NULL,
"60 Cgaracter",
/* pointer to text
NULL
*/
*/
*/
*/
*/
};
struct MenuItem MenuItemB - {

NULL,
0,55,

/* XY of Item hit box relative
140,10,
CHECKIT+ITEMTEXT+COMMSEQ+ITEMENABLED+HIGHCOMP, /*Item flags */
0,
(APTR)&IText13, /* Item render (IntuiText or Image or NULL) *1
NULL,
1* Select render */
"6",
NULL,
1* SubItem list *1
OxFFFF
};
329
3. INTUITION AND C
AMIGA
3, 1, COMPLEMENT,
20,1,
NULL,
"SO Character-,
NULL
/* front and back text pens and drawmode */
/* XY origin relative to container TopLeft *1

1* next IntuiText structure *1
/;
struct MenuItem MenuItem7 &MenuItemS,
0,44,

140,10,
CHECKIT+ITEMTEXT+COMMSEQ+ITEMENABLED+HIGHCOMP+CHECKED,
1* Item flags *1
0,
NULL,
1* Select render *1
"8",
NULL,
1* SubItem list *1
OxFFFF
);
struct IntuiText IText15 - I

3, 1, COMPLEMENT,
1* front and back text pens and drawrnode
20,1,
NULL,
"Invers,
1* pointer to text
NULL
*1
*1
*1
*1
*1
);
struct MenuItem MenuItem6 &MenuItem7,

0,33,

140,10,
CHECKIT+ITEMTEXT+COMMSEQ+MENUTOGGLE+ITEMENABLED+HIGHCOMP,
1* Item flags
0,
(APTR)&IText15, 1* Item render (IntuiText or Image or NULL)
NULL,
1* Select render
"V",
NULL,
1* SubItern list
OxFFFF
};
struct IntuiText IText16 - (

3, 1, COMPLEMENT,
20,1,
NULL,
"Bold",
1* pointer to text
NULL
*1
*1
*1
*1
*1
*1
*1
*1
*1
*/
*1
*1
);
struct MenuItem MenuItern5 - (

&MenuItem6,
0,22,

140,10,
/* Item flags
0,
(APTR)&ITextI6, 1* Item render (IntuiText or Image or NULL)
NULL,
/* Select render
"Btt,
NULL,
/* SubItern list
OxFFFF
*1
*1
*1
*1
*1
*1
*1
);

3,1,COMPLEMENT,
330
3.8
ABACUS
20,1,
NULL,
"Italic",
NULL

1* pointer to text
struct Menultem Menultem4 - I

&Menultem5,
0,11,
MENUS
*1
*1
*1
*1

140,10,
1* Item flags
0,
(APTR)&IText17, 1* Item render (IntuiText or Image or NULL)
NULL,
1* Select render
"I",
NULL,
1* Subltem list
OxFFFF
*1
*1
*1
*1
*1
*1
*1
};

3, 1, COMPLEMENT,
20,1,
NULL,
"Normal",
1* pointer to text
NULL
I;
struct MenuItem Menultem3 - {
&Menultem4,
0,0,
*1
*1
*1
*1
*1

140,10,
CHECKIT+ITEMTEXT+COMMSEQ+MENUTOGGLE+ITEMENABLED+HIGHCOMP+CHECKED,
1* Item flags *1
0,
NULL,
1* Select render *1
"N",
NULL,
1* Subltem list *1
OxFFFF
struct Menu Menu2 - I

&Menu3,
63,0,
66,0,
MENU ENABLED,
"Style",
&MenuItern3
1* next Menu structure *1

1* XY origin of Menu hit box
relative to screen TopLeft *1
1* Menu hit box width and height *1
1* Menu flags *1
1* text of Menu name *1
1* MenuItem linked list pointer *1

3, 1, COMPLEMENT,
1,1,
NULL,
"Program end",
NULL
J;
struct Menultem Menultem12 - I
NULL,
0,33,

137,10,
ITEMTEXT+COMMSEQ+ITEMENABLED+HIGHCOMP,
1* Item flags *1
0,
331
3.
INTUITION AND
AMIGA
NULL,
liE",
NULL,
OxFFFF
1* Select render *1
1* SubItem list *1
);
struct IntuiText IText20 ~ !

3, 1, COMPLEMENT,
1,1,
NULL,
"Delete",
NULL
);
struct MenuItem MenuItem11

&MenuItem12,
0,22,

137,10,
1* Item flags *1
0,
NULL,
1* Select render *1
"D",
NULL,
1* SubItem list *1
OxFFFF
);
struct IntuiText IText21 = (

3, 1, COMPLEMENT,
1,1,
NULL,
"IFF",
1* pointer to text
NULL
*1
*1
*1
*1
*1
);
struct MenuItem SubItern13 * !

NULL,
122,12,

81,10,
1* Item flags *1
0,
NULL,
1* Select render *1
"F",
NULL,
OxFFFF
);

3, 1, COMPLEMENT,
1,1,
NULL,
"ASCII",
1* pointer to text
NULL
*1
*1
*1
*1
*1
);
struct MenuItern SubItem12 = !

&SubItem13,
122,2,

1* XY of Item hit box relative
81,10,
1* Item flags *1
0,
NULL,
1* Select render *1
C,
NULL,
OxFFFF
332
3.8
ABACUS
MENUS
};
struct IntuiText IText23 = (
3,1,COMPLEMENT,
1,1,
NULL,
"Text",
1* pointer to text
NULL
*1
*/
*1
*1
*/
I;
struct Menultem SubItem11 = (
&Subltem12,
122,-8,
/* next Subltem structure *1


1* Item flags
0,
(APTR)&IText23, /* Item render (IntuiText or Image or NULL)
NULL,
1* Select render
"X",
/* alternate command-key
NULL,
OxFFFF
81,10,
*1
*1
*/
*1
*1
*1
*1
*1
*/
I;
struct IntuiText IText24 .= (
3, 1, COMPLEMENT,
1,1,
NULL,
"Save",
NULL
Ii
struct Menultem MenuItem10 = f
&Menultemll,
0,11,
*1
*1
*/
*1
/* next MenuItem structure */

137,10,
ITEMTEXTtITEMENABLED+HIGHCOMP,
1* Item flags * /
0,
NULL,
1* Select render *1
NULL,
/* alternate command-key *1
&SubItem11,
1* SubItem list *1
OxFFFF
I;
3,1, COMPLEMENT,
1,1,
NULL,
"IFF",
/* pointer to text
NULL
*1
*1
*/
*/
*/
);

NULL,
122,12,
1* next SubItem structure */

81,10,
/* Item flags *1
0,
,
;
NULL,
"I",
NULL,
OxFFFF
1* Select render */
/* no Subltem list for Subltems */
/* filled in by Intuition for drag selections *1

3,l,COMPLEMENT,
333
3.
INTUITION AND
1,1,
NULL,
"ASCII",
NULL
/* XY origin relative to container TopLeft *1

/* font pointer or NULL for defaults *1
/* pointer to text */
/* next IntuiText structure *1
struct Menultem Sublteml5 - t

&SubltemI6,
122,2,
/* next Subltem structure */

81,10,
/* Item flags *1
0,
NULL,
/* Select render */
HAH,
NULL,
OxFFFF
/* no Subltem list for Subltems
/* filled in by Intuition for drag selections
*1
*1
J;
3, 1, COMPLEMENT,
1,1,
NULL,
"Text",
/* pointer to text
NULL
I;
struct Menultem Sublteml4 - (
&Subltem15,
122,-8,
*/
*/
*/
*1
*1
/* next Subltem structure *1


81,10,
/* Item flags *1
0,
(APTR)&IText27,
Item render (IntuiText or Image or NULL) */
NULL,
Select render */
NT",
/* alternate command-key *1
NULL,
/* no Subltem list for Subltems */
OxFFFF
'*
'*
I;

3, 1, COMPLEMENT,
'* front and back text pens and drawmode
1,1,
NULL,
font pointer or NULL for defaults
"Laden",
pointer to text
NULL
/* next IntulText structure
'*
'*
*/
*/
*/
*/
*/
I:
struct Menultem Menultem9 &MenultemlO,
0,0,

/* XY of Item hit box relative

137,10,
/* hit box width and height
/* Item flags
0,
'* each bit mutually-excludes a same-level Item
(APTR)&IText28,
Item render (IntuiText or Image or NULL)
NULL,
Select render
NULL,
'* alternate command-key
&Subltem14,
Subltem list
OxFFFF
filled in by Intuition for drag selections
'*
'*
I;
struct Menu Menul t

&Menu2,
0,0,
'*
'*
*/
*/
*/
*'
*/
*1
*/
*/
*1
1* next Menu structure *1

1* XY origin of Menu hit box
relative to screen TopLeft */
57,0,
MENUENABLED,
334
1* Menu hit box width and height *1

/* Menu flags *1
ABACUS
3.8 MENUS
"Datei",
&MenuItem9
/* text of Menu name */

/* MenuItem linked list pointer */
I;
fdefine MenuList Menu1

struct NewWindow NewWindowStructure - {
275,85,
/* window XY origin relative to TopLeft of screen */
150,50,
/* window width and height */
0,1,
/* detail and block pens */
NULL,
1* IDCMP flags */
NULL,
/* other window flags * /
NULL,
1* first gadget in gadget list */
NULL,
/* custom CHECKMARK imagery * /
"Your new window",
/* window title */
NULL,
1* custom screen */
NULL,
/* custom bitmap */
5,5,
1* minimum width and height */
640,200,
/* maximum width and height */
WBENCHSCREEN
/* destination screen type */
I;
/* end of PowerWindows source generation */
3.8.4.4
Fine-tuning the source text

The source text created by PowerWindows is very long, granted, but
this is the way PowerWindows works. Unfortunately you can't tell the
program to generate a structure with all its data in one line.
Let's close this section by looking at how you can further process the
source text. In addition to the menu strip, PowerWindows creates a
Newwindow structure for us.
It always defines an Int ui Text structure together with a Menultem
structure. These are numbered for some inexplicable reasons. If you
would like to add to the program, you must change the numbers.
Unfortunately, these numbers don't give much information. It would be
better to have the names of the menu items there. You can add these to
make the insertion of new menu items easier.
The structures of the program particularly stress the importance of
having menu null arguments. Unfortunately, you may forget which
meaning refers to what. Also, you must change the structure coordinate
choices to make room.
The graphics must be inserted in the first submenu item of the third
menu (Graphic). Remove the IntuiText structures and insert the
Image structures (remember to remove the ITEMNEXT flag).
Note:
We advise that you make as few changes as possible, since it is easy to

insert an error-and this can be very hard to find later on. The simplest
method is to print out the entire program text and then mark the
changed places with a marker. If the operation is successful, you can
335
3.
INTUITION AND
AMIGA
make the new changes. But always do it between compiling. so you

have the results before your eyes.
336
3.9 THE CONSOLE DEVICE
ABACUS
3.9
The Console Device

The IDCMP sometimes only allows the receiving of data. You can
select which data should be transmitted, but the receiving of data is
limited.
The IDCMP uses two methods to read the keyboard. RAWKEY always
sends the keyboard code without changing it to the keyboard table form
(that's why the word RAW in RAWKEY). VANILLAKEY translates the
keyboard code according to the current keyboard table. This gives us the
power to change the keyboard to another country's keyboard layout
We still have a problem-no data can be sent. We must rely on the
console.device to make this possible. The console.device is a device
with which we can exchange data. It receives data from the IDCMP
through RAWKEY as well as VANILLAKEY. In addition, the console
device processes output. It displays text according to the window's size.
This ensures that no word is cut off at the end of a line. The
console.device also scrolls the text display, processes text graphically
and creates messages under certain conditions.
3.9.1
Designing Communication Direction

We need a read and write port for the console.device as we would for any
device. These ports then allow us to communicate using
IOStandardMessages. To set the ports and the standard
input/output, we need two functions that can act with Exec. The
Open_ AIIO routine calls CreatePortO and CreateStdIOO:
ConsoleWritePort = CreatePort("wgb-con-dev.write", OL);
if (!ConsoleWritePort)
{
printf("The new WritePort could not be enabled!\n");

Close_All () ;
exit (FALSE) ;
)
ConsoleWriteMsg = CreateStdIO(ConsoleWritePort);
if (!ConsoleWriteMsg)
(
printf("CreateStdIO for Write not enabled!\n");

Close_All();
exit (FALSE) ;
}
337
3.
INTUITION AND
AMIGA
The same general process holds true for reading data:

ConsoleReadPort = CreatePort("wgb-con-dev.read", OL);
if (!ConsoleReadPort)
{
printf ("The ReadPort could not be enabled! \n") ;

Close_All () ;
exit (FALSE) ;
I
ConsoleReadMsg = CreateStdIO(ConsoleReadPort);
if (!ConsoleReadMsg)
(
printf("CreateStdIO for Read not enabled!\n");

Close_All () ;
exit (FALSE) ;
I
Once these two data directions are enabled, the device can be opened.
Two fields must be initialized in the IOStdRequest structure. This
is unusual when opening a device, but the console device needs this
initialization:
ConsoleWriteMsg->io_Data
ConsoleWriteMsg->io_Length
conDevErr
(APTR) FirstWindow;
sizeof(*FirstWindow);
= Open Device ("Console.device", OL,

ConsoleWriteMsg, OL);
if (conDevErr)
(
printf("OpenDevice error!\n");
Close_All();
exit (FALSE) ;
I;
ConsoleReadMsg->io_Device
ConsoleReadMsg->io_Unit
= ConsoleWriteMsg->io_Device;
= ConsoleWriteMsg->io_Unit;
The ConsoleWri teMsg structure is attached to a window over which

the data exchange should go. After opening the window, the
ConsoleReadMsg structure receives the device and unit message.
Now the input and output can begin.
All of the keyboard input going to the specified window is first directed
to and processed by the console.device. The program then receives a
report that contains the translated keyboard codes from the
console.device.
If you want to send messages to the console.device, then you use the
ConsoleWri teMsg structure. This structure directs commands and
instructions to the device.
338
ABACUS
3.9.2
Receiving the First Data

We don't need to do anything extra in data direction to receive data.
We're ready to receive data from the keyboard. But how does it run?
The message must be sent to the console.device concerning a specific
amount of keyboard input. A StandardRequest structure sends this
instruction. The program doesn't wait until it receives an answer,
because other actions can occur within that time.
It would be simplest if you wrote a function for the command to read
keys. This function could have a name of QueueReadO
/***************************************
* Function:
Input over ConsoleDev
* =================================== *
comments:
* Author: Date:
*
*
*
---------24.10.1987 improved L&O
* Wgb
*
pointer to IOStdReq
* StdReq
*
Address of Text buffer
*
* buffer
Length of Text buffer
* length
*
***************************************/
----------
QueueRead(StdReq, buffer, Length)

struct IOStdReq *StdReq;
char
*buffer;
ULONG
length;
StdReq->io_Command = CMD_READ;
StdReq->io_Oata
= (APTR)buffer;
StdReq->io_Length = length;
SendIO(StdReq);
}
The function assigns the pointer to the IOStdReq structure which we

have decided to write to. You also assign the pointer to a buffer where
the received data will be stored later, and conclude with the buffer's
length (i.e., the number of characters that should be received):
QueueRead(consoleReadMsg, &Buffer[O], lL);
Now our loop can be on standby until GetMsg(ConsoleReadPort)

receives a response. Then a new reading routine is used:
if (GetMsg(consoleReadPort
Write (consoleWriteMag, &[0], lL);
339
3.
INTUITION AND
3.9.3
AMIGA
Displaying the characters

It would be better if we tested the characters for command codes before
output, but we'll take the easy way out and redisplay the characters on
the console.
You need a function that works basically like the read function: An
IOStdReq is filled, the write option is chosen, the text buffer
transfers, and the length is added. Then the message is sent The routine
waits until execution ends, to avoid any overlap:
1***************************************
* Functions: Output over Cons01eDev'
*
*
*
*
*
====:=~===~~~s====================
Author:
Wgb
* StdReq
* Zchar
Date:
23.10.1987
comments:
Basic L&D
*
*
'"
'"
'*"
Message port of the report *

Number of characters
*
'"
***************************************1
ConWrite(StdReq, Text, Length)
struct IOStdReq *StdReq:
char
Text[):
int
Length;
StdReq->io_Command
StdReq->io_Data
StdReq->io_Length
0010 (StdReq) ;
= CMD_WRITE;
= (APTR)Text;
= Length;
This function transfers the characters at the start of the buffer when
called. We can also display a string terminated by a null byte.
Here you use the length descriptor -IL. To test this, have the
intuition .library open a window and create a loop which
checks for a clicked close gadget. A console.device window opens
featuring complete input and output, resulting in a full window editor.
Check it out!
340
ABACUS
3.9.4
Console Device Control Sequences

Unlike IntuiText structures, console parameters may be changed,
enabled and disabled through the console.device. These control
sequences consist of simple character codes or longer character strings.
You transmit the necessary text to execute the control sequence.
The ftrst table of control characters below perform very simple tasks:
Command
BACKSPACE
UNEFEED
Character sequence Comments

Moves cursor one column left
Ox08
OxOA
Moves cursor one line down to
next line depending on the mode
function (usually to next line but
not to the fIrst column)
VERTICAL TAB
OxOB
Moves cursor one line up
FORM FEED
OxOC
Erases contents of window
CARRIAGE REfURN
Moves cursor to fIrst text column
OxOD
(but not one line down)
SHIFrIN
Disables SHIFT OUT (text mode)
OxOE
SHIFf our
OxOF
ESC
OxlB
CSI
Ox9B
Enables SHIFT OUT (graphic

mode-sets bit 7 of all character
codes)
Enables escape sequences (used
mostly with printed output)
control sequence introducerenables control sequences (used
mostly with the console device)
The last two characters mentioned above are very important. CSI
introduces a control sequence; <Esc> performs a similar task:
Label
Character Sequence Comments
Returns to original status
RESET
Esc Ox63
INSERT [N] CHARACTERS
Inserts [N] characters. If no [N]
CSI [N] Ox40
is given, N = 1
CURSOR UP [N] LINES
Moves cursor [N] lines up. If no
CSI [N] Ox41
[N] is given, N =1
CURSOR OOWN (N] LINES
Moves cursor [N] lines down. If
CSI [N] Ox42
no rNJ is ~iven, N = 1
CURSOR FORWARD
Moves cursor [N] characters to
CSI [N] Ox44
[N] CHARACfERS to the right.
If no [N] is given, N = 1
341
3. INTUITION AND C
AMIGA
Character Sequence
Label
CURSOR BACKWARD
CSI [N] Ox44
CURSOR NEXT LINE [N]

CSI [N] Ox45
Comments
Moves cursor [N] characters to
[N] CHARACTERS to the right.
If no [N] is given, N = 1
Moves the cursor to line [N]. If
no fNl is given N = 1
CURSOR PRECEDING
CSI [N] Ox46
Moves cursor to previous [N]

UNE [N] line. If no [N] is
given, N = 1
MOVE CURSOR TO CSI [N] (Ox3B [M]) Ox48
Moves cursor to line [N],
ROW [N] (COLUMN [M])
(column [M])
ERASE TO END OF
Erases window from
CSIOx4A
current DISPLA Y cursor
position to end
ERASE TO END OF LINE
CSIOx4B
Deletes window to end of line
INSERTUNE
CSIOx4C
Inserts a line
DEIEfEUNE
CSIOx4D
Deletes a line
DELETE CHARACTER [N]
CSI [N] OXSO
Deletes characters to right of
cursor
SCROLL UP [N] LINES
CSI [N] Ox53
Scrolls text [N] lines up
SCROLL DOWN [N] UNES
CSI [N] Ox54
Scrolls text [N] lines down
SET MODE
CSI Ox32 Ox30 Ox68
Changes linefeed to
carriage retumllinefeed
RESET MODE
CSI Ox32 Ox30 Ox6C
Changes linefeed to line feed
only (no added carriage return)
DEVICE SfATUS REPORT
CSI Ox36 Ox6E
Requests status report
Some applications allow combined control sequences. Multiple

sequences differ from single sequences in that they consist of multiple
groups of sequences. Another difference in multiple sequences: One
number performs the multiple functions. This number appears as an
ASCII number, rather than as a series of ASCII codes (the usual
method).
You can eliminate the use of numbers here. The console takes a
standard value given in the table. Many of the command sequences call
upon the console.device to return a message containing the requested
data (e.g., cursor position).
342
ABACUS
We suggest that you create an array from which you can recall a
sequence when you need it. This is because all of the command
sequences in the console.device consist of multiple characters. The
routine for reading the characters is sensitive to differences between
letters, numbers and control characters. Control characters are divided
into other groups. Here you need additional routines to read cursor keys,
function keys and the control sequences (which return to the user)
directed by the CSI. Here is an example of important control characters
and a command table:
fdefine
fdefine
fdefine
fdefine
#define
fdefine
#define
SPACE
BACKSPACE
LINEFEED
RETURN
CSI
DEVICE_STATUS_REPORT
WINDOW_STATUS_REQUEST
Ox20
Ox08
OxOA
OxOD
Ox9B
Ox36, Ox6E
Ox30, Ox20, Ox?1
fdefine
#define
#define
#define
CURSOR_UP
CURSOR DOWN
CURSOR_Shift_UP
CURSOR_Shift_DOWN
Ox41
Ox42
Ox54
Ox53
UBYTE Specialcharacter[]
{
SPACE,
BACKSPACE,
LINEFEED,
RETURN,
CSI, DEVICE_STATUS_REPORT,
CSI, WINDOW_STATUS_REQUEST
};
3.9.5
Reading Messages from the Console

The system of communication is almost ready. We can now transmit
and receive data over two ports. We know which "normal" characters
can be received and which control sequences are available. However, the
program still needs a few details before it can handle data.
Receiving simple keystrokes is no problem for the program. However,
messages consisting of multiple characters can make things more
complicated. Also there is danger that the program will misinterpret the
message or process the characters incorrectly.
We want to avoid these errors from the beginning and create a routine
that evaluates everything correctly. Since our ftrst concern here is with
reading, we should make sure that a large enough text buffer exists to
343
3.
INTUITION AND
AMIGA
handle the entire window. We need a reading loop that takes data from
our data buffer:
{
switch (Buffer[O])
{
1* Selection *1
QueueRead(consoleReadMsg, &Buffer[mode], 1L);
}
As always, the reading loop has a primitive assignment: See if data is
present, select the first character, select a new character, etc. The routine
treats the character as a letter or normal character that doesn't need
special handling. Then the following lines come into play:
default
ConWriteChr(consoleWriteMsg, &Buffer[O]);
Coords = FALSE;
WindowBuffer[y] [x-1] = Buffer[O];
if (WindowBuffer[y] [79] < x)
WindowBuffer[y] [79] = Xl
break;
The small processing routine above gets the first character. Then the
coordinate flag changes to FALSE, so the program knows that the
actual coordinates are no longer known. The characters are then
transferred to the window buffer. This stores all of the characters
presently in the window. Next the new character is tested to see if it
exceeds the current line length. If so, the new line length is transferred.
These methods were chosen for the following purposes: The window
buffer clears at the beginning, so that it contains only spaces (0x20).
When you enter <Return> in a line, this should be sent to DOS so the
eLI command can be executed. If all the spaces are transferred to DOS,
then the eLI cannot recognize where the command sequence ends. You
can't send a command because the eLI command word and arguments
cannot be transmitted with the added characters.
Imagine a line of command data. The line begins with the frrst character
that must be examined. Then the line in which the cursor is found
should be transmitted:
case RETURN
: ConWriteChr(consoleWriteMsg, &Buffer[O])
ConWriteChr(consoleWriteMsg, &Specialcharacter[2]);
Coords = FALSE;
if (WindowBuffer[yj (79) != 0)
{
WindowBuffer[y] [WindowBuffer[y] [79]]

Execute(&WindowBuffer[y] [0], 0, 0);
WindowBuffer[y] [WindowBuffer[y] [78]1
}
break;
344
= NULL;
= Ox20;
ABACUS
The second character reading routine processes the graphic first. This
means that the routine requires cursor placement at the beginning of the
line and in the next line. The coordinate flag is set again. A test
examines whether characters were entered in the line. If so, a null ends
the line so the ExecuteO function of DOS can be called. After that a
space ends the line.
The last character needing special attention is the <Backspace> key.
Neither the console.device task nor the buffer correction should execute:
case BACKSPACE: ConWriteChr(consoleWriteMsg,
&Buffer[O]);
ConWrite(consoleWriteMsg, &Specialcharacter[O], 2);
WindowBuffer[y][x-l] = SPACE;
Coords = FALSE;
break;
The last and most important problem (control sequence management) is

delayed. Look at the following short form:
case CSI
: mode += 1;
break
This read loop assigns a number that increments by one every time
further readings occur. From the number you can recognize how many
characters must still be processed. Multiple control sequences may be in
the buffer, because the CSI controls the function keys and cursor keys.
We will look at the same reader:
/*********************************
*
*
*
Handling of the control

sequences
*
*
*
*********************************/
else if (mode)
{
/* Check */
mode = 0;
QueueRead(consoleReadMsg, &Buffer[mode], lL);
}
The ELSE belongs to the IF statement that tests the console buffer for
a character from the device. As long as there is a message there, this
gets the data first from the buffer because the console.device only stores
the last 32 characters received. If more characters exist, the routine
simply ignores them. This is why the variable mode is used; to enter
the new character in the next location of our buffer.
Let's look at the handling of the first control sequence, which reports
the cursor position:
345
3.
INTUITION AND
AMIGA
/*********************************
* Control Sequence :
*
*
*
CURSOR POSITION REPORT
*
*
*
*
*
*********************************1
if (Buffer [mode-1] == 82)
{
Coords = TRUE;
x = 0; y = 0; i
1;
while (Buffer[i] < Ox3A & Ox2F < Buffer[i])

{
y *= 10;
y += Buffer[i]
- Ox30;
i ++;
}
++;
while (Buffer[i] < Ox3A & Ox2F < Buffer[i])
i
x *= 10;
x += Buffer[i] - Ox30;
i ++;
}
This program section goes through one character after another, and
calculates the X and Y coordinates of the cursor. The DEVICE STATUS
REPORT has the fonnat (note the ending R):
CSI Row; Line R
Next the routine evaluates the buffer character for character. The
calculation of the first value concludes when the program encounters a
semicolon. Then the X value can be calculated. It is also important that
the coordinate flag be reset. Later a test follows that sends the set flag
to the console as a report, that should include the coordinates in the data
direction. This could provide us with an endless communication system
consisting of the cursor coordinates only.
The next test shows that a control sequence is present (but not the
DEVICE STATUS REPORT). It is flrst checked to see if the cursor
keys were pressed. The console.device allows the use of <Shift> and
cursor keys. A <Shift>ed cursor key scrolls the entire window, instead
of simply changing the cursor position. This scrolling must also occur
in the text buffer.
Look at the following routine:
346
ABACUS
else
switch (Buffer[mode-l])
{
/*********************************
*
* SHIFTED CURSOR MOVE
*
*
*
*
*
*********************************/
case CURSOR Shift UP
case CURSOR Shift DOWN
Move_Down () ;
break;
Move_Up() ;
break;
/*********************************
*
*
* NORMAL CURSOR MOVE

*
*
*
*********************************/
case CURSOR_DOWN
if (y == 1) Move_Down();
break;
if (y == Ymax) Move_Up();
break;
The last two statements end with normal cursor key movements. If the
cursor has reached the top or bottom border of the screen, the screen
scrolls up or down by one line.
Two functions handle the scrolling of the window buffer. These
functions look like this:
/***************************************
*
* Functions: Text buffer processing
* ===================================
*
comments:
* Author: Date:
*
* Wgb
*
*
*
*
y
x
---------- ---------11/2/1987
Number of the line

Number of the column
*
*
*
*
*
*
*
*
*
*
***************************************1
347
3.
INTUITION AND
AMIGA
int y;
for(y=O; y<80; y ++)
Clear_Line(y);
Clear_Line (y)
int y;
{
int x;
for(x=O; x<79; x ++)
WindowBuffer[yl [xl = Ox20;
WindowBuffer[yl [78] = NULL;
f* Line end mark *1
WindowBuffer[y] [79] = NULL; f* Number of entered
characters *1
}
Move_Up 0
{
int x, y;
for(y=l; y<80; y ++)
for (x=O; x<80; x ++)
WindowBuffer[y-l] [xl
Clear_Line (79) ;
= WindowBuffer[y] [xl;
Move_Down ( )
{
int x, Yi
for(y=79; y>O; y --I
for(x=O; x<80; X ++)
WindowBuffer[y] [xl = WindowBuffer[y-l] [xl;
Clear_Line(O);
}
Two functions taken from above are of particular interest:

Clear_ BufferO and Clear_ LineO.
This function initializes the buffer at the beginning of the code. The
number of characters entered per line is set at zero, just as a zero
indicates the end of a line of text.
Clear_Line()
The Clear BufferO function calls this function. It allows the

deletion of seIected lines supported by control sequences.
It would take far too long to describe the handling and testing of every
control sequence in this chapter. However, the next routine is a generic
loop that assigns any routines that aren't handled to output status:
348
ABACUS
/*********************************
Control Sequence :
*Manipulation of any sequence
*********************************/
ConWrite(consoleWriteMsg, &Buffer[O], mode);
Coords = FALSE;
}
You can enter all of the additional control sequence functions, if you
want to. For example, you won't find a description of simple function
key assignment. It's actually very simple. Assuming that you want
these assignments permanent within the program, then you simply
establish a pointer array and add the desired texts. Here is an example for
our CLI editor:
UBYTE *Functionkeys[]
{
"dir"+RETURN,
"cd df1:"+RETURN,
"cd sys: "+RETURN,
"list"+RETURN,
"makedir ram:c"+RETURN+"copy sys:c ram:c"+RETORN,
"execute make "
};
The test for the function keys must identify these through "_" and the
number calculated. Then the pointer can be assigned to ConWr i teO;:
ConWrite(consoleWriteMasg, Functionkeys[Number], -1L);
3.9.6
Finishing the CL I Editor

When operating between Intuition, the editor window and the console,
we recommend that you open the console device through some other
method. We would like to describe that briefly here:
Although the console.device is built into the IDCMP, there are still
more options available. For total openness in manipulation we first
define an IOStdReq structure, in which we set a pointer to our
window in the value io Data. The device can be opened as usual.
Now it is connected to our Intuition window, and communication can
be carried over to the structure in use. They can be used to write as well
as read. The read and write functions that were developed at the
beginning of this chapter serve the same purpose as before.
349
3.
INTUITION AND
AMIGA
Set all of the reading elements, all of the routines to open and close,
and the text buffer manipulator (remember to define the important
variables and arrays) Now we have a neweLl editor!
Note:
The window sizing gadgets described earlier in this book can be added to
this editor. Here is a complete listing:
I*******~********************************
*
*
Program: New CLI-Ed workings
~-============
* Author:
* -----* Wgb
Date:
Comments:
10/20/1987
for testing only
*
*
*
*
*
*
*
*
****************************************1
struct Window
*FirstWindow;
struct IOStdReq *consoleWriteMsg, *conso1eReadMsg,
*CreateStdIO () ;
struct MsgPort *consoleWritePort. *conso1eReadPort.
*CreatePort () ;
struct NewWindow ConsoleWindow =
{
0, 0,
640, 101,
2, 3,
1*
1*
1*
1*
LeftEdge, TopEdge
Width, Height
DetailPen, BlockPen
IDCMP Flags
CLOSEWINDOW
GADGETUP,
WINDOWDEPTH
1* Flags
WINDOWSIZING I
WINDOWDRAG I
WINDOWCLOSE I
ACTIVATE I
SMART_REFRESH,
NULL,
1* First Gadget
NULL,
1* CheckMark
(UBYTE *) "Wgb Prod. presents BECKERshell",
NULL,
1* Screen
NULL,
1* BitMap
640, 50,
640, 200,
WBENCHSCREEN,
1* Type
};
350
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
*1
ABACUS
fldefine
#define
#define
#define
#define
#define
#define
SPACE
BACKSPACE
LINEFEED
RETURN
CSI
DEVICE_STATUS_REPORT
WINDOW_STATUS_REQUEST
Ox20
Ox08
OxOa
OxOd
Ox9b
Ox36, Ox6e
Ox30, Ox20, Ox71
#define
#define
#define
#define
CURSOR_UP
CURSOR_DOWN
CURSOR_Shift_UP
CURSOR_Shift_DOWN
Ox41
Ox42
Ox54
Ox53
UBYTE Specialcharacter[]
{
SPACE,
BACKSPACE,
LINEFEED,
RETURN,
CSI, DEVICE_STATUS~REPORT,
CSI, WINDOW_STATUS_REQUEST
};
SHORT conDevErr = FALSE;
UBYTE WindowBuffer[80] [80];
UBYTE Buffer[80];
VOID mainO
{
ULONG MessageClass;
USHORT code;
int i;
int x, y;
int Xmax = 77, Ymax
int mode = 0;
BOOL Coords;
= 11;

Open All
0;
Clear_Buffer 0 ;
ConWrite(consoleWriteMsg, &Specialcharacter[4], 3);
QueueRead(consoleReadMsg, &Buffer[Oj, 1L);
FOREVER
{

(
MessageClass
= message->C1ass;
351
3. INTUITION AND C
AMIGA
code = message->Code:
ReplyMsg(message):
{
case CLOSEWINDOW : Close_AII():

exit(TRUE):
break:
{
switch (Buffer[D])
(
case BACKSPACE
ConWriteChr(consoleWriteMsg,
&Buffer[D]) :
ConWrite(consoleWriteMsg,
&Specialcharacter[D], 2):
Coords
break:
case RETURN
FALSE:
&Buffer[D]) :
&Specialcharacter[2]):
Coords = FALSE:
if (WindowBuffer[y] [79]
!=
0)
WindowBuffer[y] [WindowBuffer[y] [79]] = NULL:

Execute (&WindowBuffer[y] [0], 0,
0):
WindowBuffer[y] [WindowBuffer[y] [79]]
Dx2D:
break:
case CST
mode += I:
break;
default
&Buffer[D]) ;
Coords = FALSE:
WindowBuffer[y] [x-I]
= Buffer[D);
if (WindowBuffer[y] [79j < x)

WindowBuffer[y) [79) = x:
break:
QueueRead(consoleReadMsg, &Buffer(mode), lL);
}
1*********************************
*
*
352
Handle the Control-
*
*
ABACUS
sequence
*
*
*
*
*********************************1
else i f (mode)
{
1*********************************
*
*
*
*
*
Control Sequence :
CURSOR POSITION REPORT
*
*
*
*
*********************************1
if (Buffer [mode-1]
==
82)
Coords = TRUE;
x = 0; y = 0; i
1;
while (Buffer[i] < Ox3a & Ox2f < Buffer[i])

{
*= 10;
Y += Buffer[i] - Ox30;
i ++;
}
i ++;
while (Buffer[i] < Ox3a & Ox2f < Buffer[i])

x *= 10;
x += Buffer[i] - Ox30;
i ++;
else
switch (Buffer[mode-1])
1*********************************
*
*
*
*
*
Control Sequence :
SHIFTED CURSOR MOVE
*
*
*
*
*********************************1
case CURSOR- Shift- UP
case CURSOR- Shift- DOWN
Move_Down () ;
break;
Move_Up() ;
break;
1*********************************
**
*
Control Sequence :
*
*
*
353
3.
INTUITION AND
AMIGA
*
*
NORMAL CURSOR MOVE
*********************************/
case CURSOR UP
if (y == 1) Move_Down!);
case CURSOR_DOWN
i f (y == Ymax)
break;
break;
/*********************************
*
*
*
*
*
Control Sequence :
Manipulation of any sequence
*
*
*********************************/
ConWrite(consoleWriteMsg, &Buffer[O], mode);
Coords = FALSE;
/*********************************
*
*
*
*
* Console Device :
New character read
*********************************/
mode = 0;
QueueRead(consoleReadMsg, &Buffer[mode], 1L);
}
/*********************************
* Console Device
*
*
*
* New cursor*
* position
*
*
*
*********************************/
if (Coords
==
FALSE)
ConWrite(consoleWriteMsg, &Specialcharacter[4], 3);

Coords
354
= TRUE;
ABACUS
/***************************************
* Function: open all
*-
** Author:
Date:
comments:
* ----* Wgb
10/20/1987
* no parameter
*
*
*
*
*
*
*
*
*
*
***************************************/

{

Close_All () ;
exit (FALSE);
}

OpenWindow(&ConsoleWindow)
{

Close_All () ;
exit (FALSE) ;
}
consoleWritePort = CreatePort("wgb-con-dev.write", OL);

if (!consoleWritePort)
{
printf("The new WritePort not enabled!\n");

Close_All () ;
exit (FALSE);
}
consoleWriteMsg = CreateStdIO(consoleWritePort);
if (!consoleWriteMsg)
(
printf("CreateStdIO for Write not enabled!\n");

Close_All ();
exit (FALSE) ;
}
355
3.
INTUITION AND
AMIGA
consoleReadPort = CreatePort("wgb-con-dev.read", OL):

if (!consoleReadPort)
{
printf("The new ReadPort not enabled!\n"):

Close_All () :
exit (FALSE):
}
consoleReadMsg = CreateStdIO(consoleReadPort);
if (!consoleReadMsg)
{
printf ("CreateStdIO for Read not enabled! \n");

Close_All ();
exit (FALSE);
}
consoleWriteMsg->io_Data
(APTR) FirstWindow:
consoleWriteMsg->io_Length = sizeof(*FirstWindow):
conDevErr = OpenDevice ("console. device", OL, consoleWriteMsg,
OL);
if (conDevErr)
{
print f ("OpenDevice error! \n") :

Close_All () :
exit (FALSE) ;
};
consoleReadMsg->io_Device
= consoleWriteMsg->io_Device;
consoleReadMsg->io_Unit
consoleWriteMsg->io_Unit;
1***************************************
*
*
*
Function: Close All

==================
*
Comments:
* Author: Date:
---------- ---------*
*
Wgb
*
*
*
No Parameter
20.10.1987
**
*
*
*
*
*
*
***************************************1
i f (!conDevErr)
CloseDevice(consoleWriteMsg):
356
ABACUS
if (consoleWriteMsg)
DeleteStdIO(consoleWriteMsg);
if (consoleReadMsg)
DeleteStdIO(consoleReadMsg);
if (consoleWritePort)
DeletePort(consoleWritePort);
if (consoleReadPort)
DeletePort(consoleReadPort);
if (FirstWindow)
if (IntuitionBase)
1***************************************
*
*
* Function: Output with the ConsoleDev*
*======
** Author: Date:
Comments:
* ------ ---------- ---------23.10.1987
* Wgb
* request Message Port
*
Number of the character
* Zchar
*
*
*
*
*
***************************************1
ConWriteChr(request, Zchar)
struct IOStdReq *request;
char
Zchar[];
ConWrite(request, Zchar, 1);
}
1*******************************
ConWriteString(request, Text)
char
Text[];
ConWrite(request, Text, -1);
}
********************************1
357
3.
INTUITION AND
AMIGA
ConWrite(request, Text, Length)

char
Text[];
int
Length:
request->io_Command = CMD_WRlTE;
request->io_Data
= (APTR)Text;
request->io_Length = Length;
DoIO(request):
}
1***************************************
*
*
*
*
* Function: input with the ConsoleDev *

=================
*
Comments:
* Author: Date:
*
---------- -----* Wgb
24.10.1987
*
*
*
*
request
whereto
Length
Message Port of report

Address of text buffers
Length of Text buffers
*
*
*
*
*
*
*
***************************************1
QueueRead(request, whereto, Length)
char
*whereto;
ULONG
Length;
request->io_Command = CMD_READ;
request->io_Data
= (APTR)whereto;
request->io_Length = Length;
SendIO(request);
}
1************************************************
int ConMayGetChr(conso1ePort, request, whereto)
struct Port
*conso1ePort:
struct IOStdReq *request:
char
*whereto;
register temp:
358
ABACUS
if (GetMsg(consolePort) == NULL)
return(FALSE);
temp = *whereto;
QueueRead(request, whereto, lL);
return (temp) ;
}
UBYTE ConGetChr(consolePort, request, whereto)

struct Port
*consolePort;
char
*whereto;
register temp;
while (GetMsg(consolePort) == NULL)
WaitPort(consolePort);
temp = *whereto;
QueueRead(request, whereto, lL);
return (temp);
}
*************************************************1
1***************************************
*
*
* Function: Text buffer positioning

================-~==
*
*
*
*
* Author: Date:
Comments:
*
---------- ---------11/2/1987
* Wgb
*
* y Number of the line
* x Number of the column
*
*
*
*
*
*
***************************************1
Clear_Buffer 0
{
int y;
for(y=O; y<80; y ++)
Clear_Line(y);
Clear_Line(y)
int y;
{
int x;
for(x=O; x<79; x ++)
359
3.
INTUITION AND
AMIGA
WindowBuffer[y] [x] = 0x20;

WindowBuffer[y] [78] = NULL; 1* End adf line marker *1
WindowBuffer[y] [79] = NULL; 1* Number of entered characters *1
}
Move_Up()
{
int x, y;
for(y=l; y<80; y ++)
for(x=O; x<80; x ++)
WindowBuffer[y-l] [x] = WindowBuffer[y] [x];
Clear_Line (79) ;
}
Move_Down ()
{
int x, y;
for(y=79; y>O; y --)
for(x=O; x<80; x ++)
WindowBuffer[y] [x) = WindowBuffer[y-l] [x];
Clear_Line (0);
}
360
3.10
ABACUS
3.10
USER-DEFINED KEYBOARD TABLES
User-Defined Keyboard Tables

We have already said that the console.device converts the RAWKEY
codes into characters. Now we'll look at which assignments, commands
or definitions are allowed by this conversion.
A table helps translate the RAWKEY codes to the proper characters. The
table indicates the original key and the character to which it should be
converted.
This keyboard table also contains sections for keys pressed together
with the <Shift>, <Alt> <Ctrl> and/or <CapsLock> keys. The
<CapsLock> codes, for example, affect uppercase letters only, but not
numbers or special characters.
Keyboard tables also allow you to assign text strings of normal or
special characters to a single key.
3.10.1
Keyboard Table Design

We now know that the translation of the RAWKEY codes occurs through
keyboard tables. We should next examine the structure of one of these
tables, and which characters such a table should contain.
The term keyboard table refers to the large table, as well as the smaller
subtables which comprise the large table.
The KeyMap structure states the contents of the subtables. It contains a
pointer to each subtable.
struct KeyMap
{
OxOO
Ox04
Ox08
OxOC
OxlO
Ox14
Ox18
OxIC
Ox20
};
00
04
08
12
16
20
24
28
32
UBYTE
ULONG
UBYTE
UBYTE
UBYTE
ULONG
UBYTE
UBYTE
*km_LoKeyMapTypes;
*km_LoKeyMap;
*km_LoCapsable;
*km_LoRepeatable;
*km_HiKeyMapTypes;
*km_HiKeyMap;
*km_HiCapsable;
*km_HiRepeatable;
361
3.
INTUITION AND
AMIGA
You can see that the KeyMap structure actually has two separate
sections-a Lo section and a Hi section.
The Lo section contains infonnation about characters with RAWKEY
codes from OxOO to Ox3F. The Hi section contains infonnation about
characters with RAWKEY codes from Ox40 to Ox6. This division exists
to keep the keyboard open for more keys which can be entered into the
Hi section. This makes enhanced keyboards possible.
Examine the RAWKEY diagram below, with which you can get the code
for any key:
Amiga 1000 keyboard

,
. . .
, .
. ...
3D
lE
IF
2D
2E
2F
lD
lE
lF
4A
- . '"
Amiga 500/2000 keyboard
1'1
J~
18
, Ho.e
lE
20
2E
....
3A
01
I'qUp -
3D
. ....
JF
4A
2F
OC
lEnd'IPqOn
10
IE,
IF
"
1-:----'--=+-"::'001'" ;
OF
lC
r 43
Figure 3.13
Let's look at each table:
km_LoKeyMapTypes
km_HiKeyMapTypes
These two tables contain one byte for every key. This byte lists the flag
infonnation described below. Unfortunately it's almost impossible to
362
3.10 USER-DEFINED KEYBOARD TABLES
ABACUS
list the combinations for every key. These combinations include nonnal
keys, <Shift>ed keys, <Ctrl> combinations and <Alt> combinations.
Because the key combinations from two qualifiers, the <Shift>, <Alt>
and the <Ctrl> key, are considered, only four characters can be used per
key. You can test to see which qualifier releases which characters. Or
you could decide upon an entire string for the key, but then you should
take into consideration that only two arrangements are allowable for the
key.
km_LoKeyMap
km_HiKeyMap
The ULONG values of these tables contain the characters that should
appear when a key is pressed. Since the ULONG value consists of four
bytes, each byte has the character code belonging to one of the four
combinations. This is because the tables handle the character as a
string. Then the ULONG value represents a pointer to the string.
km_LoCapsable
km_HiCapsable
These two tables contain no displayable character codes. We find only

set and unset bits for each RAWKEY code in these tables. A set bit
indicates that the <CapsLock> key is active, i.e., this key should be
handled like a <Shift>ed key. An unset bit displays the key in nonnal
fonn.
km_LoRepeatable
km_HiRepeatable
These two tables work much like the Capsable tables. Here again we
have a bit for each RAWKEY code. A set bit indicates key repetition
according to values set by Preferences. An unset bit disables key repeat
3.10.2
Working with the Keyboard Tables

When you want to take control of the keyboard tables, you must know
the current status of the tables. This task involves the console.device.
You can only work with the keyboard tables if the console. device is
active.
There are commands that send infonnation about the current status to
the console.device. It is possible to insert the pointer to a new keyboard
table. Unfortunately, the two operations don't execute through library
controlled functions. The entries are transmitted entirely through
IOStdReq. That is why we recommend that you write some of your
own functions.
363
3.
INTUITION AND
The CD ASKKEYMAP function gets the current KeyMap arrangement

We need the size and starting address of our KeyMap structure, into
which the data should be placed. Then we can call the request
AskKeyMap(ConWriteRequest, OwnKeyMap)
struct IOStdReq *ConWriteRequest;
struct KeyMap
*OwnKeyMap;
BYTE Error;
ConWriteRequest->io_Command - CD_ASKKEYMAP;
ConWriteRequest->io_Lenght - sizeof(struct KeyMap);
ConWriteRequest->io_Data
- &OwnKeyMap;
DoIO(ConWriteRequest);
Error
= ConWriteRequest->io_Error;
i f (Error)
return(FALSE);
else
re.turn (TRUE);
The pointer to the active keyboard tables appears in the OwnKeyMap

structure after executing the above function. This can be examined and
changed. With these methods you can, for example, change a single
character on the keyboard.
This method is enough for one or two characters, but you need
something more powerful when you have a very unusual keyboard
arrangement that cannot be activated with a few changes, or when you
want to load an entire table. Here's the alternative process. First the
keyboard values are stored in memory, either within the program text or
by loading. Then the program creates the KeyMap structure. This
structure is the primary point from which the console.device gets its
information. Then the new KeyMap must be transferred to the
console.device:
Set KeyMap (ConWriteRequest, OwnKeyMap)
struct KeyMap
*OwnKeyMap;
BYTE Error;
ConWriteRequest->io_Command = CD_SETKEYMAP;
ConWriteRequest->io_Length = sizeof(struct KeyMap);
= &OwnKeyMapi
Error = ConWriteRequest->io_Error;
364
ABACUS
if (Error)
return(FALSE);
else
return(TRUE);
This is almost the same function as listed previously. The single

difference is the different function call. If you want to make one
function perform both tasks, all you need is one parameter to extend the
function call:
DoKeyMap(WriteRequest, KeyMapPointer, CD_SETKEYMAP)
DoKeyMap(ConWriteRequest, OwnKeyMap, Mode)
struct KeyMap
*OwnKeyMap;
awORD
Mode;
BYTE Error;
ConWriteRequest->io_Command = Mode;
ConWriteRequest->io_Length = sizeof(struct KeyMap);
= &OwnKeyMap;
Error
= ConWriteRequest->io_Error;
if (Error)
return(FALSE);
else
return(TRUE);
This last function lets you address two more functions named
CD ASKDEFAULTKEYMAPandCD SETDEFAULTKEYMAP.Both
work exactly as the names suggest. These functions let us control the
default KeyMap (the KeyMap made available when the console.device
opens).
This KeyMap contains all of the characters divided into the groups Lo
and Hi. The Lo group is handled as follows:
Any keys pressed by themselves (i.e., no <Shift>, <Ctrl> or

<Alt create their normal ASCII codes.
Any keys pressed with the <Shift> key create their shifted ASCII
codes.
Any keys pressed with the <Alt> key create their ASCII codes
with the highest bit set (Ox80).
365
3.
INTUITION AND
AMIGA
Any keys pressed with the <Alt> and <Shift> keys create their
shifted ASCII codes with the highest bit set (Ox80).
Any keys pressed with the <Ctrl> key create their ASCII codes
with bits 5 and 6 deleted.
The Hi group contains only control codes that don't consist of

individual codes: We find only strings here. That is also the reason why
the control keys can be placed two in a group at most.
Keys without any qualifier
K~y
Tran::!Iat~ val!J~
<Backspace> OxOS
<Enter>
OxOD
Ox7F
<Del>
Keys with one qualifier
withQ!Jt with
Ox20
OxAO
OxOD
OxOA
OxlB Ox9B
Ox2D
OxFF
K~y
<Space>
<Return>
<Esc>
Qualifier
Alt
Ctrl
Alt
Alt
Keys of High-KeyMap
366
K~y
Vi.lI!J~
<Tab>
Ox09
Valu~ with Shift

Ox9B "Z"
Up
Down
Forward
Backward
Ox9B "A"
Ox9B "B"
Ox9B "C"
Ox9B "D"
Ox9B "T"
Ox9B "S"
Ox9B "@"
Ox9B" A"
Fl
F2
F3
F4
F5
F6
F7
FS
F9
FlO
Ox9B "0-"
Ox9B "1-"
Ox9B "2-"
Ox9B "3-"
Ox9B "4-"
Ox9B "5-"
Ox9B "6_"
Ox9B "7_"
Ox9B "S-"
Ox9B "9_"
Ox9B "10-"
Ox9B "11-"
Ox9B "12-"
Ox9B "13-"
Ox9B "14-"
Ox9B "15-"
Ox9B "16-"
Ox9B "17-"
Ox9B II IS- "
Ox9B "19- II
HELP
Ox9B "?_"
ABACUS
3.10.3
Creating Your Own Keyboard Table

You already know that the keyboard table is actually a group of smaller
tables. We'll now explain how you can create your own table, and what
to watch out for when making it
lCeyMap
Types
We first need the two table types. These contain, as explained above, a
check to see if it is a string or four single characters, and the qualifiers
which access every key. The following flags are possible:
KeyMap Types:
Nam~
:YaIJJ~
:D~S~[h21jQn
KC_NOQUAL
KC VANILLA
OL
7L
No qualifier
Standard value with <Ctrl>
KCF SHIFT
KCF ALT
KCF CONTROL
OxOlL
Ox02L
Ox04L
<Shift> key pressed

<Alt> key pressed
<Ctrl> key pressed
KCF DOWNUP
KCF DEAD
Ox08L
Ox20L
Reacts when key is released

No key reaction
KCF_STRING
Ox40L
Key displays a string
We can only reach different arrangements by choosing the type. The

four bytes have the following meanings:
KeyMap Type combinations:
Qualifier combination
KC DEAD
KCYOQUAL
KC SHIFT
KC ALT
KC CONTROL
KC ALT + KC SHIFT
KC CONTROL + KC ALT
KC CONTROL + KC SHIFT
KC VANILLA
KC STRING
Byte 1
Byte 2 Byte 3
Shift
Alt
Ctrl
Shift+Alt Alt Shift
Ctrl+Alt Ctrl Alt
Ctrl+Shift Ctrl Shift
Shift+Alt Alt Shift
(etrl also accessible)
Poiater to string
Byte 4
without
without
without
without
without
without
without
without
The option of having any table arrangement should be available in your

programs. The most frequently used flag is KC_ VANI LLA because the
<Shift> and <Alt> keys are integrated. <Ctr!> is also supported, with
bits 5 and 6 of the NOQUAL value unset
367
3.
INTUITION AND
KeyMap
AMIGA
A Lo or Hi KeyMap can be easily put together. C programmers should

do this using assembler, because the problem is easier to solve there.
You can easily complete a table divided into four groups. Each group
has a key equivalent assigned according to the above table. Every byte
contains a character that is displayed according to the key combination.
When the key assignment is handled as a string, the system replaces the
group of four bytes with a LONG value that represents the pointer to
our string. Unfortunately, not every string is displayed the same. The
pointer frrst points to another table that contains the value for this key.
The string length appears fIrst, which should appear with a normal key
press. Next an offset appears pointing from the beginning of the string
table to the string itself.
We fInd the string length listed as another value that should be given as
a <Shift>ed key press. The offset appears at the beginning of the table.
Caps able
Like a typewriter, the Amiga keyboard also has a <CapsLock> key that
acts as a continuous <Shift>. Because the keyboard has a keyboard
processor, it isn't just a matter of simulating the continuous <Shift>.
The processor executes its own form of <Shift>. The <CapsLock> key
tests to see which keys are usable and which are nOL This method has
the advantage that punctuation remains unshifted, and can be entered as
normal without having to release the <CapsLock> key.
We have 8 bytes with 64 bits for all of the keys. Each of these bits
represents a key:
Bit 0 Byte 0 RAWCODE OxOO, Bit 1 Byte 0 RAWCODE OxOl
Bit 0 Byte 0 RAWCODE Ox08 Bit 7 Byte 7 RAWCODE Ox4f.
Set all of the bits for those keys which you want <Shift>ed.
Repeatabl.e
The same method available for the Capsable table is used for the
Repeatable table. Each set bit indicates that the key should be
repeated according to the repetition time and frequency set in
Preferences.
Many keys can be excluded from the repeat function. That's important if
a key doesn't have to repeat, or if repeating could cause trouble. Usually
the <Return> key does not repeaL
368
3.11
ABACUS
3.11
MEMORY MANAGEMENT
Memory Management
Some programs may require access to a certain area of memory. The
simplest memory forms are the variables. Here you control what is
needed for program formation. Arrays are based upon these variable
forms. But you may have heard quite a bit about how limiting these can
be.
Here's where the problem arises. The program could need a variable
array double the size of the number expected for program execution.
This could have a dimension of 1000*1000 for the color values of a
high resolution graphic. Our compiler reserves a doubled variable in the
finished program 100000 times. You can imagine how much memory
that requires.
We would like to show you a short example which accesses memory
areas from the system and frees them after use. Three methods of
memory organization are demonstrated. The example itself is chosen
from Intuition. We will try to assign four text buffers for the
corresponding string gadgets.
3.11.1
Memory Organization of the Amiga

Before we can write a program, you should have a working knowledge
of the Amiga's memory layout. The Amiga has a twofold dynamic
memory management.
This table lists the memory arrangement:
Address
Arrangement
OXOOOOOOOO Chip RAM
Comments
Copy of the Kickstart ROM
Third copy of Chip RAM
Fast RAM
Ox08000000
Ox20000000 8 MByte
OxAOOOOOOO CIAs
OXCOOOOOOO 512K expansion (Amiga 500 & 2000 only)
OxC8000000 Unused
(Amiga 500 & 2000 only)
OxDCOOOOOO Real time clock
OxDFOOOOOO Custom chips
OxEOOOOOOO Unused
OxE8000000 Expansion slots
OxFOOOOOOO ROM module
0xF8000000 Kickstart ROM copy
OxFCOOOOOO Kickstart ROM
369
3.
INTUITION AND
AMIGA
The entire memory is divided into two regions. The first 512K is chip
RAM, which means it can be addressed from the custom chips. The
second region consists of fast RAM or ROM, and can be addressed from
the 68000.
The word "dynamic" comes from the fact that the 8MB fast RAM
doesn't exists in the basic Amiga: Memory can be expanded
dynamically.
When we want our program to request an area of memory from the
system, then we must specify whether it should be chip RAM or fast
RAM. If we give no specification, fast RAM is used first. If this
doesn't succeed, the memory is is taken from chip RAM.
Note:
Some older programs may cause incompatibility problems with

memory expansion. Without memory expansion you are left with only
chip RAM. This memory can also be addressed from the custom chips.
Adding memory expansion adds fast RAM to the system. The system
may crash when older application programs attempt access to the
memory in this region.
3.11.2
First Steps with AllocMemO

Our first try at memory control uses an Exec function named
AllocMemO. This function allocates a memory area according to
specific size and criteria. You assign two arguments and receive the
address of the added memory in return. If no memory can be found with
the specified size and attributes, the system returns a null pointer
(error).
MemoryBlock = AllocMem(ByteSize, Requirements);
DO
-198
DO
01
The flags MEMF FAST and MEMF CHIP describe the attributes of the
memory area. MEMF_PUBLIC indicates memory usable for different
tasks. The MEMF_PUBLIC flag is currently unimplemented. However~
you should use it anyway to maintain compatibility with later Amiga
systems. MEMF_CLEAR clears the area of memory (Le., fills the area
with null bytes)., which means it should be filled with zeros. The
memory area must be a minimum of eight bytes long.
The following program reserves memory for four text buffers:
/***************************************
*
* Memory Management : AllocateMemory
* ===================================
*
370
*
*
*
*
3.11 MEMORY MANAGEMENT
ABACUS
* Author: Date:
*
---------10/16/1987
* Wgb
*
comments:
----------
only for test

purposes
*
*
*
*
***************************************/
iinclude <exec/memory.h>
tdefine MemoryType MEMF CHIP
MEMF CLEAR
OBYTE *UndoBuffer, *FileBuffer, *DiskBuffer,

*SuffBuffer, *AllocMem();
main ()
{
UndoBuffer ~ AllocMem(512L, MemoryType);

if (!UndoBuffer)
{
printf("Problems with the Ondo buffer!\n");

FreeMemory();
exit (FALSE) ;
}
FileBuffer
AllocMem(30L, MemoryType);
i f (!FileBuffer)
{
printf("Problems with the FileBuffer!\n");

FreeMemory();
exit(FALSE);
}
DiskBuffer
= AllocMem(512L,
MemoryType);
i f (!DiskBuffer)
{
printf("Problems with the DiskBuffer!\n");

FreeMemory();
exit(FALSE);
}
SuffBuffer = AllocMem(lOL, MemoryType);

if (!SuffBuffer)
{
printf("Problems with the SuffBuffer!\n");

FreeMemory();
exit(FALSE);
}
printf("Memory reserved!\n");
FreeMemory();
371
3.
INTUITION AND
exit(TRUE);
}
FreeMemory ()
{
if
if
if
if
(UndoBuffer)
(FileBuffer)
(DiskBuffer)
(SuffBuffer)
FreeMem(UndoBuffer,
FreeMem(FileBuffer,
FreeMem(DiskBuffer,
FreeMem(SuffBuffer,
512L);
30L);
512L);
10L);
printf("Memory free again!\n");

}
Program
Description
The include file <exec/memory.h> contains the memory type

definition. We link the two files together and then define a label. Then
we need the four pointers to the memory areas. The function is supplied
with a value for the required memory.
The main program tries to allocate the memory areas. If this doesn't
work it branches to the free routine and the program stops. The free
routine works similar to the Close AllO function. Every pointer is
checked to see if it points to a memory block. After this is done the
memory will be freed. This avoids the error that occurs when memory
is freed that was not allocated.
To release the memory we use another Exec function:
FreeMem(MemoryBlock, ByteSize);
-210
A1
DO
The disadvantage of this method is that we must free every memory

area, and know the memory size. That is unnecessary work for the
programmer in both computation time and program size.
3.11.3
Improvement using AllocEnt ryO

Often multiple memory areas have different needs. The AllocMemO
function is usable for this, but not a very good solution. A structure
would be helpful. This structure contains all of the memory areas with
their features. We then assign the pointer to AllocEntryO and it
attempts to allocate all of the memory areas.
The basic structure is named MemList:
struct MemList
{
Oxoo
OxOE
372
00
14
struct Node ml_Node;

UWORD ml_NumEntries;
ABACUS
OxlO
Ox18
};
16
24
struct MemEntry ml_ME[l);
The old node is needed at the beginning, as usual. The

ml NumEntries function states the number of entries on the
memory list. The second MemEn try structure gives additional
information to the fIrst memory block.

struct MemEntry
{
union
ULONG meu_Reqs;
APTR meu_Addr;
}
Oxoo
Ox04
me_Un;
04 ULONG me_Length;
Ox08
};
08
DefInition:
me_un me_Un
me_Reqs me_Un.meu_Reqs
me_Addr me_Un.meu_Addr
The value me Un contains the defInition of the features of this memory

block at the beginning. The starting address appears here according to
the allotment. The length of the memory area is reserved in
me_Length.
From these two structures, MemList and MemEntry, we must

assemble a structure before we can use AllocEntryO. This can look
something like this:
struct Memoryneeds
{
struct MemList Head;

struct MemEntry UndoBuffer;
struct MemEntry FileBuffer;
struct MemEntry DiskBuffer;
struct MemEntry SuffBuffer;
} Memory;
This self-contained structure is filled with the desired values. Then

AllocEntryO can be called. It tries to get the memory, and we fInd
the data in the individual MernEntry structures.
MemList
DO
AllocEntry(Entry);
-222
AO
FreeEntry(Entry);
-228
AO
373
3.
INTUITION AND
AMIGA
Here is the complete listing:

/**************************************
*
*
: MemoryEntries *
* ================================== *
* Memory management
*
*
Comments:
*
* Author: Date:
---------- ---------*
*
10/16/1987 only for test *
* Wgb
purposes
*
*
*
*
**************************************/
#include <exec/memory.h>
#define MemoryType MEMF_FAST
struct MemList *MemoryPtr, *AllocEntry();
APTR UndoData, FileData, DiskData, SuffData;
struct Memoryneeds
{
struct MemList Head;

struct MemEntry UndoBuffer;
struct MemEntry FileBuffer;
struct MemEntry DiskBuffer;
struct MemEntry SuffBuffer:
} Memory;
maine)
{
Memory.Head.ml_NumEntries
= 5:
Memory.Head.ml_me[O].me_Length = 0;
Memory.UndoBuffer.me_Reqs
= MemoryType;
Memory.UndoBuffer.me_Length
512L;
Memory.FileBuffer.me_Reqs
= MemoryType;
Memory.FileBuffer.me_Length = 30L;
Memory.DiskBuffer.me_Reqs
= MemoryType;
Memory.DiskBuffer.me_Length
512L;
Memory.SuffBuffer.me_Reqs
MemoryType;
Memory.SuffBuffer.me_Length
10L:
MemoryPtr
(struct MemList *)AllocEntry(&Memory):
if ULONG)MemoryPtr & (131

{
printf("The list cannot be posted!\n");

exit (FALSE) :
}
else
374
ABACUS
UndoData = MemoryPtr->ml_me[l).me_Addr;
FileData = MemoryPtr->ml me[2) .me Addr;
DiskData = Memoryptr->ml=me[31.me=Addr;
SuffData = MemoryPtr->ml_me[4].me_Addr;
}
printf ("Memory allocated! \n");

FreeEntry(MemoryPtr);
printf("Memory is free again!\n");
exit(TRUE);
}
3.11.4
The Final Solution

The two preceding functions are from the Exec library. Since this
chapter is about C and Intuition, we've saved the memory management
functions supported by Intuition until now.
The two functions are called All 0 c Remembe r 0 and
FreeRememberO. As the names suggest, Intuition helps us by
marking the reserved areas. We assign the function the size of the
desired memory area, the memory type and the pointer to a Remember
structure. Intuition either returns an error message or the address of the
memory area. The Remember structure is specifically for memory
management under Intuition, and "remembers" the used memory areas.
If you want to release the memory area, assign FreeRememberO the
pointer to the Remember structure. This releases all the previously
reserved blocks.
Here are the two functions with their general syntax:
MemBlock
DO
= AllocRemember(RememberKey,
-396
AO
Size, Flags);
DO
Dl
FreeRemember(RememberKey, ReallyForget);
-408
AO
DO
The Remember structure looks like this:

struct Remember
{
OxOO
Ox04
Ox08
OxOC
};
00
04
08
12
struct Remember *NextRemember;

ULONG RememberSize;
UBYTE *MemorYi
375
3. INTUITION AND C
Structure
description
The beginning of the structure is a pointer to another Remembe r

structure. Multiple memory blocks are thus linked to a list.
RememberSize tells the size of the memory, while *Memory
represents the address.
The trick is that you first provide a pointer to a Remembe r structure
that is not yet present, called RememberKey, and this is set to zero.
This pointer is passed to AllocRememberO. and then the first
Remember structure becomes active. The Remember list gets larger
with every call to RememberKey.
FreeRememberO's task is very simple. The function goes back and
forth in the list and releases every area of memory. The memory that
the list itself occupies can be released if the programmer wishes.
The advantage to this system lies in the easy release of all of the
different areas of memory. and in the organization maintained by
Intuition. Here's the program:
/**************************************
* Memory Management
: Remember
==========================~=======
* Author:
*
* Wgb
Date:
16.10.1987
Comments:
only for test
purposes
*
*
*
*
*
*
**************************************1
#define MemoryType MEMF_CHIP
MEMF_CLEAR

struct Remember
*RememberPtr = NULL;
VOID *OpenLibrary();
UBYTE *UndoBuffer, *FileBuffer, *DiskBuffer.
*SuffBuffer. *AllocRemember();
main ()
{

OpenLibrary("intuition.library". OL)
printf("No Intuition Library found!\n");
376
3.11
ABACUS
MEMORY MANAGEMENT
exit(FALSE);
}
UndoBuffer
MemoryType) ;
FileBuffer
MemoryType);
DiskBuffer
MemoryType);
SuffBuffer
MemoryType);
AllocRemember(&RememberPtr, 512L,
AllocRemember(&RememberPtr,
30L,
AllocRemember(&RememberPtr, 512L,
AllocRemember(&RememberPtr,
10L,
printf("Memory reserved!\n"};
FreeRemember(RememberPtr, TRUE);
printf ("Memory free again! \nn) ;
}
377
4.
Operating
System.
Programming
4. OPERATING SYSTEM PROGRAMMING
ABACUS
4.
Operating System
Programming
This chapter contains a large, real-world application written in C. We've
included this listing for two reasons:
I.}
You'll be able to use this editor, which is much more flexible

than ED for entering your own C source codes;
2.}
This listing demonstrates how to combine different components

of the operating system.
We selected a text editor as a demonstration program because it uses a

relatively large number of the operating systems functions. You don't
have to worry about programming the operating system; your
programming task just involves memory management, text output and
keyboard input As we said above, you can use this editor to write your
own C programs. Later in the chapter we'll show you how to combine
assembler code with C source code, and how to search for errors using
the debugger.
In addition to the normal capabilities that any editor has, our editor
includes a few extra features for creating program source code in general.
This means that you can use the editor for writing source code in most
programming languages, not just in C.
The editor was compiled with the Aztec C compiler. However, the
source code is flexible enough that it should compile with any
compiler. When you find special compiler functions or utility program
calls, the editor code is written so that other compilers can use this.
4.1
Planning the Editor

When you begin a new program, you should first thoroughly determine
what the program should do. Write down the task you want the fmal
product to perform, then break it down into the segments needed to
perform this task. This helps you see logical errors before
programming, and helps eliminate large programming changes during
the development stages. You should write down all the functions
needed, problems that must be solved, and possible solutions for these
problems. This list will be used later in the develooment stage. We'll
381
AMIGA
start this brainstonning session with a list of four requirements of our

editor:
1.)
fast;
2.)
flexible;
3.) programmable;
4.)
user-friendly.
Fast
Let's begin with the first requirement The editor should be fast. Placing
the source text in memory plays a major role in speed. The more
organized the memory, the faster text can be manipulated. You can
allocate editing memory with a given size, and load the text into that
memory range as a series of blocks. The advantage to this is a
minimum of memory management. In addition, source files load and
save quickly.
Buffers
Problems arise when you try inserting characters in the text. If you
insert a character in a large text, the editor becomes unbearably slow
because a relatively large memory area must be moved. It would be
better if you could copy the line containing the entry into a buffer and
insert the character there. Keeping this buffer small speeds up the
memory movement. The buffer is then copied into the text if the user
moves the cursor out of the line. This method works well enough if
you have spare memory available and want to edit a short text.
However, it can take a long time to shift the edited text, especially for
files longer than 30K.
There is also a disadvantage during scrolling. To move from one line to
the next you must search for the end of the line. This takes somewhat
longer than computing the beginning of the next line.
Arrays offer another option for placing text in memory. You can
determine the maximum number of characters per line and the
maximum number of lines, then define an array with the corresponding
dimensions: line field [maxline] [maxcolumn]. That way you
have fast access to individual lines and characters within a line.
Characters can be inserted directly in the text. This method of insertion
is fast because you only move the characters of the current line. This
method also has disadvantages:
382
If you want to insert new lines, you must move the entire text.
The array size is preset. If you want a larger array, you must first
change the size of the array, thereby deleting its contents.
Not all of the lines use the maximum line length. Trial and error
has shown that only a few lines reach maximum length, while
most lines only use a third to one half of the allowable line
width.
ABACUS
Linked lists
4.1 PLANNING THE EDITOR
Because of these disadvantages you can choose another method and store
the lines as linked lists. Then available memory limits the maximum
number of lines instead of an array, and changing the pointer moves
you to the next line. The lines have no set length so short lines require
little memory. The resulting memory management is complicated as
you'll see in the development stage, but the bottom line is a good
compromise between speed and memory.
After you know how your lines are constructed, you must consider from
where you'll get the memory for each line. The operating system can
supply the memory at any time. You can utilize this memory by
dividing blocks of memory into smaller units (the operating system
cannot produce these smaller units). It is better if you allocate a block
of a predetermined size (e.g., 5K) from the operating system and divide
this up yourself, according to the present need. Write the functions so
you don't need to think about the "why" later when you want to add a
new line to the program. We'll go into more detail when. we discuss
memory management.
User-friendly
Now for a few ideas to make the program more user-friendly. An Undo
function will remove the changes just made to a line. To create the
Undo function you must store the original text in a buffer, delete the
changed line and restore the line's original contents from the buffer.
A feature that has always bothered us when working with editors and
word processors is when the program cannot sense text changes. When
leaving an editor it usually asks, "Are you sure that you have saved the
text?" whether or not changes have been made since the last save. The
editor would be a friendlier program if it only asked this question after
the text has been changed since the last save. To determine whether the
text was changed, a flag must be set whenever a change is made. We
mention this now because if you forget to set the flag in a few places in
the code, the program won't work. The error will be hard to find if you
don't know about this flag.
The editor should support multiple windows. This is important when
doing a lot of work if you want to move blocks of text between
different source codes. When you only have one window, this results in
a lengthy load and save. But you can't just open more windows. Each
window requires its own editing memory, its own cursor and other
things. This means that you cannot globally save all of the important
information about the text, like number of lines, cursor position, etc.
Instead it must be saved to a structure for which exactly one structure
exists for each editor window. There is still a pointer to the actual
structure, whose window is still active. This means more work for you
at the beginning of the development stage, but once you've
implemented a second window, providing for the others becomes easier.
An interesting capability of the Amiga is the option of assigning a
character string to any key. You can, for example, specify that the
Amiga sends a "ae" or ''\344'' (octal code for a) every time you press
383
AMIGA
<Alt><A>. Or you can specify whether the <Return> key sends a

linefeed (LF), a carriage return (CR) or a combination of the two
(CRLF). You can also assign special characters to any key. The
possibilities given here will be described later in this chapter.
Flexibility
This program should be flexible. That is, instead of the programmer

laying out certain unchangeable parameters in the source code, the user
should be able to change some parameters to suit his own needs.
Tabs
Your editor should be able to insert true tabs instead of just blank
spaces. Structured programming involves inserting spaces to indent
lines, which wastes memory. If you own an Aztec C compiler, you
could use the Z editor which inserts true tabs. However, Z is a very
difficult editor to use and learn. It would be better if you could set the
tabs anywhere you want This is difficult to program, but this tab
system makes the editor more flexible.
Auto indent
Automatic indentation could be indirectly tied in with tabs. When you

press <Return> at the end of a line, the cursor should move to the same
first column as the previous line (i.e., the first character that is neither a
blank nor a tab). This auto indent saves the user's inserting spaces or
tabs at the beginning of each line for structured programming.
The editor should also display special and control characters. When ED
encounters unusual characters it displays the message "file contains
binary" and stops. Control characters use the lower half of the ASCII
character codes. Because these characters aren't displayable on the
Amiga, you could display these in a different color and add a constant
value to the ASCII code of the character. This then returns a visible
character. The character with ASCII code 1 appears as a red "A" in the
text (Code("A") = 65 = 1 + 64). You could display the reserved C
keywords in bold print This would make the program easier to read.
Folding
The final user-friendly feature we could include in the editor is folding.

Folding lets you make a section of program code invisible without
actually deleting the section. For example, if you have two program
lines that you wish to view, but the two lines are separated by other
program codes, you can fold the unnecessary text away.
Programmable We now tum to the programmability of the editor. All of the editor's
functions should be accessible from simple commands. You can enter
multiple commands in one line by placing semicolons between each
command sequence. In addition, command sequences can be repeated by
entering a number before a sequence (either single commands or
multiple commands). So far this editor seems to do everything that ED
does. We're going to take this editor a few steps further.
First we allow the use of variables which can provide numbers for
cursor X/y coordinates, strings or command sequences. If you place a
numeric variable before a command, this command executes the number
of times given in the variable. In addition you can have expanded
384
ABACUS
4.1 PLANNING THE EDITOR
control structures, FOR-TO-DO, WHILE-DO and IF-THEN-ELSE

constructs. Command sequences can be assigned to any function key.
More ideas for commands could be implemented later, such as searching
and replacing blocks of text (from line to line).
It would also be convenient to have access to CLI commands while in
the editor. Then, for example, you would no longer need to save your
text, exit the editor and call the compiler.
Data structures Once you have the complete list of ideas you must now think about the
data structures which make up the editor. Data structures correspond to
the important things which you must think about before you begin
programming. All of the Amiga's data structures accessed by the
operating system are of similar design; you only think about the lists.
This makes manipulation of objects easier. The following line is the
first data structure in your editor:
struct Zline
{
struct Zline *succ;

struct Zline *pred;
UBYTE flags;
UBYTE len;
/* UBYTE liner] */
pointer to the next line

pointer to previous line
Diverse Flags
length of the line
The two pointers later construct the linked list. You need two pointers
because you want to move to either the previous line or the next line.
The Amiga also has functions for doubled linked lists, which you can
also use for your lines. If (flags & 128) is unequal to zero, the line
appears in the buffer because the user changed this. This flag is needed
because you must display the buffer if you move to a line where this
flag is seL We will define more flags as the need arises.
Line length
The len variable specifies the line length including the CR, LF or
CR/LF that indicates the end of the line. By this method you can have a
line without an end of line character if the line is longer than the
screen's width. Instead of this line continuing off to the right of the
screen, the line ''wraps'' and text appears on the next screen line without
being split by a CR, LF or CR/LF. Word processors handle lines in
this way.
The line itself directly follows len. Because of the changing lengths,
you cannot include it in the structure. The length of the structure
including the line appears as follows:
EntireLength
(sizeof(struct line) + line.len + 1) & -2.
EntireLength must be an even number, because the pointer in the

Amiga must point to an even address in memory. This is a rule when
working with the 68000 microprocessor. If you disobey this rule, you'll
get a Guru Meditation number 00000003 (address error).
'385
4.
OPERATING SYSTEM PROGRAMMING
AMIGA
Memory blocks The next structure configures the memory blocks containing the text
lines. A linked list connects the blocks of memory. The code needs a
pointer to a list of all the available memory segments in this block, the
length of the block, and the number of unused bytes. The structure
looks like the following:
struct Memoryblock
{
Next memory block

struct Memoryblock *succ;
struct Memoryblock *pred;
Previous memory block
ULONG length;
Maximum number of free bytes
Momentary number of free bytes
ULONG free;
struct FList freelist;
List of free pieces
};
The first two variables insert the memory blocks in a double linked list
The length variable gives the block size without the memory block
structure. This is the equivalent of the number of free bytes present in
an unused block (the length of a memorysection structure). Free
states how many bytes are free in this block. Now the problem arises of
where these free bytes are. These blocks can exist in any location, and
are rarely in sequence. When a line is deleted, a free section of memory
must not lie on another piece of free memory. Because every line is at
least 10 bytes long (sizeof(struct line)!), you have 10 bytes in
which you can construct a list of all of the free memory pieces in a
block. Each element of this list has the following construction:
struct Memorysection
{
struct Memorysection *succ;

struct Memorysection *pred;
UWORD len;
};
Management
The first two elements are chosen with the operating system list in
mind. They are handled as a double linked list for a set of management
functions in the Exec library. Because you use the same chain, you
can also use these functions for your lists without having to write your
own. The element len contains the length of the entire memory
section, including the Memorysection structure.
Next we need an FList structure, which has the same design as the
list header of an operating system list:
struct FList
{
struct Memorysection *head;

struct Memorysection *tail;
struct Memorysection *tailpred;
};
The structure should be familiar to you from the Exec lists (see the
Appendix for an explanation of Exec . library). It is often said that
386
4.1
ABACUS
PLANNING THE EDITOR
head points to the fIrst element of the list, tailpred to the last
element of the list, and t ail is equal zero. This arrangement
simplifIes insertion and deletion in the lists.
This may sound like a lot of effort just to manage a line display. We
could specify the memory for each line from the operating system, but
this tends not to work too well. Remember that a text can consist of
1000 or more lines. Every time you change a line, you must release the
old memory and allocate new memory. Once you've spent some time
editing one text, a lot of gaps remain in memory.
When you want a new line, the program should fIrst search for a
memory section that has exactly the desired length. If it cannot find
such a piece, take the memory from the largest memory section
available. Using the largest section as a pattern, be sure that no smaller
memory sections are reduced without needing reduction. For example,
you need 14 bytes and the next largest memory piece has 16 bytes. If
you distribute this, you get two memory pieces, one 14 bytes long and
the other 2 bytes. These two bytes are leftovers. They can't be placed in
the list of free memory pieces because you need at least 10 bytes for the
Memorysection structure.
When you have no more room, allocate a new memory block and insert
it in the block list.
That brings us to the last structure, the Edi tor structure. One exists
for every editor window, and extracts all the important data needed for
the window:
struct Editor
{
struct Editor *succ;

struct Editor *pred;
struct Window *window;
Pointer to the editor window
struct BList block;
List of the memory blocks
struct ZList zlines;
List of the lines
UBYTE buffer[MAXWIDTH];
Buffer to edit
UBYTE tabstring[MAXWIDTH];
Tabulators
UWORD num_lines;
Number of lines
struct Zline *actual;
Pointer to the actual line
struct Zline *top;
Pointer to top line,
which is visible in the screen
UWORD toppos;
Number of the top line
UWORD xpos,ypos;
Cursor-Position
UWORD changed:l;
It is 1 in case the text is changed
UWORD insert:l;
It is 1 incase of Insert mode
};
Two more structures join the Edi tor structure, which represent the
list heads:
387
4.
AMIGA
struct BList
{
struct Memoryblock *head;

struct Memoryblock *tail;
struct Memoryblock *tailpred;
};
struct ZList
{
struct'Zline *head;
struct Zline *tail;
struct Zline *tailpred;
};
The individual elements of the Edi to r structure are easy to prepare.
388
4.2
ABACUS
4.2
DEVELOPMENT
STAGES
Development Stages
We want to begin developing the modules from which we can create the
editor. The following source code opens the libraries which let the
Amiga access operating system functions. The Amiga displays a
message if the access is successful. Remember that you can only call
functions of a library if you have opened it with OpenLibrary. It is
also good practice to close all opened libraries.
/* <Editor.c> Version 0.0 */
/************
* Includes:
************/
finclude <intuition/intuition.h>
/***********
* Defines:
***********/
#define REV 33L
/**********************
* External Functions:
**********************/
void CloseLibrary();
/*********************
* Global Variables:
*********************/
struct Library *IntuitionBase

NULL, *GfxBase
struct Library *DosBase = NULL;
NULL;
/*****************
* Main program:
*****************/
389
AMIGA
main ()
{
if ( ! (IntuitionBase =
OpenLibrary(nintuition.library",REV)
goto Ende;
if ( !(GfxBase
goto Ende;
= OpenLibrary("graphics.library",REV)
if ( ! (DosBase = OpenLibrary(ndos.library",REV)
goto Ende;
printf(nEverything is open\n");
Ende:
if (DosBase) CloseLibrary(DosBase);
if (GfxBase) CloseLibrary(GfxBase);
if (IntuitionBase) CloseLibrary(IntuitionBase);
The #include functions add the data structure definitions needed to

access the operating system routines. If you omit these commands, the
C compiler displays error messages because of unknown structures.
You should declare all functions that you use within a program at the
beginning of the program. This is so the C compiler can determine
whether you have specified the value of a function and a variable of the
correct type.
Delays
Compiling takes some time, since the compiler needs to read the
include files. This long delay also bothered the developers of the Aztec
C compiler, so they added an option which precompiles include files.
This precompilation links the files faster than with normal includes.
For this reason, write a prelist file that contains only the #incl ude
instructions of your source text. The edi tor. prelist file located
in the pre directory, looks like the following in our example:
<pre/editor.prelist>
Make a directory named PRE and save this file into it. To compile this
file you must inform the compiler that it should create a file using the
precompiled includes. The Aztec C compiler uses the +H option to
declare this:
cc +Hpre/editor.pre pre/editor.prelist
This compiler command compiles the file containing the include

instructions with the name "editor.prelist"inthepre
directory. The file "edi tor. pre" is created in the same directory.
390
4.2 DEVELOPMENT STAGES
ABACUS
This file can be combined with the +I option when compiling the
program:
cc +Ipre/Editor.pre src/Editor.c
The src (source) directory should contain the file Editor.c. The
compiler should now work faster. The make utility program offers
another possibility for simplifying the process. The make utility is
used when a program consists of multiple modules that are linked
together. It is better to divide large programs into smaller program
modules. These modules are then compiled and put together with the
linker. When modules are changed, the programmer must only compile
the edited modules instead of the entire program.
Compiler and linker module control is handled by the make utility.
You enter make, which reads the makefile. The makefile
contains the modules your program consists of, and how these should
be compiled and linked. If you call make, it checks the system date and
time for any changes to sections of your program. It then automatically
recompiles edited modules and links the compiled modules into a
finished program.
You use the makefile to tell the make utility which files of your
program depend on which other files. The object file of one module
depends on the source text, and the finished program depends on the
object files of all of the modules. Because of this dependence and the
date and time when the file was created, make tests for changes and
takes the appropriate steps to create the program. It is therefore
important that the time and the date are correctly set in your Amiga.
To make this somewhat clearer, let's look at how make would be used
to compile your editor. For this we type the names of the files which
comprise your editor:
Editor
src/Editor.o
src/Editor.c
pre/Editor.pre
pre/Editor.prelist
The include files also belong in this category, but they already appear in
pre/Editor.pre. The finished editor program depends on the object
module src/Editor.o. The libraries also playa role, but you don't
change them. You need the dependency of each file to configure the
makefile in case you make any changes in the process of program
development
The object module src/Editor.o depends on the source text
src/Editor.c and on the precompiled includes pre/Editor.pre;
these depend on pre/Editor.prelist, which contains the names
of the includes that should be precompiled. When you want to
document the dependence for the make utility, enter the filenames in
391
the first column of a line, followed by a colon. Following that enter the
filenames on which this file depends. For this first editor the dependence
looks like the following:
Editor:
sre/Editor.o
sre/Editor.o:
sre/Editor.e pre/Editor.pre
pre/Editor.pre: pre/Editor.prelist
The files scr /Edi tor.c and pre/Edi tor .prelist depend on
other smaller files. So that make knows how to compile or link all of
these fIles, you must enter the instructions in the make fIle. Type
these directly under the line that describes the dependence. The
commands cannot begin in the first column, because this is reserved for
the dependents. You should also have at least one line of instructions
after each one. You can enter any instructions behind each dependent.
The make utility executes these until it either encounters a dependent
or the end of the make fIle.
Linking
Now enter the instructions that link your editor created together. To
make the finished program from the object modules, the following
command must be executed:
In sre/Editor.o -Ie
-0
Editor
For compilation of the source text:

ee +Ipre/Editor.pre sre/Editor.e
To create the precompiled include you add two instructions. When you
create these, as you did above, the C compiler creates an object file,
although this wouldn't be necessary because you are only interested in
the precompiled includes. The assembler is also called, which uses
additional time. Combine the call of the assmebler with the -A option.
The assembler fIle which the compiler created is written to the RAM
disk and then deleted:
ee -A +Hpre/Editor.pre pre/Editor.prelist -0 ram:Ed.asm
delete ram:Ed.asm
Your makefile for the creation of the editor should look like the
following, yours may differ slighlty depending on the structure of your
directories (i.e is the pre directory in the src directory):
Editor: sre/Editor.o
In sre/Editor.o -Ie
-0
Editor
sre/Editor.o: sre/Editor.e pre/Editor.pre

ee +Ipre/Editor.pre sre/Editor.e
ee -A +Hpre/Editor.pre pre/Editor.prelist -0 ram:Ed.asm
delete ram:Ed.asm
392
4.2
ABACUS
DEVELOPMENT STAGES
Make creates the file whose name is found first in the makefile, in
our case Ed ito r. If we want to create another file (e.g.,
src/Edi tor. 0) state the name when calling make from the eLI:
make src/Editor.o
Let's use make just a bit more. Because our program consists of
modules, we can create the makefile before we create all of the
modules. First we want the object module, from which our editor is
made, to be defined in a variable:
OBJ=src/Editor.o
The variable must be directly at the beginning of the line, followed by

an equal sign and the listing of the object module. This has another
purpose besides forming the makefile. Because you'll want to use
the debugger later to check your program for errors, you should link the
program with the -w option. This option ensures that when you use
the debugger, you can access the names of the functions and global
variables (which simplifies the debugging). Now you have two link
instructions in your makefile, from which the names of the object
modules emerge. The Edi tor creates a link instruction while the
Debug joins a symbol table with the -w option. We use the variable
OBJ so we only need to define the object module once and enter only
the variable to the link instructions:
OBJ=src/Editor.o
Editor: $ (OBJ)
In $(OBJ) -lc
-0
Debug: $ (OBJ)
In -w $(OBJ) -lc
Editor
-0
Editor
To be able to use the variable, it must be entered in parentheses with a

"$" preceding the parentheses. Be sure that the same program is created
by both link instructions, and that you have not forgotten a module for
the instruction.
When you want to link the editor with the -w option, you call ma ke
as follows:
make Debug
The make utility determines that the file Debug doesn't exist and
directs all of the steps to the creation of file specified in the
makefile.
But now we come back to the 20 modules which comprise your

program. If you want to state how this is created from the source text
for every object module, you waste a lot of memory and create
unnecessary work. To prevent this, ma ke uses rules. One rule states
393
AMIGA
how a file with the same name and a different extension (.0) is created
from a file with another extension (.c):
.c.o:
/* Call compiler
c to 0*1
This process can also be exchanged. The method is different from

stating the dependents. The first extension, which must begin with a
period (.), is written at the beginning of the line. The second extension
follows directly behind it, and it also must begin with a period. The
line ends with a colon.
The following lines list the instructions for putting the dependents to
use. Because you can't directly state filenames with the rules, make
offers two variables, both of which contain the filenames that should be
used for this rule. $@ is equivalent to the entire filename and $ * is the
name of the file without an extension or pathname. The filename has
the same extension as the second extension of the rule. Take
src/Editor.o as an example. $@ = "src/Editor.o" and $* =
"Edi tor". The following rule states a pre file is created from a
.prelist file:
.prelist.pre:
cc -a -0 ram:$*.preasm +H$@ pre/$*.prelist
delete ram:$*.preasm
The difference for the above from defined dependent is that the file that
the RAM disk creates has the name Editor.preasm when the file
pre/Editor.pre is be created. Now you can remove the instructions
that follow your dependents for the Edit 0 r . pre file.
The next rule that we want to define states how an object file is created
from a source text. Here you should ensure that your program contains
only a main module in which the mainO function is defined, if it
consists of multiple modules. It allows the C compiler to save some
memory by setting the +B flag for all other modules. While main
begins with a jump to the initialization routine, the other modules do
not require this jump; the initialization routine must only be at the
beginning of main. We will define the rule so that it can create all of
the modules out of the main module:
.c.O:
cc +B +Ipre/Editor.pre -0 $@ src/$*.c
Your main module represents the exception to the rule. It is sufficient

to write a corresponding instruction behind the dependent definition for
the src/Editor.o module. Then ifmake doesn't find an instruction
for a dependent, it uses the corresponding rule. Your makefile is
now fmished and looks like the following listing:
394
4.2 DEVELOPMENT STAGES
ABACUS
<make file>
.c.o:
.prelist.pre:
cc -A -0 ram:$*.preasm +H$@ pre/$*.prelist
OBJ=sre/Editor.o
Editor: $ (OBJ)
In $(OBJ) -lc
-0
Debug: $ (OBJ)
In -w $(OBJ) -Ie
Editor
-0
Editor
src/Editor.o: sre/Editor.e pre/Editor.pre
Here is another file for the editor that will be used in developing our
editor. The file defines the editor's data structures:
<sre/Editor.h>
/************
* Includes:
************1
/*******************
* Data structures:
*******************/
struet Zline
(
struet Zline *succ;

struet Zline *pred;
UBYTE flags;
UBYTE len;
};
tdefine ZLF USED 128
This file contains only the definitions for lines that you will change.
The makefile must also be changed accordingly:
395
AMIGA
<makefile>
.c.o:
.prelist.pre:
OBJ=src/Editor.o
Editor: $ (OBJ)
In $(OBJ) -lc
-0
Debug: $ (OBJ)
In -w $(OBJ) -lc
Editor
-0
Editor
pre/Editor.pre: pre/Editor.prelist src/Editor.h

src/Editor.o: src/Editor.c src/Editor.h pre/Editor.pre
Editor.h is used at the moment from Editor.pre and Editor.o.

To understand better what happens when make is called, enter
Editor.c, Editor.h, Editor.prelist and the makefile.
Now you must copy make to your disk and call make editor. The
m a k e utiliy then establishes that Ed ito r depends on
src/Editor.o, which doesn't yet exist Make notices that it must
create s r c / Ed ito r.o and determines that this depends on
src/Editor.c, src/Editor.h and pre/Editor.pre. While
the first two files already exist, the third doesn't. Make also notices that
it should create pre/Editor.pre and shows on what these depend.
pre/Editor.pre depends on pre/Editor.prelist and
src/Edi tor.h. Because both of these exist, make can create the me
pre/Editor.pre following the rule. prelist .pre, in which the
following command is sent to the eLI:
cc -A -0 ram:Editor.preasm +Hpre/Editor.pre
pre/Editor.prelist
The unnecessary me ram:Edi tor.preasm can be deleted:

delete ram:Editor.preasm
After pre/Edi tor.pre exists, make creates src/Edi tor.o and

executes the following command:
There is also a rule to make an object file from a * . c file, but because
we have given an extra instruction, the rule is ignored. To conclude, the
editor itself is created after the object file exists:
In src/Editor.o -lc
396
-0
Editor
ABACUS
4.2
DEVELOPMENT
STAGES
With this you have not only finished the first version of your program,
but you also have a stable base on which you can build the additional
programming of the editor. In large programming projects the make
utility and the makefile can save a great deal of time and effort. Be
sure that you understand how the make utility functions or see your C
manual for more information on the make utility that came with your
compiler.
397
4. OPERATING SYSTEM PROGRAMMING AMIGA C FOR ADVANCED PROGRAMMERS
4.3
Step by Step
This section describes the step by step construction of our editor
program from the preceding example Editor. c source text. On the
optional diskette each version of the editor is contained in its own
directory, you may wish to do the same with your editor.
4.3.1
Opening a Window
First the editor should open a window. You could copy one of the
example programs from Chapter 2 and edit it to suit your needs.
Because you want to be able to open more windows, you must write
your own routines to open a window and add this to the Editor
structure. You need a function that closes a window again and releases
the Editor structure.
Next let's consider how to open a window. This function should fll"St
try to obtain memory for the Editor structure. If it succeeds, a
window opens and the pointer to this window is entered in the structure.
Then the structure must be initialized, especially for the memory block
and the line lists. The function for closing the window closes the
window then releases the memory occupied by the Edi tor structure.
That would theoretically be all you need. However, if you did it !his
way, you would have to make some changes when you wanted to add
more windows. Every window has a message port through which it
sends its messages to the operating system. Because the
IntuiMessage structure also determines which window receives the
message, the message port will be enough for our purposes. Using the
message port saves you memory.
Your own
MessagePort
398
You must also make it clear to Intuition that you want a single
message port for all of your windows. The program must clear the
IDCMP flags in the Newwindow structure before calling
OpenWindow. Because the window should contain no input, it
receives no message port from Intuition. If a window opens, you enter
the address of your message port, which you initialized from
CreatePort, in the UserPort field of the Window structure. Then
call ModifyIDCMP and define the input that you want to get. Because
Intuition has already found a message port for the window, it doesn't
use a new one. You must test for the closing of the window and set the
UserPort field of the window structure to zero before you call
Closewindow, so your message port will be closed.
4.3
ABACUS
Key display
STEP BY STEP
The next step of our editor will be to display the pressed key. The
operating system has two possibilities: VANILLAKEY and RAWKEY.
While you get the ASCII code of the corrseponding key with
VANILLAKEY, you get scan codes with RAWKEY. RAWKEY returns
which key was pressed, and not which letter the key corresponds to.
Work with RAWKEY in your editor because we can determine whether
the function or cursor keys were pressed. These are not passed by
VANILLAKEY because these keys don't send an ASCII code.
First complete the structure that you need for your editor, this is entered
in Editor.h:
<src/Editor.h>
1************
* Version 1A
* Includes:
************1
1***********
* Defines:
*
***********1
ide fine MAXWIDTH 80
1*******************
**
Data structures:
*
*******************1
struct Zline
{
struct Zline *succ;

struct Zline *pred;
UBYTE flags;
UBYTE len;
};
idefine ZLF_USED 128

struct ZList
{
struct Zline *head;

struct Zline *tail;
};
{

UWORD len;
};
399
4.
AMIGA
struct FList
{

struct Memorysection *tailpred:
};
struct Memoryblock
{

ULONG length;
ULONG free;
struct FList freeliste:
};
struct BList
{

};
struct Editor
{

struct Window *windowi
struct BList block;
UBYTE buffer[MAXWIDTH];
UWORD num lines;
struct Zllne *actual;
struct Zline *top;
UWORD toppos;
UWORD xpos,ypos;
UWORD changed:l;
UWORD insert:l;
}i
struct EList
{
struct Editor *head;

struct Editor *tail:
struct Editor *tailpred;
}i
The structures in the above code have already been described in Section
4.1, except for the last one. The EList structure manages all of the
editor windows, just as the BList structure manages all of the
memory blocks and the ZList structure manages all of the lines. Next
follows the source text of the editor:
<src/Editor.c>
/* Editor.c VIA */
/************
*
400
Includes:
4.3
ABACUS
STEP BY ST E P
************/
#include "src/Editor.h"
/***********
* Defines:
***********1
#define REV 33L
1**********************
**
External Functions:
**********************1

void CloseLibrary(),NewList(),AddTail(),CloseWindow(),free();
void DeletePort(),ModifyIDCMP(),ReplyMsg();
struct Editor *malloc(),*RemHead();
struct MsgPort *CreatePort();
struct IntuiMessage *GetMsg();
ULONG WaitO;
1*********************
**
Global Variables:
**********************1
struct Library *IntuitionBase
NULL;
= NULL,
*GfxBase
= NULL,
*DosBase
struct EList editorList;

struct NewWindow newEdWindow
100,40,440,156,
AUTOFRONTPEN,AUTOBACKPEN,
REFRESHWINDOW I MOUSEBUTTONS I RAWKEY 1 CLOSEWINDOW,
WINDOWSIZING I WINDOWDRAG I WINDOWDEPTH
WINDOWCLOSE
SIZEBBOTTOM
I SIMPLE_REFRESH I ACTIVATE,
NULL, NULL,
(UBYTE *) "Editor",
NULL, NULL,
100,40,640,200,
WBENCHSCREEN
};
struct MsgPort *edUserPort
= NULL;
1***************
** Functions:
*
*
401
4.
AMIGA
***************/
/***********************************
**
OpenEdi tor ()
* Open Editor window and

* initalize Editor structure
* Returns pointer to Editor structure

* or is zero in case of an error
************************************/
struct Editor *OpenEditor()
{
register struct Editor *ed = NULL;

register struct Window *wd;
register ULONG flags;
/* IDCMPFlags saved, sets it to zero in structure
=> use your own UserPort! */
flags = newEdWindow.IDCMPFlags;
newEdWindow.IDCMPFlags = NULL;
/* Memory for editor structure: */
if (ed = rnalloc(sizeof(struct Editor)
/* Window open: */
if (wd = OpenWindow(&newEdWindow
(
ed->window = wd;
/* UserPort established: */
wd->UserPort = edUserPort;
ModifyIDCMP(wd,flags);
/* Parameter initialization: */
NewList(&(ed->block;
NewList(&(ed->zlines;
ed->num lines = 0;
ed->actual = NULL;
ed->top = NULL;
ed->toppos = 0;
ed->xpos = 1;
ed->ypos = 1;
ed->changed = 0;
ed->insert = 1;
}
else
{
freeed) ;
ed = NULL;
newEdWindow.IDCMPFlags
return (ed);
flags;
/***************************
* CloseEditor(ed)
402
4.3
ABACUS
STEP BY STEP
* Close editor window.
* ed
Editor structure.
***************************/
void CloseEditor(ed)
struct Editor *ed:
{
/* Close window: */
ed->window->UserPort = NULL:
CloseWindow(ed->window):
free (ed) :
/*****************
* Main program:
*****************/
main 0
{
register struct Editor *ed:

BaaL running = TRUE;
register struct IntuiMessage *imsg;
ULONG signal,class;
UWORD code,qualifier:
APTR iaddress;
register UWORD n = 1;
/* Libraries open: */
if ( ! (IntuitionBase = OpenLibrary("intuition.library",REV)
goto Ende;
if ( ! (GfxBase = OpenLibrary ("graphics library" , REV) ) )
goto Ende;
if ( ! (DosBase = Open Library ("dos.library",REV )
goto Ende;
/* UserPort open *1
if (! (edUserPort = CreatePort(NULL,OL) goto Ende;
1* Editor list initialization: */

NewList(&editorList):
/* Open first editor window: *1
if (ed = OpenEditor(
AddTail(&editorList,ed):
else
goto Ende;
dJ
{
signal
= Wait (11 edUserPort->mp_SigBit):
while (imsg
= GetMsg(edUserPort
class
imsg->Class;
code
imsg->Code;
qualifier = imsg->Qualifier;
403
iaddress
= imsg->IAddress;
ReplyMsg(imsg);
1* Event processing: *1
switch (class)
(
case RAWKEY:
printf("RawKey-Event Number %3d code: Code
Qualifer = $%4x. \n", n++, code, qualifier);
break;
= $%4x,
case CLOSEWINDOW:
running = FALSE;
break;
default:
printf(IINot a processable event: %lx\n",class);
} 1* of case *1
1* of while (GetMsg( *1
} while (running);
Ende:
1* Close all editor windows: *1

while (ed = RemHead(&editorList
Clo seEdi tor (ed);
1* Close UserPort: *1
if (edUserPort) DeletePort(edUserPort);
1* Close Libraries: *1
if (DosBase) CloseLibrary(DosBase);
if (GfxBase) CloseLibrary(GfxBase);
if (IntuitionBase) CloseLibrary(IntuitionBase);
The OpenEditor function works exactly as described. First open the

libraries in the main program, then a single MessagePort
(edUserPort) is entered in the user port of the window with
OpenEdi tor. Then the editor list is initialized and a window opens
and is connected to the editor list (AddTail). The main loop follows,
which waits for input in the main loop {do {... } while (running.
The operating system signals an event. Messages are received until no
more are present; it could be that the editor only received one signal,
but multiple messages. In the switch case function they are
differentiated according to the message type and handled accordingly. In
this case with RAWKEY only the code of the pressed key is displayed, a
number (n) counted, CLOSEWINDOW set to FALSE and the program
ended. To conclude all of the windows must be closed again, then the
message pon and lastly the libraries.
A few observations about the program:
404
To write the program to run on all compilers, you should use the
variable types from <exec/types.h> and not the normal C
4.3
ABACUS
STEP BY ST E P
standard data types, because the data type integer is implemented

once as a 16-bit number and once as a 32-bit.
When you want to save casting (type conversion), you can

declare the functions as you want. Especially functions, that the
pointer to objects returns, can be declared so that the pointer
returns all of the possibilties, like it did in the program with
malloc, RemHead and GetMsg.
You should also declare all of the functions that you use in your
program, if only as void. Otherwise you could easily overlook
a function, and the program could stop because you needed a
32-bit pointer but only had a 16-bit pointer available. There is
also no problem if you must compute or return with a 32-bit
integer instead of with a 16-bit integer.
The goto statement in conjunction with the error handler is

better than IF statements. The function listed below, which
could follow every error in the main program, is very inefficient
compared to just using goto:
i f (error)
{
C!oseWindow(xxx);
C!oseLibrary(yyy);
exit (FALSE) ;
The functions NewList, AddTail and RemHead are for

Exec lists. When you look at the structures of lists and nodes
(see the Appendix for a description of Exec), you determine that
these are only minimally more extensive than your structures.
The functions named work with your structures because they
don't access additional elements. It would be different if you also
used the Enqeue function because this also uses the priority
field of the Exec node, which your structure does not have.
Compile and link the program. Run it and press a couple of keys. The
program returns a message when you press a key, and one when you
release a key. When you press one key and hold it down, something
unexpected happens: Once a certain amount of time passes (set by
KeyRepeatDelay), the program displays the same message multiple
times. This continues until you release the key. The only difference
between these messages is that the ninth bit of the qualifier is set,
which means that this key transmits the key repeat function. The editor
allows key repeat with both VANILLAKEY and RAWKEY modes.
405
4.
4.3.2
From
RAWKEY
AMIGA
to ASCII
The question remains how we can convert RAWKEY characters into

normal ASCII characters. A user-defined conversion table is out of the
question, because this won't work on Amigas using alternate keyboard
maps. The operating system also has a function that handles keyboard
layout.
The RawKeyConvert function is part of the console device. This
function can be accessed by selecting a pointer to console base without
opening the console device first, because the function can be called like
a library function. You receive the pointer to the console base by
calling:
OpenConsole("console.device",-lL,IOStd,OL);
and selecting the pointer from the io Device field of the IOStdReq
structure. Then you can call RawKeyConvert as a normal function.
In addition to a pointer to an InputEvent structure, you must assign
a pointer to a buffer whose length and a pointer to a KeyMap structure
has been set to zero, because the standard keyboard table should be used
for the conversion. You must fill the Inputevent structure with the
RAWKEY code and the qualifier from the IntuiMessage structure
before the call. To allow ASCII output instead of RAWKEY output, the
followings changes are necessary:
Define the maximum length of keyboard input as 128 characters:
#define MAXINPUTLEN 128L
Declare the new OpenDevice and RawKeyConvert functions:

ULONG OpenDevice();
SHORT RawKeyConvert();
Now two global variables appear that we already mentioned. We set the
ie_Class field to RAWKEY in the InputEvent structure so that
you don't have to do this explicitly in the program:
struct IOStdReq ioStdReq;
UBYTE inputBuffer[MAXINPUTLEN];
UWORD inputLen = 0;
struct InputEvent inputEvent =
{
0,
IECLASS RAWKEY, 0,
0,0
};
406
4.3
ABACUS
STEP BY STEP
The characters from the RawKeyConvert function pass to the input

buffer, stored with the corresponding key. Define one variable at the
beginning of the main program:
SHORT 1;
After you open the libraries in the main program, select the pointer to
the console device:
if ( !(OpenDevice{"console.device",-lL,&oiStdReq,OL)
ConsoleDevice = ioStdReq.oi Device;
else
goto Ende;
Finally, construct the handling of RAWKEY messages in the main loop:

case HAWKEY:
if {! (code & IECODE UP PREFIX
-
inputEvent.ie Code = code;

inputEvent.ie=Qua1ifier = qualifier;
if ({I = RawKeyConvert{&inputEvent,
inputBuffer,MAXINPUTLEN,NULL)
) > 0)
{
inputBuffer[l) = 0;
printf{"%3d> Length = %d: ",n++,l);
for (1 = 0; inputBuffer[l); 1++)
printf{"%2x", (UWORD)inputBuffer[l);
if (inputBuffer[O] >= 32 && inputBuffer[O) <= 127)
printf(" %s",inputBuffer);
printf ("\n");
break;
First test whether the corresponding key is pressed or released; the

second case (IECODE_UP_PREFIX) isn't currently of interest. Then
the values from code and qualifier transfer into the InputEvent
structure. RawKeyConvert is then called. Save the return value in 1
(not in input Len) and test whether this value is greater than O. If 1 is
less than 0, an error occurs (buffer too small). If 1 equals zero, the key
returns no characters.
H the returned value is greater than zero, it gives the length of the string
stored with the key. A null byte appears behind the string so that you
can output the string with printf. Normally keys are assigned a
single character so the value one is usually returned as the length, but
there are exceptions. The program returns a successive number and then
the length of the string. Then follows the string itself, first in hex
notation then as a true ASCII string, but only if the first character lies
between a space (32) and a delete (127), and is a printable character. If
you don't encounter this check, your output appears in the CLI window
because the CLI converts certain ASCII codes to cursor movements.
407
Type it in, compile, link, run and examine. While testing the normal
keys don't give any unusual results, the function keys return three or
four characters at a time. You must bear this in mind with your
keyboard check. This is also the reason that VANI LLAKEY returns no
function key information. The IntuiMessage can only return one
character, not three at once. When you examine all of the special
characters and compare them with the codes supplied in the
documentation (see the AmigaOOS manual), you may find that some
codes may not match. There may be some errors in your output.
The complete source text for the section of the editor described above
can be found in the VO .1 directory on the optional disk for this book.
The complete listing for readers who do not have the optional disk is
presented in Appendix A. If you own an Aztec C compiler you can
compile this by changing to the VO.1 directory and entering make
Edi tor. Before this you must ensure that the make utility exists in a
place accessibe to the CLI. For example, you could copy make to the
RAM disk:
copy make to ram:
path ram: add
Enter the following to compile the program on the optional diskette:

cd
va.l
make Editor
The single difference from the paths discussed here is that the pre
directory is elsewhere on the disk, in the main editor directory. Because
pre is used for all of the versions of the editor, you should make sure
that the file :pre/Editor.pre exists for each new version of the
editor. Delete the file pre/Editor.pre before you call make for
each new version.
4.3.3
~ernory ~anagernent
You must have a method of saving and joining lines together before
you can load and display a file. You need a function for this which
provides a new Zline structure, into which you can copy a string and
the current line. The function header should look like the following:
struct ZLine *newZline(len)
len;
UWORD
len represents the length of the string, including the end of line
character. Because you can only use memory pieces of even lengths, the
length must appear in the function as an even number. Because this
happens often, create a corresponding define:
408
ABACUS
4.3
STEP BY STEP
#define EVENLEN(x) (X)+1)&-2)
Before you implement the function, consider how this should operate. If
no memory block is found in the memory list and the text memory is
also empty, a new block with 10K appears and a line is selected.
Otherwise a search begins for an available line. If this happens, a
garbage collection executes and the memory is released before a second
search executes. If this also happens, not enough memory blocks exist
for further line storage. The program uses a new memory block of only
SK. That is why you should work with two different block sizes, so
that only one memory block will be needed for texts less than 10K in
length. The following flowplan explains our method:
If no memory block exists in the block list
select a new block (length = 10K)
select line
done
else
select line
if none is found
execute garbage collection
select line
if none is found
select new block (5K)
select line
done
The next function uses a new memory block:

struct MemoryBlock *newBlock(len)
ULONG len;
If a new memory block can be used, this memory block structure is

initialized and a pointer returns to the block, otherwise a null is returned
for the block. The memory block must then be inserted in the memory
block list. For this you need a pointer to the current Editor structure:
struct Editor *actualEditor;
The memory block accesses AlloeMem instead of the standard C

function malloe. This route is actually better than using malloe,
because malloe places the used memory in a list so that it can release
all memory not specifically released at the end of the program. This list
uses memory. Since all of the memory should be released before the end
of the program anyway, use AlloeMem.
The same reasons apply for not using the AlloeRemember function,
which the Intuition library provides for memory management. The
advantage to this function is its recall of all memory that you allocated
allowing easy release. The Remember structure must be saved with
Al I oeRemember, which uses additional memory.
The functions presented so far have needed little explanation. This next
function requires some description, however:
409
struct Zline *getZline(len)

UWORD len;
This function searches all the memory blocks of the current editor for a
memory piece that can accept the line. The line length is given in
Line, which must be an even value within the function. It searches for
a memory block. then determines which block is large enough to accept
the text. Since you have not added memory blocks, we can distribute .
the memory pieces in a more optimal way:
First try to find a memory piece that has the exact length of the text If
this doesn't happen, search for the largest memory piece available. This
avoids making a small piece smaller. If a memory piece is so small that
after the new line is cut from it, not enough memory exists for the
memory piece structure, and at least two bytes remain, the search was
unsuccessful. Because the rest must be at least as large, you can insert
another line there.
Remember that each memory block consists of two things: The lines
that lie in it, and the free memory pieces from which room for new
lines is taken.
The following function is the last accessed from newZline:
struct Memoryblock *garbageCollection(len)
UWORD len;
This function reorganizes the free memory inside of the memory blocks
so that only one memory piece exists with the maximum size per
memory block. Each memory piece needs ten bytes of management
information (sizeof(Memorysection. Multiple memory pieces
in a memory block allocate a little free memory. Garbage collection
executes memory block by memory block, until a memory block with
a free memory piece the size of len bytes is found, or until the end of
the memory block list occurs. If the search was successful, a pointer to
the corresponding memory block returns (otherwise, zero is returned).
The pointer can eventually be used to keep the GetZline function
from searching through every memory block.
All of these functions appear in a module called Memory.c. The entire
module is in the following listing:
<src/Memory.c>
/************
**
Includes:
************/
#include
#include
#include
#include
410
<exec/types.h>
<intuition/intuition.h>
<exec/memory.h>
"src/Editor.h"
4.3
ABACUS
STEP BY STEP
1***********
*
* Defines:
************1
idefine BIGBLOCK (200L-sizeof(struct Memoryblock
#define BLOCKSIZE (lOOL-sizeof(struct Memoryblock
1**********************
** External
Functions:
***********************1
struct Memoryblock *AllocMem();
void NewList(),AddTail(),Remove(),FreeMem();
1*********************
** External
Variables:
**********************1
extern struct Editor *actualEditor;
1***************
*
*
*
*
***************1
*
*
Functions:
1*******************************************
**
freeMemoryblock(blk)
* frees the memory of memory

* blocks free, without regard for
* lines that lie in it.
*,
* blk Memoryblock.
********************************************1
A
void freeMemoryblock(blk)
struct Memoryblock *blk;
(
Remove (blk);
FreeMem(blk,blk->Iength + sizeof(struct Memoryblock:
1**********************************************
**
newerBlock(len):
**
*
used memory for new memory block with len length

and returns the pointer
* in case of error => NULL.
**
*
len '" Length of memory block.
411
4.
AMIGA
C FOR
**********************************************/
struct Memoryblock *newerBlock(len)
ULONG
len;
{
register struct Memoryblock *blk;

register struct Memorysection *spst;
/* Allocate memory: */
if (blk = AllocMem(len+sizeof(struct Memoryblock),MEMF_CLEAR
(
/*Initialize memory block structure: *1

blk->length = len;
blk->free = len - sizeof(struct Memorysection);
NewList (& (blk->freeliste;
spst = (struct Memorysection *) (blk + 1);
AddTail(&(blk->freeliste),spst);
spst->len = (UWORD)blk->free;
return (blk);
1***********************************************
* searchMemorysection(block,len)
* Searches for memory piece with len length in the block block
* and returns the pointer or zero in case
* of no room.
* block A Memoryblock.
* len = Length of line (even).
************************************************/
struct Memorysection *searchMemorysection(block,len)
struct Memoryblock
*block;
UWORD
len;
(
register UWORD minl,l;

register struct Memorysection *st,*fnd
NULL;
/* Now searches through if enough memory: */

i f ((ULONG) len <= block->free)
(
/* Minimal length in case found Length <> len */

minI = len + sizeof(struct Memorysection) + 2;
/* Searches through list: */
for (st = block->freeliste.head; st->succ; st
if ((1 = st->len) == len)
st->succ)
/* Optimal size found: */

fnd = st;
break;
else
if ((1 >= minI) && (1 > ((fnd)? fnd->len
/* Search for largest block: */
fnd = st;
412
0)
ABACUS
4.3
STEP BY STEP
return (fnd);
1********************************************
**
**
ConvertSpstToZline(blk,st,len)
Converts memory piece in line.

* If the piece was longer than the line,
* the piece is shortened, otherwise the piece is removed
* from the list.
**
blk " memory block of memory piece.

* st " memory piece.
* len = Length of the line
*********************************************1
struct Zline *ConvertSpstToZline(blk,st,len)
struct Memoryblock
*blk;
register struct Memorysection
*st;
register UWORD
len;
{
register UWORD 1;
register struct Zline *z;
if (st->len
==
(1
= EVENLEN(len)))
1* Found piece fits exactly: *1

Remove(st);
z = struct Zline *)st;
}
else
{
1* .distribute piece: *1
z = (struct Zline *)
UBYTE *)st) + sizeof (struct
+= sizeof (struct Zline)) );
st->1en
1;
t
blk->free
1;
Memorysection) + st->len - (1
z->flags = 0;
z->len' = len;
return (z);
1***********************************************
**
**
getZline (len)
Tries to get line with len length and,

* searches through all of the memory blocks.
* returns pointer to line or
* zero if no room.
* len = Length of the line.

***********************************************1

UWORD
len;
{
413
4.
AMIGA
C FOR
register awORD I;
register struct Memoryblock *blk,*fblk
register struct Memorysection *st,*fst
=
=
NULL;
NULL;
1* make even length: *1

1
EVENLEN(len);
1* run through list of blocks: *1
for (blk = actualEditor->block.head; blk->succ; blk

>succ)
if (st = searchMemorysection(blk,l
if (st->len == 1)
= blk-
1* found optimal memory piece: *1

fblk = blk:
fst = st;
break:
}
else
if (st->len> fst)? fst->len :
1* searches for largest memory piece: *1

fblk = blk:
fst = st:
i f (fst)
return (ConvertSpstToZline(fblk,fst,len;
else
1* no piece found: *1
return (NULL);
1*********************************************
* optimalBlock(block)
* Searches through all of the memory pieces of the block

* and makes one out of all of the memory pieces
* that lie one after another.
* block
Memoryblock.
**********************************************1
void optimalBlock
(blk)
{
register struct Memorysection *stl,*st2,*stle;

for (stl
= blk->freeliste.head;
stl->succ; stl
stl->succ)
1* Certain pointer directly behind stl: *1

stle = (struct Memorysection *)
(UBYTE *)stl) + stl->len + sizeof(struct
Memorysection;
for (st2 = blk->freeliste.head: st2->succ; st2
if (stle == st2)
st2->succ)
1* memory piece st2 lies behind stl: *1

stl->len += st2->len + sizeof(struct Memorysection);
414
ABACUS
4.3
STEP BY ST E P
blk->free += sizeof(struct Memorysection);

Remove(st2);
(UBYTE *)stl)+stl->len+sizeof(struct
Memorysection;
1* begin at the beginning of the list: *1

st2
(struct Memorysection *) &(blk->freeliste.head);
1****************************************
**
garbageCollection(len)
**
*
*
*
*
cleans the memory until at least

len bytes are free. returns pointer.
to memory block where there is room
or zero if no room.
zero if no room.
**
len
= number
of bytes that should be accomodated
*****************************************1

UWORD len;
{

struct Memorysection *st,spst;
register UBYTE *ptr,*end;
register struct Zline *z,*fz;
register UWORD n;
for (blk = actualEditor->block.head; blk->succ; blk = blk>succ)
1* Only garbage collection if more than one memory piece! *1
if st = blk->freeliste.head->succ)? st->succ : FALSE)
{
1* test for beginning and end of block memory: *1

ptr = (UBYTE *) (blk + 1);
end = ptr + blk->length;
1* lines lie at the beginning: *1
while (ptr < end)
{
1* searches for lines that lie at next ptr: *1

fz = NULL;
for (z = actualEditor->zlines.head; z->succ; z
z-
>succ)
if z >= ptr) && (z < end
i f z < fz) II (fz = NULL
fz = z;
i f (fz)
i f (fz != ptr)
{
1* note old pointer: *1
z = fz;
st = (struct Memorysection *)ptr;
415
4.
AMIGA
spst = *st;
1* writes line after ptr: *1

for (n = (EVENLEN(fz->len) + sizeof(struct Zline
1;
n; n--)
*UWORD *)ptr)++ = *UWORD *)fz)++;
1* pointer correction: *1
fz = (struct Zline *)st;
fz->succ->pred = fz;
fz->pred->succ = fz;
if (actualEditor->actual == z)
actualEditor->actual = fz;
if (actualEditor->top == z)
actualEditor->top = fz;
st->succ = spst.succ;
st->pred = spst.pred;
st->len = spst.len;
spst.succ->pred = st;
spst.pred->succ = st;
1* push memory pieces together: *1

optimalBlock(blk);
else
1* set ptr high: */

ptr += EVENLEN(fz->len) + sizeof(struct Zline);
else
1* no more lines in block! *1

break;
/* of while * /
1* Test if there is enough memory: *1
if (searchMemorysection(blk,len
return (blk);
1* of if (at least 2 blocks) *1
return (NULL);
1*****************************************
**
newZline (len)
**
get room for new line,

* fits characters in len.
* returns pointer to new line,
* or zero if error.
*
*
*
len
= number
of characters.
*****************************************1
struct Zline *newZline(len)

UWORD
len;
{
register
register
register
register
416
struct Zline *z;

struct Memorysection *st;
UWORD 1;
ABACUS
4.3
STEP BY STEP
if (actualEditor->block.head->succ)
1* blocks already exist: *1
if .(z = getZline(len))
return (z);
else
1* Garbage-collection:line not directly usable *1
if (blk = garbageCollection(l - EVENLEN(len)))
if (st = searchMemorysection(blk,l))
return (ConvertSpstToZline(blk,st,len));
else
1* may not actually be encountered 111 *1
return (NULL);
else
1* no reults for Garbage-Collection ->pick new block *1
if (blk = newerBlock(BLOCKSIZE))
(
AddTail(&(actualEditor->block),blk);
return (getZline(len));
}
else
return (NULL);
else
1* no block in the block list: *1

if (blk = newerBlock(BIGBLOCK))
(
}
else
return (NULL);
Some functions appear in the above code that haven't yet been
described. This is because some jobs can be done better with functions,
and you don't have to know all the details on every function. Here are a
couple of words for the additional functions:
void freeHemoryblock(blk)
This function releases the memory occupied by Memoryblock blk.

There is no consideration if other lines lie within a block.
CloseEdi tor uses this function to free the entire memory of the
editor.
struct Hemorysection *searchMemorysection(block,len)
struct Memoryblock
*block;
UWORD
len;
This function searches the given block for a memory piece that either
has the length len, or for the largest existing memory piece. Either a
pointer to the found memory piece or null returns, which means none
of sufficient length could be found. Be very careful that blk - >f ree
contains the sum of the lengths of the individual memory pieces of this
block, so that a simple check determines if enough space is available. If
the length could not be found, only the larger memory pieces are of
interest: those (sifeof(struct Memorysection) +2) bytes larger
417
than len, because the found memory piece must be distributed between
a line and a memory piece. When no more room exists for the
structures, no distribution occurs.
The following line may be difficult to understand:
if 1 >= minl) && (1 > fnd)? fnd->len : 0)
For those who have trouble with the above line, here is a small tip. If
fnd equals zero, no memory piece was found and cannot be tested to
see if 1 is greater than the length of the memory pieces already found
(fnd). In any case, the second comparison should be true. The
expression fnd) ? fnd->len : 0) equals zero if fnd is zero.
Otherwise fnd ->len (the length of the memory piece fnd). Because
1 is always positive, 1 is always greater than 0, letting you bypass
the special case when fnd is still zero.
struct Memoryblock
*blk;
*st;
register UWORD
len;
When you want to use a new line, you must convert a memory piece
into a line, or at least take a line from a memory piece. The function
mentioned earlier lowers the amount of work required for this
conversion. If the memory piece was as long as the line, it is removed
from the list of all of the memory pieces. Otherwise the line is removed
from the end of the memory piece with the corresponding length. In
each case the number of free bytes of the block belonging to the
memory piece decreases. It then returns a pointer to the line.
The following function appends memory pieces that lie directly behind
one another:
void optimalBlock (blk)
You'll always need this function when you lengthen a memory piece or
insert a new memory piece in the list. You must be sure that no two
memory pieces are directly behind one another, because this costs you
10 bytes (sizeof(struct memorypiece.
The following flowplan describes the garbageCollection
function:
FOR all blocks
are the~e at least 2 memory pieces in this block?
move all lines in this block to their beginning
join all memory pieces into one
is there enough room in this block?
return the pointer to the block
else: return zero
418
4.3
ABACUS
4.3.4
STEP BY STEP
Testing the new functions

Now we should test the functions created in the memory module. We
will create a module used to test our newly created functions. The
following t est. c module allows you to call all of the important
functions and display the occupied memory:
<src/Test.c>
/************
**
Includes:
************/
ilinclude "src/Editor.h"
/***********
**
Defines:
************/
/**********************
**
External Functions:
**********************/
struct Zline *newZline();

UBYTE *gets(),getchar();
UWORD strlen () ;
void strncpy(), AddTail(),deleteZline();
/*********************
**
External variables:
**********************/
/***************
* Functions: *
***************/
void print ()
{

printf ("%s\n", z+l);
z->succ)
419
4.
AMIGA
void print blocks()
register struct Memoryb1ock *b1k;

printf("B1ock list:\n\n");
for (b1k = actua1Editor->block.head; blk->succ; b1k
>succ)
printf("B1ock %8lx, Lenght = %ld, Free = %ld\n",
b1k,b1k->length,b1k->free);
printf("\nn);
void input 0
(
UBYTE 5(80);

register UWORD 1;
printf("New Line: HI;
i f (gets (s) )
(
1 = strlen (5);
i f (s[1-1) != , ')
(
s[l++) = 13;
s[l) = 0;
if (z
newZline(l
strncpy(z+l,s,l);
AddTai1(&(actua1Editor->zlines),z):
else
printf ("Error in new line!!! \n");
else
printf("Error with gets!!!\n");
void Test 0
{
register UBYTE c;
printf(">");
while c = getchar()
!= 27)
while (getchar() != 10) ;

switch (c)
{
case 'a':
input ();
break;
case 'p':
print ();
break;
case 'b':
print blocks () ;
break;
420
/* read one CR!
*1
= b1k-
ABACUS
4.3
STEP BY STEP
printf (">");
The print function displays the entire text that you have entered. The
print_blocks function specifies the starting address,line lengths
and the number of free bytes of all of the memory blocks currently in
the editor. inputO expects line input that is added to the end of the list
of all lines. Otherwise, as you do later in the real editor, a null byte
represents the end of line, with which the line forms the output of all of
the lines as simply as possible. There the printf function expects a
pointer to a string as a parameter, which ends with a null byte.
These functions are called from the Test function which waits for line
input from a user. You can add a line by pressing <8> (Append), print
the text by pressing <p> (print), and read a list of all of the memory
blocks by pressing <b> (Blocks). After you enter a letter you must
press the <RetUrn> key, because we still use the CLI window for the
input routine and it is line oriented (input will be sent only after
pressing <Return. The <Return> is not used in the switch case.
Now you must change your program so that the Te s t function is
automatically called. Change the Editor module as follows. First
define the external functions:
void Test () ;
void freeMemoryblock();
Then change the CloseEditor function so that all memory is

released again:
struct Editor *00;
{

/* Free first memory block: */
while (blk = (struct Memoryblock *)RemHead(&(ed->block)
freeMemoryblock(blk);
/* Then Close window: */
ed->window->UserPort = NULL;
CloseWindow(ed->window)i
free (ed) ;
And in the main program add the Test function before the main loop
and before the wai t function:
Test ();
Finally, change the makefile so that the two new modules are also
compiled. To do this, expand the line in which OBJ is defined:
QBJ=src/Editor.o src/Memory.o src/Test.o
421
4.
AMIGA
At the end of the make file insert the following two lines:
src/Memory.o: src/Memory.c src/Editor.h pre/Editor. pre
src/Test.o: src/Test.c src/Editor.h pre/Editor.pre
The last change that you should make is set the BIGBLOCK and
BLOCKSIZE defines to smaller values so that you don't have to
enter 10K of text later before you get a second memory block. Change
10K into 120 bytes and 5K into 60 bytes. This also lets you check
what happens when you want to insert a line that is longer than an
entire empty block.
Compile, link and test. When inserting a line that is larger than the
block, a new block is used, even though the line cannot fit there.
Because you set the block size to 5K for later, the following happens: if
only 22 bytes are free in a block, a line with 21 or 22 characters fits
there; or if you insert one with less than 10 characters, then there must
be room for a Memorysection structure.
4.3.5
Deleting Lines
Until now you have only tested the functions for inserting a new line.
To be able to test the functions optimalBlock and garbage
Collection, you need a function that can delete any line. This
function first removes the line from the list, searches for the block
where it lies and converts the line into a memory piece. Then the block
is optimized. The function for the editor looks like this:
void deleteZline
(line)
register struct Zline *line;
{
,
register struct Memorysection *st;
= NULL;
/* remove line from list: */

Remove (line) ;
/* set pointer to this line to zero: */
if (actualEditor->actual == line)
actualEditor->actual = NULL;
if (actualEditor->top == line)
actualEditor->top = NULL;
/* determine in which block the line lies: */
for (blk = actualEditor->block.head; blk->succ; blk = blk>succ)
if ((line> blk) && (line < (((UBYTE *) (blk + 1 + blk>length )
{
fblk = blk;
break;
422
ABACUS
4.3
STEP BY STEP
i f (fblk)
{
1* convert line into memory peices, insert in list: *1

fblk->free += ( struct Memorysection *)line)->len
= EVENLEN(line->len) );
AddTail(&(fblk->freeliste),line);
optimalBlock(fblk);
if (fblk->length == (fblk->free + sizeof(struct
Memorysection)
1* Block contains no more lines: Delete! *1
freeMemoryblock(fblk);
If the line was the last line in the memory block, this is deleted. Before
you write a few additional functions for testing, a note about
ConvertSpst ToLine. Because it was easy to program, we created
the function so that the line ends if the line to be inserted is smaller
than the memory piece. This means that the lines in a block extend
from back to front.
You have already determined that the garbage collection
moves all the lines to the beginning of the block. It's not too late to fIx
this back-to-front problem. Change the ConvertSpstToLine
function so that the lines extend from front to back:
struct Memoryblock
*blk;
*st;
register UWORD
len;
{
register UWORD 1;
z = (struct Zline *)st;
if (st->len -= (1 = EVENLEN(len)
Remove(st);
else
{
1* distribute piece: *1
UBYTE *)st) += (1 += sizeof (struct Zline;
( st->succ '"' struct Memorysection *)z)->succ )->pred
( st->pred = struct Memorysection *) z) ->pred ) ->succ
st->len = struct Memorysection *)z)->len - 1;
}
blk->free
-=
= st;
= st;
1;
z->flags = 0;
z->len z len;
retum (z);
}
A small disadvantage to this solution: The pointer for the chaining

must be adjusted. This gives you a chance to play with the many
423
4.
AMIGA
C FOR
possibilities of C. The assembler code that changes the pointer is fairly

efficient Look at the assembler code. Compile the Memory module as
follows:
cc +B +Ipre/Editor.pre src/Memory.c -a -t
-0
ram:sp.asm
The created assembler code can be found on the RAM disk under the
name sp.asm. Because the assembler code contains the C commands as
comments. it's not too difficult to find the place. Search for the label
_ConvertSpstToZline, then cursor down until you see the above
lines as comments. Directly under that you should fmd the assembler
code for these lines. This code looks like this:
;
struct Memorysection *)z)->succ )->pred
= st;
move.l (a3),aO
;a3 = z
;a2 = st;
move.l aO, (a2)
move.l a2,4(aO)
( st->pred = struct Memorysection *)z)->pred )->suoc
move.14(a3),aO
move.l aO,4(a2)
move.l a2, (aO)
= st;
(st->succ
Notice how the assembler code looks is similar for both functions. You
can improve it. For example, the following section from the
garbageCollection function moves a piece of memory:
/* push line down after ptr: *1
for (n = (EVENLEN(fz->len) + sizeof(struct Zline
;
;
;
1;
n; n--)
move.l dS,aO
move.l #O,dO
move.b 9(aO),dO
add.w #l,dO
belr.l #O,dO
add.w #10,dO
move.w dO,d6
lsr.w #l,d6
bra .74
;dS = fz
;d6 = n
;a2 = ptr
.73
;
*UWORD *)ptr)++ = * UWORD *)fz)++;

move.l dS,aO
add.l #2,dS
move.l a2,al
add.l #2,a2
move.w (aO), (al)
.71
sub.w #l,d6
.74
tst.w d6
bne .73
.72
;
Here you can improve the loop labeled. 73. You can replace the five
instructions directly after label. 73 with one (MOVE.W (An)+ ,(Am)+) if
you place the variable f z in an address register instead of in a data
register You can append the three instructions after label .71 together
424
4.3
ABACUS
STEP BY STEP
using a DBRA instruction. If you later decide to remove this function,

then you can place it at 71 with improvements.
Now you will need to add functions to the test. c module, so you can
delete lines. It would also be interesting to know how the memory of
the individual memory blocks is distributed. Look at the following
functions:
void print memory()

register UBYTE *ptr,*end;
register struct Memorysection *st;
struct Zline *fz;
struct Memorysection *fst;
printf (IIMemory construction: \n");
for (blk = actuaIEditor->block.head; blk->succ; blk = blk>succ)
{
printf (II\nBlock %Slx, Lenght = %ld, Free = %1d\n",

blk,blk->length,blk->free);
while (ptr < end)
{
fz = NULL;
fst = NULL;
1* search for the line that is next to the ptr: *1

for (z = actuaIEditor->zlines.head; z->succ; z = z->succ)
if (z < fz)? fz : (struct Zline *)end
fz = z;
1* No line found or line> prt -> search memory section *1

i f fz > ptr) I I ! fz)
{
for (st = blk->freeliste.head; st->succ; st = st->succ)

i f st >= ptr) && (st < end
if (st < fst)? fst : (struct Memorysection *)end
fst = st;
1* In case memory section> ptr -> error!! I!! *1

if fst > ptr) II ! fst)
{
printf (" Unused memory section! \n");

i f (fz)
i f (fst)
i f (fst < fz)
ptr = (UBYTE *)fst;

else
ptr = (UBYTE *)fz;
else
ptr = (UBYTE *)fz;
else
i f (fst)
425
4.
AMIGA
ptr = (UBYTE *)fst;

else
break;
else
{
1* Memorysection: *1
printf(fOMemory section (%d)\n",fst->len);
ptr += sizeof(struct Memorysection) + fst->len;
else
{
1* Zline *1
printf(fOLine: %s\n",fz+l);
ptr += sizeof(struct Zline) + EVENLEN(fz->len);
printf ("\n");
void delete_line()
{

UWORD nr;
printf("Number of line that should be erased: [0 nJ ");
scanf (fO%d fO , &nr);
while (getchar() != 10);
1* read one CR! *1
for (z = actualEditor->zlines.head; z->succ && nr; z = z>succ, nr--);
i f (z->succ)
{
printf(fODelete: %s\n",z+l);
deleteZline (z) ;
The print memory function looks complicated, but all it does is

display the lines and memory sections in the order in which they appear
in the block. When a function determines that the block contains a
piece of memory that is neither a line nor memory section, an error
message is displayed. The function tries to trap the error that the pointer
pt r sets at the beginning of the next line or memory piece. The
pointer ptr points to the start of the memory area that is searched, and
end points to the end of the same.
The second function erase_line looks much simpler. This requires
a number to be input and deletes the line that corrseponds to the
number, starting the count with zero. Unfortunately the while loop
with getchar must exist, because otherwise an end of line remains.
This makes sure that you can enter input after the <Return> key is
426
ABACUS
4.3
STEP BY ST E P
pressed. There could be a more elegant solution, but it is only a test

program designed for functionality. In addition, the test functions will
not be in the finished program.
The switch case ofthe Test function needs the following lines so
that you can examine the new test functions:
case's':
print_memory () ;
break;
case 'd':
delete line 0 ;
break;-
Compile the program and examine the individual modules. Add new
lines, delete others and constantly examine the memory arrangement.
Maybe you will find some things that can be improved. The complete
source text and the compiled editor can be found in directory VO. 2 of
the optional disk for this book.
4.3.6
Text Output in a Window

This section implements functions that let you display text in the
editor's own window instead of the CLI window. As in almost
everything, the Amiga and its operating system offer different ways of
reaching the same goal.
First you must open the window, then specify the window's size and
other qualities. Chapter 3 described window parameters. You have a
number of these parameters available.
The output window currently in use is a perfectly normal window.
Look at the data in the Newwindow structure of the main program,
which includes the following numbers:
100,40,440,156
This is the starting position of the window. We will change this as

soon as we have most of the editor functions tested and running.
The following line ensures that your window appears in the same colors
as the Workbench window:
AUTOFRONTPEN,AUTOBACKPEN
The rDCMP flags follow next:

REFRESHWINDOW I MOUSEBUTTONS I RAWKEY I CLOSEWINDOW I NEWSIZ,
427
The program needs information about redisplaying window data. This

information states whether a mouse button or key has been pressed,
whether the user has clicked on the window's close gadget, and whether
the window's size has changed (more on window size in a moment).
Next you should assign the window flags:
WINDOWSIZING 1 WINDOWDRAG 1 WINDOWDEPTH 1 WINDOWCLOSE 1
SIZEBOTTOM
Window
sizing
The window should have facilities for enlarging, reduction and

movement. You don't want to exit with the close gadget, otherwise the
CLOSEWINDOW flag would be meaningless. The size gadget for
enlarging or reducing appears in the lower right corner of the window.
The ACTIVATE flag ensures that the window is active at the same time
it is opened.
The SIMPLE REFRESH flag is one we haven't mentioned until now.
This flag determines the operating system's next task if the user moves
another window over the editor window. This action wipes out the
contents of the window hidden by the overlapping window. When you
remove the overlapping window, the operating system must refresh
(redraw) the deleted contents of the window. Chapter 3 mentioned three
ways of screen refreshing:
SIMPLE REFRESH
Redraws the window.
SMART_REFRESH
Recreates the window from window data

available in RAM.
SUPER BITMAP
Stores text and graphic contents of window in

a buffer for later recall.
SIMPLE_REFRESH or SMART_REFRESH works quite well for any

window used for simple text display. Since the editor window will stay
at the front most of the time (Le., won't be covered too often),
SIMPLE_REFRESH is enough in most cases. In addition, the
SIMPLE_REFRESH window uses less memory than the
SMART_REFRESH window.
The SIMPLE_REFRESH window has one disadvantage that you'll see

when you start scrolling. The ScrollRaster function moves the
screen contents, thus scrolling the window. If you scroll in a
SIMPLE_REFRESH window with another window covering it, some
problem occurs with refreshing the formerly covered section of the
window. If you wish, replace S IMP L ERE F RES H with
SMART_REFRESH. SMART_REFRESH movesfaster, but also uses
more memory.
Clipping
428
There is another problem with text output in your window. You do not
want to write in the border of your window. The GIMMEZEROZERO
ABACUS
4.3
STEP BY STEP
flag provides you with your own RastPort for input inside of the
window. The window border has its own RastPort as well, so that you
cannot write in the border. But this also requires more memory, and the
output is slower because you must manage two RastPorts from
Intuition. You want the simplest, more efficient way, so clipping
makes sure that your text does not run into the screen border.
Two functions exist for the actual text output. One comes from the
graphics library and is called Text; the second comes from
Intuition and is called PrintIText. You assign Text a pointer to
the RastPort, the text and the length. PrintIText needs one pointer
to the RastPort, one pointer to the IntuiText structure and X and Y
coordinates. You can specify the text's writing mode, color, position
and the font parameters. The text to which the IntuiText structure
points must end with a null byte.
At first glance the Intuition functions appear to be more elegant than
the graphic functions, which look like sloppy programming. The
elegance of Intuition functions come with a price-time. Which special
functions do we need to output text in our editor? We need to determine
the output position (Le., Move from graphics library). We
want to display control codes in red and change the text color back
(SetAPen from graphics library). We also want reserved C
words to appear in bold print (SetSoftStyle). Those are all the
functions we need from graphics library. We chose the more
primitive functions because they execute faster than normal operating
system functions.
How can we make sure that text doesn't overwrite the window's border?
Simple. We first establish the total size of the window within the
borders, measured by character height and width. Then we determine
when no more characters should be displayed. We place the width and
height of the window in the Editor structure, because this value can
be shared by multiple windows. Since these two values change as the
window size changes. we must reset the window using the NEWSIZE
flag listed in the I DCMP flags. Every time a corresponding report
returns, the program recalculates window height and width. A
SIMPLE_REFRESH window returns a REFRESHWINDOW report, and
redisplays the contents of the window.
The right and bottom borders still remain. These borders don't belong to
either the window border or the text (the window width is not divisible
by the letter width). The same goes for the height. The following
illustration shows exactly how we solved this problem:
429
fll
I4-CN-.!
ch
1.
4-
Basel ine
yoff
_ _ _ _u
yru
Figure 4.1
xrl
xoff
xrr
The letters are not proportional to make it easier to view. The letter
above the window shows an example of the height and width of the
characters (cw =character width, ch = character height). Because we
don't use any proportional character sets in this example, these values
remain the same for all letters. The window width in characters (wcw =
WindowCharacterWidth) comes from subtracting the width of the
window (window->Width) from the width of the left border
(Window->BorderLeft) and right border (window>BorderRight), then dividing the result by the width of one
character (cw).
The height is calculated in much the same way:
wch
(window->height - window->BorderTop
- window->BorderBottom) I chi
The width of the left and top screen borders specify the top left position
of the window. We save the two values to the Edi tor structure as
xoff (X offset) and yoff (y offset). The same goes for cw and ch.
which can change from window to window if you change the character
set using Preferences, then open another window. A pointer to the
RastPort of the window belongs in our Editor structure so that we
don't always have to access the window structure. We know that
graphic output on the Amiga is through the RastPorts.
You know what to do about the borders. The editor may never need
this, but it will keep graphic garbage to a minimum. Each time the
program receives a REFRESHWINDOW report it deletes the contents of
430
ABACUS
4.3
STEP BY STEP
both borders. We need additional information for this (e.g., the border
positions). The above illustration listed additional coordinates with xrl
(X, Border, left comer), xrr (X, border, right corner), and yru (y,
border, bottom corner). We need the top edge of the border, which is
given in the drawing by Y. Y provides itself in part, so no calculations
are needed yet. Display the text line by line and count the Y
coordinates. If either the window fills up or no more text is available,
the Y coordinate of the line appears directly beneath the last text line.
This corresponds exactly to the top corner of the border.
There are more details given in the program. When positioning the text,
be careful that the text always appears relative to its baseline (see
illustration above). We must increase the Y position to match the
distance of the baseline from the top edge of the letter before calling
Move.
Now combine the individual functions that handle text output into one
function:
void printAll ()
This function displays the entire text. Next the program tests the top
line in the window as long as the line appears, until the window fiUs or
the end of the text occurs. The printLine function passes control to
a pointer to the line and the current Y position. Then the two borders
are deleted, provided they are greater than zero.
The next function displays the line left-justified at the Y position:
void printLine(line,y)
This is not as easy as it sounds. Our editor works with true tabs, which
would appear as little graphic boxes. When we let the output run from
the console.device, the tabs are displayed correctly. The console. device
uses too much memory overhead, which makes the output slower and
won't let us start anything. We must convert the tabs ourselves.
Another function performs this task:
void convertLineForPrint(text,length,width,buffer)
This function expects a pointer to the text (not the Line structure) and
its length. You must assign a pointer to a buffer in which the converted
lines are written, because spaces fill in the buffer to the end of the line.
This function doesn't display a visible CR/LF, because the CR or LF
should not appear at the end of a line (for aesthetic reasons, if for
nothing else). If the line doesn't end with a CR, LF or CR/LF, the
function fills the rest of the line with dots (CHR$ (183) ) This also
applies to lines that don't completely fit in the window because they are
too long. A point at the end of a line means that the lines continues,
and multiple points mean that the line doesn't conclude with a normal
end of line character.
431
4.
AMIGA
We can now convert and display the line using the Text graphic
function. Because a line can include spaces, or because the line is
narrower than the window ,the line could be indented for fonnatting
reasons. This formatting becomes inefficient. Displaying spaces
requires the same amount of time as displaying a letter or number. A
faster method would be to replace any larger number of spaces by
drawing a rectangle the same size as the spaces to be displayed. The
following function perfonns this task:
void printAt(buffer,length,x,y)
The printAt function also handles control character output Control

codes are displayed in red. For example, CRR$ (1) appears as a red "A"
("A" = CRR$ (1 I 64). A pointer to the text controls length and
the X coordinate position according to xoff.
One remaining text display function controls output, wcw, wch,
xoff, etc.:
void initWindowSize(ed)
The variable ed points to the Editor structure whose value should be

tested. Before we look at these functions, we should review our list of
the features that our editor should have.
Folding
Folding refers to the ability to fold text into areas in memory where the
line wouldn't nonnally be visible. Usually you would search through
hundreds of lines to find a routine. Folding allows you to move
routines into their own areas, marked by the following string:
I*#FOLD: Routine_Name *1
Each fFOLD statement acts as a line pointer which indicates a set of

folded lines behind the statement. The pointer must be activated to get
to the folded segment. This takes less time than scrolling through a full
listing. Because the maximum number of lines that fits in the screen
decreases, 50 (same as 400, the maximum resolution) divided by 8
(character height), we can also detennine the size of this field.
We fIrst want to expand the include fIle Edi tor.h:
#define
#define
#define
#define
#define
MAXHEIGHT 50
FGPEN 1L
BGPEN OL
CTRLPEN 3L
CONTROLCODE(c) (c)&127)<32)
struct Editor
{

struct BList block;
UBYTE buffer[MAXWIDTH):
432
4.3
ABACUS
STEP BY STEP
UBYTE tabstring [MAXWIDTH] :

UWORD num lines:
struct Zline *actual:
struct Zline *zlinesptr[MAXHEIGHT]:
UWORD toppos:
UWORD xpos,ypos;
UWORD changed:l;
UWORD insert:l;
struct RastPort *rp:
UWORD xoff,yoff;
UWORD CW, Chi
UWORD wcw,wch:
UWORD xrl,xrr,yru,bl;
};
In addition to the maximum window height, the standard foreground

color (FGPEN = ForegroundPen), the background color (BGPEN =
BackgroundPen) and the color for the control codes (CTRLPEN) are
defined. We also define what a control code is (CONTROLCODE)-a
character with a code less than 32, or greater than or equal to 128 and
less than 160. The Amiga ASCII table displays these characters as
rectangles. If you use your own character set in which you have
assigned your own character to these codes, you can change the
definition of CONTROLCODE.
The edi tor structure changes slightly. Instead of the top pointer there
is now the zlinesptr array. zlinesptr[O] serves the same
function as top had before. Additional elements for text output are
added to the end of the structure as discussed previously. The two
functions of the Memory. c module that make use of actual must
be rewritten. In the function garbageCollection the lines where
zprt appears must be defined as struct line **zptr. Search
for the following lines:
And add the following lines:

for (n = 0, zptr = actualEditor->zlinesptr:
n <: actualEditor->wch; n++, zptr++)
i f (*zptr = z)
{
*zptr = fz;
break:
The function deleteZline also needs a change. The lines listed

below:
if (actualEditor->actual == zline)
actualEditor->actual = NULL:
require the addition of zprt and n (UWORD). Add to the above lines
the following:
433
for (n = 0, zptr = actualEditor->zlinesptr;

n <= actualEditor->wch; n++, zptr++)
if (*zptr == line)
{
*zptr = NULL;
break;
Now we come to the changes in the main E di tor. c module.

printAll is not the only function that we are going to add:
void initWindowSize(),SetDrMd(),SetAPen(),SetBPen();
void printAll(), BeginRefresh(),EndRefresh();
The following two new functions test for one line before or after the
current line. These may seem unnecessary, but the folding extension
will need them later:
1*********************************************
* nextLine (line)
* Returns pointer to next line

* (and executes folding)
* zero if not next line
* line
line structure.
*********************************************1
struct Zline *nextLine(line)

{

i f (line)
{
if (z
line->succ)
i f (! (z->succ) )
NULL;
return (z);
else
return (NULL);
1*******************************************
**
prevLine (line)
* Returns pointer to previous line

* (and executes folding)
* zero of no previous line
*
*
* line
line structure
*
*
*******************************************1
434
4.3
ABACUS
STEP BY STEP
struct Zline *prevLine(line)

{

i f (line)
{
if (z
line->pred)
i f (! (z->pred
z = NULL;
return (z);
}
else
return (NULL);
The OpenEdi tor function was also expanded:

(
register
register
register
register
register
register
register
struct Editor *ed = NULL;

struct Window *wd;
struct RastPort *rp;
ULONG flags;
UBYTE *ptr;
struct Zline **zptr;
UWORD n;
1* IDCMPFlags saved, sets it to zero in structure

=> use your own UserPort! *1
1* Allocate memory for memory structure: *1

if (ed = malloc(sizeof(struct Editor)
1* Window open: */
{
ed->window = wd;
ed->rp = (rp = wd->RPort);
1* Set write mode: *1

SetDrMd(rp,JAM2);
SetAPen(rp,FGPEN);
SetBPen (rp, BGPEN) ;
1* UserPort established: *1
wd->UserPort z edUserPort;
1* Parameter initialization: *1
ed->num lines z 0;
ed->actual = NULL;
ed->toppos = 0;
ed->xpos z 1;
ed->ypos = 1;
ed->changed z 0;
ed->insert = 1;
435
/* zlinesptr-Array initialization: */
for (n = 0, zptr = ed->zlinesptr; n < MAXHEIGHT; n++,
zptr++)
*zptr = NULL;
/* Tab initialization: */
ptr = ed->tabstring;
'*ptr++ = 1;
for (n = 1; n < MAXWIDTH; n++)
i f (n % 3)
*ptr++ = 1;
else
*ptr++ '" 0;
ed->wch = 0;
initWindowSize(ed);
else
(
free(ed);
ed = NULL;
return (ed);
flags;
Tab initialization gives the user a couple of predefined tabs. Two new
flags must be inserted in the routine which checks the report types in
the main program:
case NEWSIZE:
initWindowSize(actualEditor);
print ();
break;
case REFRESHWINDOW:
BeginRefresh(actualEditor->window);
printAll () ;
EndRefresh(actualEditor->window,TRUE);
break;
The functions BeginRefresh and EndRefresh should be called

after a REFRESHWINDOW event so Intuition knows that the window
was restored. The functions also ensure that only the sections of the
screen that need it have been renewed (see Chapter 2 for information).
The print function originated in the test. c module. Before we
assign this, let's implement a new module named output. c:
<src/Output. c>
/************
* Includes:
*************/
436
43
ABACUS
STEP BY STEP
/***********
* Defines:
************/
Itdefine CR 13
#define LF 10
#define TAB 9
/**********************
**
External Functions:
**********************/
void SetAPen(),RectFill(),Text(),Move();
struct Zline *nextLine();
/*********************
**
External Variables:
*********************/

/***************
**
Functions: *
****************/
/**********************************
* initWindowSize(ed)
* Initializes the width/height

* statement of the actual windows.
* Restores the zlineptr field
*
**
ed
Editor structure.
**********************************1
void initWindowSize
(ed)
register struct Editor *ed;
{
register
register
register
register
w
rp
struct Window *w;

struct Zline *z,**zptr;
UWORD n,newh;
= ed->window;
= ed->rp;
ed->xoff
ed->yoff
(UWORD)w->BorderLeft;
(UWORD)w->BorderTop;
437
4.
ed->cw
ed->ch
AMIGA
(rp->TxWidth)? rp->TxWidth : 8;
(rp->TxHeight)? rp->TxHeight : 8;
(w->Width - ed->xoff - w->BorderRight) / ed->cw;

ed->wcw
(w->Height- ed->yoff - w->BorderBottom)/ ed->ch;
newh
ed->xrl
ed->xrr
ed->yru
ed->bl
=
=
=
ed->xoff + ed->wcw*ed->cw;
w->Width - w->BorderRight - 1;
w->Height - w->BorderBottom - I:
rp->TxBaseline;
/* Now restore the zlineptr field: *1

if (newh != ed->wch)
{
if (newh < ed->wch)

{
/* Pointer to Null set: *1

zptr = ed->zlinesptr + newh + 1;
for (n = newh; n < ed->wch; n++)
*zptr++ = NULL;
else
{
zptr = ed->zlinesptr + ed->wch;

z
= *zptr++;
for (n = ed->wch; n < newh; n++)
*zptr++ = (z = nextLine(z;
ed->wch = newh;
1****************************************
**
convertLineForPrint(line,len,buf)
* Convert line so that this can be

* displayed simply.
* Tabs converted into spaces
*
**
line" Zline.
* len = the length.
* w : maximum wiidth.
* buf "buffer, in which the converted line
*
is stored.
****************************************/
void convertLineForPrint (line, len,w,buf)

UBYTE
*line;
register WORD
len;
register UWORD
w;
register UBYTE *buf:
{
register UBYTE *tab;

register awORD 1 = 1;
register UBYTE lastchar;
/* determine if ends with CR or LF: *1
i f (len)
438
ABACUS
4.3
tab
STEP BY STEP
line + len - 1;
CR)
i f (*tab =
{
len--;
lastchar
I.
else i f (*tab
LF)
len--;
lastchar = I ' ;
if len) && (*--tab
len--;
else
lastchar
else
lastchar
CR
183;
183;
/* Zline convert: */
tab = actualEditor->tabstring;
while len--) && (1 < w
i f *buf = *line++) = TAB)
cD
{
*buf++ =
1++;
';
tab++;
while *tab) && (1 < w;

else
{
buf++;
tah++;
1++;
1* determine if the line is not completely converted: */

if
(len >= 0)
lastchar = 183:
/* line filled from last character to end: */

while (1++ <= w)
*buf++ = lastchar;
/************************************
* printAt(buf,len,x,y)
**Prive
character string buf a position

* (x,y) in a window. The character* string is changed!
*
*
*
*
buf A character sring.

len = the length.
x = Pixel-X-Position.
y = Pixel-Y-Position.
*************************************/
void printAt
(buf,len,x,y)
439
4.
AMIGA
C FOR
register UBYTE *buf;

len;
register UWORD x,y;
WORD
{
struct RastPort *rp = actualEditor->rp;

register UWORD cw = actualEditor->cw;
register UWORD x2;
UWORD y2 = Y + actualEditor->ch - 1;
UBYTE *ptr;
ULONG 1;
while (len> 0)
i f (*buf = , ')
{
1* output spaces: *1
x2
x-I;
while *buf =
, ') && len--)
buf++;
x2 += cw;
SetAPen(rp,BGPEN);
RectFill(rp, (ULONG)x, (ULONG)y, (ULONG)x2, (ULONG)y2);
SetAPen(rp,FGPEN);
x = ++x2;
}
else
{
1* Also output text: *1

Move(rp, (ULONG)x, (ULONG)y + actualEditor->bl);
ptr = buf;
if (CONTROLCODE(*buf
{
SetAPen(rp,CTRLPEN);
while (CONTROLCODE(*buf) && len--)
*buf++ 1= 64;
1 = buf - ptr;
Text (rp,ptr,l);
SetAPen(rp,FGPEN);
else
{
while (!CONTROLCODE(*buf) && (*buf != , ') && len--)

buf++;
1 = buf - ptr;
Text (rp, ptr, 1) ;
x += cw*l;
1************************************
* printLine(line,y)
* Displays Zline at pixel oosition

*y on the screen.
440
ABACUS
4.3
STEP BY STEP
* line A Zline structure.

* y = pixel positon.
************************************1
void printLine
(line,.y)
register UWORD
y;
{
static UBYTE buf[MAXWIDTH];

register UWORD w;
convertLineForPrint(line+l,line->len,w
>wcw,buf) ;
printAt(buf,w,actualEditor->xoff,y);
= actualEditor-
1**********************************
* printAll
* redisplays the entire screen.
**********************************/
void printAll ()
{
register UWORD y,ch,count;

register struct Zline **z;
register struct RastPort *rp:
y
ch
= actualEdi tor->ch;
count
Ip
actualEditor->yoff;
= actualEditor->wch;
actualEditor->rp:
/* Output the line: *1

for (z = actualEditor->zlinesptr; count-- && (*z != NULL):
z++, y += chI
printLine(*z,y);
1* erase right and bottom border: *1

SetAPen(rp,BGPEN);
if (actualEditor->yru >= y)
RectFill(rp, (ULONG)actualEditor->xoff, (ULONG)y,
(ULONG)actualEditor->xrr, (ULONG)actualEditor->yru);
if actualEditor->xrr >= actualEditor->xrl)
&& (y > actualEditor->yoff
RectFill(rp, (ULONG)actualEditor->xrl, (ULONG)actualEditor>yoff,
(ULONG)actualEditor->xrr, (ULONG)y-l);
SetAPen(rp,FGPEN):
Now here are a few comments about the above functions and the special
additions:
'
The ini tWindowSize function performs a security check

with the size of the character set and a null.
441
If the window was enlarged or reduced, either additional pointers

are calculated or the unnecessary pointers are set to zero. The
nextLine function will make the implementation of folding
easier. Then we only need to change the nextLine and
prevLine functions. All other functions that access these two
functions can work with folding.
convertLineForPrint searches for the end of line and sets

the variable lastchar. This variable contains the characters
that fill in the rest of the line. The tab variable acts as the
pointer to the end of line. Be sure that there is at least one
character (usually a space or a period) displayed behind the actual
text of the line. If the window is 70 characters wide, the line is
only 69 characters wide.
In printAt the x2 and y2 variables contain the coordinates of

the bottom right comer for the rectangle that must be displayed
instead of blank spaces. The background color must be assigned
to the rectangle beforehand. Otherwise the program displays a
white (visible) rectangle.
The many type conversions are necessary because we must

assign LONG values to the operating system functions.
The function printLine defines the buffer as static,

otherwise it takes up too much room on the stack. In addition,
other local variables have space reserved on the stack, but no
more than 80 bytes at a time.
Calling the Re c t F i 11 function with the coordinates that

separate the left comer from the right or the top comer from the
bottom make the Amiga run faster. The check in printAll is
very important to this. Besides the definition of xrl, xrr and
yru, the borders may be assigned negative widths. (i.e., if the
window's width or height is divisible by the width or height or a
character). In that case no border can be displayed.
So far, so good. We need changes to the test. c module. First declare

a few external functions:
void printAll();
Now rewrite the print function so that it supplies the values to the
pointer established in the zlinesptr array:
void print ()
{

register UWORD n;
for (n = 0, z = actualEditor->zlinesptr; n <z actualEditor>wch;
442
ABACUS
4.3
STEP BY STEP
n++, z++)
printf("Line %d an Adresse %61x.\n",n,*z);
There is a new function for initializing the zlinesptr field. It won't

be needed later in the actual program, so we placed them in the
test. c module:
void init zlinesptr()
{
register struct Zline *z,**zptr;

register UWORD n;
for (n ~ 0, z ~ (struct Zline *) &(actualEditor->zlines.head),
zptr ~ actualEditor->zlinesptr; n < MAXHEIGHT; n++)
*zptr++ = (z ~ nextLine(z));
Finally the actual test program needs a change. After inserting and
erasing lines, zlinesptr must be re-initialized. Furthennore, the
printAll function must be called by the command p (print). The
corresponding branch of the switch case application looks like the
following:
case 'a':
input ();
init zlinesptr();
break;
case 'p':
print ();
printAll () ;
break;
case 'b':
print blocks () :
break;
case
151:
print memory () ;
break;
case 'd':
delete_line ();
init zlinesptr();
break;
Next add the following line and object definition (OBJ= ) to the
make file.
src/Output.o: src/Output.c src/Editor.h pre/Editor.pre
The complete editor source can be found in the Vo . 3 directory of the

optional diskette for this book. After compiling start the test editor and
enter more lines by pressing a (append). Exit the test module by
pressing <Esc>. You can move the window, enlarge it, reduce it, etc.
Make sure that the window display remains consistent.
Now we'll concentrate on making the editor a more efficient program.
You remember that the C compiler compiles the expression *var++
443
incorrectly. Remove the increment (++) and process this separatelythe C compiler creates a more efficient object code. Compile the
following test program to create its assembler source text:
main 0
{
register char *ptr;

char string[ID];
register short n;
for (n = 0, ptr
*ptr++ = 0;
= string;
n < 10; n++)
for (n = 0, ptr
*ptr = 0;
= string;
n < 10; n++, ptr++)
Both for loops increment the pointer

using the following assignment:
ptr~
Compile the program
cc test.c -a -t
You get the assmebler source text with the name Test.asm, which
looks like the following using our Aztec C compiler:
main 0
;
;
public main
main:
- link as,it.2
movem.l .3,-(sp)
register char *ptr;
char string[ID];
register short n;
;
;
;
for (n = 0, ptr = string; n < 10; n++,

move.l itO,d4
lea -ID(a5),aO
move.l aO,a2
bra .1
.6
*ptr++ = 0;
move.l a2,aO
add.l In, a2
clr.b (aD)
.4
add.w In,d4
.1
cmp.w 1n0,d4
bIt .6
.5
;
for (n
move.l itO,d4
lea -10(a5),aO
move.l aD, a2
bra .11
.10
;
ptr
= string;
*ptr
clr.b (a2,
444
= 0,
= 0;
n < 10; n++, ptr++)
4.3
ABACUS
STEP BY ST E P
.8
add.w #1,d4
add.l #1,a2
.11
cmp.w #10,d4
bIt .10
.9
,
.12
movem.l (sp)+,.3
unlk as
rts
.2 equ -10
.3 reg d4/a2
public .begin
dseg
end
Look at the two loop bodies in particular, for the first loop from
label. 6 to label. 7, and for the second loop from label. 1 0 to
label. 11. You see yourself that the first loop needs four commands
while the second has three commands. Therefore it is especially
important that these commands be run through as often as the loop
counter allows. This command becomes more powerful the more often
you execute the loop.
It is worthwhile to examine the assembler source text. There the
compiler manufacturer tries to make their product as efficient as
possible. The loop counter procedure is discussed later in this book.
4.3.7
The Cursor
The editor should let the user move the cursor to any point in a source
text for easy editing. The following section shows how to expand the
editor for cursor movement using either cursor keys or mouse. When
the cursor reaches the bottom or top border of the window, the window
contents must scroll to allow access to the next or previous lines in the
window. This vertical cursor movement is important, since almost
every source text has more lines than can fit into just one window.
Horizontal scrolling must be available as well. When the user reduces
the size of the editor window, lines that are wider than the reduced
window must also be accessible through scrolling.
Folding
We must consider folding when dealing with cursor movement.

Remember when folding, two lines that lie directly under each other in
the editor window do not have to follow directly after one another in the
text. When a line contains a fold starting mark, all lines following this
mark, up to the corresponding fold end mark are invisible. The next
visible line in the window is the one following the fold end mark. As
445
4.
AMIGA
with the zlinesptr array, an array is needed that contains the line
number for every field that is visible in the editor window.
The editor specifies the cursor's line position through the zlinesptr
array. In addition to the true line position, the zlinesnr array needs
the window position. This also allows testing without the
implementation of folding, before including the folding routine.
Tabs
Tabs present the next problem. The editor Z (included with the Aztec C
compiler) forces the cursor to jump over multiple columns if it lands
on a tab. Cursor movement depends on the character on which the
cursor is found. This is why the cursor stands on the last character of a
line instead of behind a line. The cursor position should be fully
selectable, even though it causes a few problems with character input
(e.g., what happens if you overwrite a tab with another character).
Scrolling
Horizontal scrolling is slightly different from vertical scrolling. When

the contents of a window scrolls to the right, the printAll function
must redisplay the left border. Then a problem arises if the last character
of the output is either a space or a dot. This creates a line which is
wider than the window (see the 1 a s t c h a r variable in the
ConvertLineForPrint function). That means that when we scroll
only one column, dots appear in the first column. We must confess that
we noticed this problem when we first examined this. We got around it
by converting one more column than was actually needed, and omitting
this last character when displaying the line.
Another problem occurs when redisplaying the right border if the last
character is a space or a dot. When we move the window contents to the
right, the character in the last column must be overwritten.
The new right border output occurs when we scroll to the left. When
accessing printAll, the editor must be instructed to view the
window with a wide left border. This displays the text at the right
border. We must display one more column than was scrolled because
the last column contains only spaces and dots.
A number of functions could be created that would solve the problem.
However, when creating functions for fancy output, it's easy to lose the
perspective of exactly what is affected from the changes. We avoided
making the output functions too complicated, and leave you to make
any modifications to the scrolling that you might want.
Now we return to folding. A fold starting mark defines the beginning of
a fold. A typical starting mark, which appears in the current level of
text, looks like this:
I*#FOLD: *1
The fold ending mark appears at the end of the text the programmer
wants folded, and looks like this:
446
4.3
ABACUS
STEP BY STEP
I*#ENDFD*I
Both marker strings must appear at the beginning of a line (no leading
spaces). You may have noticed that the marks begin with the two
characters that normally start comments in C (! *). This is necessary
because the C compiler must skip these lines. Later we'll show you
how to change the marks so that the programmer can use the folding
technique in other languages.
The words fFOLD and fENDFD must be entered in capital letters. The
editor will not fold lines using lower case lettering (e.g., fFold or
:ftendfd are unacceptable to the editor).
The colon following the word fFOLD lets you add comments or
function names. This can tell the user the purpose of the program text
within the fold. A completed fold displays the line as seen above and
the commentary. For example:
I*#FOLD: InnerRoutine *1
Fold levels
To read a fold, move the cursor onto the starting mark. Press
<Ctrlxf> (f as in "fold"). The editor displays the contents of the fold.
You can have multiple levels of folding (inserting folds within folds).
When creating folds, the editor defines the level of each set of lines and
the depth of the fold in which the lines lie. The first line in a text lies
at level zero. A fold starting mark increments the level by one, and a
fold ending mark decrements the level by one.
The following example displays the fold level on the left margin. The
lines on the right correspond to the lines within the fold level:
Line 1
1
1
Line 3
I*#FOLD: Line 4 *1
Line 5
I*#ENDFD Line 6*1
Line 7
I*#ENDFD Line B *1
Line 9
I*#FOLD: Line 10 *1
I*#FOLD: Line 11 *1
Line 12 I*#ENDFD Line 13 *1
Line 14
I*#ENDFD Line 15 *1
Line 16
-
o I*#FOLD: Line 2 *1
2
2
1
o
o
1
2
2
1
The Edi tor structure defines two variables named minfold and
maxfold. Only the lines whose levels lie between minfold and
maxfold are displayed in the window.
<Ctrl><f> enters the next fold level in two ways:
447
4.
AMIGA
When the cursor lies at a fold starting mark, minfold and

maxfold are set to the level of the following line so that you
see only the corresponding fold.
Otherwise only maxfold increases by one so that you can only

see the fold that lies one level higher. in addition to the fold
currently in view.
You can exit the current fold by pressing <Ctrl><e> (e as in "exit").

You cannot exit a fold using the cursor keys; the cursor stays inside of
the fold currently being accessed.
<Ctrl><e> decreases maxfold by one. When minfold is greater
than maxfold. it changes to the same value as maxfold. The
example text mentioned above can be accessed by the following
process:
o
o
o
o
o
o
o
1
1
1
1
o
o
1
1
1
1
1
1
448
minfold andmaxfold are initially equal to zero so that the

text from its top level looks like this:
Line 1
/*#FOLD: Line 2 *1
Line 9
/*#FOLD: Line 10 */
Line_16
-
You see only the lines whose fold level is between minfold
and maxfold. or equal to zero. Press <Ctrl><f> while the
cursor is in line_l. Most of the text unfolds. revealing the
following:
Line 1
/*#FOLD: Line 2 *1
Line 3
/*#FOLD: Line 4 */
Line 7
I*#ENDFD Line 8 */
Line 9
/*#FOLD: Line 10 */
/*#FOLD: LIne~ll */
Line 14
/*#ENDFD Line 15 */
Line 16
-
Pressing <Ctrl><f> again while the cursor is in 1 in e 1

unfolds the entire text, since maxfold becomes equal to-2.
Press <Ctrl><e> twice to display the lines with a fold level of
zero. Move the cursor to line_2 and press <Ctrl><f>. The
following fold contents appear:
Line 3
/*#FOLD: Line 4 */
Line 7
/*#ENDFD Line_8 *1
ABACUS
4.3
STEP BY STEP
This is confusing. Shouldn't folding display the current fold, instead of

these extra folds? When you set maxfold to the maximum line level
while minfold stays at zero, you'll see fold marks in the text even if
they weren't selected. This is why the program displays all of the
current fold's start and end marks in black characters. When you come
across a fold start mark in white characters. this means that you can
access the fold.
Implememting We need four functions to move the cursor in the proper directions:
cursorRight, cursorLeft, cursorUp, cursorDown
These need no arguments, and return either lRUE (the cursor moves) or
FALSE (the cursor remains stationary, or reaches the end ofa line). We
have no use for this return value at the moment. However, if you wish
to add more capabilities to the editor later on, knowing whether or not
the cursor actually moved can be useful. This knowledge can be used in
conjunction with interrupt criteria for loops.
These cursor movement functions call other functions that scroll the
editor window contents in the corresponding direction:
scrol11eft, scrollRight, scrollDown, scroll Up
Cursor movement functions must call opposing scrolling functions.

For example, cursorRight calls scrollLeft. We assign the
number of lines or columns that should be scrolled as an argument. The
argument type is UWORD. All scrolling functions are of type void
(they return no result).
The next function receives the key presses and accesses the
corresponding cursor functions. This function must also handle
keyboard input and output:
void handleKeys(buf,len)
UBYTE
*buf;
UWORD
len;
Cursor display Cursor movement is useless without displaying a visible cursor. The
Cursor function displays the cursor in COMPLEMENT mode:
void Cursor ()
When we call this function before the Wait function in the main loop.
the cursor appears when the editor is inactive. When the program is
busy, the cursor disappears and cannot be used.
Now that you know the important functions for the new module, add
these items to the Edi tor. h file:
ide fine FOLDPEN 21
idefine Z1INESPTR(n) aktuellerEditor->zeilenptr[n]
#define Z1INESNR(n) aktuellerEditor->zeilennr[nl
449
#define ZLF FSE 64

#define ZLF=FOLD 63
FOLDPEN is the color in which the fold marks are displayed (usually
black). The ZLINESPTR and ZLINESNR functions should ease access
to the elements of corresponding fields. Instead of:
actualEditor->linesptr[nl
write:
ZLINESPTR(n)
The last two defines belong to the line structure flags in the data
structure Line. When the flag ZLF FSE is set, it handles the line as a
fold mark. It doesn't matter whetheilt is handled as a fold start or fold
end mark (FSE = Fold Start End). ZLF FOLD specifies the fold level
of the line (for non-folded lines ZLF FOLD=O). Be careful that the fold
start marks themselves don't belong to the fold. The maximum fold
level is 63.
The expanded Edit 0 r structure looks like this:
struct Editor
(

struct BList block;
UBYTE buffer[MAXWIDTH);
UWORD num lines;
struct Zline *zlinesptr[MAXHEIGHT];
UWORD zlinesnr[MAXHEIGHT];
UWORD toppos,leftpos;
UWORD xpos, ypos;
UWORD wdy;
UWORD changed:l;
UWORD insert:l;
UWORD xoff,yoff;
UWORD xscr,yscr;
UWORD CW, Chi
UWORD wcw,wch;
UWORD minfold;
UWORD maxfold;
Ii
The following elements are new:

zlinesnr[)
This contains the numbers of all of the lines that are visible in the
window (see also zlinesptr).
450
4.3
ABACUS
STEP BY STEP
leftpos
When the window scrolls to the left, the first column of the window
changes to the next column up. leftpos specifies the number of the
column directly to the left of the visible window. leftpos has a
default value of zero, which means that the first visible column in the
window has the number one.
wdy
This specifies the line position of the cursor with a value between zero
and (actualEditor->wch-l).
xscr, yscr
These variables supply the right bottom border of the window

(necessary for scrolling). For text output xscr - (xrl - 1) and
yscr == (y - 1) (measured in pixels).
min fold, maxfold
Specifies the minimum and maximum fold levels that determine which
folds appear on the screen and which don'L
The following listing for the Cursor module contains all of the
previously described cursor functions:
<src/Cursor.c>
/************
*
*
Includes:
************/
/***********
** Defines:
************1
fdefine
#define
#define
#define
#define
#define
#define
CSI Ox9B
CUU 'A'
CUD 'B'
CUF 'C'
CUB '0'
SU 'S'
SO 'T'
fdefine CFOLD 6
#define CENOF 5
/**********************
*
451
4.
AMIGA
C FOR
***********************/
void SetDrMd(),RectFill(),printAlI();
struct Zline *nextLine(),*prevLine();
1*********************
* External Variables:
**********************/
extern struct Editor *actuaIEditor;
extern UWORD SplitDec;
/***************
**
Functions: *
****************1
*
1************************************
**
**
*
*
*
*
restoreZlinenptr:
Restore the zlinesptr/nr field
for all lines in window plus
the next line.
The first line must be
correct!
*************************************1
void restoreZlinenptr()
{
register UWORD n;
register struct Zline **zptr,*z;
register UWORD *pnr;
UWORD znr;
for (n
actuaIEditor->wch, zptr = actuaIEditor->zlinesptr,

pnr = actuaIEditor->zlinesnr,
z *zptr++, znr = *pnr++; n; n--)
*zptr++
*pnr++
(z
= nextLine(z,&znr));
znr;
1**********************************
** Cursor:
*
* Set or erase the cursor

* at the actual position.
***********************************/
void Cursor 0
{
register struct RastPort *rp;
4Sl
4.3
ABACUS
STEP BY STEP
register LONG x,y;

register LONG cw,ch;
SetDrMd(rp : actuaIEditor->rp,COMPLEMENT);
x
(actuaIEditor->xpos - actuaIEditor->leftpos - 1)
*(cw = actuaIEditor->cw) + actuaIEditor->xoff;
y = actuaIEditor->wdy
*(ch : actuaIEditor->ch) + actuaIEditor->yoff;
=
RectFill(rp,x,y,x + cw-1,y + ch-1);

SetDrMd(rp,JAM2);
/*******************************
* scroIIRight(num):
* Scroll the window contents

* num characters to the right.
* nurn
= number of characters.
*******************************1
void scroIIRight(nurn)
register UWORD
num;
{

UWORD wcw,xoff,leftpos;
actuaIEditor->leftpos
-=
num;
if (nurn >= actuaIEditor->wcw)

printAll () ;
else
{
1* Scrolling: *1
ScrollRaster(actuaIEditor->rp,
-num * (LONG) actualEditor->cw, OL,
(LONG)actualEditor->xoff, (LONG)actualEditor->yoff,
(LONG)actualEditor->xscr, (LONG)actuaIEditor->yscr);
1* now list printAII(): */
wcw : actualEditor->wcw;
actuaIEditor->wcw = num + 1;
SplitDec = 1;
y
actualEditor->yoff;
actualEditor->ch;
count = actualEditor->wch;
ch
1* OUtput line: */
for (z = actuaIEditor->zlinesptr; count-- && (*z != NULL);

z++, y += chI
printLine(*z,y);
SplitDec = 0;
/* Display right border beside last character */
453
4.
AMIGA
C FOR
leftpos = actuaIEditor->leftpos;
xoff
= actuaIEditor->xoff;
actuaIEditor->wcw
= 2;
actuaIEditor->leftpos += (y = wcw - 2);
actuaIEditor->xoff
+= actuaIEditor->cw * y;
y
count
actuaIEditor->yoff;
actua1Editor->wch;
1* OUtput line: *1
for (z = actua1Editor->z1inesptr; count-- && (*z != NULL);
z++, y += ch)
printLine(*z,y);
actuaIEditor->wcw
= wcw;
actuaIEditor->leftpos= leftpos;
actuaIEditor->xoff = xoff;
1*******************************
**
scrollLeft(num):

* num characters left.
* num = number of characters.
*******************************1
void scrollLeft(num)
register UWORD num;
{

actuaIEditor->leftpos += num;
if (num >= actuaIEditor->wcw)
printAll () ;
else
{
1* Scrolling: */
ScroIIRaster(actuaIEditor->rp,
num * (LONG)actuaIEditor->cw,OL,
(LONG)actuaIEditor->xoff, (LONG)actualEditor->yoff,
(LONG)actuaIEditor->xscr, (LONG)actua1Editor->yscr);
1* now list printAII(): */
wcw
actuaIEditor->wcw;
1eftpos = actuaIEditor->1eftpos;
xoff
= actuaIEditor->xoff;
actuaIEditor->wcw
= num + 1;
actuaIEditor->leftpos += (y = wcw - num - 1);
actuaIEditor->xoff
+= actuaIEditor->cw * y;
y
actuaIEditor->yoff;
actuaIEditor->ch;
count = actuaIEditor->wch;
ch
454
4.3
ABACUS
STEP BY STEP
1* Output line: *1
for (z = actualEditor->zlinesptr; count-- && (*z != NULL);
z++, y += ch)
printLine(*z,y);
actualEditor->wcw
= wcw;
actualEditor->leftpos= leftpos;
actualEditor->xoff = xoff;
1*******************************
**
**
scrollUp(num):
Scroll the window contents

* nurn lines up.
* Nurn lines must
* existl
**
nurn = number of lines.
*******************************1
void scrollUp (nurn)

register UWORD nurn;
{
register struct Zline *z,**zptr,**zold;

register UWORD n,web = actualEditor->web,y;
UWORD znr,*pnr,*oldnr;
i f (nurn
>= web)
1* Display window without Scrolling *1

1* First look for top line: *1
n = wch;
i f (nurn > wch)
{
z = ZLlNESPTR(n);
znr = ZLINESNR(n);
while (++n < nurn)
z = nextLine(z,&znr);
}
else
{
z = ZLINESPTR(n - 1);
znr = ZLINESNR(n - 1);
1* z points before the top line of the window *1

for (n = 0, zptr = actualEditor->zlinesptr,
pnr = actualEditor->zlinesnr; n <= wch; n++)
*zptr++ = (z = nextLine(z,&znr;
*pnr++ = znr;
1* Now redisplay the window: *1

printAll();
else
(
4SS
4.
AMIGA
1* Scrolling: *1
ScrollRaster(actualEditor->rp,
OL,num * (LONG)actualEditor->ch,
(LONG)actualEditor->xscr, (LONG)actualEditor->yscr);
1* zlinesptr/nr recalculated: *1
zptr = actualEditor->zlinesptr;
zold = &(ZLlNESPTR(num;
pnr = actualEditor->zlinesnr;
oldnr = &(ZLINESNR(num;
for (n = 0; n <= wch - num; n++)
{
*zptr++ = *zold++;
*pnr++ = *oldnr++;
z = *--zold;
zold = zptr - 1;
znr = *--oldnr;
oldnr = pnr - 1;
while (n <= wch)
1*
mark for output!
*1
*pnr++ = znr;
n++;
1* Output new line: *1

for (n = num, y = actualEditor->yoff+ (wch-num) *actualEditor-
>ch;
(n && (*zold != NULL; n--, zold++, y += actualEditor>ch)
printLine(*zold,y);
actualEditor->toppos += num;
1*******************************
* scrollDown(num):
*
*
*
*

num lines down.
Num lines must
exist!
* num = number of lines.

********************************1
void scrollDown (num)
register UWORD num;
{

register UWORD n,wch = actualEditor->wch,y;
i f (num >= wch)
{
1* Display window without Sciolling *1
456
ABACUS
4.3
STEP BY ST E P
/* search backwards fo rtop line: */

for (z = ZLINESPTR(O), znr = ZLINESNR(O), n = num - wch; n;
n--)
z = prevLine(z,&znr);
/* z points to first line after bottom windo edge. */
for (n = 0, zptr = &(ZLlNESPTR(wch, pnr =
&(ZLlNESNR(wch;
n <= wch; n++)
*zptr-- = z;
*pnr-- = znr;
/* Now redisplay the window: */
printAll();
}
else
{
/* Scrolling: */
OL,-num * (LONG)actualEditor->ch,
/* zlinesptr recalculated: */
zptr = &(ZLlNESPTR(wch;
zold = &(ZLINESPTR(wch - num;
pnr = &(ZLlNESNR(wch;
oldnr = &(ZLINESNR(wch - num;
for (n = 0; n <= wch - num; n++)
{
*zptr-- = *zold--;
*pnr- = *oldnr--;
for (z = *++zold, znr = *++oldnr, n = num; n; n-)
{
*zptr-- = (z = prevLine(z,&znr;
*pnr-- = znr;
/* Output new line: */
for (n = num, y = actualEditor->yoff; (n && (*zold !=
NULL) );
n--, zold++, y += actualEditor->ch)

printLine(*zold,y);
actualEditor->toppos -= num;
/****************************************
**
**
cursorLeft:
moce cursor one position

* left, False returned if cursor,
* is in first column.
*****************************************/
457
4.
BOOL cursorLeft ()
{
if (actuaIEditor->xpos > 1)
{
actualEditor->xpos--;
if (actuaIEditor->xpos <= actuaIEditor->leftpos)
scrollRightUWORD)1);
retum (TRUE);
}
else
return (FALSE);
1*****************************************
**
cursorRight:
*
* moce cursor one position
* right, False returned if cursor,
* is in last column.
******************************************1
BOOt cursorRight ()
{
if (actuaIEditor->xpos < MAXWIDTH)

{
actualEditor->xpos++;
if (actuaIEditor->xpos
>= (actuaIEditor->leftpos + actuaIEditor->wcw
scrollLeft UWORD) 1);
return (TRUE);
}
else
return
(FALSE);
1*****************************************
cursorUp:
* Move
cursor one line up.

returned when
on first line.
* False
*
*
*****************************************1
BOOL cursorUp ()
{
UWORD help;
if (actuaIEditor->wdy)
actuaIEditor->wdy--;
else
if (prevLine(ZLlNESPTR(O),&help
scroIIDownUWORD)1);
else
retum (FALSE);
458
4.3
ABACUS
actuaIEditor->ypos
STEP BY STEP
ZLINESNR(actuaIEditor->wdy);
return (TRUE);
1******************************************
* cursorDown:
*
*
*
*
Move cursor one line down.

False returned when
on last line.
*******************************************1
BOOL cursorDown()
{
if (ZLlNESPTR(actuaIEditor->wdy + 1
{
if (++actualEditor->wdy >= actualEditor->wch)

(
scrolIUpUWORD)l);
actualEditor->wdy--;
actuaIEditor->ypos
ZLlNESNR(actuaIEditor->wdy);
return (TRUE);
else
return (FALSE);
1**********************************
**
halfpageUp:
* Move cursor on hal f page

* up. The text is scrolled
* half page up
* False returned when

* in first line.
***********************************1
BOOL halfpageUp()
{
register UWORD max,n;

UWORD help;
max = actualEditor->wch 1 2;
for (n = 0, z = ZLINESPTR(O); n < max; n++)
if (! (z = prevLine(z,&help)
break;
i f (n)
(
scroll Down (n) ;

actualEditor->ypos
= ZLlNESNR(actualEditor->wdy);
return (TRUE);
459
4.
AMIGA
else
return (FALSE);
1**********************************
* halfPageDown:
* Move cursor on half page

* down. "The text is scrolled
* half page up
* oben gescrollt.
* on last line.
**********************************1
BOOL halfPageDown()
{

UWORD help;
max = actualEditor->wch 1 2;
for (n = 0, z = ZLINESPTR(actualEditor->wch - 1); n < max; n++)
i f (! (z
= nextLine(z,&help)
break;
i f (n)
{
scrollUp (n) ;
actualEditor->ypos
ZLINESNR(actualEditor->wdy);
return (TRUE);
else
return (FALSE);
/*******************************
* handleKeys(buf,len):
* Handles keyboard messages.
*
*
*
buf
len
buffer with character.

of characters.
= number
*******************************1
register WORD
len;
{
register UBYTE first;

while (len> 0)
{
first = *buf;
460
4.3
ABACUS
STEP BY STEP
OOf++;
len--;
if first
== CSI) && (len> 0
len--;
switch (*buf++)
{
case CUU:
cursorUpO;
break;
case CUD:
cursorDown();
break;
case CUF:
cursorRight () ;
break;
case CUB:
cursorLeft () ;
break;
case SU:
halfPageDown 0;
break;
case SD:
halfPageUp() ;
break;
default:
buf--;
len++;
break;
)
else if first == CFOLD) && (actuaIEditor->maxfold <

ZLF FOLD
-{
actuaIEditor->maxfold++;
if (z = ZLlNESPTR(actuaIEditor->wdy
i f z = z->succ) ->succ)
.
1* If there is a following line: *1
if (ZLINESPTR(actuaIEditor->wdy)->flags & ZLF_FOLD)
< (z->flags & ZLF FOLD
&& z->flags & ZLF~)OLD) = actualEditor->maxfold
1* When this begins a new fold: *1

actuaIEditor->minfold = actualEditor->maxfold;
ZLINESPTR(O) = z;
ZLINESNR(O)
= ZLINESNR(actuaIEditor->wdy) + 1;
actuaIEditor->wdy = 0;
restoreZlinenptr();
printAll () ;
actuaIEditor->ypos = ZLINESNR(actuaIEditor->wdy);
}
else if first
== CENDF)
&& (actualEditor->maxfold
actualEditor->maxfold--;
if (actuaIEditor->minfold > actualEditor->maxfold)
actuaIEditor->minfold = actualEditor->maxfold;
if (ZLINESPTR(O)->flags & ZLF_FOLD) > actualEditor>maxfold)
461
4.
AMIGA
I I ZLlNESPTR(O)->flags & ZLF_FOLD) < actualEditor>minfold) )

{
ZLINESPTR(O) = prevLine(ZLlNESPTR(O),&(ZLlNESNR(O);
actualEditor->wdy = 0;
restoreZlinenptr();
printAll () ;
actualEditor->ypos = ZLINESNR(actualEditor->wdy);
else
putchar (first);
Here are a few specifics about some of the functions:
462
The de fine s determine the characters controlling cursor

movement and folding. The labels are taken from AmigaDOS
sources. You must be careful with the check in handleKeys
that the cursor movement character sequences are two to three
characters long. For now a CSI comes from the characters given
in the defines.
The restoreLinesptr function recalculates the linesptr

and the zlinesnr arrays. This function is necessary to enable
redisplay of window contents after a fold is encountered.
The scroll functions allow scrolling by more than one line or

character. They allow more characters or lines to be scrolled than
can fit on the screen at a time. The vertical scroll functions work
well for searching for a specific text
A for loop operates in the scrolling functions instead of using

the printAll function in the scroll functions. printAll
erases the borders that are unnecessary to scrolling, and takes too
much time.
The Zlinesptr and the zlinesnr arrays must be

recalculated for scrollUp and scrollDown. To avoid
recalculating the entire array, the program moves the contents
needed after the scrolling, just as the lines are scrolled.
In addition to testing the folding, cur s 0 r Up and

cursorDown must check to see if the next or previous line
exists in the current level. The next Line and prevLine
functions must be expanded correspondingly to perform this task.
These functions supply a pointer to the current line number,
which must then be raised or lowered appropriately.
To test the vertical scrolling for more than one line, the
halfPageUp and halfPageDown functions scroll the text a
4.3
ABACUS
STEP BY STEP
half window page up or down. The cursor position relative to the

window remains unchanged.
When the user presses <Ctrl><f>, the editor must determine

whether the cursor is at a fold start mark. The fold level of the
current line is less than the level of the following line within the
fold (assuming that a following line exists). The following line
is not checked with nextLine because the next line appears in
the text, regardless of the fold. When the level of the following
line corresponds to the maximum fold level plus one, the editor
handles the current line as a fold start mark because the next line
is invisible.
When a fold is encountered, it must be determined that the above

line does not belong to the fold just exited before calling
restoreLinesptr. Otherwise the window contents make no
sense.
Now we can expand the other modules. Two new functions must be
declared in the main Edi tor module:
void handleKeys(),Cursor();
The nextLine and prevLine functions must be modified for the

folding:
struct Zline *nextLine(line,pnr)
register UWORD
*pnr;
{

register UWORD minfold = actuaIEditor->minfold;
register UWORD maxfold = actualEditor->maxfold;
i f (z = line)
i f (z->succ)
cD
{
if z = z->succ) ->succ)
i f z->flags & ZLF FOLD) < minfold)
{
1* Following lines lie outside of fold *1
z = NULL;
break;
else
*pnr += 1;
else
{
1* no following line */
z = NULL;
break;
}
while z->flags & ZLF FOLD) > maxfold);

else
z '" NULL;
463
return (z);
The function prevLine looks exactly like it, only "->succ" is

replaced throughout by "- >pred":
struct Zline *prevLine(line,pnr)
register UWORD
*pnr;
{

register UWORD minfold = actuaIEditor->minfold;
register UWORD maxfold = actuaIEditor->maxfold;
i f (z = line)
i f (z->pred)
<b
{
i f z = z->pred) ->pred)
if z->flags & ZLF_FOLD) < rninfold)

{
1* previous line lies outside fold */
z = NULL;
break;
else
*pnr += 1;
else
{
1* no previous line *1
z = NULL;
break;
}
while z->flags
else
&
ZLF_FOLD) > maxfold);
= NULL;
return (z);
The OpenEdi tor function initializes the other elements of the editor:
ed->leftpos
ed->wdy
ed->minfold
ed->maxfold
= 0;
=
=
=
0;
0;
0;
1* zlinesptr/nr-Array initialization: */
for (n = 0, zptr = ed->zlinesptr, pnr a ed->zlinesnr;
n < MAXHEIGHT: n++, zptr++, pnr++)
*zptr = NULL;
*pnr = 1;
The main program now requires two new variables in place of SHORT
1:
464
4.3
ABACUS
STEP BY STEP
ULONG mouseX,mouseY
The main loop also requires some changes. Here is the current listing of
the main loop:
d>
(
1* Set cursor
Cursor();
signal
*f
Wait(IL edUserPort->mp_SigBit);
1* erase cursor again: *f

Cursor();
while (imsg
{
GetMsg(edUserPort
= imsg->Class;
class
= imsg->Code;
code
iaddress
= imsg->IAddress;
mouseX
imsg->MouseX;
mouseY
= imsg->MouseY;
ReplyMsg (imsg);
switch (class)
{
case RAWKEY:
i f (! (code & IECODE_UP_PREFIX
(
inputEvent.ie Code
= code;
inputEvent.ie-Qualifier = qualifier;
if inputLen-= RawKeyConvert (
&inputEvent,inputBuffer,MAXINPUTLEN,NULL)
) >= 0)
handleKeys(inputBuffer,inputLen);
break;
case MOUSEBUTTONS:
{
register WORD x,y;
if (mouseX <= actualEditor->xoff)

x = 0;
else
x = (mouseX - actualEditor->xoff)
I actualEditor->cw;
if (++x >= actualEditor->wcw)
x = actualEditor->wcw - 1;
x += actualEditor->leftpos;
if (x > MAXWIDTH)
x = MAXWIDTH;
if (mouseY <= actualEditor->yoff)
y = 0;
else
y
= (mouseY
- actualEditor->yoff)
1 actualEditor->ch;
465
if
y < actuaIEditor->wch)
&& (actuaIEditor->zlinesptr[y]
actualEditor->xpos ~ x;
actuaIEditor->wdy = y;
actuaIEditor->ypos = actuaIEditor->zlinesnr[y];
case NEWSIZE:
1* check if cursor still in window! *1

if (actuaIEditor->xpos > actuaIEditor->leftpos
+actuaIEditor->wcw)
actuaIEditor->xpos = actuaIEditor->leftpos
+ actuaIEditor->wcw;
if (actuaIEditor->wdy >= actuaIEditor->wch)
actuaIEditor->ypos = actuaIEditor->zlinesnr
[(actuaIEditor->wdy = actuaIEditor->wch - 1)1;
break:
case REFRESHWINDOW:
BeginRefresh(actuaIEditor->window):
printAll () ;
EndRefresh(actualEditor->window,TRUE);
break:
case CLOSEWINDOW:
running = FALSE:
break;
default:
printf("Not a processable event: %lx\n",class);
} 1* of case *1
} 1* of while (GetMsg( *1
while (running);
The MOUSEBUTTONS check lets the user place the mouse on the
window by clicking on that screen character, as with most word
processors.
The memory module remains unchanged. However, the output
module must be modified to display fold markers in alternate colors.
First, a printLine needs a new global variable to make sure the last
column is not displayed:
1*********************
* Global
*
Variables:
*********************1
UWORD SplitDec = 0;
466
4.3 STEP BY STEP
ABACUS
The initWindowSize function must recalculate the zlinesptr

and the zline snr arrays. This requires the definition of two additional
variables:
UWORD znr,*pnr'
The array calculation looks like this:

1* Now restore the zlineptr field: *1
{
if (newh < ed->wch)

{
1* Pointer to Null set: *1

*zptr++ = NULL;
else
{

= *zptr++;
z
pnr = &(ed->zlinesnr[ed->wch]);
znr
= *pnr++;
{
*zptr++ = (z = nextLine{z,&znr;
*pnr++ = znr;
ed->wch
newh;
The convertLineForPrint function needs changes as well.

Output must begin in the column determined by leftpos, not just
the first column of the window. A counter (skip) to leftpos does
this check, frrst seeing if skip has already reached zero. If skip equals
zero the character is written. Otherwise, skip decrements by one. We
skip the first skip character:
register UWORD skip
actualEditor->leftpos;
The determination of lastchar stays the same, but the conversion is

changed:
1* line conversion: *1
while ({len--) && (I < w
i f ({*buf = *line++) = TAB)
d:J
{
tab++;
i f (skip)
skip--;
else
{
1++;
*buf++ =
';
467
} while *tab) && (1 < w;

else
(
tab++;
if (skip)
skip--;
else
{
1++;
buf++;
The rest of the function is unchanged.

The printAt function must have additional parameters dealing with
character color. These parameters are FGPEN or FOLDPEN. Both
parameters are important since after a space, text color cannot be sure
whether or not the text is a fold mark. The foreground color must be
changed for spaces, and then reset:
void printAt (buf,len,x,y,fgpen)
w:lRD
len;
register UWORD
x,y;
ULONG
fgpen;
Throughout this function

SetAPen (rp, FGPEN) :
is replaced by:
SetAPen(rp,fgpen);
The print Line function can be expanded so that it recognizes fold

marks and changes text color accordingly. FGPEN must be disabled at
the end of the function so that we can leave the program at any place
where the foreground color is FGPEN:
void printLine
(line,y)
register UWORD
y;
{
static UBYTE buf[MAXWIDTH1:

register UWORD w;
register struct Zline *z:
register ULONG pen;
convertLineForPrint(line+l,line->len,w = actualEditor>wcw,buf) ;
1* If start/end marker of fold => FOIDPEN *1

if (line->flags & ZLF FSE)
if z
{
468
= line->succ)->succ)
4.3
ABACUS
STEP BY STEP
if z->flags & ZLF FOLD) <= actuaIEditor->maxfold)

(
SetAPen(actuaIEditor->rp,FOLDPEN);
pen = FOLDPEN;
else
pen
FGPEN;
else
(
pen = FOLDPEN;
printAt(buf,w - SplitDec,actuaIEditor->xoff,y,pen);
/* Old write color first again: */
if (pen != FGPEN)
SetAPen(actuaIEditor->rp,FGPEN);
else
printAt(buf,w - SplitDec,actuaIEditor->xoff,y,FGPEN);
Changes must still be made to the test. c module before we can test
the new extensions. The line number displayed by the print function
comes from the zlinesnr array:
void print ()
{

register UWORD n,*pnr;
for (n = 0, z = actuaIEditor->zlinesptr,
pnr = actuaIEditor->zlinesnr;
n <= actuaIEditor->wch; n++, z++, pnr++)
printf("Zline %d an Adresse %6Ix. \n", *pnr, *z);
The number of lines should be adapted for the input and delete line
functions, by adding the following for input:
actuaIEditor->num_lines++;
andfor delete_line:
actuaIEditor->num_lines--;
The ini t zlinesptr function must determine the fold level of all
of the lineS:
void init zlinesptr()

register UWORD n,*pnr,fold = 0;
UWORD znr;
for (z
= actuaIEditor->zlines.head;
z->succ; z
= z->succ)
469
z->flags
z->flags & -ZLF_FOLD) 1 (fold & ZLF_FOLDi
i f (z->len >: 8)
{
if *ULONG *) (z+l=='/*#F') && (*ULONG

*) (z+1)+1)=:'OLD:'
{
fold++i
z->flags 1= ZLF_FSEi
}
if *ULONG *) (z+l=='/*#E') && (*ULONG

*) (z+l)+l)=='NDFD'
{
fold--;
z->flags 1= ZLF_FSEi
for (n = 0, z = actualEditor->zlines.head,
zptr = actualEditor->zlinesptr,
pnr = actualEditor->zlinesnr, znr = 1;
n < MAXHEIGHTi n++)
*zptr++ = Zi
*pnr++ = znri
z = nextLine(z,&Znr)i
The fold mark test looks rather "wild," but it works faster than the
string compare function from the standard library. This happens because
four characters are constantly compared with one another (ULONG =4
bytes). Currently it works in our test module because we can allow
ourselves some leeway in test programming. Later, if we allow any
other strings as fold marks, we use the strcmp or strncmp
functions. because there is no guarantee that the corresponding character
string is exactly eight characters in length.
A printAll call at the end of the function test ensures that the
window contents are updated. Compile and link this version of the
editor.to test this version of the editor. We will use the Amiga's ability
to redirect input from a file. The optionl diskette contains a Test Text
file to do this. The file contains a linefeed, an a, and another linefeed at
the end of each line. Using redirection, this file can be loaded into our
test version of the editor using the append (a) command. An escape
character and a linefeed are located at the end of the file to let the test
editor know when we are finished appending text. Please see Apendix A
for complete information on altering a standard text file suitable for
redirection into our test editor.
You'll find the TestText source, the editor source text and compiled
editor in the VO. 4 directory of the optional disk.
Mter you have created a test file, enter the following to start the editor
and use the TestText source as redirected input. see the optional
diskette or Appendix A for the TestText listing:
470
4.3
ABACUS
STEP BY STEP
Editor <Test Text
The end of Test Text contains an escape code, which indicates the end
of the source file. You can now examine the editor and some of its
features, including folding.
Debugging
The debugger is an important part of C development. The only way to

become proficient at using the debugger is to sit down and use it.
Therefore, we have an exercise to help you .learn debugging: The
cursor's Y position should be displayed from the debugger.
Prepare to recompile the editor, but this time enter the following:
make debug
The makefile ensures that the finished program contains information

about variable names and symbol tables. This helps the programmer
find certain locations in the program. After compilation, enter the
following to start the debugger from the Aztec disks, this depends on
how you have installed the Aztec C compiler, db may be in a different
directory:
SYS3:bin/db
This loads a file named. dbini t that contains only one instruction
(al). This lets you begin a new task while in the debugger. Start the
editor as explained above. The debugger notices that a new task is in
operation. It checks to see if the program contains a symbol table, and
loads the symbol table if it exists.
Mter all loading ends, the debugger displays the frrst instruction:
jrnp
.begin
We want the debugger to display the Y position of the cursor. This

display occurs only after cursor movement. The Y position cursor
movement occurs primarily in the cursorUp and cursorDown
functions. We must know the location in memory containing the
position number, and access this location for the Y position display.
The Y position is an element of the Edi tor structure, and the
cursorUp and cursorDown functions change this position (see the
disassembled list below for functions in detail). Enter the following
debugger instruction:
u cursorUp
The debugger displays the memory in disassembled form from the

address cursorUp on. The list looks something like this:
cursorUp
-cursorUp+4
:cursorup+8
_cursorUp+c
_cursorUp+e
link
a5,i-2
movea.l actualEditor,aO
tst.w 252 (aD)
beq.s
cursorUp+18
movea.l :actualEditor,aO
471
cursorUp+l2
-cursorUp+l6
-cursorUp+18
-cursorUp+lc
-cursorUp+20
-cursorUp+24
:=cursorUp+28
cursorUp+2a
-cursorUp+2c
-cursorUp+2e
-cursorUp+32
-cursorUp+36
:=cursorUp+38
cursorUp+3a
:=cursorUp+3c
cursorUp+3e
:=cursorUp+40
cursorUp+44
-cursorUp+46
-cursorUp+4a
-cursorUp+4c
-cursorUp+4e
:=cursorUp+S2
_cursorUp+S6
cursorUp+Sc
:=cursorup+Se
subq.w
bra.s
pea
movea.l
move.l
jsr
addq.w
tst.l
beq.s
move.w
jsr
addq. w
bra.s
moveq
unlk
rts
movea.l
moveq
move.w
as!.l
movea.l
adda.l
movea.l
move.w
moveq
bra.s
#l,2S2(aO)
cursorUp+40
::2 (as)
actualEditor,aO
ca(aO),-(a7)
~revLine
#S,a7
dO
_cursorUp+3a
#l,-(a7)
scrollDown
i2, a7
cursorUp+40
iO,do
as
_actualEditor,aO
#0, dO
2S2(aO),dO
#l,dO
dO,al
actualEditor,al
-actualEditor,a6
Ica(al),2S0(a6)
#l,dO
_cursorUp+3c
The zlinesnr array supplies the Y position. We're looking for a

program section that reads a value from an array, then writes the value
as an element of the Edi tor structure. This section appears at the
address _cursorUp+40. First a pointer to the current Editor
structure is needed after AO. Then wdy passes to DO; this value
becomes the index of the zlinesnr array (thUS the logical shift left
multiplication (asl)). The value of the array passes to the ypos
variable at address cursorUp+S6. That means that ypos lies 250
hex bytes from the beginning of the structure.
Some knowledge in assembler is required to discover which assembler
memory location represents which C variable. Another way of
obtaining this value: Count up the elements in the Edi tor structure,
according to their memory needs in bytes. Once we determine the
location of ypos, we must allocate an area at which we can set a
breakpoint, which ensures display of the Y position. After the Y
position changes, the program jumps to address cursorUp+3c.
This address contains the UNLK assembler instruction, which concludes
every assembler subroutine in Aztec C. Set the breakpoint at the
following address:
bs cursorUp+3c ipd 250+*actualEditorig
About the arguments in the above line: The pd (Print Decimal) forces
the debugger to display the contents of the given memory location in
decimal notation). The memory location corresponds to the Y position
of the current editor, 250 bytes after the address to which the current
editor points. The g means that the debugger should restart the editor.
We don't want to stop the editor, but instead want to have the Y
472
4.3
ABACUS
STEP BY STEP
position displayed. Set the next breakpoint in the cursorDown

subroutine, and at the end of the program. Enter:
u cursorDown
Continue to enter u until you find the UNLK instruction. It will appear
at address _ cursorDown+66. Set the breakpoint as before:
bs cursorDown+66 ;pd 250+*actualEditor;q
This concludes the preparations. Press <g:> (GO) to start the editor.
Everything runs as usual, until you press the <Cursor up> key. Then
the debugger window comes to the foreground and activates. The cursor
does not appear in the debugger window following a question mark.
One catch: The editor is no longer in the foreground, nor is it active.
Reduce the size of the debugger window so that it doesn't block the
editor window but is still large enough to clearly display the Y
position. Enlarge the editor window so that you can see the entire
debugger window. Now reactivate the editor window by either clicking
on the window border or the cursor (clicking anywhere else moves the
cursor to another position).
Every time you press the <Cursor up> or <Cursor down> key, you
must reactivate the editor window (this is tolerable, but not very
elegant). Move the cursor up a couple lines onto folds, and see if the
position displayed is true. Enter the fold and scroll around a little. Once
you have done some scrolling, return to the beginning of the text and
see if this has the line number one. You'll find an error in the
calculation of the Y position.
With some logical thought we can determine how the error occurred.
Everything runs properly, as long as we remain in the window without
accessing any folds. This is fine, since line numbers are stored in the
zlinesnr array. As soon as this array changes, however, an error
occurs. The next Line function (which calculates the line numbers)
initializes the line numbers, returning zero. Two options are available
for seeking errors:
1.)
List and read the source text until you find the error;
2.)
Continue executing the program through the debugger and

narrow the error range down until you can easily locate the error.
Let's start by searching the source. Load the cursor. c module into
an editor and look at the program locations which affects folding. You
won't see any obvious errors there-we didn't. Since the problem seems
to occur when scrolling, search for the scrollUp and scrollDown
functions. They don't seem to contain any errors, either. Before we
continue testing, look at the prevLine function, which is accessed by
the scrollDown function. Notice that in this function the line
473
4.
AMIGA
numbers increment by one (exactly like in nextLine (*pnr++,

instead of decrement by one (*pnr--).
Change this, save the file, and recompile the editor. The editor in
directory VO.4 on the optional diskette contains this error, so you can
check it with the debugger. When the error is removed, you can test the
editor again if you have the desire.
4.3.8
Text Input
The editor should be able to handle text entry and editing. Functions
listed in this section will give the editor capabilities of entering,
inserting and deleting characters and lines in a source text. Let's begin
with the "how" of writing these functions.
Insert mode
Before inserting a character, the corresponding line must be copied into

the buffer of the Edit 0 r structure. Then all of the characters under the
cursor or to the right of the cursor are moved to the right, and the new
character appears at the current cursor position. This occurs in insert
mode only. If the user tries to insert characters past the right border of
the screen, the screen flashes: You cannot move past the right border of
the full size window.
Overwrite
mode
When the editor is in overwrite mode, any characters entered from the
keyboard are placed in the text. If any characters already exist in the
same line, the new characters replace the old. The cursor moves one
character to the right for each new character.
Control characters playa special role when inserting, particularly CR
(carriage return = 13), LF (linefeed = 10), BS (Backspace = 8), DEL
(Delete = 127), and TAB (9). CR and LF create a new line at the current
cursor position. BS and DEL delete characters. TAB inserts multiple
spaces.
Tabs
Tabs require special attention. Since the editor's main purpose is for
entering program text, it doesn't really matter whether tabs are spaces or
not. The Aztec C compiler treats tabs as blank spaces. Consider the
following:
When we copy a line to the buffer, the tabs change into spaces,
just as we did in the convertLineForPrint function.
When the line transfers from the buffer back into the program
line, the spaces change back into tabs.
The two items mentioned above offer formatting with a minimum of

memory requirements. The sole disadvantage of this method: No
474
ABACUS
4.3
STEP BY STE P
multiple spaces (the editor views these as true spaces, not tabs) or userspecified tabs allowed. A flag named actualEditor->tabs solves
this problem. When this flag is set to one, spaces convert to tabs.
When it is set to zero, tabs are treated as normal control characters
Ctrl><i for input and output (i.e., no conversion).
Now we must write the buffer back into the line structure. This occurs
when the cursor moves out of the current line. The function needed
executes when a message is processed within the editor's main loop.
This function writes the buffer contents back to the line if the cursor
exits this line. A variable named bufypos either contains the number
of the line found in the buffer, or contains zero if no lines exist in the
buffer. The variable actual points to the line structure of the line
found in the buffer. In addition, the ZLF USED flag in the line
structure must be set when copying a line In the buffer. This flag
determines whether a line from the output appears in the buffer.
Now we can begin to create the specifics of the new Edit. c module:
void getLineForEdit (line, znr)
register UWORD
znr;
This function copies the line line to the buffer and sets all of the
corresponding variables. The convertLineForPrint function
performs line conversion. The actualEditor->leftpos must
bave been preset to zero, allowing the conversion of all characters and
the return to original values later. This function requires some
revision-the converted line's length is needed for placing in
actualEditor->buflen, and we must know lastchar (from
actualEditor->buflastchar). For character deletion, the last
character in the buffer must be replaced by lastchar, so that the line
displays correctly on the screen.
BOOL saveLine(line)
struct Zline *line;
Save the line found in the buffer to the line structure, which points to
actualEditor->act ual. Eventually a new line must be used in
case the length of the original line differs from the new line. In case the
old or the changed line represents a fold marker, the levels of all of the
following lines must correspond to the old or changed line. If the line
cannot be rewritten, FALSE returns.
void savelfCursorMoved()
The main loop of the editor calls this function after processing each
message. This function determines whether the cursor has left the actual
input line, and. if so, calls the saveLine function.
The following functions control text processing. Since these are
implemented as individual functions, the implementation of a command
language can be added fairly easily, because individual instructions only
475
call their corresponding functions. Each function returns FALSE if it

cannot be executed.
BOOL undoLine ()
When a line is found in the buffer, this removes changes and displays
the original line.
BOOL deleteChar ()
Erases the character under the cursor and moves the rest of the line one
character to the left (enabled by pressing the <Delete> key).
BOOL backspaceCbar ()
Erases the character to the left of the cursor and moves the rest of the
line one character to the left provided insert mode is enabled. If insert
mode is disabled, a space appears to the left of the cursor, and the cursor
moves one character to the left.
BOOL insertCbar(c)
UBYTE c;
Inserts character c at the current cursor position provided insert mode is

enabled. If insert mode is disabled, the new character appears at the
current cursor position. This function also processes tabs.
BOOL insertLine (c)
UBYTE c;
Inserts a linefeed at the current cursor position. If the cursor is in the

middle of a line, this line breaks into two lines at the cursor. Character
c is either CR or LF, according to which key was pressed. The cursor
moves to the beginning of the next line.
The insertLine function acted as a basis for auto indent. Auto
indent moves the cursor to the starting column of the new line after
pressing the <Return> key. That is, if you started a line in column six
of a line and pressed <Return> while auto indent is active, the cursor
moves to column six of the new line. This also occurs if you press
<Return> to break a line while the cursor is in the middle of the line.
Resetting the autoindent flag of the Edi tor structure disables
auto indent.
BOOL deleteLine()
Erases the line in which the cursor currently lies. You must be careful
if the cursor lines in a fold at the end of a text. If you try to delete all of
the lines, other folded lines still exist. In this case this function reduces
minfold until the line found appears in the top line of the window.
The function tries to leave the cursor at the current position, scrolling
up the boltom section of the window's contents after the line is deleted.
476
4.3 STEP BY STEP
ABACUS
All of the functions are called from the handleKeys function, which
appeared in Section 4.3.7. The source text of our new module looks
like this:
<src/Edit.c;>
/************
**
Includes:
*************/
#include "sre/Editor.h tl
/***********
**
Defines:
***********/
#define CR 13
#define LF 10
#define TAB 9
1* Lenght of fold start to end */

#define FSE LEN 10
1* Number of characters, significant for the fold markers: *1
#define FSE_SIG 8
1**********************
**
External Functions:
**********************/
UWORD convertLineForPrint(),strncmp(),recalcTopOfWindow();
void deleteZline(),Insert(),printAII(), restoreZlinenptr();
void DisplayBeep(),printLine(),AddHead(),scroIIRight(),
scrol!Up() ;
void ScroIIRaster();
struct Zline *newZline(),*nextLine(),*prevLine();
BOOL cursorLeft(),cursorRight(),cursorDown(), eursorHome();
1*********************
** External
Variables:
*********************/

/*********************
* Global variables:
*********************1
UBYTE foldStart[]
UBYTE foldEnde[]
"/*IIFOLD:*/";
"/*IIENDFD*/";
477
4.
AMIGA
C FOR
1***************
* Functions:
***************1
1********************************
*
* getLineForEdit(line,znr)
*
* Copy line from editor to the
* the buffer.
**
*
line 1\ Zline.
znr : line number of the line.
********************************1
void getLineForEdit (line,znr)

register UWORD
znr;
{
register UWORD leftpos;
1* get leftpos: *1
leftpos = actualEditor->leftpos;
actualEditor->leftpos = 0;
1* convert line: *1
actualEditor->buflen =
convertLineForPrint(line+l,line->len,MAXWIDTH + 1,
actualEditor->buffer,&actualEditor->buflastchar);
1* Zline as "used"
: *1
line->flags 1= ZLF USED;
actualEditor->actual = line;
actualEditor->bufypos : znr;
1* establish Leftpos: *1
actualEditor->leftpos = leftpos;
1*******************************
*
*
cursorOnText:
*
*
*
*
make sure that the cursor

is found on text.
*******************************1
void cursorOnText()
{
register struct Zline **zptr;

register UWORD 1;
1* makes sure that the cursor is really on text: *1
zptr = &(ZLlNESPTR(l = actualEditor->wdy;

while *zptr == NULL) && 1)
{
zptr--i
478
4.3
ABACUS
STEP BY ST E P
1--;
}
actua1Editor->wdy = 1;
actua1Editor->ypos = ZLINESNR(l);
1*************************
**
deconvertLine():
* Deconver line.
**
*
bUf A target buffer.

buffer A source pointer.
* buf1en= the 1enght.
**************************1
UWORD deconvertLine(buf, buffer,buf1en)
UBYTE
*buf, *buffer;
UWORD
buf1en;
{
register UBYTE *pI,*p2,*tab,*fb = NULL;

register WORD 1;
1* deconvert line: *1
pI = buffer;
p2 = buf;
tab= actua1Editor->tabstring;
1
= buflen;
if (actualEditor->tabs)
1* Spaces converted to tabs: *1
while (1)
{
i f *p2 = *pl++)
")
i f (fb =
fb =
NULL)
p2;
else
i f (fb)
fb = NULL;
1* fb points to first space

*1
1* other characters are copied. (first blank)
*1
i f ! *++tab) && (fb

{
*fb = TAB;
fb++;
p2 = fb;
else
p2++;
1--;
else
while (1)
{
479
*p2 = *pl;
pl++;
p2++;
1--;
1* Spaces at end deleted: *1

if (actualEditor->skipblanks)
while p2 > buf) && * (p2-l) =
p2--;
return
C (UWORD)
.)
II (* (p2-l)
(p2 - buf);
1****************************************
* getFoldInc(line)
*
* determine if line is fole start

or end mark and gives 1 or -1
* back.
* Else a 0 is returned.
* line
result line.
****************************************1
getFoldInc(line)
struct Zline
*line;
UWORD
register UBYTE *p1;

register UWORD 1;
1* fold marker at start of line! */

pI = (UBYTE *) (line + 1);
I = line->len;
while (1 && *pl = .) II (*p1 == TAB))
{
pl++;
1-;
1* Type of fold markers (-> 1): *1

1*
1*
1*
1 = Start,
-1
End and
0 no fold marker.
*/
*1
*/
if (1 >= FSE SIG)

if (strncmp(pl,foldStart,FSE SIG)
1 = 1;
else if (strncmp(pl,foldEnde,FSE SIG)

1 = -1;
else
1 = 0:
else
1 = 0; 1* has no fold markers */
return (I):
/*******************************
**
480
saveLine(line)
0)
0)
=0
TAB)
4.3
ABACUS
STEP BY S T E P
* Save line.
* line
old line-Structure.
*******************************1
BOOL saveLine(line)
struct Zline *line;
{
IE'
static UBYTE buf[MAXWIDTH + 2];

*/
register UBYTE *pl,*p2,*tab,*fb
register WORD l,len;
struct Zline *pred,*old,**zptr;
/* Only for security: */
if (actualEditor->actual
return (FALSE);
==
/* +2 for CR and
= NULL;
NULL)
/* deconvert line: */
p2 = buf + (len = deconvertLine(buf,actualEditor->buffer,
actualEditor->buflen));
/* firm place, or line ends with CR or LF, */
/* and increases len correspondingly. *1
p2 = buf + len;
if (1 = line->len)
{
pI = (UBYTE *) (line + 1
i f (*pl == CR)
+ line->len - 1;
len++;
*p2 = CR;
}
else if (*pl
==
LF)
len++;
if 1 > 1) && (*--pl
==
CR)
len++;
*p2++ = CR;
*p2
LF;
1* calculate new line: */

if (EVENLEN(len) != EVENLEN(line->len
{
old = line;
if (line = newZline(len
{
Insert (&actualEditor->zlines,line,old);
line->f1ags = old->flags & -ZLF USED;
deleteZline (old) ;
/* now possible to convert pointer: */
for (1 = 0, zptr = actualEditor->zlinesptr;
1 <= actualEditor->wch; 1++, zptr++)
if (*zptr == old)
{
481
4.
AMIGA
*zptr = line;
break;
else
return (FALSE);
else
{
line->len
= len;
line->flags &= -ZLF_USED;
1* write line : *1
p2 = buf;
1
= len;
while
(1)
*pl = *p2;
pl++;
p2++;
1-;
1* reconstruct folding: *1
1* Was old or is new line on fold mark? *1
= getFoldlnc(line);
if line->flags & ZLF FSE) I I 1)
1* next the ZLF FSE flag set to

if
(1)
1:
*1
line->flags 1= ZLF_FSE;
else
line->flags &= -ZLF_FSE;
1* calculate how the fold level of the following lines *1

1* : level += 1.
*1
i f old = line->succ) ->succ)
{
if (line->flags+l) & ZLF FOLD)

ZLF FOLD
1* Old line was fld start: *1
1 -= 1;
else if line->flags & ZLF FOLD)
ZLF FOLD
1* Old line was fold end: *1
1 += 1;
(old->flags &
old->flags+l) &
i f (1)
{
1* calculate fold level fo raIl following lines: *1

while (old->succ)
{
= (old->flags & -ZLF FOLD)

I old->flags + 1) & ZLF FOLD);
old->flags
old = old->succ;
restoreZlinenptr();
printAll () ;
482
4.3
AB.4.CUS
STEP BY STEP
cursorOnText () ;
1* line output: *1
1* First calculate Ypos:
*1
1 <z actualEditor->wch; 1++, zptr++)
if (*zptr == line)
{
printLine(line,actualEditor->yoff + actualEditor->ch*l);
break;
1* Delete reference: *1
actualEditor->buflen
= 0;
actualEditor->bufypos = 0;
1* Text was changed! *1
actualEditor->changed = 1;
return (TRUE);
1***************************************
**
savelfCursorMoved:
**
Save line in buffer,

* if the cursor is moved.
****************************************1
void savelfCursorMoved()
{
register UWORD bufypos;

if (bufypos = actualEditor->bufypos)
if (actualEditor->ypos != bufypos)
if (! saveLine(actualEditor->actual
DisplayBeep(NULL);
1******************************************
**
undoLine:
*
*
*******************************************1
void undoLine ()
{

if (z = actualEditor->actual)
{
483
4.
AMIGA
actualEditor->buflen : 0;
z->flags &=
~ZLF_USED;
1* line output: *1
printLine(z,actualEditor->yoff
+ actualEditor->ch*actualEditor->wdy):
1***************************************************
* getBufferPointer(pptr,pinc,prest):
**
Initialize pointer to actualEditor->buffer.

* The buffer is initialized.
* FALSE is returned i f the buffer
* cannot be initialized.
* pptr ~ Pointer in buffer at cursor position.

* pinc ~ Offset to buffer start.
* prest~ buflen - inc.
***************************************************1
800L getBufferPointer(pptr,pinc, prest)

UBYTE
**pptr;
UWORD
*pinc,*prest;
{
register UBYTE *ptr;

register WORD inc, rest;
/* Buffer initialized: *1
if (actualEditor->actual == NULL)
{
register UWORD wdy;

if (ZLINESPTR(wdy = actualEditor->wdy
getLineForEdit(ZLlNESPTR(wdy),ZLINESNR(wdy;
else
if (actualEditor->num lines == 0)
/* first line: *1
if (z
= newZlineUWORD)
* UBYTE *) (z + 1 = LF;
AddHead(&actualEditor->zlines,z);
ZLlNESPTR(O) = z;
ZLINESNR(O) = (actualEditor->num lines
1);
getLineForEdit(ZLINESPTR(O),ZLINESNR(O:
}
else
return (FALSE);
else
1* Cursor nott in line! */

return (FALSE):
/* Calculate pointer to cursor position: *1
484
4.3
ABACUS
STEP BY STEP
inc = actuaIEditor->xpos - 1;
ptr = actuaIEditor->buffer + inc;
rest= actuaIEditor->buflen - inc;
1* What if inc> buflen? *1

if (inc < MAXWIDTH)
{
if (rest < 0)
{
register UBYTE *p;

p = actuaIEditor->buffer + actuaIEditor->buflen;
actuaIEditor->buflen -= rest;
while (rest)
{
*p =
p++;
I;
rest++;
else
return (FALSE);
*pptr = ptr;
*pinc = inc;
*prest= rest;
return (TRUE);
1************************************
**
deleteChar:
* Delete character under the cursor.

* False is returned if no more
* characters.
*************************************1
BOOL deleteChar()
{
UBYTE *ptr;
UWORD inc,rest;
register UBYTE *pl,*p2;
register UWORD 1;
if (! getBufferPointer(&ptr,&inc,&rest))
return (FALSE);
i f (rest)
{
pI = ptr + 1;
p2 = ptr;
1
= --rest;
while (1)
{
*p2 *p1;
p1++;
p2++;
1-;
485
4.
*p2
AMIGA
= actua1Editor->buflastchar;
1* Lenght - 1 *1
actua1Editor->buflen--;
printLine (actua1Editor->actua1, actua1Editor->yoff
+ actua1Editor->ch*actua1Editor->wdy);
return (TRUE);
else
return (FALSE);
/******************************
**
backspaceChar:
* Delete character before cursor.

* False is returned when
* Cursor in start of line
*******************************1
BOOL backspaceChar()
{
UBYTE *ptr;
UWORD inc, rest;
register UWORD 1;
return (FALSE);
i f (inc)
{
if (actua1Editor->insert)
{
= ptr;
pl
p2
--ptr;
rest;
while (1)
=
1
{
*p2
*pl;
pl++;
p2++;
1--;
*p2
= actua1Editor->buf1astchar;
1* Length - 1 *1
actua1Editor->buflen--;
inc--;
else
{
*--ptr =
inc--;
rest++;
cursorLeft () ;
486
';
4.3
ABACUS
STEP BY STEP
printLine(actualEditor->actual,actualEditor->yoff
+ actualEditor->ch*actualEditor->wdy);
return (TRUE);
}
else
return (FALSE);
1******************************************
* insertChar (c)
**
Inserts character in buffer.

* FALSE returned if character
* cannot be inserted.
**
c= character.
******************************************1
BOOL insertChar (c)

register UBYTE c;
{
UBYTE *ptr;
WORD inc, rest;
WORD xadd = 1;
if (! getBufferPointer(&ptr,&inc,&rest
return (FALSE);
if c
==
TAB) && (actualEditor->tabs
if (actualEditor->insert)
{
1* Insert spaces: *1
register UBYTE *p1,*p2;
register UWORD num,l;
p1
= actualEditor->tabstring + actualEditor->xpos - 1;
= MAXWIDTH
num = 1;
- actualEditor->xpos;
while (1 && (*++p1

{
xadd++;
1-;
num++;
1* num = Number of spaces to insert: *1
if (actualEditor->buflen + num > MAXWIDTH)

num = MAXWIDTH - actualEditor->buflen;
1* Write rest of line : *1

p1
p2
ptr + rest;
pl + num;
1
= rest;
while (1)
=
=
*--p2 = *--pl;
1-;
487
4.
AMIGA
C FOR
pI = ptr;
1
= num;
while (1)
{
*pl = ':
pl++;
1-;
actualEditor->buflen += num;
rest += num;
}
else
{
1* move cursor only: *1

register awORD max;
tab = actualEditor->tabstring + actualEditor->xpos - I:
max = MAXWIDTH - actualEditor->xpos;
while (max && (*++tab
{
xadd++;
max-i
}
else
{
1* besides a normal character: *1

if (actualEditor->buflen < MAXWIDTH)
{

register awORD 1;
pI
p2
ptr + rest;
pI:
1
rest;
while (1)
{
=
=
*p2 = *--pl;
p2--;
1-:
*ptr
c:
1* Length + 1 *1
actualEditor->buflen++:
rest++:
else
1* Word-Wrap here later *1

return (FALSE):
else
{
*ptr '" c;
i f (rest = 0)
{
actualEditor->buflen++;
488
4.3
ABACUS
rest
STEP BY STEP
1;
/* move cursor: */
while (xadd)
{
if (! cursorRight()) break;
xadd--;
/* output line: */
print Line (actualEditor->actual,
actualEditor->yoff + actualEditor->ch*actualEditor->wdy);
return (TRUE);
/**************************
**
insertLine(c):
*
*
*
*
new line inserted before

actual line.
FALSE returned if
error encountered.
* c
line end (CR/LF).
**************************/
BOOL insertLine(c)
UBYTE
c;
{

UBYTE *ptr;
WORD inc,rest,len;
UWORD autoindent;
register UWORD 1;
/* No new line inserted, when actuale line
has no fold markers!!! */
if z = actualEditor->actual) == NULL)
z = ZLINESPTR(actuaIEditor->wdy);
if (!z I I ! (z->flags & ZLF_FSE))
{
/* get line in buffer: */

return (FALSE);
/* calculate auto indent: */
if (actualEditor->autoindent)
(
pi
actualEditor->buffer;
1 = inc;
while
(l
&& *p1 == , ') II (*pl
TAB) ))
p1++;
489
1-;
i f (1)
autoindent : pI - actualEditor->buffer + 1;
else
1* insert spaces in line: *1
autoindent : I:
else
autoindent : I:
p2 : buf + (len = deconvertLine(buf,actualEditor>buffer, inc) );
*p2= c;
len++;
if (z = newZline(len
{
z->flags : (actualEditor->actual->flags & ~ZLF_USED);

pI = (UBYTE *) (z + I):
p2 : buf;
1 : len;
while
(1)
*pl = *p2;
p1++;
p2++;
1--;
1* line connect: *1
Insert (&actualEditor->zlines,z,actualEditor->actuaI>pred) ;
actualEditor->num lines++;
actualEditor->changed : 1;
1* write rest of buffer line *1
1* remember auto indent! *1
pI : actualEditor->buffer:
1
: autoindent:
while (--1)
pl++;
1* Old contents remain! *1
p2 : ptr;
1 = rest;
while (1)
{
*pl : *p2;
p1++;
p2++;
1-;
}
actuaIEditor->buflen : pI - actuaIEditor->buffer;
1* rest filled with lastchar: *1

1 : inc + 1 - autoindent;
while (1)
{
490
ABACUS
4.3
STEP BY STEP
*pl = actualEditor->buflastchar;
p1++;
1--;
1* firm place, or new line fold marker: *1

if (1 = getFoldlnc(z
{
register struct Zline *old;

register UWORD fold;
z->flags 1= ZLF_FSE;
1* calculate fold level for all following lines: *1

if old = z->succ)->succ)
while (old->succ)
{
old->flags = (old->flags & -ZLF FOLD)

1 old->flags + 1) & ZLF~OLD);
old = old->succ;
1* fit min fold and maxfold: *1

if 1 > 0) II z->flags & ZLF_FOLD) > 0
{
'fold = (z->flags & ZLF FOLD) + 1;

if (actualEditor->minfold > fold)
actuaIEditor->minfold = fold;
if (actualEditor->maxfold < fold)
actualEditor->maxfold = fold;
ZLINESPTR(actualEditor->wdy) = z;
actualEditor->wdy =
recalcTopOfWindow(actualEditor->wdy);
1* if Cursor calculate line => zlinesptr[O]: *1

if (actualEditor->wdy == 0)
ZLINESPTR(O) = z;
restoreZlinenptr();
1* redisplay window: *1
actualEditor->xpos = autoindent;
if (actualEditor->xpos <= actualEditor->leftpos)
{
1* If scrolling necessary: printAll *1

actualEditor->leftpos = actualEditor->xpos - 1;
(++actualEditor->wdy >= actualEditor->wch)
it
(
ZLlNESPTR(O) = ZLINESPTR(l);
ZLINESNR(O) = ZLINESNR(l);
restoreZlinenptr();
actuaIEditor->ypos = ZLlNESNR(actualEditor->wdy);
printAll () ;
}
else
i f (l)
491
4.
AMIGA
C FOR
1* If fold marker: redisplay window: *1

if (++actuaIEditor->wdy >= actuaIEditor->wch)
(
ZLINESPTR(O) = ZLINESPTR(1);
ZLINESNR(O) = ZLINESNR(1);
restoreZlinenptr();
printAlI () ;
}
else
(
1* it was scrolled: *1
(
1* Cursor was on last line: *1

register UWORD nr;
scroIIUpUWORD) 1);
nr = (--actuaIEditor->wdy) - 1;
printLine(ZLINESPTR(nr),actuaIEditor->yoff
+ actuaIEditor->ch*nr);
}
else
(
1* CUrsor near middle of window: *1

register UWORD nr;
if (ZLINESPTRnr = actuaIEditor->wdy) + 1
{
1* move rest of window down: *1

ScrollRaster(actuaIEditor->rp,
OL, (LONG)-actuaIEditor->ch,
(LONG)actuaIEditor->xoff,
(LONG)actuaIEditor->yoff
+ actuaIEditor->wdy*actuaIEditor->ch,
(LONG)actuaIEditor->xscr,
(LONG)actuaIEditor->yscr);
}
nr--;
return (TRUE);
}
else
return (FALSE);
}
else
(
cursorHome 0;
cursorDown 0 ;
492
4.3
ABACUS
STEP BY STEP
return (FALSE);
1****************************
**
**
deleteLine:
delete line
* that the cusor is on.
* FALSE returned if
* it cannot be deleted
*****************************1
BOOL deleteLine()
{
register struct Zline *line,*next,*prev;

register UWORD 1 = 0;
UWORD nextnr,prevnr;
if (line
ZLINESPTR(actualEditor->wdy))
1* mark the deleted line: *1

if (line == actualEditor->actuEl)
{
actualEditor->buflen = 0;
actualEditor->num_lines--;
1* Reconstruct folding: *1
{
1* Fold level for all following lines: *1
i f z = line->su=) ->succ)
(
if (line->flags+1) & ZLF FOLD)

ZLF FOLD))
/* line was Fold start: *1
1 = -1;
else if line->flags & ZLF_FOLD)
ZLF FOLD))
1* line was Fold end: *1
1 = 1;
else
(z->flags
&
z->flags+l) ,
1 = 0;
if (1)
while (z->succ)
{
z->flags = (z->flags & -ZLF_FOLD)

I z->flags + 1) & ZLF_FOLD);
z = z->su=;
493
4.
1* Next
prevnr=
prev =
nextnr=
next =
AMIGA
FOR ,ADVANCED PROGRAMMERS
and previous line marked: *1

prevLine (line, &prevnr) ;
ZLlNESNR(actualEditor->wdy);
nextLine (line, &nextnr) ;
1* Security check: *1
while prev = NULL) && (next
>minfold) )
NULL) && (actualEditor-
1 = 1;
1* Window completely output *1
actualEditor->minfold--;
prevnr=
prev =
nextnr=
next =
ZLINESNR(actuaIEditor->wdy);
nextLine (line, &nextnr) ;
i f (prev II next) break;
if (actualEditor->minfold) continue;
if next = actualEditor~>zlines.head)->succ
next = NULL;
else
nextnr = 1;
NULL)
break;
1* delete line *1
deleteZline(line);
1* pointer to actual line set for recalc: *1
i f (next)
{
ZLlNESPTR(actualEditor->wdy)
ZLINESNR(actuaIEditor->wdy)
= next;
= nextnr;
else
{
ZLlNESPTR(actualEditor->wdy) = prev;
ZLINESNR(actualEditor->wdy) = prevnr;
1* Restore window contents: *1

if (l)
(
actualEditor->wdy = recalcTopOfWindow(actualEditor->wdy);
restoreZlinenptr();
printAllO;
cursorOnText 0 ;
}
else
{
1* with scrolling: *1
register UWORD oldwdy;
oldwdy = actualEditor->wdy;
494
4.3
ABACUS
STEP BY STEP
restoreZlinenptr();
if (next)
{
1* push up rest of window: */

OL, (LONG)actualEditor->ch,
(LONG)actualEditor->xoff,
(LONG)actualEditor->yoff
+ actualEditor->wdy*actuaIEditor->ch,
(LONG)actuaIEditor->xscr,
(LONG)actualEditor->yscr);
printLine(ZLINESPTR(actualEditor->wch - 1),
actualEditor->yoff
+ actuaIEditor->ch*(actuaIEditor->wch - 1));
)
else
i f (oldwdy -- actualEditor->wdy)
{
1* scroll top section of window down: *1

OL, (LONG)-actualEditor->ch,
(LONG)actualEditor->yoff,
(LONG)actualEditor->xscr,
(LONG)actuaIEditor->yoff
+ (actualEditor->wdy + 1)
*actualEditor->ch - 1);
printLine(ZLlNESPTR(D),actuaIEditor->yoff);
}
else
printLine(ZLlNESPTR(oldwdy),actualEditor->yoff
+ actualEditor->ch*oldwdy);
cursorOnText () ;
return (TRUE);
}
else
return (FALSE);
}
Now a few remarks about the functions:
The character strings which represent the fold marks are defined
as global variables at the beginning of the module. FSE_SIG

specifies the number of characters needed for the fold mark; in .
this case /*.fFOLD: represents the fold start and /*.fENDFD
represents the fold end. This character string can be changed so
that the editor can accomodate other programming languages
(e.g., ; .fFOLD: could be used in assembler source). FSE_LEN
supplies the character string's total length. This length must be
identical for both starting marks and ending marks. The entire
length will be needed later when we automate the omission of
text segments. Then the editor must insert a line before and after
495
the fold mark. The user must remember to add the closing
comment mark after the fold marks.
The getLineForEdi t function shows a change in the

convertLineForPrint function. It returns the length of the
converted line. The fifth parameter requires a pointer to a UBY1E
variable in which lastchar is saved.
Notice the new function named cursorOnText. It checks to

see whether the cursor is in an existing line or if it was already
moved past the end of text or fold. In the latter case the cursor
moves to the last line in the window.
The deconvertLine function performs the opposite function

of the Output module's convertLineForPrint function.
It checks each tab (tab = 0) for preceding spaces (fb! =0). If
leading spaces exist, the function replaces these spaces with tabs.
The convertLineForPrint function has yet another talent

When the program encounters a set actualEditor>skipblanks flag, it deletes all unnecessary spaces at the end
of a line. If you try to edit text entered with a preset right margin
less than the right border of the full size window (i.e., a line
width less than 80 columns), you cannot delete the dots
following the formatted line (more on this later).
The getFoldInc function tests to see whether a line

represents a fold mark, and returns zero if not. Otherwise
getFoldInc returns a value of 1 for a fold start mark and -1
for an ending fold mark. These values are compatible with the
saveLine function.
The saveLine function controls setting the fold level if the

line is a fold marker.
The following observations concern the above items. If a normal line is

changed to a fold start mark, then the fold levels of all of the following
lines must be incremented by one.
If a fold start mark is changed to a normal line and no other fold marks
exist after this, fold levels of all of the following lines must be
decremented.
These increment and decrement rules also apply to the fold end mark. If
a line is a fold mark before as well as after a change, things can be
complicated. However, creating a table of values, which returns the
value by which all successive folds must be increased or decreased,
makes things much simpler. This value depends on the status of a line
(normal line or fold mark). The process works like this: The
getFoldInc function handles a line changed to a fold mark and notes
the return value. A fold mark changed to a normal line forces a
496
4.3
ABACUS
STEP BY STEP
comparison between the fold level of the current line and the fold level
of the next line. In this case we get the negative value relative to
getFoldInc (-I for the start and +1 for the end). Now we add the two
values, and receive the value that must be added to the fold level of all
subsequent lines.
So much for folding. The changed flag from the Editor structure
detennines whether something in the text has been changed. If the user
changes a line with saveLine, the changed flag must be seL
The line is displayed at the end of the function.
The getBufferPointer function helps convert or copy a

line to the editor buffer. It initiIaizes the three variables that
simplify editing, a pointer to the character at which the cursor is
found, the number of characters to the left of the cursor and the
length of the rest of the line. If the cursor follows the last
character of the line, the program inserts enough spaces to keep
one space directly to the left of the cursor. The insertChar
function inserts a character at the current cursor position.
In overwrite mode, pressing the <Tab> key moves only the

cursor to the right. If you move the cursor in this way, don't
change text and don't type spaces over the texL
The insertLine function prohibits division of a fold mark by

pressing <Return>. This would create two new fold marks from
the one fold mark. We chose not to solve this, but feel free to
expand the program to work in this case. Instead you can enter a
new fold mark and press <Return>, and the fold level, change will
remain harmless.
Text output in the window is another matter. If you insert a new line,
the window contents are moved out of order. It's possible to use the
printAll function, but that would be too slow. Besides, we want the
editor to allow scrolling. The insertLine function should contain
the following:
When the window contents must be scrolled horizontally, or when the
newly insert line is a fold marker, printAll displays the window
contents. This would not be necessary with horizontal scrolling, but if
we scroll the window contents horizontally and then vertically,
printAll is not much slower. Otherwise it scrolls vertically and the
section of text at the cursor position scrolls down. When the cursor is
on the bottom line, the entire contents of the window scroll up.
Scrolling must also be enabled in the deleteLine function.

The cursor remains in the current line and the rest of the text
scrolls up. When the cursor is in the last line and deleteLine
is invoked, the section above it scrolls down. If there is no text
497
4.
in the top half of the window. the Y position of the cursor

decrements.
Another problem arises in this function when the cursor is in a fold at
the end of the text (minfold > 0), from which all. lines are deleted.
Then the editor cannot access the lines in this fold-this is why the
deleteZLine function reduces minfold until it finds a line.
The fold marker can also be erased with deleteZLine. Then the fold
level of all subsequent lines is changed accordingly.
The changes to the modules are fairly extensive, so let's begin with
Edi tor. h. We removed the toppos variable from the Edi tor
structure because the zlinesnr array already performs the same task.
The following elements are new:
UBYTE
UWORD
{)WORD
{)WORD
{)WORD
buf1astchar;
buf1en,bufypos;
tabs
: 1;
skipb1anks : 1;
autoindent : 1;
The Output. c module is next, into which we must add the

convertLineForPrint function. Here's the altered function
headel:
{)WORD convertLineForPrint(1ine,len,w,buf,lc)
UBYTE
*line;
register WORD
len;
register UWORD
w;
register UBYTE
*buf;
UBYTE
*1c;
We need another local variable:

UWORD realLen;
These variables determine the length of the converted line. The

following conversion remains almost as it was. However, we must
specify the line length before the rest of the line fills with last char:
1* Save the length of the line *1
rea1Len = 1 - 1;
Now the rest of the line fills with lastchar and the function returns
the length:
*lc .. 1astchar;
return (realLen);
We made so many changes to the printLine function that we chose

to reprint the function in its entirety below. This function displays the
buffer if the ZLF_USED flag of the line was set If the line does not
exist (line =null), the corresponding window area is deleted (as in the
deleteLine function):
498
4.3
ABACUS
STEP BY STEP
void printLine
(line,y)
register UWORD
y;
{
static UBYTE buf[MAXWIDTH];

i f (line)
{

register UWORD w;
register ULONG pen;
UBYTE lc;
/* check if line was processed: */
if (line->flags & ZLF USED)
/* copy buffer after buf, since printAt changed this! */

if (MAXWIDTH > aetualEditor->leftpos)
{
w = MAXWIDTH - aetualEditor->leftpos;
p2 = aetualEditor->buffer + aetualEditor->leftpos;
pl = buf;
while (w)
{
*pl = *p2;
p1++;
p2++;
W--i
else
pl = buf;
/* fill rest with lastchar: */
p2 = buf + (w = actualEditor->wcw);
Ie = actualEditor->buflastehar;
while (pl < p2)
{
*pl = Ie;
p1++;
else
eonvertLineForPrint(line+l,line->len,
w = aetualEditor->wcw,buf,&le);
/* If start/end marker of fold => FOLDPEN */
{
i f z = line->succ) ->suee)
{
if z->flags & ZLF FOLD) <= aetualEditor->maxfold)

{
SetAPen(actualEditor->rp,FOLDPEN);
pen = FOLDPEN;
}
else
pen = FGPEN;
else
499
SetAPen(actualEditor->rp,FOLDPEN);
pen = FOLDPEN;
.}
1* Old write color first again: *1

if (pen != FGPEN)
}
else
printAt(buf,w - SplitDec,actuaIEditor->xoff,y,FGPEN);
}
else
{
1* case no line => erase in window: *1

struct RastPort *rp
= actuaIEditor->rp;
SetAPen(rp,BGPEN);
RectFill(rp,(ULONG)actualEditor->xoff,
(ULONG)y,
(ULONG)actualEditor->xscr,
(ULONG)y + actuaIEditor->ch - 1);
SetAPen (rp, FGPEN) ;
}
The Cursor. c module has also undergone some changes. The

recalcTopOfWindow function is new. This function determines the
line position of the top line of the window and places the value in the
linesptr and linesnr arrays. This function initializes the
line spt r array after a text alteration, or after changing minfold or
maxfold. The cursor remains at the old position.
UWORD recalcTopOfWindow(y)
register UWORD
y;
{
register struct Zline *z,*zp;

register UWORD znrp;
UWORD znr,oldy = y;
z = ZLlNESPTR(y);
znr = ZLlNESNR(y);
while (y)
{
lIP
.. z;
znrp = znr;
if z = prevLine(z,&znr
break;
y-;
if (y)
{
ZLlNESPTR(O) .. zp;
ZLINESNR(O) .. znrp;
}
else
{
500
;: NULL)
4.3
ABACUS
ZLINESPTR(O)
ZLINESNR(O)
STEP BY STEP
z;
= znr;
return (oldy - y);
This function keeps the cursor in the current line when encountering a
fold and when deleting a line. The new line position is returned from
this function, which usually matches the old line position. Now when
encountering a fold, the cursor can be relatively far down in the
window, but not very much text is visible. Then the cursor must be
moved up where the new position accesses this function.
The program segment for fold handling is placed before the
handleKeys function:
1***************************
**
enterFold:
* STEP one fold in.
***************************1
void enterFold ()
(

if (actuaIEditor->maxfold < ZLF FOLD)
{
if (z : ZLINESPTR(actuaIEditor->wdy))
if z : z->succ)->succ)
1* If there is a following line: *1
if (ZLINESPTR(actuaIEditor->wdy)->flags & ZLF FOLD)
&& z->flags & ZLF:FOLD) == actuaIEditor->maxfold)
1* When this begins a new fold: */
actuaIEditor->minfold = actuaIEditor->maxfold;
ZLINESPTR (0) = z;
ZLINESNR(O)
actualEditor->wdy = 0;
actualEditor->wdy = recalcTopOfWindow(actuaIEditor->wdy);
restoreZlinenptr();
printAll () ;
actualEditor->ypos = ZLlNESNR(actualEditor->wdy);
1*****************************
* exitFold:
** Step
one fold out.
*****************************1
501
void exitFoldO
{
register UWORD wdy;

if (actualEditor->maxfold)
{
actualEditor->maxfold--;
if (actualEditor->minfold > actualEditor->maxfold)
aqtualEditor->minfold = actualEditor->maxfold;
wdy = actualEditor->wdy;
if (ZLINESPTR(wdy) == NULL)
actualEditor->wdy = wdy = 0;
if (ZLlNESPTR(wdy
if (ZLINESPTR(wdy)->flags & ZLF FOLD) > actualEditor>maxfold)
II ZLINESPTR(wdy)->flags & ZLF FOLD) < actualEditor>minfold
ZLlNESPTR(wdy) =
prevLine(ZLlNESPTR(wdy),&(ZLlNESNR(wdy);
actualEditor->wdy = recalcTopOfWindow(wdy);
restoreZlinenptr();
printAllO;
cursorOnText 0 ;
The last function shows how the recalcTopOfWindow function is

inserted. In addition to the existing functions for cursor movement, an
additional function is needed for placing the cursor at the beginning of a
line, based on the function used in insertLine:
BOOL cursorRome()
{
actualEditor->xpos = 1;
if (actualEditor->leftpos)
scrollRight(actualEditor->leftpos);
return (TRUE);
The following new defines are necessary to this version of the

editor:
idefine
ide fine
idefine
idefine
ide fine
ide fine
ide fine
idefine
idefine
idefine
idefine
idefine
502
CROME 'A'
UNDO '?'
CFOLD 6
CENDF 5
TABMODE 20
WRITEMODE 15
AUTOINDENT 1
DELLINE 2
DEL 127
BS 8
LF 10
CR 13
4.3
ABACUS
STEP BY STEP
Examine the following listing for handleKeys, and you'll notice

some changes:
register WORD
len;
{

while (len> 0)
(
first = *buf;
buf++;
len--;
if first == CSI) && (len> 0
(
len--;
switch (*buf++)
{
case CUU:
cursorUpO;
break;
case CUD:
cursorDown 0 ;
break;
case CUF:
cursorRight 0 ;
break;
case CUB:
cursorLeft 0 ;
break;
case SU:
halfPageDown () ;
break;
case SD:
halfPageUp () ;
break;
case' ':
i f (len> 0)
{
len-;
swi tch (*buf++)
(
case CHOME:
cursorHome () :
break;
default:
if (! insertCharUBYTE)CSI
DisplayBeep(NULL):
buf -= 2;
len += 2;
break;
else
goto noCSI:
case UNDO:
if len> 0) && (*buf
{
==
'-'
buf++;
len--;
503
undoLine():
break;
}
1* Else: default *1
default:
noeS!:
1* No command sequence! *1
DisplayBeep(NULL);
OOf--:
len++:
break:
1* switch (first) *1
}
else
switch (first,
{
case CFOLD:
enterFold () :
break;
case CENDF:
exi tFold () :
break:
case TABMODE:
actuaIEditor->tabs = !actuaIEditor->tabs:
printAlI () :
break;
case WRITEMODE:
actuaIEditor->insert = !actuaIEditor->insert;
break;
case AUTOINDENT:
actuaIEditor->autoindent= !actuaIEditor->autoindent;
break:
case DELLINE:
if (! deleteLine(
DisplayBeep(NULL);
break:
case DEL:
if (! deleteChar(
DisplayBeep(NULL);
break:
case BS:
i f (! backspaceChar () )
DisplayBeep(NULL);
break;
case LF:
case CR:
if (! insertLine(first
DisplayBeep(NULL);
break;
default:
if (! insertChar(first
DisplayBeep(NULL);
1* switch (first) *1
1* while (len) *1
}
The ftrst switch case instruction handles the CSI sequences as

before. It is inconvenient that these CSI sequences do not have a set
504
ABACUS
4.3
STEP BY STEP
length, some are two and three characters long. This instruction
provides another swi tch case instruction which processes all of the
sequences that begin with CSI ' '. This is currently only CSI ' , 'A',
which is sent through <Shift><Cursor Left>. This could be expanded
later to include user defined sequences.
The second switch case instruction handles control characters that
call their own functions at that moment Later we will access a section
of the command line.
Let's look at an overview of editor commands available so far:
Keys
Cursor keys
Shift Cursor up/down
Shift Cursor left
Help
CtrIA
CtrlB
Ctrl E
CtrlF
CtrIO
Ctrl T
Delete
Backspace
Comments
moves the cursor by one character/line
scrolls text a half page up or down
sets the cursor at the beginning of the line
undo
turns Auto Indent on/off

erases the line the cursor is in
leaves a fold
enters a fold
switches between insert and overwrite mode
turns tabs on/off
as usual
as usual
The functions from the new Edi t module must be declared:

void undoLine();
void cursorOnText();
BOOL insertChar(),deleteChar(),backspaceChar(),
insertLine(),deleteLine();
The deleteZline function in the Memory. c module has been

changed for two reasons:
The contents of the zlinespt rand zlinesnr arrays remain

unchanged
The line is not erased from the list of the lines.
We have also eliminated the following lines following

Remove (line) to keep deleteZLine fully operational:
1* remove line from list: (keep these two line)*1
Remove(line);
1* set pointer to this line to zero:
(delete these)*1
if (actualEditor->actual == line)
if (*zptr == line)
{
50S
*zptr = NULL;
break;
A few changes still need to be made to the main Edi tor. c module.
The additional elements of the Edi tor structure must be initialized in
the OpenEdi tor function:
ed->buflen
= 0;
ed->bufypos
= 0;
ed->buflastchar= , ';
ed->tabs
= 1;
ed->skipblanks = 1;
ed->autoindent = I:
It may surprise you when testing the cursor that the cursor runs on
(keeps going after you release the cursor key) horizontally. This
happens with almost all the functions that move the cursor or insert
text, and when you set the keyboard repeat to very high speed from
Preferences.
This effect can be prevented by checking to see if something has
changed after the program processes a keypress. If this is the case, and if
all of the RAWKEY messages have set the REPEAT flag, the first
message is removed. The REPEAT flag is set from the operating
system when a RAWKEY message is encountered instead of a keyboard
repeat, because the user has pressed the key. Messages that do not set
the REPEAT flag are not deleted because the user may sometimes press
keys faster than the editor can process them.
The current messages are deleted until only one remains. This is
processed during the next execution of the main loop. In principle we
could also delete the first message, but then the cursor movement would
be uneven. The fact that the cursor moves one character farther than you
may want it to move is the price paid for the fast and smooth cursor
movement.
The RAWKEY check looks like the following:
case RAWKEY:
{
register struct IntuiMessage *im1,*im2;

if (! (code & IECODE UP PREFIX
{
inputEvent.ie Code
= code;
if inputLen-= RawKeyConvert(
) >= 0)
handleKeys (inputBuffer, inputLen);
/* prevent running: */
iml = (struct IntuiMessage *)
edUserPort->mp_MsgList.lh_Head;
506
4.3
ABACUS
STEP BY S T E P
while (im2 = (struet IntuiMessage *)

iml->ExecMessage.mn_Node.ln_Suee)
if
if
if
if
if
(im2->ExecMessage.mn Node.ln Suee == NULL) break;

(iml->Class != RAWKEY) break;
(! (iml->Qualifier & IEQUALIFIER REPEAT break;
(im2->Class != RAWKEY) break; (! (im2->Qualifier & IEQUALIFIER_REPEAT break;
1* Message removed: *1
lml = GetMsg(edUserPort);
ReplyMsg (lml);
lml
(struct IntulMessage *)
break;
The following function must be called directly after the s wit c h

case assignment:
savelfCursorMoved();
This function must be declared as void in the external function list
To compile this version, look for the name of the object file edi t . 0
in the makefile, and add the following line to the end of the
makefile:
sre/Edit.o: sre/Edlt.e sre/Editor.h /pre/Editor.pre
The optional diskette for the book contains the source text in the v 0 . 5
directory. After you compile the new program you have quite a bit to do
because there are many options to test. Start the program using the
Test Text file as described in the last section. An error that we found
in the program during development has been left in the program for
you. You should be able to find it for it yourself.
Debugging
Have you found it? If you want to add characters to the end of a line in
the overwrite mode, the editor does not enter the new characters when
you exit the line again. We looked over the source text and couldn't find
an error anywhere. This is where the debugger helps us. Compile the
editor using make debug. Start the debugger with db and then the
editor, as we have already described (your drive setup may differ):
sys3:bin/db
Editor <TestText
The debugger displays the editor as usual. Now think about how to
proceed further. The error occurs when we want to add a character at the
end of the line using the insertChar function. Set a breakpoint at
this function:
507
4.
AMIGA
bs insert Char
and start the editor by pressing <g> (Go). Activate the editor window
and press <Ctrl><O> to enter overwrite mode. Search for the end of a
line, or for a blank line. Move the cursor there and press a key (e.g.,
<8. The editor stops and the debugger displays the beginning of the
insertChar function. Now you need to find the error! The reason
could be that the length of the buffer was not increased correctly when
you want to add a character. It would be best to move through this
function in single step mode.
First the LINK and MOVEM instructions perform general initialization.
Then register D4 receives the inserted character and treats the character
as a register variable. The local variable -a (A%) is filled with the
value one, which it handles as xadd. The getBufferPointer
function call occurs. Press <t> to bypass this function, not <s>.
Pressing <t> executes a command completely and executes a subroutine
of BSR or JSR with a subroutine call, before program control returns
to the debugger.
Now we look at the values written in the p t r. inc and res t
variables:
p2dx a5-8
Before the getBufferPointer function call. three addresses are

placed on the stack, which are the addresses of the three variables
sought:
-8 (as)
-6 (as)
-4 (as)
Because variables in C always appear on the stack in the order in which

they were defined, we can assume the following association:
rest = -8 (as)
inc = -6(a5)
ptr = -4 (a5)
To display this we must display two words at address -8 (as) and then
a long word, which the debugger calls p2dX. In addition, two decimal
and one hexadecimal long word must appear from the start address. The
values for inc and rest are both zero if the cursor moves to an empty
line. The value of ptr is not of interest, but look where ptr points:
db *(a5-4)
The line in the debugger window consists of an address, and equal sign.
and a 20. ptr points to a string that is made up only of spaces
( $ 20 ) . Let the debugger continue in single step mode until everything
is correct. It follows the testing of the return values and then checks to
see if D4 contains the value 9. D4 is the inserted character. and 9
508
4.3
ABACUS
STEP BY S T E P
corresponds to a tab. Because you pressed a nonnalletter, the program

continues.
A value passes from the Edi tor structure and the first bit is marked
from there and tested. If this bit equals zero, the editor is in overwrite
mode and the program continues. When you disassemble the program
from here, you should see the following:
insertChar+184
-insertChar+l88
-insertChar+l8a
-insertChar+18e
-insertChar+190
-insertChar+194
insertChar+ 198
:=insertChar+1ge
movea.l
move.b
c1r.w
beq.s
movea.l
addq.w
move.w
-4(aS),aO
d4, (aO)
-8 (as)
insertChar+1ge
-actualEditor,aO
#1,76(aO)
#1,-8(aS)
This short program segment corresponds to the following lines which

allow character insertion in overwrite mode. These lines appear at the
end of the insertChar function:
*ptr = c;
i f (rest = 0)
{
rest = 1;
The characters at the location indicated by pt r are saved. The re s t

variable should be set to zero (instead of memory location -8 (AS) ).
Here's the error: The single equal sign between rest and 0 in the
second line. That line should appear as follows:
(rest =
0);
Correct the error and recompile the program if you wish.

C's ability to write instructions appears here as a hindrance to
programmers. However, this same ability lets the programmer create
intricate functions.
Local
variables
Before going on, we want to discuss a difficult subject with you-local

variables. You already know that a block is a set of functions and
statements which are enclosed in a pair of curly braces. Local variable
definitions usually appear at the beginning of a block. So far in this
book, local variables have been used only in conjunction with functions
in programs. Local variables can be defined at the beginning of a
function. In fact, you can define local variables in any block.
Local variables help clarify the variable structure, provided the variables
are not treated as register variables. Register variables cause problems
because of the limited number of accessible registers (in controllers five
to seven). This can be fixed by placing local variables in blocks: Define
register variables in just the block that requires them, and place all of
509
the necessary instructions in a separate block. This releases the

corresponding registers outside the block, and can be handled by the C
compiler in another way. Since more register variables can be used, the
finished program runs faster. Look at the slow and fast functions
listed below:
slow!)
{
register int
register int
register int
for (al = a,
a,b,c;
al,bl,cl;
a2,b2,c2;
bl ~ b, cl
c; al > 0; al--)
c; a2 > 0; a2--)
for (a2 = a, b2
b, c2
fast ()
{
register int a,b,c;

i
register int al,bl,cl;
for (al = a, bl
b, cl = c; al > 0; al--)
%
register int a2,b2,c2;

for (a2 = a, b2
b, c2
%
c; a2 > 0; a2--)
These serve the same purpose. The difference is in the use and
placement of register variables. Compile each to get the following
assembler source texts:
; slowO
; {
public slow
slow:
link as,#.2
movem.l .3,-(sp)
register int a,b,c;
register int al,bl,cl;
;
; for (al = a, bl
-)
move.w d4,d7
move.w dS,a2
move.w d6,a3
bra .7
.6
.4
sub.w #1,d7
.7
.s
510
tst.w d7
bgt .6
= b,
cl = c; al > 0; al-
ABACUS
4.3
; for (a2
= a,
b2
STEP BY STEP
b, c2 = c; a2 > 0; a2--
move.w d4,-2(aS)
move.w dS,-4(aS)
move.w d6,-6(a5)
bra .11
.10
;
.8
sub.w U,-2(a5)
.11
tst.w -2(a5)
bgt .10
.9
;
;}
.12
.2
.3
movem.l (sp)+,.3
unlk as
rts
equ-6
reg d4/d5/d6/d7/a2/a3
; fast ()
;{
public fast
fast:
link as, 13
movem.l .14,-(sp)
; register Int a,b,c;
;
; {
= a,
for (a1
bl
= b,
c1
= c;
a1 > 0; a1-
= c;
a2 > 0; a2--
-)
move.w d4,d7
move.w dS,a2
move.w d6,a3
bra .18
.17
;
.15
sub.w U,d7
.18
tst.w d7
bgt .17
.16
; {
"
for (a2
= a,
b2
= b,
c2
move.w d4,d7
move.w d5,a2
move.w d6,a3
bra .22
.21
;
.19
511
sub.w #l,d7
.22
tst.w d7
bgt .21
.20
;
;}
.23
movem.l (sp) +, .14
unlk as
rts
.13 equ 0
.14 reg d4/dS/d6/d7/a2/a3
dseg
end
These instructions clearly show which variables are register variables

and which aren't. The fIrst for loop of the slow function has more
obvious registers, but the second for loop writes the register values
into memory. This loop defines too many register variables, which
means that the C compiler must convert some of them into normal
variables. The fast function uses register variables exclusively, which
are all available when the function executes.
The coming functions require local variable defmitions within blocks
for inclusion in the editor source. You may want to experiment with
local variables on your own. They can often help make a program faster
while keeping the source text readable, without resorting to assembler
routines.
4.3.9
Command Line
File access must still be created for our editor to be complete. Since
filenames must be entered, a command line should be included in the
editor, like the ED editor program. In addition, a status line should
exist, which displays the current cursor position and other information.
On to the command line. This should serve a purpose similar to the one
in ED: The command line should be accessible by pressing the <Esc>
key, and should accept two-letter commands. ED'S command set isn't
confIgured in an easy to remember order. For example, look at ED'S
block commands: BS marks the start of a block, but you must enter m
to insert a block, instead of BI.
Our own editor should have commands which have a first letter
indicating the general command grouping. Let's consider which
command groups would be needed by an editor. First priority would be
512
ABACUS
4.3
STEP BY STEP
loading and saving data. We also need commands for cursor control, text
block manipulation, setting mode flags, text deletion, word searching
and exiting the program. Also, commands may be needed if you choose
to extend the editor to a programmable mode. Here are a few commands
we came up with by brainstorming (the ones marked with an asterisk
are implemented in our version of the editor):
ASCII file
commands
AF ["Filename"]
AL ["Filename"]
AS ["Filename"]
saves text within a folded section

loads an ASCII text file
saves the entire text
Block
commands
BC (Block copy)
BD (Block delete)
BE (Block end)
BF (Block fold)
BH (Block hide)
BM (Block move)
BP (Block paste)
BS (Block start)
BX (Block cut)
copies block to buffer

deletes block
marks end of block
folds block
deletes block marker
moves block
inserts block contained in buffer
marks beginning of block
deletes block from text to buffer
Cursor
commands
CB (Cursor bottom)
CD (Cursor down)
CE (Cursor end)
CH (Cursor home)
CL (Cursor left)
CN (Cursor next)
CP (Cursor previous)
CR (Cursor right)
CS (Cursor start)
CT (Cursor top)
CU (Cursor up)
moves cursor to end of text/fold *

moves cursor one line down
*
moves cursor to end of line
*
moves cursor to start of line
*
moves cursor one character left
*
moves cursor to start of next line *
moves cursor, start of previous line *
moves cursor one character right *
*
moves cursor to start of text/fold *
moves cursor one line up
*
Delete
commands
DB (delete back)
DC (delete char)
DL (delete line)
backspace
delete character
delete line
Exchange
commands
EN ["String! "String2"] replacesnextocc~oe

EP ["String! "String2"] replaces previous occurrence
Find
commands
FN ["String"]
FP ["String"]
IF command
IF(condition)assigment
*
*
*
*
*
searches for next occurrence of string

searches backwards
513
4.
Line
commands
LA
AMIGA
"String"
LI "String"
LJ (line join)
LS (line. split)
inserts string as new line behind

current line in text (append)
inserts string as new line before
cment line in text (insert)
adds one line to following line
splits line at cursor position
M [A-Z] [=instructions]
Macro
command
Quit command Q
execute or set macro
Repeat
RI? instruction
command
Set commands Sa
SA
Sb
SB
Si
So
St
ST
repeats instruction until stopped
Tab commands TC (Tab clear)

TS (Tab set)
erases tab at cursor position

sets tab at cursor position
eXit command x
saves changed text and exits
Undo
command
undo changes in current line
exits editor
Auto Indent off

Auto Indent on
SkipBlanks off
SkipBlanks on
insert mode
overwrite mode
tabulators off
tabulators on
*
*
*
*
*
*
*
*
*
The editor commands, with the exception of the Set commands, accept
commands entered in either uppercase or lowercase.
This section lists the development process of the most important editor
features, such as saving and loading text. Some of the features already
exist from the previous versions (e.g., cursor movement).
The first item needed is a command line that works independently from
the normal editing functions. A string gadget placed at the lower left
border of the window serves this purpose. Use of a string gadget means
that you can enter text in the gadget immediately, without clicking on
it to activate it. The string gadget becomes active the moment the user
presses the <Ese> key. The main difference between our command line
and ED'S <Esc> commands is that you can edit the text in the string
gadget.
The editor must execute the commands of the command line as soon as
the user presses the <Return> key. The RELVERIFY flag indicates a
pressed <Return> key to the editor. The editor then receives a
514
ABACUS
4.3
STEP BY STEP
GADGETUP message each time the user presses <Return> while in a

string gadget. It must then determine which command was entered and
execute the appropriate function. Because we have a complete range of
commands, we could have the editor compare the command line with all
known commands. There is no need to do this, since our commands are
sorted according to group, and all commands in one group begin with
the same letter. The editor reads the first character, moves to that
command group, then executes the function called by the second
charaCtel'.
A pointer to the second letter of the command is given in this function

because the first is already known. If the function can execute the
command, this pointer returns to the character after the command,
otherwise it returns an error message. The executeCommand
function calls these commands. This function skips spaces and looks
for semicolons (which allow the execution of multiple commands in
one line).
The command line also allows the user to enter multiple commands
enclosed in curly braces, with a number preceding the first brace. The
editor executes the commands within braces the number of times
indicated by the preceding number. The following command sequence
moves the current line and the nine lines after it three characters to the
left:
CS;10{3DC;CD};10CU
The cursor then returns to the original line. Two items of interest in
this command:
Info line
1.)
The savelfCursorMoved function must be called because

the cursor may have been moved.
2.)
The user must be able to stop a command in process without

losing data. Command execution stops when the user presses any
key.
The info line appears along the bottom border of the window, to the
right of the command line. This line actually contains a minimum
amount of information, to keep the info line small and maximize the
size of the text area:
X and Y position of the cursor

Number of lines in the text
minfold status
maxfold status
Flag status
515
4.
AMIGA
If we limit the maximum width for each number to four digits (two
digits for minfold or maxfold), the maximum is 34 characters. The
info line looks something like this:
X=nnnn Y=nnnn #=nnnn [nn,nnj ICATB
The X and Y indicate the cursor's present coordinates. The # represents

the total number of lines. The first number in brackets display
minfold, and the second number displays maxfold. The five
characters at the end of the info line represent the flags:
I:
C:
A:
T:
B:
insert I O:Overwrite
changed
autoindent.
tabs.
skipblanks.
Lower case characters indicate a disabled flag, while upper case

characters indicate an enabled flag. We set the width of the command
line relative to the window's width. This avoids any overlapping
between the info line and the command line if the user reduces the size
of the window. The minimum width of the window must be set
accordingly-the command line cannot have a negative width if you
make the window too small.
Assembler and Functions in the editor display a value or group of values in the info
line without constantly redisplaying the info line. Because speed is
C code
important for displaying the cursor positio~, this is a good opportunity
to write a subroutine in assembler and append it to our program. To
keep the assembler programming to a minimum, we'll tell the program
to convert a number to a string in assembler programming.
The C library's ftoa function performs this task, but ftoa requires a
floating point number. The C compiler converts the integer number
into a floating point number, then converts the floating point number
to the necessary string. Our function converts an integer number
directly, where we can also determine the length of the character string.
The Aztec C compiler allows easy combination of assembler and C
code. Instead of the function header we first place the =!tasm compiler
instruction at the beginning of a line. Then we enter our assembler
program and end it with an =!tendasm instruction. The compiler now
knows that an assembler program starts after =!tasm and ends before
=ltendasm. The compiler initially bypasses this assembler code. The
assembler, which is called from the C compiler, compiles this code.
The assembler's routine address must be indicated so that the assembled
code may be accessed from C functions. If you want the routine
accessed under the default name cn_utoa (CoNvert Unsigned To
516
ABACUS
4.3
STEP BY STEP
Ascii), you must include the following line at the beginning of the
routine:
global _cn_utoa
_cn_utoa:
The global statement defines the label_cn_utoa. This label can

now be accessed like a global variable or a function. The underscore
character preceding the function name is necessary because the C
compiler precedes all of the labels with an underscore when compiling a
C program into an assembler program. If you have already seen what
the C compiler creates for assembler programs, you have probably
noticed this.
The parameters that are defined first are placed on top of the stack. The
following function call:
cn_utoa(string,value, length);
returns parameters as the following addresses on the stack if value and

length are of type UWORD:
O(SP) A return address
4 (SP) A string
8(SP) = value
10 (SP) = length
We cannot use all of the free registers in our assembler program. The
registers that the C compiler uses for reference to local (AS) and global
(A4) variables may be changed, like the register that uses register
variables (D4-D7, A2, A3). However, there are more than enough
registers for us, especially since we only need four segments for our
subroutine. Here is the complete assembler program, which will be
located in Output. c:
1******************************
*
*
cn_utoa(buf,val,len):
Convert UWORD in ASCII.
buf
val
len
buffer for ASCII.
= Value
= Maximum
Lenght.
******************************/
void cn_ utoa () ;

#asm
cseg
public _cn_utoa
517
cn utoa:
; 4(sp)
; 8(sp)
;10(sp)
lea
move. I
moveq
move.w
move.w
lea
bra.s
4(sp),aO
(aO)+,a1
#O,dO
(aO)+,dO
; = Value
(aO),dl
; = maximum Lenght
0(a1,dl.w),aO ; ~ End of Buffers
.in
.lp:
divu
swap
add.b
move.b
clr.w
swap
Buffer,
Value,
= maximum Lenght.
#10, dO
dO
#'O',dO; dO.w
dO,-(aO)
dO
dO
= VAlue
MOD 10
; Zero-Flag is 1, when dO.1 = 0
.in:
dbeq
dl, .lp
; length is value = 0
; or String full
beq.s
. i b ; Value = 0 => Rest is
; full of spaces
bra.s
.r ; String full!
.bl:
move.b
#' ',-(aO)
.ib:
dbra
.r:
dl, .bl
rts
#endasm
The program fHIs in the string starting from the end. The number to be
converted divides by ten each time, and the remainder from the division
is written in the string as numerals. The rest of the string fills with
blanks, as long as memory is available.
The values must be displayed properly in the info line. We need the
position at which this display must begin. Add to that the position of
the gadget and the position of the Int ui Text structure (which the
info line contains). The Intuitext structure determines the character
set which keeps the info line as small as possible We recommend
Topaz_80, which permits 80 characters per line. We must also
choose this character set for the output of the values. The
P r i n t I Te x t function allows character set assignment in the
Intui Text structure. This function also requires a null byte at the
end of the string. It would be better to use the text function from the
graphics library, because you can enter the length of the character
string. Consider that the value must be copied to the info line string,
because the border is also redisplayed when the window is redisplayed. If
we display the new value over the old value using PrintIText or
518
ABACUS
4.3
STEP BY STEP
Text in the window, the new output appears the same as the old
because it is always found in the string.
We want to copy or convert the new value (xpos, ypos, etc.) to the
info line, then use Text to display the corresponding section of the
info line. For this we must first set the character set with SetFont.
Then OpenFont must open the character set for access by the main
program. The value display functions look like the following (the
assembler program _cn_utoa should precede these functions in the
Output. c module!):
I*#FOLD: printXpos *1
1*****************************
** printXpos:
** Give Xpos of cursor.
******************************/
void printXpos ()
{
register struct TextFont *oldFont;

register struct Editor *ae = actualEditor;
cn utoa(ae->gi Text + INFO XPOS INC,ae>xpos, (UWORD)INFO_XPOS_LEN);oldFont = ae->rp->Font;
SetFont(ae->rp,infoFont);
Move(ae->rp,ae->ip xpos,ae->ip topedge);
Text(ae->rp,ae->gi-Text + INFO-XPOS INC, (ULONG) INFO XPOS LEN);
SetFont (ae->rp, olctFont);
}
I*#ENDFD*I
I*#FOLD: printYpos *1
1*****************************
* printYpos:
* Give Ypos of cursor.
*****************************/
void printYpos ()
{

cn utoa(ae->gi Text + INFO YPOS INC,ae>ypos, (UWORD)INFO_YPOS_LEN);oldFont = ae->rp->Font;
Move(ae->rp,ae->ip ypos,ae->ip topedge);
Text(ae->rp,ae->gi-Text + INFO-YPOS INC, (ULONG) INFO YPOS LEN);
SetFont (ae->rp, olctFont);
}
/*#ENDFD*/
519
4.
AMIGA
C FOR
/*#FOLD: printNumLines */
/*****************************
* printNumLines:
*
*
*
*
Give number of lines

of text
*****************************1
void printNumLines()
{

en utoa(ae->gi Text + INFO NUMOFLINES INC,ae->num lines,
(UWORDlINFO_NUMOFLINES_LEN);
oldFont = ae->rp->Font;
Move (ae->rp,ae->ip numzlines,ae->ip topedge);
Text (ae_>rp,ae_>gi_Text + INFO NUMOFLINES INC,
(ULONG)INFO NUMOFLINES LEN);
SetFont(ae->rp,oldFont); }
l*itENDFD*1
l*itFOLD: printFold */
1********************************
* print Fold:
**
Gives minfold and max fold display.
********************************1
void printFold ()
{
register struet TextFont *oldFont;

register struct Editor *ae = aetualEditor;
en_utoa(ae->gi_Text + INFO_MINFOLD_INC,ae->minfold,
(UWORD) INFO MINFOLD LEN);
en_utoa(ae->gi_Text + INFO_MAXFOLD_INC,ae->maxfold,
(UWORD)INFO_MAXFOLD_LEN);
Move (ae->rp,ae->ip minfold,ae->ip topedge);
Text (ae->rp,ae->gi=Text +
INFO MINFOLD INC, (ULONG) INFO MINFOLD LEN);
Move (ae->rp,ae->ip maxfold:ae->ip topedge);
Text (ae->rp,ae->gi-Text +
INFO MAXFOLD INC, (ULONG) INFO MAXFOLD LEN);
SetFont (ae::>rp,oldFont); }
/*#ENDFD*I
/*#FOLD: printFlags *1
1*****************************
**
*
printFlags:
* At Ypos of cursor.
520
4.3
ABACUS
STEP BY ST E P
*****************************1
void printFlags()
{

register struet Editor *ae = aetualEditor;
register UBYTE *fs = ae->gi_Text + INFO_FLAGS_INC;
if (ae->insert)
*fs++ = 'I';
else
*fs++ = '0';
if (ae->ehanged)
*fs++ = 'C';
else
*fs++ = 'e';
if (ae->autoindent)
*fs++ = 'A';
else
*fs++ = 'a';
i f (ae->tabs)
*fs++ = 'T';
else
*fs++= 't';
if (ae->skipblanks)
*fs'" 'B';
else
*fs = 'b';
Move(ae->rp,ae->ip flags,ae"':>ip topedge);
Text(ae->rp,ae->gi-Text +
INFO FLAGS INC, (ULONG)INFO FLAGS LEN);
SetFont(ae->rp,oldFont);}
I*#ENDFD*I
I*#FOLD: print Info *1
1****************************
**
print Info:
* Gives new
Infoline.
****************************1
void printlnfo()
{
printXpos () ;
printYpos () ;
printNumLines();
printFold () ;
printFlags () ;
}
I*#ENDFD*I
All of these functions operate in the same manner: First the new value
is copied into the string, the character set is changed and the changed
section of the info line is redisplayed. The normal character set returns.
The defines compute the positions of the individual values, and
521
form a new pair of elements in the Ed ito r structure. The

initWindowSize function initializes these new elements, which
contain the pixel positions for all of the values. The new defines are
defined in editor. h, SO we must declare the new SetFont function
in the Output module:
void SetFont () :
We also need a pointer to the character set used for the info line, and
which is defined and enabled in the main module:
extern struct TextFont *infoFont;
The positions of each value in the info line must be calculated in the
ini tWindowSi ze function. This happens before restoration of the
linesptr array:
/* Position for Infoline: */
register ULONG xinc,xsize;
xinc = w->Width + ed->G Info.LeftEdge + ed->gi IText.LeftEdge
- 1;
xsize= infoFont->tf_XSize;
ed->ip xpos = xinc + INFO XPOS INC * xsize:
ed->ip~os = xinc + INFO=YPOS=INC * xsize;
ed->ip nurnzlines = xinc + INFO NUMOFLINES INC * xsize;
ed->ip-rninfold = xinc + INFO MINFOLD INC
xsize;
ed->ip-maxfold = xinc + INFO-MAXFOLD-INC * xsize;
ed->ip-flags = xinc + INFO FLAGS INC * xsize;
ed->ip-topedge = w->Height
ed->G Info.TopEdge
+ ed->gi IText.TopEdge
+ (ULONGlinfoFont->tf_Baseline - 1;
This concludes the changes made to the Output. c module. The

Edi tor. c module defines the gadgets needed for the info and
command lines. Now let's look at the source text for the Command. c
module:
<src/command.c>
/************************************
Module: Command
*
* Command-Interpreter for
Editor.
************************************/
/*#FOLD: Includes
/************
**
*
522
Includes:
*/
4.3
ABACUS
STEP BY ST E P
************/
#include <stdio.h>
/*#ENDFD*/
/*#FOLD: Defines */
/***********
** Defines:
************/
#define TAB 9
#define LF 10
#define CR 13
#define UC(c) (c >= 'a') && (c <= 'z'? (c & OxDF): c)
#define DIGIT(c) c >= '0') && (c <= '9'
#define SKIPBLANKS(str) while (*str == , ') str++;
/*#ENDFD*/
/*#FOLD: External Functions */
/**********************
**********************/
IDOL
cursorLeft(),cursorRight(),cursorUp(),cursorDown(),cursorHome();
BOOL cursorEnd(),deleteChar(),deleteLine(),
backspaceChar(),insertLine();
BOOL saveLine();
void printAll(),printFlags(),fclose(),Insert(),
strncpy(),printNumLines();
void restoreZlinenptr(),SetWindowTitles(),
initFolding(),printYpos();
void DisplayBeep(),saveIfCursorMoved(),CopyMem();
struct Zline *newZline(),*prevLine(),*nextLine();
FILE *fopen ();
int fwrite(),getc();
UBYTE *getLastWord();
ULONG fseek ();
UWORD recalcTopOfWindow();
/*#ENDFD*I
I*#FOLD: External Variables *1
/*********************
*********************/

extern struct MsgPort *edUserPort;
I*#ENDFD*/
/*#FOLD: Globale Variables */
/*********************
**
Globale Variables:
*********************/
523
4.
AMIGA
struct jmpEntry
{
UBYTE c;
UBYTE * (*fkt) ();
};
I*#ENDFD*I
1***************
*
*
Functions:
*
****************1
I*#FOLD: saveASCII *1
1*********************************
* saveASCII(name)
**
Save text under name to

* Diskette
**
name
File name.
*********************************1
BOOL saveASCII(name)
UBYTE
*name;
{
register FILE *file;

if
if
(actuaIEditor->bufypos)
(! saveLine (actuaIEditor->actual
DisplayBeep(NULL);
return (FALSE);
(file
if
{
fopen (name, "w"
z = actuaIEditor->zlines.head;
while (z->succ)
{
i f (z->len)
if (fwrite(z + 1, (int) z->len,l, file) != 1)

{
1* Error! *1
fclose (file);
return (FALSE);
z
= z->succ;
actuaIEditor->changed
printFlags () ;
fclose (file);
return (TRUE);
else
return (FALSE);
524
= 0;
ABACUS
4.3
STEP BY STEP
I*#ENDFD*I
I*#FOLD: loadASCII *1
1***********************************
* loadASCII(name)
* Load text from disk. The text

* is inserted after the actual line
* in the text.
**
name
File name.
***********************************1
BOOL loadASCII(name)
UBYTE
*name;
{
UBYTE buf[MAXWIDTH];
register UBYTE *ptr,*tab;
int e;
UWORD rm = aetuaIEditor->rm;
register FILE *file;
register UWORD len;
register struet Zline *z,*zn;
if (file
fopen(name,"r"
1* display name in title line: *1

strnepy(aetuaIEditor->filename,name,
sizeof(aetuaIEditor->filename;
SetWindowTitles(aetuaIEditor->window,aetuaIEditor>filename,-lL);
= ZLINESPTR(aetuaIEditor->wdy
(z
if
{
aetuaIEditor->ehanged
printFlags 0 ;
while (!feof(file
{
tab
ptr
len
= aetuaIEditor->tabstring;
= buf;
= 0;
1* read one line: *1

while (!feof(file) && (len < rm
{
if e = gete(file
break;
else
*ptr++ = (UBYTE)e;
if e
==
EOF)
== TAB) && (aetuaIEditor->tabs
cb
{
tab++;
len++;
while *tab) && (len < rm;
else
l
525
4.
AMIGA
tab++;
len++;
i f (c == LF)
break;
else if (c == CR)
if (!feof(file) && (len < rm
{
if *ptr = (UBYTE)getc(file == LF)

ptr++;
else
/* set file pointer back: */
fseek(file,-lL,1);
break;
}
/* of if (TAB) *1
1* of while */
if (len = ptr - buf)
{
1* Wordwrap? *1
if (!actualEditor->skipblanks)
if len >= rm) && (c != CR) && (c != LF
{
tab = getLastWord(buf,len);
fseek(file, (long) tab - ptr,1);
len = (ptr = tab) - buf;
/* save line now: *1
actualEditor->1ineptr =
if (zn = newZline(1en
Z;
1* Garbage-Collection
z = actualEditor->lineptr;
zn->len = len;
CopyMem(buf,zn + 1, (ULONG)len);
Insert (&actualEditor->zlines,zn,z);
z = zn;
actualEditor->lineptr = NULL;
else
{
fclose (file);
return (FALSE);
1* calculate fold level: *1

ini tFolding () ;
/* redisplay window: */
if ZLINESPTR(O) == NULL) && (actualEditor->zlines.head>succ) )
{
ZLINESPTR(O)
ZLINESNR(O)
= actualEditor->zlines.head;
=
1;
restoreZlinenptr();
526
*1
4.3
ABACUS
STEP BY STEP
printAll () ;
printNumLines();
fclose (fil~);
return (TRUE);
else
return (FALSE);
)
/*IIENDFD*/
/*************************************************
* command-Functions:
**
This processes a function group
*
*
*
*
All functions return a pointer to

the next command, if no error
is encountered, otherwise a negative pointer
to the faulty character is returned.
**
Returna a NULL, so this

* serves as break criteria.
* A -1 ends the program.
*************************************************/
/*IIFOLD: commanciASCII */
UBYTE *commanciASCII(str)
register UBYTE
*str;
{
UBYTE *error lL - str,buf[80);

register UBYTE *name;
switch (UC(*str))
{
case 'L':
str++;
SKIPBLANKS (str)
if (*str = ,n,)
{
1* take file name from command line */

str++;
name = buf;
while *name = *str) && (*str != "")
&& (name < buf + sizeof(buf) - 1))
name++;
str++;
*name = 0;
if (*str
str++;
"")
name = buf;
else
1* use file name form load: */

name = actuaIEditor->filename;
527
4.
AMIGA
if (!loadA5CII(name
return (error);
break;
case '5':
str++;
SKIPBlANK5 (str)
if (*str = '"')
{
1* get file name from command line *1

str++;
name = buf;
while *name = *str) && (*str != .... )
&& (name < buf + sizeof(buf) - 1
name++;
str++;
*name = 0;
if (*str =
str++;
'"')
name = buf;
else
1* use file name from load: *1

name = actualEditor->filename;
if (!saveA5CII(name
return (error):
break:
default:
return (NULL - str);
return (str);
}
I*#ENDFD*I
I*#FOLD: commandCursor *1
UBYTE *commandCursor(str)
register UBYTE
*str;
{
switch (UC(*str
{
case 'R':
if (!cursorRight(
return (NULL):
break;
case 'L':
if (!cursorLeft(
return (NULL);
break;
case 'U':
i f (! cursorUp 0 )
return (NULL);
break;
case 'D':
i f (! cursor Down () )
528
4.3
ABACUS
STEP BY STEP
return (NULL);
break;
case 'E':
i f (!cursorEnd(
return (NULL);
break;
case '5':
case 'H':
i f (!cursorHome(
return (NULL);
break;
case 'N':
i f (!cursorDown(
return (NULL);
i f (!cursorHome(
return (NULL);
break;
case 'P':
i f (! cursorUp () )
return (NULL);
i f (! cursorHome () )
return (NULL);
break;
case 'T':
i f (!cursorTop(
return (NULL);
break;
case 'B':
i f (!cursorBottomO)
return (NULL);
break;
default:
return (str + 1);

}
I*#ENDFD*I
I*#FOLD: commandDelete *1
UBYTE *commandDelete(str)
register UBYTE
*str;
{
switch (UC(*str
{
case 'L':
i f (! deleteLine(
return (NULL);
break;
case 'e':
i f (! deleteChar(
return (NULL);
break;
case 'B':
i f (! backspaceChar(
return (NULL);
break;
default:
529
return (str + 1);

}
/*#ENDFD*/
/*#FOLD: commandLine *1
UBYTE *commandLine(str)
register UBYTE
*str;
{
switch (UC(*str
{
case '5':
i f (! insertLine UBYTE) 0
return (1 - str);
break;
default:
return (str + 1);
}
I*#ENDFD*I
/*#FOLD: commandQuit *1
UBYTE *commandQuit(str)
register UBYTE
*str;
{
return (-1L);
}
I*#ENDFD*I
/*#FOLD: commandSet *1
UBYTE *commandSet(str)
register UBYTE
*str;
{
switch (*str)
{
case 't':
actualEditor->tabs = 0;
printAll () ;
printFlags();
break;
case 'T':
actualEditor->tabs = 1;
printAll () ;
printFlags () ;
break;
case 'i':
case 'I':
actualEditor->insert = 1;
printFlags();
break;
case '0':
case '0':
actualEditor->insert = 0;
printFlags () ;
break;
case 'b':
actualEditor->skipblanks= 0;
printFlags () ;
break;
case 'B':
actualEditor->skipblanks= 1;
printFlags () ;
break;
530
4.3
ABACUS
STEP BY ST E P
case 'a':
actualEditor->autoindent= 0;
printFlags () ;
break;
case 'A':
actualEditor->autoindent= 1;
printFlags () ;
break;
case 'r':
case 'R':
{
register UWORD rm;

if (*++str
==
'=')
str++;
(DIGIT (* str
if
{
rm = 0;
while (DIGIT(*str
{
rm = 10*rm + (*str - '0');

str++;
i f (rm > MAXWIDTH)
rm = MAXWIDTH;
else if (rm < 10)
rm = 10;
)
else
1* Fehler *1
else
rm
= MAXWIDTH;
actualEditor->rm = rm;
--str;
break;
default:
return (str + 1);
}
I*#ENDFD*I
I*#FOLD: commandExit *1
UBYTE *commandExit(str)
register UBYTE
*str;
{
if (actualEditor->changed)
if (saveASCII(actualEditor->filename
return (-lL);
else

else
return (-lL);
}
I*#ENDFD*I
531
/*#FOLD: commandBracket */
UBYTE
*executeCommand();
1* declaration due to recursion!
UBYTE
*commandBracket(str)
register UBYTE
*str;
{
register
UBYTE *ptr;
(ptr = executeCommand(str
if
{
if
(ptr!= -IL)
(ptr >= Ox80000000)
/* Pointer behind command line '}':
ptr =
11 - ptr;
else
/* pointer to error: */
ptr = NULL - ptr;
if
*1
else
{
1* search for ending '}': */

ptr =
str;
while (*ptr && (*ptr!= , ) , ) )
{
(*ptr
if
lIl1)
d:>
{
ptr++;
while (*ptr
&&
(*ptr != "";
ptr++;
(*ptr =
0)
ptr = NULL;
else
ptr++;
if
return (ptr);
}
I*IIENDFD*/
I*IIFOLD: messageWaiting *1
/*********************************
* messageWaiting:
* Test if there is a message

* in the UserPort.
* Returns TRUE if yes.
*********************************1
BOOL messageWaiting()
{
register struct IntuiMessage *im;

im = (struct IntuiMessage *)edUserPort->mp MsgList.Ih Head;
while (im->ExecMessage.mn Node.ln SuccI
{
-
532
ABACUS
4.3
STEP BY STEP
if im->Class == RAWKEY)
&& ! (im->Code & IECODE UP PREFIX)) return (TRUE);
if (im->Class == NEWSIZE)-return (TRUE);
im = (struct IntuiMessage *)im->ExecMessage.mn_Node.ln_Succ;
return (FALSE);
}
/*#ENDFD*/
/*#FOLD: Functions table */
struct jmpEntry jmpTable[)
{
'A',&commandASCII,
'C',&commandCursor,
'D',&commandDelete,
'L',&commandLine,
'Q',&commandQuit,
'S',&commandSet,
'X',&commandExit,
'{',&commandBracket
};
/*#ENDFD*/
/*#FOLD: executeCommand */
/****************************************
* executeCommand(str)
* Executes the commands to which str,

* points. The end is marked by
* a Null byte.
* The function returns the following:

* 0: no error
* -1:
Program end
* else: pointer to error in string.
* or:
negative pointer to '}'
*****************************************/
UBYTE *executeCommand(str)
register UBYTE
*str;
{
register UBYTE *ptr,c;

register UWORD count,n;
register UBYTE * (*fkt) ();
while (!messageWaiting() && *str)
{
SKIPBIANKS(str)
if
(*str =
'}')
return (NULL - str) ;
i f (DIGIT (*str)
{
count = 0;
while (DIGIT(*str))
{
count = lO*count + (*str - '0');

str++;
533
if
(count =
count = 1;
else
count
0)
= 1;
1* search function, the command executes: */

c = UC(*str);
fkt = NULL;
for (n = 0; n < sizeof(jmpTable)/sizeof(struct jmpEntry);
nt+)
if (c
~=
jmpTable[n].c)
fkt = jrnpTable [n] fkt;

break;
i f (fkt)
while (!rnessageWaiting() && count--)

{
if (ptr
(*fkt) (str +
if (ptr >= Ox80000000)

i f (ptr = -lL)
1* Program end *1
return (ptr);
else
1* Error encountered: *1
return (NULL - ptr);
else
1* If count : z 0: Next command. */
1* else repeat *1
if (count ~ 0)
str = ptr;
else
1* break *1
return (NULL);
else
1* Unknown command *1
return (str);
savelfCursorMoved();
SKIPBlANKS(str)
if (*str * = ' ; ' )
str++;
else if (*str ~
return (NULL else if (*str)
return (str);
return (NULL);
}
I*#ENDFD*I
534
, } ')
str);
ABACUS
4.3
STEP BY ST E P
A few remarks about the new module:
The include file stdio. h must be added to the modules, since

file processing requires the standard C file functions fopen,
fclose, etc.
The defines contain useful macros which convert a character

to upper case (uC=upper case), to differentiate when a character is
handled as a numeral (DIGIT), and bypass blank spaces in a
string (SKIP BLANKS).
Because of the design of the command set, we don't need an

elaborate switch case function to differentiate between
commands. The fIrst letter of the command and the address of the
corresponding function are placed in an array. The data structure
of the array is jmpEntry. The array is defined after the
command functions because the functions must already be defIned
when the fIeld is initialized (the C compiler must be able to
recognize the function addresses).
The saveASCII function expects a pointer to a filename as a

parameter, and saves the entire text under this name to disk. The
function returns FALSE if an error occurs.
The same data for saveASCII applies to the loadASCII

function, although it is more complicated because text must be
separated into individual lines. The text to be loaded is inserted
after the current line of the existing text. If no text exists, the
text loads as usual. The variable z points to the previous line.
When loading, individual characters are read until the program reaches
an end of line character, or when the line is too wide. Because the
function getc (which reads the characters) works internally with a
buffer, the method is no slower than if we loaded an entire text block
into a buffer and removed the characters from there. The maximum
number of characters per line that can be loaded is a changeable
parameter-the new variable rm in the Editor structure serves this
purpose. rm displays the right margin.
A set skipblanks flag breaks lines if a line is too long to fIt on the
screen. This break shifts the word causing the problem to the next line.
SkipBlanks is also known as word wrap. The getLastWord
function is implemented in the Edit module. It returns a pointer to the
beginning of the last word of a string.
The editor had an error in it since its fIrst stages of development. The
garbageCollection function in the Memory module reorganizes
a memory block. All of the lines can be moved so that this function
535
4.
AMIGA
checks the actual variable and the zlinesptr array, and moves a
pointer. A pointer to the last line borrows data from the loadASCII
function, which must also be checked. The Editor structure uses this
zlinesptr array. The garbageCollection function checks this
pointer. When we have a local variable that points to a line, we save
this in zlineptr before calling the newLine function, which calls
the garbageCollection function. After the call we access
zl inept r again, and can be certain that this pointer points to the
same line. The error occurs in the saveLine function, when the
newLine function is called while a local variable contains a pointer to
a line. We will correct this error.
After all of the lines load, ini tFolding recalculates the fold levels
of all of the lines, and the window contents are redisplayed.
The functions that execute individual commands are called

comrnandXYZ, where XYZ represents the command group. A
pointer to the second letter of the command is given in this
function. A pointer is returned which usually points to the
location following the command. However, special cases can
arise: When the function returns zero, this serves as the break
criteria for the command interpreter. A value of -1 exits the editor
(Le., X and Q). When the program otherwise returns a negative
value, an error occurs. You get a pointer to the error when you
multiply the return value by -lor when you pass it zero.
The currently implemented commands consist of corresponding

function calls, which include other modules. The ASCII
commands are somewhat more extensive because a character
string must be handled. The s r=nn function in the Set
commands sets the flags. You can set the right margin for word
wrap from this function. The value can be a value from 10 to
MAXWIDTH.
536
Bracketed commands call the comrnandBracket function,

which recursively calls the executeComrnand function. The
executeComrnand function returns if an error occurs, or if
brackets appear in the command line. In the latter case, execution
continues following the closing bracket (}). Problems can occur
if the user stops the bracketed command during execution. In this
case command execution should continue after the closing
bracket (}). The comrnandBracket function searches for a
closing bracket (}) and returns a pointer to the following
characters. The user can stop the command without problem by
pressing a key, because this break criteria stays the same.
4.3
ABACUS
STEP BY ST E P
The messageWaiting function checks for a user-invoked

command break. This function executes either if a key is pressed
Shift>, <Ctrl> or <Alt>, because these only send RAWKEY),
or if the window size changes. In the latter case, the output
becomes incorrect, because some of the values calculated for
output in the ini tWindowSi ze function probably changed.
The executeCommand function controls the execution of each

command. It expects a pointer to the command sequence as a
parameter. It returns zero if no error occurs (similar to the
command functions), -Ion program end, or a pointer to the
error. If the returned value is negative, it treats the negative
number as a pointer to a closing bracket (}). When the
commandBracket function calls the executeCommand
function, it handles the number as the end of a bracketed
command sequence. When the function is called from the main
program, it handles the number as an error because no opening
bracket ({) occurred in the command sequence.
Because pointers in C are handled as values without prefixes, we can't

test any pointers less than zero. Instead we must test the pointer for a
value greater than Ox800000000 (currently the smallest negative
number available).
Fixing the bug Next we will correct the error mentioned above, which lies in the
Me m0 r y . c module. Insert the following lines in the
garbageCollection function, so that the actualEditor>actual pointer makes the proper test:
if
(actualEditor->lineptr:: z)
actualEditor->lineptr
fz;
Now add the following includes to the Edi tor. h me:

#include <stdio.h>
#include <intuition/intuitionbase.h>
The positions and lengths of each value of the info line must be stated
as defines:
#define
#define
#define
#define
#define
#define
#define
#define
#define
#define
#define
#define
INFO XPOS INC 2

INFO XPOS LEN 4
INFO-YPOS-INC 9
INFO-YPOS-LEN 4
INFO-NUMOFLINES INC 16
INFO-NUMOFLINES-LEN 4
INFO-MINFOLO INC 22
INFO-MINFOLO-LEN 2
INFO-MAXFOLO-INC 25
INFO-MAXFOLD-LEN 2
INFO-FLAGS INC 29
INFO=FLAGS=LEN 5
537
The INC defines state the number of characters in the value

removed from the beginning of the info line, and the _LEN defines
state the number of characters in the value. The following elements
extend the Editor structure:
UBYTE gc_SIBuffer[256];
struct StringInfo gc_GadgetSI;
struct Gadget G Command;
UBYTE gi Text[361;
struct IntuiText gi IText;
struct Gadget G Info;
ULONG ip xpos; ULONG ip- ypos;
ULONG ip-nurnzlines;
ULONG ip=minfold;
ULONG ip maxfold;
ULONG ip=flags;
ULONG ip topedge;
UBYTE filename [80];
UWORD rm;
struct Zline *lineptr;

};
The first six elements contain the gadget structures for the command
and info lines. Next follow the positions of the individual values of the
info line, the current filename initialized from the loadASCII
function, the right margin and the help pointer to a line viewed by the
garbageCollection function.
We now tum our attention to the main module Edi tor. c, which
requires the addition of another include file:
#include <intuition/intuitioribase.h>
Two new defines follow, they state the argument template returned
when the user enters Editor ? from the eLI:
#define UT1 "USAGE: Editot [Flags] <Filename>"
#define UT2 "Flags: [-aJ[-AJ[-b][-BJ[- iIIJ[- 010][- rIR[=nll [-
t1 [-T1"
Since the editor now has built in text loading capability. you can start
the editor from the eLI by entering the editor name. a space and the
filename:
editor myfile
You can also enable and disable flags from the eLI as the editor loads.
For example, instead of loading the editor, pressing <Esc> and entering
SA to enable auto indent, the following loads the editor and sets the
auto indent flag:
editor -a
538
4.3
ABACUS
STEP BY STEP
Let's look at the other external functions:

void strncpy(),CloseFont(),puts();
void printXpos() ,printYpos() ,printlnfo() ,DisplayBeep();
struct TextFont *OpenFont();
BOOL handleKeys(),ActivateGadget(),loadASCII();
UBYTE *executeCommand() ,*commandSet ();
Remember that the handleKeys function is now of type BOOL. You

remember the reason for this when we changed the Cursor. c module.
We defined two gadgets with global variables. These gadgets control the
command and info lines. The G_command string gadget represents the
command line, while the G_Info boolean gadget represents the info
line:
UBYTE gc SIBuffer(256);
UBYTE gc=undoBuffer(256);
struct Stringlnfo gc GadgetSI =
gc_ SIBuffer,
gc_UndoBuffer,
0,
256,
0,
0,0,0,0,0,
0,
NULL,
NULL
);
SHORT gc_BorderVectors[4)
struct Border gc Border =
{0,0,350,0);
-1,-1,
1,0,JAMl,
2,
gC_BorderVectors,
NULL
);
struct Gadget G_Command =

(
NULL,
2,-9,
-302,9,
GADGHCOMP I GRELBOTTOM I GRELWIDTH,
RELVERIFY I BOTTOMBORDER,
STRGADGET,
(APTR) &gc_Border,
NULL,
NULL,
0,
(APTR)&gc_GadgetSI,
2,
NULL
);
SHORT gi BorderVectors(14) = {-1,0,1,10,0,lO,0,0,298,0,298,1,283,l};
539
4.
struct Border gi_Border
AMIGA
0,0,
1,0,JAMl.
7,
gi_BorderVectors,
NULL
}i
struct TextAttr TOPAZ80

TOPAZ EIGHTY,O,O
};
UBYTE gi Text[36] = "X= 1 Y=

struct IntuiText gi_IText =
1 i=
0 [ 0. 0] lcATB";
1,0,JAM2,
4,2,
&TOPAZ80.
gi_Text.
NULL
};
struct Gadget G Info
&G Corrrnand.
-i98,-10,
280,10,
GADGHNONE I GRELBOTTOM I GRELRIGHT,
BOOLGADGET,
(APTR) &gi Border,
NULL,
&gi IText.
0, -
NULL,
1,
NULL
};
0,0,640,200,
REFRESHWINDOW I MOUSEBUTTONS t RAWKEY I CLOSEWINDOW I NEWSIZE
GADGETUP,
WINDOWSIZING I WINDOWDRAG I WINDOWDEPTH r WINDOWCLOSE I
SIZEBBOTTOM
NULL, NULL.
NULL,
NULL, NULL,
320,50,640,200,
WBENCHSCREEN
);
struct TextFont *infoFont
540
NULL;
4.3
ABACUS
STEP BY ST E P
The window allows GADGE TUP events, so the program receives a

report when the user presses the <Return> key in the string gadget, and
thus executes a command sequence. Since we want to be able to open
more editor windows later, we must copy sections of this gadget to the
Edi tor structure, because no two windows can use the same gadgets.
We don't need to copy all of the structures: A border structure; the
command line buffer (S I Bu f fer) is needed so we can enter two
different command sequences; the Stringlnfo structure; the info line
text; the Intuitext structure, which points to this; and the two
gadget structures, for two windows.
The gadgets and other structures are copied into the OpenEdi tor
function in the Edi tor structure. This happens directly after the
memory allocation for the Editor structure, because the gadgets must
be joined to the NewWindow structure so that they also appear in the
window. In addition, remember to set the structure pointers to one
another, because this no longer makes sense after copying them into the
Edi tor structure. The following OpenEdi tor function avoids
initializing the rest of the parameters since they are the same:
struct Editor *OpenEditor ()
{

ULONG flags;

{
1* Gadgets initialization: *1
ed->gc SIBuffer[O] = 0;
ed->gc=GadgetSI
= gc_GadgetSI;
= G Command;
ed->G Command
ed->gI IText
= g1 IText;
ed->G Info
= G Info;
strncpy(ed->gi_Text,gi_Text,sizeof(ed->gi_Text);
1* set pointer: *1
ed->gc GadgetSI.Buffer
= ed->gc_SIBuffer;
ed->G Command.SpecialInfo = (APTR) &(ed->gc GadgetSI):
ed->g1 IText.IText
= ed->gi Text;
= &(ed->G Command);
ed->G Info.NextGadget
ed->G-In fo. GadgetText
= &(ed->gI IText);
newEdWindow.FirstGadget
= &(ed->G Info);
newEdWindow.Title
= ed->filename;
ed->filename[O]
'"' 0;
1* Window open: *1
{
541
4.
AMIGA
ed->window = wd;
/* Write mode set: */
SetDrMd(rp,JAM2);
SetAPen(rp,FGPEN);
SetBPen(rp,BGPEN);
NewList(&(ed->block));
NewList (& (ed->zlines));
ed->num lines = 0;
ed->actual
= NULL;
ed->buflen
= 0;
ed->bufypos
= 0;
ed->buflastchar= , ';
ed->leftpos
= 0;
ed->xpos
= 1;
ed->ypos
= 1;
ed->wdy
= 0;
ed->changed
= 0;
ed->insert
= 1;
ed->tabs
= 1;
ed->skipblanks = 1;
ed->autoindent = 1;
ed->minfold
= 0;
ed->maxfold
= 0;
ed->rm
= MAXWIDTH;
/* zlinesptr/nr-Array initialization: */
for (n = 0, zptr = ed->zlinesptr, pnr = ed->zlinesnr;
n < MAXHEIGHT; n++, zptr++, pnr++)
*zptr = NULL;
*pnr = 1;
/* Tab initialization: */
*ptr++ = 1;
i f (n % 3)
*ptr++ = 1;
else
*ptr++ = 0;
register struct Editor *oldae;

ed->wch = 0;
initWindowSize(ed);
542
ABACUS
4.3
STEP BY STEP
oldae
= actualEditor;
actualEditor = ed;
printInfo () ;
actualEditor = oldae;
else
{
free(ed);
ed = NULL;
newEdWindow.IDCMPFlags = flags;
return (ed);
The info line is displayed at the end of OpenEdi tor so that this
agrees with the additional values. The actualEditor variable turns
towards the re-opened editor because the output functions assume that
act ualEdi tor points to the info line of the current editor.
Our main program requires some major additions. The main function
expects the argc and argv arguments, which encompass all of the
arguments given to the program from CLI access. At the beginning of
the program, before the libraries open, the program tests to see whether
the editor was called with Edit 0 r ? If so, the program displays the
argument template and returns to the CLI. The puts function displays
a string without the format specifications and escape sequences of the
printf function.
After the libraries are open the character set for the info line is accessed,
and a pointer to it is placed in the infoFont variable. Before the
editor is opened, the program determines the maximum height of the
screen and enters this in the Newwindow structure, giving the window
the maximum size when it is opened. The ActiveScreen variable of
the IntuitionBase structure acts as a pointerto the active screen.
Mter the editor window opens, the program reads the arguments entered
from the CLI (if any) and executes these arguments through the
command line. The order of the arguments must follow the template.
Otherwise the program ends and an error message appears. The one error
that loads the editor anyway occurs when the editor tries to find the
filename stated from the CLI and fails.
The main loop stays the same as before. However, the handleKeys
function returns a value of type BOOL. If this value is FALSE, the
program stops. MOUSEBUTTON and NEWSIZE messages display the
cursor position with printXpos and printYpos, if these are
changed.
543
4.
AMIGA
The GADGETUP message check is new. This began execution of the

command line. If an error occurs, the cursor of the gadget moves to the
location of the error and the string gadget becomes active. Because the
GADGETUP check does not know which gadget caused this message,
you can access the command line by clicking on the command line with
the mouse. This allows the user to execute a command sequence that
has already been entered.
The end of the program has also changed. All of the variables that were
tested are set to zero. The reason for this can be found in a utility
program named SHELL. This program is an improvement to the CLI
that allows the definition of commands as resident (the commands are
retained in memory, regardless of how often they are executed). The
advantage here is that the command does not have to be loaded first. In
addition, the resident command requires less memory than if you copied
the command to the RAM disk as a disk file. If the command was
copied to the RAM disk, it would have to be loaded before it could
execute. If it is defined as resident, it appears in memory once, and only
becomes active when the user calls it.
Commands or programs to be defined as resident must fulfill certain
criteria. Never assume that variables are initialized because their
contents can be derived from an earlier call. To get around this potential
problem, all important variable contents return to their starting values
when the program ends. This ensures that the editor finds initialized
starting values every time the user executes the editor. Now on to the
new main function:
main(argc,argv)
int argc;
UBYTE
*argv [1 ;
{
register struct Editor *ed:

BOOL running = TRUE;
register struct IntuiMessage *imsg:
ULONG signal, class, mouseX, mouseY;
UWORD code, qualifier:
APTR iaddress;
1*
if
Was Editor started with "Editor ?" , the Text output *1

(argc =
2)
if
(*argv[l] = '?')
{
puts (UT1);
puts (UT2):
goto Ende;
1* Libraries open: *1
i f ( ! (IntuitionBase = OpenLibrary("intuition.library",REV)
goto Ende;
i f ( ! (GfxBase = OpenLibrary ("graphics library" , REV) ) )
goto Ende;
544
ABACUS
4.3
STEP BY STEP
i f ( ! (DosBase = OpenLibrary ("dos. library", REV)
goto Ende;
if ( ! (OpenDevice("console.device",-lL,&ioStdReq,OL)
ConsoleDevice = ioStdReq.io Device;
else
goto Ende;
1* Font for Infoline open: *1

if (! (infoFont = OpenFont(&TOPAZ80) goto Ende;
1* UserPort open *1
if (! (edUserPort = CreatePort(NULL,OL) goto Ende;
~
Max height for window calculated: *1
newEdWindow.Height = newEdWindow.MaxHeight =
struct IntuitionBase *)IntuitionBase)->ActiveScreen>Height;
1* Editor list initialization: *1

NewList(&editorList);
1* Open first editor window: *1
i f (ed = OpenEdi tor 0
AddTail(&editorList,ed);
actualEditor = ed;
}
else
goto Ende;
1* Command line reading: *1

i f (argc >= 2)
{
while
< argc)
(n
(*argv[n] =
if
._.)
(commandSet(argv[n] + 1) >= Ox80000000)
if
(
DisplayBeep{NULL);
break;
}
else
{
{loadASCII (argv [n] ) )

printAll () ;
else
DisplayBeep(NULL);
if
1*
File name end input!
*1
n++;
break;
n++:
i f (n < argc)
{
545
4.
AMIGA
C FOR
1* Error! *1
puts (UTI);
puts (UT2);
goto Ende;
I*#FOLD:
Main loop *1
d>
(
1* Set cursor: *1
Cursor ();
signal = Wait(IL edUserPort->mp_SigBit);
/* erase cursor again: *1
Cursor ();
while (imsg = GetMsg(edUserPort
{
class
code
qualifier
iaddress
mouseX
mouseY
=
imsg->Class;
imsg->Code;
imsg->Qualifier;
= imsg->IAddress;
imsg->MouseX;
imsg->MouseY;
ReplyMsg(imsg) ;
switch (class)
{
case RAWKEY:
{
register struct IntuiMessage *iml,*im2;

if (! (code & IECODE_UP_PREFIX
{
inputEvent.ie Code
= code;
) >= 0)
running = handleKeys(inputBuffer,inputLen);
1* run-on suppressed: *1
edUserPort->mp MsgList.lh Head;
while (im2 = (struct IntuiMessage *)
iml->ExecMessage.mn_Node.ln_Succ)
if
if
if
if
if
(im2->ExecMessage.mn_Node.ln_succ 2~ NULL) break;

(iml->Class != RAWKEY) break;
(! (iml->Qualifier & IEQUALIFIER REPEAT break;
(im2->Class != RAWKEY) break; (! (im2->Qualifier & IEQUALIFIER_REPEAT break;
1* Message reply: *1
iml = GetMsg(edUserPort);
ReplyMsg(iml);
546
4.3 STEP BY STEP
ABACUS
iml = (stroct IntuiMessaqe *)

edUserPort->mp_MsqList.lh_Head;
break;
}
case MOUSEBUTTONS:
{
register WORD x,y;

i f (mouseX <= actualEditor->xoff)
x = 0;
else
x = (mouseX - actualEditor->xoff)
1 actualEditor->cw;
x += actuaIEditor->leftpos;
if (x > MAXWIDTH)
x = MAXWIDTH;
y = 0;
else
y = (mouseY - actuaIEditor->yoff)
1 actualEditor->ch;
if y < actualEditor->wch)
&& (actualEditor->zlinesptr[y))
actualEditor->xpos = x;
actuaIEditor->ypos = actualEditor->zlinesnr[y);
printXpos();
printypos () ;
case NEWSlZE:

if (actualEditor->xpos > actualEditor->leftpos
+actualEditor->wcw)
actualEditor->xpos = actuaIEditor->leftpos
printXpos 0 ;
{
actuaIEditor->ypos = actualEditor->zlinesnr
[(actualEditor->wdy = actuaIEditor->wch - I)};
printYpos () ;
break;
case REFRESHWINDOW:
BeqinRefresh(actuaIEditor->window);
547
4.
AMIGA
printAlI () ;
EndRefresh(actuaIEditor->window,TRUE);
break;
case GADGETUP:
{

register SHORT pos,hw;
if (ptr = executeCommand(actuaIEditor->gc SIBuffer
if (ptr == -IL)
running = FALSE;
else
{
if
(ptr >= Ox80000000)

ptr =
NULL - ptr:
1* set cursor in StringGadget to error *1

pes = ptr - actualEditor->gc SIBuffer:
actuaIEditor->gc GadgetSI.BufferPos = pes;
if (pes < actuaIEditor->gc GadgetSI.DispCount)
actuaIEditor->gc GadgetSr.Disppos = 0:
else
actuaIEditor->gc GadgetSI.DispPos = pes

- actuaIEditor->gc_GadgetSI.DispCount 1 2:
1* Gagdet aktivated: *1
ActivateGadget(&(actuaIEditor->G Command),
actuaIEditor->window~NULL):
DisplayBeep(NULL);
break:
case CLOSEWINDOW:
running = FALSE;
break:
default:
printf("Not a processable event: \lx\n",class):
1* of case *1
savelfCUrsorMoved();
} 1* of while (GetMsg( *1
} while (running):
I*#ENDFD*I
Ende:
1* Close all editor windows: *1

CloseEditor(ed):
if (edUserPort)
{
DeletePort(edUserPort);
edUserPort = NULL:
1* Font closed: *1
548
4.3
ABACUS
STEP BY STEP
if (infoFont)
{
CloseFont(infoFont);
infoFont = NULL;
1* Close Libraries: *1
i f (DosBase)
{
CloseLibrary(DosBase);
DosBase = NULL;
}
i f (GfxBase)
{
CloseLibrary(GfxBase);
GfxBase = NULL;
}
if (IntuitionBase)
{
IntuitionBase = NULL;
eLI
arguments in
the editor
The commandSet function reads the CLI command so that you can
set flags from the CLI instead of in the editor's command line (we
mentioned this earlier). When entering the set flags from the CLI.
remember to separate each argument with a space. and enter a dash
instead of an S. The following CLI command loads the editor. sets the
right margin at 60, disables both SkipBlanks and tabs, and loads the
ASCn me named Floatingtext .ASC into the editor:
Editor -r=60 -b -t Floatingtext.ASC
There are also a few changes in the Cursor. c module. First let's
examine some of the new Defines:
'define
'define
#define
#define
CEND '@'
ESC 27
TAB 9
REPCOM 7
Because flag switching now occurs through the command line, we can
remove the control code flag switching. Delete the De fin e s
TABMODE, WRITEMODE, and AUTOINDENT.
The following external functions are also new. We need these
commands for the info and command lines:
void printXpos(),printYpos(),printFold(),printFlags();
BOOL ActivateGadget();
UBYTE *executeCommand();
The following functions affect the cursor position or other info line
values (cursorLeft, cursorRight. cursorUp, cursorDown.
549
4.
AMIGA
halfPageUp, halfPageDown, cursorHome,enterFold,

ex i t F old) SO they must call the corresponding functions
(printXpos, printYpos, printFold). Three new functions
control cursor movement, which place the cursor at the end of the line,
the beginning of the text or at the end of the text:
I*#FOLD: cursorEnd *1
1***************************************
**
cursorEnd:
* Set the cursor on the last line.
***************************************1
BOOL cursorEnd()
{

register UWORD len,pos;
if (actuaIEditor->bufypos)
pos ~ actuaIEditor->buflen + 1;
else
if (z ~ ZLINESPTR(actuaIEditor->wdy
{
1* remove CR from lenght: *1

len
ptr
= z->len:
~
UBYTE *) (z + 1
+ len - I:
i f (len)
i f (*ptr
=~ CR)
len--;
else if (*ptr == LF)
i f (--len)
if (*--ptr ~ CR)
len--;
if (actuaIEditor->tabs)
{
ptr
tab
pos
UBYTE *) (z + 1;
= actuaIEditor->tabstring;
= 1;
while len--) && (pos < MAXWIDTH

if (*ptr++ ~ TAB)
cb
{
tab++;
pos++;
} while *tab) && (pos < MAXWIDTH;
else
{
tab++;
pos++;
else
pos
}
else
550
len + 1;
4.3
ABACUS
STEP BY STEP
return (FALSE);
actualEditor->xpos = pos;
1f (pos >= (actualEditor->leftpos + actualEditor->wcw
scrollLeft(pos + 1 - (actualEditor->leftpos
+ actualEditor->wcw;
else if (pos <= actualEditor->leftpos)
scrollRight(actualEditor->leftpos + 1 - pos);
printXpos () ;
return (TRUE);
}
I*#ENDFD*I
cursorEnd is the most complicated function because the end of the

line must be calculated from the tabs. The cursorTop function
counts the lines that are in the top half of the window and scrolls the
window down by this value. The cursorBottom function works in
the same manner except that the program must be prepared for the
possiblity of no text in the bottom of the window. Insert these three
functions following the cursorHome function.
The handleKeys function now returns a value of type BOOL:

BOOL handleKeys(buf,len)
In the second switch case function, at the location where the CSI
sequence check occurs, insert three lines following the check for
CURSORHOME:
case CEND:
cursorEnd () ;
break;
In the second switch case function, at the location where control

code checking occurs, insert the following lines before the check for
CFOLD:
case ESC:
1* Rest of buf transferred: *1

{
register UBYTE *sibuf;

register UWORD cnt;
sibuf = actualEditor->gc_SIBuffer;
while (len)
{
*sibuf++
len--;
*buf++;
*sibuf = 0;
cnt = (sibuf - actualEditor->gc_SIBuffer);
actualEditor->gc GadgetSI.BufferPos
= cnt;
actualEditor->gc=GadgetSI.NumChars
= cnt;
sst
actuaIEditor->gc_GadgetSI.DispPos = 0;
if (!ActivateGadget(&(actualEditor->G_Command).
actuaIEditor->window,NULL
DisplayBeep(NULL);
break;
case REPCOM:
{

if (ptr = executeCommand(actualEditor->gc SIBuffer
i f (ptr == -IL)
return (FALSE);
else
{
if
(ptr >= Ox80000000)

ptr = NULL - ptr;
1* Cursor in StringGadget in error set *1

pes = ptr - actualEditor->gc SIBuffer;
actualEditor->gc GadgetSI.BufferPos = pes;
if (pes < actualEditor->gc_GadgetSI.DispCount)
actualEditor->gc GadgetSI.DispPos = 0;
else
actualEditor->gc GadgetSI.DispPos - pes
- actualEditor->gc_GadgetSI.DispCount 1 2;
ActivateGadget (& (actuaIEditor->G Command).
actualEditor->window~NULL);
DisplayBeep(NULL);
1* return now: *1
return (TRUE);
break;
When Esc is pressed the rest of the characters entered from the
keyboard are written to the active command line. This character transfer
makes the most sense since you can configure the keyboard. A key
could be configured so that when pressed, it sends <Esc> then a
command sequence. All the user has to do is press <Return> to enter
the sequence.
The second switch case function handles the case if the user
pressed <Ctrl><G> (REP COM), which repeatedly executes the command
currently in the command line. In the case of an error, the cursor of the
command line stops at the error and activates the window, just like the
GADGETUP check in the main program.
You must erase the TABMODE, WRITEMODE and AUTOINDENT
checks, because these flags can now be set through the command line.
552
4.3
ABACUS
STEP BY ST E P
This keeps the user from pressing the wrong key, and allows the user to
insert additional control codes directly into the text. A special feature of
our editor is that we can read text which consists of any characters.
When examining the editor we determined that the <Return> key sends
a CR. However, other CLI commands (e.g., TYPE) cannot make use of
lines ending with just a CR. We always add a linefeed (LF) as the end
of line so that text written from the editor can be used in other
programs. The CR/LF check looks like the following code:
case LF:
case CR:
i f (! insert Line UBYTE) LF
DisplayBeep(NULL);
break;
This concludes the Cursor. c module. Now we can tum to the final
changes needed in the Edi t c module. Start by adding a few new
external functions:
void printXpos(),printYpos(),printFold();
void printNumLines(),printFlags();
BOOL cursorEnd();
This module has the calls to the functions corresponding to changes

that occur in the info line (printXpos, printYpos, printFold,
printNumLines and printFlags).
A local pointer to a line in lineptr of the Editor structure must
be saved by the saveLine function before the newLine function is
called. The corresponding program segment is as follows:
1* use new line: *1
{
actuaIEditor->lineptr = line;
/* Garbage-Collection */
{
old = actuaIEditor->lineptr;
Insert (&actualEditor->zlines,line,old);
line->flags = old->flags & -ZLF USED;
deleteZline (old) ;
/* now possible to convert pointer: */
1 <= actuaIEditor->wch; 1++, zptr++)
if (*zptr == old)
{
*zptr = line;
break;
}
else
553
actualEditor->1ineptr = NULL;
return (FALSE);
else
{
line->len
= len;
Remember to place the printYpos and printFlags calls in this

function, because changed is set. A call for printNumLines also
belongs in the getBufferPointer function because a first line is
redisplayed. The deleteChar and backspaceChar functions
change so that TRUE is always returned. The difference between TRUE
and FALSE is unfortunately not confirmed here. This often occurs
owing to an interruption during the execution of a command sequence.
A small change in the insert Line function must be made,
otherwise the LS (line split) command will not execute correctly:
p2 = buf + (len = deconvertLine(buf,actualEditor>buffer,inc ;
i f (*p2 = c)
len++;
This lets the editor divide lines without inserting characters. The end of
the function redisplays the info line's values:
printXpos 0 ;
printypos () ;
printNurnLines();
printFlags () ;
If minfold and maxfold have changed, printFold must be called
at the corresponding place.

A small error exists in the deleteZline function that we first
noticed when implementing the info line. Remember that we had
miscalculated the Y position, and since no info line existed at that time,
the debugger was the only way to find the error. Something similar has
happened in the deleteZline function, and at the location
determined by the pointer to the next and previous lines:
prevnr = ZLlNESNR(actualEditor->wdy);
prey = prevLine(line,&prevnr);
nextnr = ZElLENNR(actualEditor->wdy);
next
= nextLine(line,&nextnr);
We assumed that the next line had the same number as the current line,
because the current line was erased. The above command sequence must
be changed as follows:
554
ABACUS
4.3
prevnr=
prev =
next ::
nextnr=
STEP BY STEP
nextLine (line, &nextnr);
The fact that nextnr doesn't initialize when nextLine executes

causes no problem, because the value from nextnr only increments to
one in the nextLine function.
The getLastWord function is new. This function returns the pointer
to the last word from a given string. Insert this function before the
insert Char function:
/*ItFOLD: getLastWord *1
1**************************************
*
*
*
*
*
getLastWord(str,len)
Returns pointer to last
word in a string.
str A String.
len = buffer length.
**************************************1
UBYTE *getLastWord(str,len)
register UBYTE
*str;
register UWORD
len;
(
register UBYTE c,*lwb,*lwn,*end;

lwb = NULL;
1* Last Word is space */
lwn = NULL;
1* Last Word normal */
end = str + len - 1;
/* pointer to string end */
for (str += len; len; len--)
{
if c = *-str)
= , .)
lwb = str;
break:
}
else if (! c >= 'A') && (c <= 'Z')

II (c >= 'a') && (c <= 'z')
II (c >= '0') && (c <= '9')
II (c >= 192) II (c = '_')
i f (lwn == NULL)
lwn = str:
}
if (lwb == NULL)
if (lwn)
lwb = lwn;
else
lwb = end;
else
i f (lwn)
if end - lwb) > MAXWIDTH / 5)
if end - lwn + 5) < (end - lwb
lwb = lwn;
return (lwb + 1);
/*ItENDFD*/
555
4.
AMIGA
The function searches for the last character preceded by a blank (1 wb =

last word blank), as well as the last character not preceded by a letter, a
number, an underline or an international character (1wn =last word
normal). The 1 wb value is the default. If 1wb is too far from the end of
string (1/5 of the maximum width = 16 characters), and if a proper
character is found for 1wn, the function gives lwn first priority
provided it is at least five characters closer to the end of string than
lwb. Words divided with a hyphen are split at the hyphen, as long as
the first section of the word is longer than five characters. A negative
number like "-123" is ignored, since no characters exist to the left of
the minus sign.
The insertChar function needs some fine-tuning. When you insert
tabs at the beginning of a line and the line reaches the maximum width,
no more tabs can be inserted. However, the cursor moves anyway. To
avoid this, the xadd variable can be reset if the line length goes
beyond the maximum width. Expand the line xadd -= as
follows:
if (actualEditor->buflen + num > MAXWIDTH)

{
xadd -= num + actualEditor->buflen - MAXWIDTH;

num = MAXWIDTH - actualEditor->buflen;
WordWrap executes before inserting a character in the line:

if ! actualEditor->skipblanks)
&&
(actualEditor->xpos >= actualEditor->rm
1* Word-Wrap when SkipBlanks == 0 *1

actualEditor->xpos = getLastWord(actualEditor->buffer,
actualEditor->buflen)
- actualEditor->buffer + 1;
if (! insertLineUBYTE) 0
return (FALSE);
DisplayBeep(NULL);
cursorEnd () ;
return (FALSE);
The original function returns after the character is inserted in the line.
The ini tFo1ding function should be inserted at the end of the Edit
module:
556
ABACUS
4.3
STEP BY STEP
/*flFOLD: initFolding */
/**********************************
* initFolding:
* calculate the fold level of all
* lines from text start.
**********************************/
void initFoldingO
{

register UWORD level,l;
z = actuaIEditor->zlines.head;
level = 0;
while (z->succ)
{
z->flags = (z->flags & ~ZLF FOLD)

-
(level & ZLF_FOLD);
z->flags 1= ZLF FSE;

level += 1;
z
= z->succ;
/*lIENDFD*/
This completes the final changes. Remove the test. 0 module from
the makefile. Replace it with the Command. 0 module and compile
it. You can find source text and the finished program on the optional
disk for this book, in the v 0 6 directory. Appendix A contains the
complete source text for all of the modules..
Now you can load any text; don't use the TestText file. Access the
editor from the CLI and load the Command. 0 file. If loading from the
optional disk for this book, you'd enter the following CLI commands:
cd VO.6
Editor src/Command.c
We added folds to this and all other modules for maximum readability.
You now have an overview of all of the functions in this module.
Pressing <CtrlxF> moves you deeper into a fold, and pressing
<Ctrl><E> moves you toward the top fold level.
From this point, you're on your own. You now have all the elements
for programming an editor. In addition, you have a finished product for
editing your own source texts. We mentioned a few extra commands
that could be added later, and listed some of the elements of a
programmable editor. With this knowledge and some extra study, you
could add extended features to your own editor, such as menus, block
functions and more.
557
4.
4.4
AMIGA
Editor Command Set

The following overview lists the commands available in the final
version of the editor listed in this book.
4.4.1
Editor Keyboard Commands
Cursor
movement
<Cursor up/down!left/right>
<Shift><Cursor right/left>
<Shut><Cursorup/down>
<Delete>
<Backspace>
<Tab>
Control keys
<Ctrl><B>
<Ctrl><E>
<Ctrl><P>
<Ctrl><G>
<Ctrl><H>
<Ctrl><1>
<Ctrl><J>
<Ctrl><M>
moves cursor one line/character in

that direction
cursor to beginning/end of current line
cursor one-half page up/down
deletes character under cursor
deletes character to left of cursor
inserts tab
deletes current line
exits current fold
enters fold
re-executes current command
backspace
tab
linefeed
carriage return
4.4.2
Editor Command Line Commands
ASCII
commands
AL ["Filename"]
loads a source text named "filename"
AS ["Filename")
saves a source text named "filename"
CB (Cursor bottom)
CD (Cursor down)
CE (Cursor end)
CH (Cursor home)
moves cursor to end of text/fold

moves cursor one line down
moves cursor to end of line
Cursor
movement
558
4.4
ABACUS
EDITOR COMMAND
SET
CL (Cursor left)
CN (Cursor next)
CP (Cursor previous)
CR (Cursor right)
CS (Cursor start)
CT (Cursor top)
CU (Cursor up)
moves cursor one character left

moves cursor to start of next line
moves cursor to start of previous line
moves cursor one character right
moves cursor to start of current line
moves cursor to start of text/fold
moves cursor one line up
DB (delete back)
DC (delete char)
DL (delete line)
backspace
delete character
delete line
LS (line split)
divides line into two lines at current

cursor position
Quit command
exits editor
Set commands
Sa
SA
Sb
SB
Si
So
ST
St
Auto Indent off

Auto Indent on
SkipBlanks off
SkipBlanks on
Insert mode
Overwrite mode
Tabs on
Tabs off
Delete
commands
Line
commands
eXit command x
saves changes and exits editor
559
ABACUS
ApPENDIX A VERSIONI MAKEFILE
Appendix A Editor listings

This section contains complete editor listings of the first version and
the final version of the editor program, for users that wish to type in
the editor source code. Changes in versions 2 thru 5 are outlined in the
text. The optional disk available for this books contains the complete
source code for all six version of the editor.
Version 1 makefile
.c.o:
cc +B +Ipre!Editor.pre -0 $@ src/$*.c
prelist. pre:
OBJ=src/Editor.o
Editor: $ (OBJ)
In $(OBJ) -lc
-0
Editor
Debug: $ (OBJ)
In -w $(OBJ) -lc
-0
Editor
pre/Editor.pre: pre/Editor.prelist src/Editor.h

src/Editor.o: src/Editor.c src/Editor.h pre/Editor.pre
561
ApPENDIX A EDITOR LISTINGS
Version 1 Editor.h
/************
* Includes:
. ..
************/
/***********
**
Defines:
************/
#define MAXWIDTH 80
#define EVENLEN (x) (x) +1) &-2)
/*******************
* Data structures:
*******************/
struct Zline
{
struct Zline *succ;

struct Zline *pred;
UBYTE flags;
UBYTE len;
};
#define ZLF USED 128

struct ZList
{
struct Zline *head;

struct Zline *tail;
};
{

UWORD len;
};
struct FList
{

};
struct Memoryblock
562
AMIGA
ABACUS
ApPENDIX A VERSIONl EDITOR.H

ULONG length;
ULONG free;
struct FList freeliste;
};
struct BList
{

};
struct Editor
{

struct BList block;
UBYTE buffer[MAXWIDTH1;
UBYTE tabstring[MAXWIDTH);
UWORD num lines;
struct Zline *top;
UWORD toppos;
UWORD xpos,ypos;
UWORD changed:1;
UWORD insert:1;
}:
struct EList
{
struct Editor *head:

struct Editor *tail:
};
563
Version 1 Editor.c
1* Editor.c VO.1 *1
1************
* Includes:
************1
1***********
* Defines:
***********1
#define REV 33L
#define MAXINPUTLEN 128L
1**********************
**********************1
void DeletePort(),ModifyIDCMP(),ReplyMsg();
ULONG Wait(),OpenDevice();
1*********************
* Global Variables:
*********************1
struct Library *IntuitionBase = NULL, *GfxBase
struct Device *ConsoleDevice = NULL;
= NULL,
*DosBase
= NULL;

struct Editor *ActualEditor;
struct NewWindow newEdWindow =
{
100,40,440,156,
REFRESHWINDOW I MOUSEBUTTONS I RAWKEY I CLOSEWINDOW,
WINDOWSIZING I WINDOWDRAG I WINDOWDEPTH I WINDOWCLOSE I SIZEBBOTTOM
NULL, NULL,
(UBYTE *)"Editor",
NULL, NULL,
100,40,640,200,
WBENCHSCREEN
};
struct MsgPort *edUserPort = NULL;

564
ApPENDIX A VERSION 1 EDITOR.C
ABACUS
UBYTE inputBuffer[MAXINPUTLEN]:
UWORD input Len = 0;
struct InputEvent inputEvent
{
0,
IECLASS_RAWKEY,O,
0,0
};
1***************
* Functions:
***************1
1***********************************
*
*
*
*
*
OpenEditor ()
Open Editor window and
initalize Editor structure
Returns pointer to Editor structure
or is zero in case of an error
***********************************1
{

register struct Window *wd:
register ULONG flags;

1* Memory for editor structure: */

1* Window open: *1
(
ed->window = wd:
1* UserPort established: *1
ModifyIDCMP(wd,flags):
1* Parameter initialization: *1
ed->num lines = 0:
ed->actual = NULL:
ed->top = NULL:
ed->toppos = 0;
ed->xpos = 1;
ed->ypos = 1:
ed->changed = 0:
ed->insert = 1;
}
else
{
free (ed) ;
ed = NULL:
565
AMIG-A
return (ed);
flags;
1***************************
* CloseEditor(ed)
* ed A Editor structure.
***************************1
struct Editor *ed;
(
1* Close window: *1
ed->window->UserPort = NULL;
CloseWindow(ed->window);
free (ed);
1*****************
* Main program:
*****************1
main()
{

BOOL running = TRUE;
ULONG signal, class;
UWORD code, qualifier;
APTR iaddress;
SHORT 1;
1* Libraries open: *1
if ( I (IntuitionBase = OpenLibrary("intuition.library",REV)
goto Ende;
if ( I (GfxBase
OpenLibrary("graphics.library",REV)
goto Ende;
if ( I (DosBase
OpenLibrary ("dos.library", REV)
goto Ende;
if ( I (OpenDevice("console.device",-IL,&ioStdReq,OL)
ConsoleDevice = ioStdReq.io Device;
else
goto Ende;
1* UserPort open *1
if (I (edUserPort = CreatePort(NULL,OL) goto Ende;
1* Editor list initialization: *1
1* Open first editor window: *1
if (ed = OpenEditor(
(
ActualEditor = ed;
else
goto Ende;
566
ABACUS
ApPENDIX A VERSION
1 EDITOR.C
d:>
{
signal
Wait(1L edUserPort->mp_SigBit);
class
= GetMsg(edUserPort
= imsg->Class;
code
while (imsg
{
qualifier
iaddress
imsg->Code;
imsg->Qualifier;
= imsg->IAddress;
ReplyMsg (imsg) ;
/* Event processing: */
switch (class)
{
case RAWKEY:
if (! (code & IECODE_UP~REFIX
{
inputEvent.ie_Code
= code;
inputEvent.ie_Qualifier = qualifier;
if ((l = RawKeyConvert (
) >= 0)
inputBuffer[l] = 0;
printf(II%3d> Length = %d: ",n++,l);
for (1 = 0; inputBuffer[l]; 1++)
printf("%2x", (UWORD)inputBuffer[l]);
if (inputBuffer[O] >= 32 && inputBuffer[O] <= 127)
printf(II %s",inputBuffer);
printf("\n");
break;
case CLOSEWINDOW:
running = FALSE;
break;
default:
} /* of case */
/* of while (GetMsg(
while (running);
*/
Ende:
/* close all editor windows: */
CloseEditor(ed);
/* Close UserPort: */
if (edUserPort) DeletePort(edUserPort);
/*
if
if
if
Close Libraries: *1
(DosBase) CloseLibrary(DosBase);
(GfxBase) CloseLibrary(GfxBase);
(IntuitionBase) CloseLibrary(IntuitionBase);
567
Version 4 TestText
The JI characters in the following fIle represent the linefeed character,
decimal 1O. The character at the end of the fIle represents the Escape
character, decimal 27. These characters and the "a" must be added to the
fIle since redirection is used to load the test fIle into Version 4 of the
editor for testing. We used the replace command from the Abacus
Amiga DOS Toolbox to replace a single linefeed with
"linefeed,a,linefeed" (10,97,10) and then placed an "escape,linefeed,
linefeed" (27,10,10) at the end of the file.
Notice the folding marker comments in this text, make sure to add these
to your Test Text fIle. The redirection feature of AmigaDOS allows
the a's and linefeeds in this fIle to be interpeted as input for this version
of the editor. Remember that an Escape character (decimal 27) must be
at the end of the fIle.
ctl!
1* TestText for testing the editor */i

ctl!
I*#FOLD: Includes */i

ctl!
I************i
as!
*i
as!
* Includes:'lI
as!
*'lI
as!
************/'ll
as!
':II
ctl!
#include <exec/types.h>':II
ctl!
#include <intuition/intuition.h>'lI
ctl!
#include "src/Editor.h"':II
as!
/*#ENDFD Includes */'lI
as!
/*#FOLD: Defines */'ll
as!
/***********'lI
a'll
*'lI
as!
* Defines:'lI
as!
*i
ctl!
***********/i
as!
'lI
568
ApPENDIX A VERSION
ABACUS
a'll
#define REV
TESTTEXT
33L~
a'l[
#define MAXINPUTLEN 128L~

a'll
I*#ENDFD Defines */'J[
a'll
I*#FOLD: External Functions
a'll
*/~
I**********************~
a'll
*'11
a'll
* External Functions:'1I
a'll
*'.I!
a'l[
**********************/'11
a~
'11
a'll
void Test(),print();~
a'll
'11
a'll
struct Library *OpenLibrary();'1I
aCJ[
struct Window *OpenWindow();'1I

a'll
void CloseLibrary(),NewList(),AddTail(),CloseWindow(),free();'ll
a'll
void DeletePort(),ModifyIDCMP(),ReplyMsg(),freeMemoryblock();'.I!
a'll
void initWindowSize(),SetDrMd(),SetAPen(),SetBPen(),printAll();'J[
a'J[
void handleKeys(),Cursor();'ll
a'll
struct Editor *malloc(),*RemHead();'ll
a'l[
struct MsgPort *CreatePort();'ll

aCJ[
struct IntuiMessage *GetMsg();'1I

a'J
ULONG Wait (),OpenDevice();'1I
a'J
SHORT RawKeyConvert();'.I!
a'll
'.I!
a'll
I*#ENDFD External Functions */'11
'11
'lI
569
AMIGA
Version 6 makefile
.c.o:
cc +B +I/pre/Editor.pre -0 $@ src/$*.c
.prelist.pre:
cc -A -0 ram:$*.preasm +H$@ /pre/$*.prelist
OBJ=src/Editor.o src/Memory.o src/OUtput.o src/Cursor.o src/Edit.o\
src/Command.o
Editor: $ (OBJ)
In $(OBJ) -lc
-0
Debug: $ (OBJ)
In -w $(OBJ) -lc
Editor
-0
Editor
/pre/Editor.pre: /pre/Editor.prelist src/Editor.h

src/Editor.o: src/Editor.c src/Editor.h /pre/Editor.pre
cc +I/pre/Editor.pre src/Editor.c
src/Memory.o: src/Memory.c src/Editor.h /pre/Editor.pre
src/OUtput.o: src/Output.c src/Editor.h /pre/Editor.pre
src/Cursor.o: src/Cursor.c src/Editor.h /pre/Editor.pre
src/Edit.o: src/Edit.c src/Editor.h /pre/Editor.pre
src/Command.o: src/Command.c src/Editor.h
/pre/Editor.pre
..
570
ABACUS
ApPENDIX A VERSION
6 EDITOR.H
Version 6 Editor.h
1************
* Includes:
************1
#include <stdio.h>
1***********
* Defines:
***********1
#define
#define
#define
#define
#define
#define
#define
#define
MAXWIDTH 80
MAXHEIGHT 64
BGPEN OL
FGPEN lL
FOLDPEN 2L
CTRLPEN 3L
EVENLEN (x) (x)+1) &-2)
CONTROLCODE(c) (c)&127)<32)
#define ZLINESPTR(n) actualEditor->zlinesptr(n)

#define ZLINESNR(n) actualEditor->zlinesnr[n)
#define
#define
#define
#define
#define
#define
#define
#define
#define
#define
#define
#define
INFO XPOS INC 2

INFO XPOS LEN 4
INFO=YPOS=INC 9
INFO YPOS LEN 4
INFO-NUMOFLINES INC 16
INFO-NUMOFLlNES-LEN 4
INFO-MINFOLD INC 22
INFO-MINFOLD-LEN 2
INFO-MAXFOLD-INC 25
INFO-MAXFOLD-LEN 2
INFO-FIAGS INC 29
INFO=FIAGS=LEN 5
/*******************
* Data structures:
*******************/
struct Zline
{
struct Zline *succ;

struct Zline *pred;
UBYTE flags;
UBYTE len;
};
571
AMIGA
#define ZLF USED 128

#define ZLF FSE 64
#define ZLF-FOLD 63
struct ZList
{
struct Zline *head;

struct Zline *tail;
};
{

UWORD len;
};
struct FList
{

};
struct Memoryblock
{

ULONG length;
ULONG free;
struct FList freeliste;
}:
struct BList
{

);
struct Editor
{
struct Editor *succ:

struct BList block:
UBYTE buffer[MAXWIDTH + 1];
UBYTE buflastchar;
UWORD buflen,bufypos:
UWORD num_lines;
struct Zline *zlinesptr[MAXHEIGHT];
UWORD zlinesnr[MAXHEIGHT];
UWORD leftpos;
UWORD xpos,ypos;
UWORD wdy;
UWORD changed
I:
UWORD insert
1;
UWORD tabs
1;
572
ApPENDIX A VERSION
ABACUS
6 EDITOR.H
UWORD skipblanks : 1;
UWORD autoindent : 1;
struet RastPort *rp;
UWORD xoff,yoff;
UWORD xser,yser;
UWORD ew,eh;
UWORD wew, weh;
UWORD minfold;
UWORD maxfold;
UBYTE gc_SIBuffer[256j;
struct StringInfo ge GadgetSI;
struct Gadget G Command;
UBYTE gi Text[36];
struct IntuiText gi_IText;
struct Gadget G Info;
ULONG ip_xpos; ULONG ip ypos;
ULONG ip=nurnzlines;
ULONG ip minfold;
ULONG ip-maxfold;
ULONG ip-flags;
ULONG ip=topedge;
UBYTE filename [80] ;
UWORD rm;
struct Zline *lineptr;
};
struct EList
{
struct Editor *head;

struct Editor *tail;
};
573
AMIGA
Version 6 Editor.c
1*********************************************
*
*
*
Editor
Text editor for book Adv C.
1988 DATA BECKER
ABACUS
*********************************************1
Includes *1
1************
I*#FOLD:
* Includes:
************1
/*#ENDFD*/
/*#FOLD: Defines *1
1***********
* Defines:
************1
#define
#define
#define
#define
T]"
REV 33L
MAXINPUTLEN 128L
UT1 "USAGE: Editor [Flags] <Filename>"
UT2 "
Flags: [-a] [-A] [-b] [-B] [- ill] [- 010] [- rIR[=n]] [-t] [-
I*#ENDFD*/
I*#FOLD: External Functions *1
/**********************
**********************1

void DeletePort(),ModifyIDCMP(),ReplyMsg(),freeMemoryblock();
void initWindowSize(),SetDrMd(),SetAPen(),SetBPen(),printAll();
void Cursor(),savelfCursorMoved(),strncpy(),CloseFont(),puts();
void printXpos(),printYpos(),printlnfo(),DisplayBeep();
void BeginRefresh(),EndRefresh();
ULONG Wait (),OpenDevice();
struct TextFont *OpenFont();
BOOL handleKeys(),ActivateGadget(),loadASCII();
UBYTE *executeCommand(),*commandSet();
I*#ENDFD*/
I*#FOLD: Global Variables
1*********************
Global Variables:
*********************1
574
*1
ABACUS
ApPENDIX A VERSION
struct Library *IntuitionBase = NULL, *GfxBase

struct Device *ConsoleDevice = NULL:
NULL, *DosBase
6 EDITOR.C
NULL:

struct Editor *actualEditor:
/*#FOLD: Window und Gadgets *1
UBYTE gc SIBuffer[256]:
UBYTE gc-UndoBuffer[256];
struct StringInfo gc GadgetSI
gc SIBuffer,
gc- UndoBuffer,
0, -
256,
0,
0,0,0,0,0,
0,
NULL,
NULL
};
SHORT gc BorderVectors[4]
struct Border gc Border =
{
-1,-1,
1,0,JAMl,
{0,0,350,0}:
2,
gc_BorderVectors,
NULL
}:
struct Gadget G_Command
NULL,
2, -9,
-302,9,
GADGHCOMP
GRELBOTTOM
RELVERIFY
BOTTOMBORDER,
GRELWIDTH,
STRGADGET,
(APTR) &gc Border,
NULL,
NULL,
0,
(APTR)&gc GadgetSI,
2,
NULL
}i
SHORT gi_BorderVectors[14]
struct Border gi Border =
{
-
{-1,0,-1,10,0,10,0,0,298,0,298,1,283,1};
0, 0,
1,0,JAMl,
7,
gi_BorderVectors,
NULL
};
struct TextAttr TOPAZ80
575
AMIGA
C FOR

TOPAZ EIGHTY,O,O
};
UBYTE gi_Text[36] = "X= 1 Y=

struct IntuiText gi IText =
{
1,0,JAM2,
4,2,
&TOPAZ80,
gi_Text,
1 #=
a [ 0,
0] lcATB";
NUlL
};
struct Gadget G_Info =

{
&G Command,
-i98,-10,
280,10,
GADGHNONE I GREIBOTTOM I GRELRIGHT,
BOOLGADGET,
(APTR)&gi_Border,
NULL,
&gi IText,
0, NULL,
1,
NUlL
};
0,0,640,200,
REFRESHWINDOW I MOUSEBUTTONS I RAWKEY I CLOSEWINDOW I NEWSIZE I GADGETUP,
WINDOWSIZING I WINDOWDRAG I WINDOWDEPTH I WINDOWCLOSE I SIZEBBOTTOM
I SIMPLE REFRESH I ACTIVATE,
NULL, NULL,
NULL,
NULL, NULL,
320,50,640,200,
WBENCHSCREEN
};
struct TextFont *infoFont

/*#ENDFD*/
= NULL;
struct MsgPort *edUserPort
= NULL;

UBYTE inputBuffer[MAXINPUTLEN};
UWORD inputLen = 0;
struct InputEvent inputEvent ~
{
0,
IECLASS RAWKEY,O,
0,0
};
/*#ENDFD*/
576
ApPENDIX A VERSION
ABACUS
6 EDITOR.C
/*IIFOLD: Functions */
/*#FOLD: nextLine */
/*********************************************
**
nextLine(line,pnr)
* Returns pointer to next line,

* and executes folding
* Zero, if no next line.
**
line ~ line-structure.
* pnr ~ to linenumber.
**********************************************/
struct Zline *nextLine(line,pnr)
register UWORD
*pnr;
(

register UWORD minfold = actualEditor->minfold;
register UWORD maxfold = actualEditor->maxfold;
i f (z = line)
i f (z->succ)
d:>
{
if z = z->succ)->succ)
if z->flags & ZLF FOLD) < minfold)
{
/* Following lines lie outside of fold */
z = NULL:
break;
else
*pnr += 1;
else
{
1* no following line *1
z = NULL:
break:
while z->flags & ZLF_FOLD) > maxfold):
else
z
= NULL:
return (z);
}
/*#ENDFD*/
/*#FOLD: prevLine */
/*******************************************
**
prevLine(line,pnr)
**
Returns pointer to previous line,

* and executes folding
* Zero, if no previous line.
**
line ~ Zline-structure.
* pnr ~ to line number.
*
577
AMIGA
*******************************************1
struct Zline *prevLine(line,pnr)
register UWORD
*pnr;
{

register UWORD min fold = actualEditor->minfold;
register UWORD max fold = actualEditor->maxfold;
i f (z = line)
i f (z->pred)
d::>
{
if ((z = z->pred)->pred)
if ((z->flags & ZLF FOLD) < minfold)
{
/* previous line lies outside fold *1
z = NULL;
break;
else
*pnr -= 1;
else
{
/* no previous line */
z = NULL;
break;
while ((z->flags & ZLF_FOLD) > maxfold);
else
z = NULL;
return (z);
}
I*#ENDFD*/
/*#FOLD: OpenEditor *1
1***********************************
* OpenEditor ()
* Open Editor window and

* initalize Editor structure
* Returns pointer to Editor structure

or is zero in case of an error
*
*
***********************************1

{

ULONG flags;

=> use your own UserPort! */
578
ApPENDIX A VERSION 6 EVITOR.C
ABACUS

{
/* Gadgets initialization: */
ed->gc SIBuffer[Oj = 0;
ed->gc-GadgetSI
= gc GadgetSI;
= G Command;
ed->G Command
ed->g1 IText
= g1 IText;
ed->G Info
:: G Info;
strncpy(ed->gi_Text,gi_Text,sizeof(ed->gi_Text;
/* set pointer: */
ed->gc_GadgetSI.Buffer
= ed->gc_SIBuffer;
ed->G Command.SpecialInfo = (APTR) &(ed->gc GadgetSI);
= ed->gi Text;
ed->g1 IText.IText
ed->G Info.NextGadget
= &(ed->G Command);
ed->G-Info.GadgetText
:: &(ed->gl IText);
newEdWindow.FirstGadget
= &(ed->G Info);
newEdWindow.Title
= ed->filename;
ed->filename [OJ
= 0;
/* Window open: */
if (wd:: OpenWindow(&newEdWindow
{
ed->window = wd;
/* Write mode set: */

SetDrMd(rp,JAM2);
SetAPen(rp,FGPEN);
SetBPen(rp,BGPEN);
ed->num lines :: 0;
ed->actual
= NULL;
ed->buflen
= 0;
= 0;
ed->bufypos
ed->buflastchar= ' ';
ed->leftpos
= 0;
ed->xpos
:: 1;
ed->ypos
:: 1;
ed->wdy
~ 0;
= 0;
ed->changed
ed->insert
= 1;
ed->tabs
= 1;
ed->skipblanks = 1;
ed->autoindent = 1;
ed->minfold
= 0;
ed->rnaxfold
= 0;
= MAXWIDTH;
ed->rm
register stiuct Zline **zptr;
579
AMIGA
ApPENDIX A EDITOl.' LISTIlIiGS
/* zlinesptr/nr-Array initialization: */
for (n = 0, zptr = ed->zlinesptr, pnr ~ ed->zlinesnr;
n < MAXHEIGHT; n++, zptr++, pnr++)
*zptr
*pnr
= NULL:
= 1;
1* Tab initialization: *1
*ptr++ = 1;
i f (n % 3)
*ptr++ = 1;
else
*ptr++ = 0;
register struct Editor *oldae;

ed->wch = 0;
initWindowSize(ed):
oldae
actualEditor
printlnfo():
actualEditor
= actualEditor:
ed;
oldae;
else
{
free (ed) :
ed = NULL;
newBdWindow.IDCMPFlags
return (ed);
flags;
I*#ENDFD*I
I*#FOLD: CloseEditor *1
/***************************
* CloseEditor(ed)
* ed
Editor structure.
***************************1
struct Editor *ed;
{

/* Free first memory block: *1
while (blk = (struct Memoryblock *)RemHead(&(ed->block)
freeMemoryblock(blk);
580
ABACUS
ApPENDIX A VERSION (; EDITOR.C
1* Then Close window: *1

ed->window->UserPort ~ NULL;
CloseWindow(ed->window);
free(ed);
}
1*/tENDFD*1
/*/tENDFD*/
/*/tFOLD: Main program */
/***************** .
** Main program:
*
*****************1
main (argc, argv)

int argc;
UBYTE
*argv[];
{

BooL running ~ TRUE;
ULONG signal,class,mouseX,mouseY;
UWORD code, qualifier;
APTR iaddress;
/*
if
Was Editor started with "Editor ?" , the Text output *1

(argc =
2)
if (*argv[l] = '?')
{
puts(UT1);
puts (UT2);
gotl) Ende;
1* Libraries open: */
if ( ! (IntuitionBase ~ OpenLibrary("intuition.library",REV)
goto Ende;
if ( ! (GfxBase ~ OpenLibrary("graphics.library",REV)
goto Ende;
if ( ! (DosBase ~ OpenLibrary ("dos.library", REV )
goto Ende;
if ( ! (OpenDevice("console.device",-lL,&ioStdReq,OL)
ConsoleDevice ~ ioStdReq.io Device;
else
goto Ende;
1* Font for Infoline open: */

if
(1
(infoFont
OpenFont(&TOPAZ80) goto Ende;
1* UserPort open *1
if
(1
(edUserPort
CreatePort(NULL,OL) goto Ende;
/* Max height for window calculated: *1

newEdWindow.Height ~ newEdWindow.MaxHeight
struct IntuitionBase *)IntuitionBase)->ActiveScreen->Height;
/* Editor list initialization: *1

1* Open first editor window: */'

i f (ed
OpenEditor () )
581
AMIGA
actualEditor = ed;
}
else
goto Ende;
/* Command line reading: */
if (argc >= 2)
{
while
(n
< argc)
(*argv[n]
if
I_I)
(commandSet(argv[n] + 1) >= Ox80000000)
if
(
DisplayBeep(NULL);
break;
}
else
(
(loaciASCII(argv[n]
printAllO;
else
DisplayBeep(NULL);
if
1*
File name end input! * /
n++;
break;
n++;
if (n < argc)
(
1* Error! */
puts(UT1);
puts (UT2) ;
goto Ende;
/"ilFOLD:
Main loop */
<b
(
/* Set cursor: */
CUrsor 0 ;
signal = Wait(1L edUserPort->mp_SigBit);
/* erase cursor again: */
Cursor ();
while (imsg = GetMsg(edUserPort
{
class
= imsg->elass;
= imsg->Code;
iaddress
= imsg->IAddress;
code
582
ABACUS
ApPENDIX A VERSION
mouseX
mouseY
6 EOITOR.C
imsg->MouseX;
imsg->MouseY;
ReplyMsg (imsg) ;
/* Event processing: */
switch (class)
{
case RAWKEY:
{
register struct IntuiMessage *iml,*im2;

if (! (code & IECODE_UP_PREFIX
{
inputEvent. ie_Code
= code;
inputEvent.ie Qualifier = qualifier;
) >= 0)
running = handleKeys(inputBuffer,inputLen);
/* run-on suppressed: */
while (im2 = (struct IntuiMessage *)
iml->ExecMessage.mn_Node.ln_Succ)
if
if
if
if
if
(im2->ExecMessage.mn_Node.ln_Succ == NULL) break;

(im1->Class != RAWKEY) break;
(! (im1->Qualifier & IEQUALIFIER_REPEAT break;
(im2->Class != RAWKEY) break;
(! (im2->Qualifier & IEQUALIFIER_REPEAT break;
/* Message reply: */
im1 = GetMsg(edUserPort);
ReplyMsg (im1);
im1 = (struct IntuiMessage *)
break;
case MOUSEBUTTONS:
{
register WORD x,y;

if (mouseX <= actualEditor->xoff)
x = 0;
else
x = (mouseX - actua1Editor->xoff)
/ actua1Editor->cw;
x += actualEditor->leftpos;
if (x > MAXWIDTH)
x = MAXWIDTH;
y = 0;
else
583
AMIGA
y = (mouseY - actuaIEditor->yoff)
I actuaIEditor->ch;
if ((y < actuaIEditor->wch)
&& (actuaIEditor->zlinesptr[y]
actuaIEditor->xpos = X;
actuaIEditor->ypos = actuaIEditor->zlinesnr[yJ;
printXpos () ;
printYpos () ;
case NEWSIZE:
initWindowSize(actuaIEditor);

if (actuaIEditor->xpos > actuaIEditor->leftpos
+actuaIEditor->wcw)
actuaIEditor->xpos = actuaIEditor->leftpos
printXpos ( ) ;
{
actuaIEditor->ypos = actuaIEditor->zlinesnr
[(actuaIEditor->wdy = actuaIEditor->wch - 1)J;
printYpos () ;
break;
case REFRESHWINDOW:
BeginRefresh(actuaIEditor->window);
printAll () ;
EndRefresh(actuaIEditor->window,TRUE);
break;
case GADGETUP:
{

i f (ptr = -1L)
running = FALSE;
else
{
if
(ptr >= Ox80000000)

ptr =
NULL - ptr;
1* set cursor in StringGadget to error *1

pos = ptr - actuaIEditor->gc_SIBuffer;
actuaIEditor->gc GadgetSI.BufferPos = pos;
if (pos < actuaIEditor->gc_GadgetSI.DispCount)
actuaIEditor->gc GadgetSI.DispPos = 0;
else
actua1Editor->gc_GadgetSI.DispPos = pos
584
ABACUS
ApPENDIX A VERSION
6 EDITOR.C
- actuaIEditor->gc_GadgetsI.DispCount / 2;
1* Gagdet aktivated: */
ActivateGadget(&(actuaIEditor->G Command),
actuaIEditor->window~NULL);
DisplayBeep(NULL);
}
break;
case CLOSEWINDOW:
running = FALSE;
break;
default:
1* of case */
savelfCUrsorMoved();
} 1* of while (GetMsg(
*/
} while (running);
/*#ENDFD*/
Ende:
/* Close all editor windows: *1
CloseEciitor(ed);
if (edUserPort)
{
DeletePort(edUserPort);
edUserPort = NULL;
1* Font closed: *1
if (infoFont)
(
CloseFont(infoFont);
infoFont = NULL;
1* Close Libraries:
*/
i f (DosBase)
{
CloseLibrary(DosBase);
DosBase = NULL;
}
i f (GfxBase)
{
CloseLibrary(GfxBase);
GfxBase = NULL;
}
if (IntuitionBase)
{
IntuitionBase = NULL;
}
I*#ENDFD*I
585
Version 6 Edit.c
1****************************************
*
*
*
Module: Edit
Functions to edit the text
****************************************/
/*#FOLD: Includes
/************
* Includes:
************/
*/
/*#ENDFD*/
/*#FOLD: Defines */
/***********
* Defines:
***********/
#define CR 13
#define LF 10
#define TAB 9
/* Lenght of fold start and end: */
#define FSE LEN 10
/* Number of characters, significant for fold marking: */
#define FSE SIG 8
/*#ENDFD* / /*#FOLD: External Functions */
/**********************
**********************/
UWORD convertLineForPrint(),strncmp(),recalcTopOfWindow();
void deleteZline(),Insert(),printAll(),restoreZlinenptr();
void DisplayBeep(),printLine(),AddHead(),scrollRight(),scrollUp();
void ScrollRaster(),printXpos(),printYpos(),printFold();
void printNumLines(),printFlags();
struct Zline *newZline(),*nextLine(),*prevLine();
BOOL cursorLeft(),cursorRight(),cursorDown(),cursorHome(),cursorEnd():
/*#ENDFD*/
/*#FOLD: External Variables */
/*********************
*********************/
/*#ENDFD*/
/*#FOLD: Global Variables *1
1*********************
* Global Variables:
*********************/
UBYTE foldStart[]
586
"/*#FOLD:*/";
A PPENDIX A VERSION 6 EDIT.C
ABACUS
UBYTE foldEnde []
/*!fENDFD*/
"/*!fENDFD*/";
/***************
* Functions:
***************/
/*#FOLD: getLineForEdit */
/********************************
* getLineForEdit(line,znr)
* Copy line to edit into
* the buffer.
* line 1\ Zline.
* znr : line number of the line.
********************************/
void getLineForEdit (line, znr)
register struct Zline *linei
register UWORD
znr;
{
register UWORD leftpos;

/* get leftpos: */
leftpos : actualEditor->leftpos;
actualEditor->leftpos : 0;
1* convert line: */
convertLineForPrint(line+l,line->len,MAXWIDTH + 1,
actualEditor->buffer,&actualEditor->buflastchar):
/* mark line as "used" : */
line->flags I: ZLF USED;
actualEditor->actual : line;
actualEditor->bufypos : znr;
/* establish Leftpos: */
actualEditor->leftpos : leftpos;
}
/*#ENDFD*/
/*#FOLD: cursorOnText */
1*******************************
* cursorOnText:
* make sure that the cursor
* is found on text.
*******************************/
void cursorOnText()
{

register UWORD I:
/* makes sure that 'the cursor is really on text: */
zptr: &(ZLINESPTR(l : actualEditor->wdy;
while *zptr:: NULL) && 1)
{
zptr--:
1-;
actualEditor->wdy
actualEditor->ypos
I;
ZLINESNR (1) :
587
AMIGA
I*#ENDFD*I
I*#FOLD: deconvertLine *1
1*************************
*
*
*
*
*
deconvertLine():
Deconverts line.
buf "target buffer.
buffer" source pointer.
buflen= buffer length.
*************************1
UWORD deconvertLine(buf, buffer,buflen)
UBYTE
*buf, *buffer:
UWORD
buflen;
{
register UBYTE *pl,*p2,*tab,*fb

register WORD 1;
= NULL;
pl = buffer:
p2 = buf:
tab: actualEditor->tabstring;
1
= buflen:
1* Spaces converted to tabs: *1
while (1)
{
if *P2 :: *pl++)
")
if (fb = NULL)
fb :: p2:
}
else
{
i f (fb)
fb
NULL;
1* fb points to first space *1

1* other characters are copied. (first blank)
if ! *++tab) && (fb
{
*fb :: TAB;
fb++;
p2 :: fbi
else
p2++:
1-;
}
else
while
(1)
*p2 = *pl:
pl++:
p2++:
1-:
1* Spaces at end deleted: *1

if (actualEditor->skipblanks)
588
*1
ABACUS
ApPENDIX A VERSION
while p2 > buf)

p2--;
&&
, ') II (*(p2-1)
*(p2-1)
6 EDIT.C
TAB )
return awORD) (p2 - buf;

}
/*#ENDFD*/
/*#FOLD: getFoldInc */
/****************************************
* getFoldInc(line)
* Tests if line is a fold start
* or end mark and gives 1 or -1
* back.
* Else a 0 is returned.
* line ~ corresponding line.
****************************************/
awORD getFoldlnc(line)
struct Zline
*line;
{
register UBYTE *pl;

register awORD 1;
/* Fold marker at beginning of line! */
1
= line->len;
while (1 && *p1 = , ') II (*p1 = TAB)
{
p1++;
1-;
/* Type of fold markers (-> 1): */

/* 1:: Start,
End and
/* -1
*/
/* 0 no fold marker.
if (1 >= FSE SIG)
if (strncmp(pl,foldStart,FSE SIG)
1
1;
*/
*/
==
else if (strncmp(pl,foldEnde,FSE SIG)

1 = -1;
0)
0)
else
1 = 0;
else
1 = 0;
/* has no fold markers * /
retum (1);
}
/*#ENDFD*/
/*#FOLD: saveLine */
/*******************************
* saveLine(line)
* Saves line again.
* line ~ old line structure.
*******************************/
BOOL saveLine(line)
struct Zline *line;
{

register UBYTE *p1,*p2,*tab,*fb
register WORD l,len;
struct Zline *pred,*old,**zptr;
/* +2 fuer CR und LF */
=
NULL;
589
AMIGA
C FOR
/* Only for security: */

if (actualEditor->actual == NULL)
return (FALSE);
p2 = buf + (len = deconvertLine(buf,actualEditor->buffer,
actualEditor->buflen;
/* determine if line ended with CR or LF, *1
1* and increases len correspondingly. *1
p2 = buf + len;
if (1 = line->len)
{
pI = UBYTE *) (line + 1
i f (*pl = CR)
+ line->len - 1;
len++;
*p2 = CR;
}
else if (*pl == LF)

{
len++;
if 1 > 1) && (*--pl
==
CR
len++;
*p2++ = CR;
*p2 = LF;
1* use new line: */

{
actualEditor->lineptr = line;
1* Garbage-Collection
old = actualEditor->lineptr;
Insert(&actualEditor->zlines,line,old);
line->flags = old->flags & -ZLF USED;
deleteZline (old) ;
-
1* now possible to convert pointer: */

1 <= actualEditor->wch; 1++, zptr++)
if (*zptr = old)
{
*zptr = line;
break;
else
{
return (FALSE);
else
{
590
*/
ABACUS
line->len
.. len;
1* write line back: *1

pI
p2
=
=
(UBYTE *) (line + 1);

buf;
= len;
while (1'
{
*pl .. *p2;
p1++;
p2++;
1--;
}
1* reconstruct folding: *1
1* was old or is new line a fold marker? *1
1 = getFoldlnc(line);
i f line->flags & ZLF FSE) II 1)
1* next the ZLF FSE flag set to
i f (1)
1:
*1
line->flagsl= ZLF FSE;

else
line->flags &= -ZLF_FSE;
1* calculate how the fold level of the following lines *1

1* must be changed: level += 1.
if old
= line->succ)->succ)
*1

1* Old line was fold start: *1
1
-=
(old->flags & ZLF_FOLD
1;
else i f line->flags & ZLF FOLD) =

1* Old line was fold end:-*I
1 += 1;
old->flags+l) & ZLF_FOLD
i f (1)
{
1* change fold level of following lines: *1

while (old->succ)
{

I old->flags + 1) & ZLF FOLD);
old = old->succ;
restoreZlinenptr();
printAll () ;
cursorOnText () ;
printypos () ;
1* Display lines for security: *1

1* First calculate Ypos:
*1
1 <= actuaIEditor->wch; 1++, zptr++)
if (*zptr == line)
591
AMIGA
C FOR
printLine(line,actualEditor->yoff + actualEditor->ch*l);
break;
1* Delete reference: *1
= 0;
1* Text was changed! *1
printFlags () ;
return (TRUE);
}
I*#ENDFD*!
I*#FOLD: savelfCursorMoved *1
1***************************************
* savelfCursorMoved:
* Saves line in buffer,
* if the cursor is moved.
***************************************1
void savelfcursorMoved()
{
register UWORD bufypos;

if (bufypos = actualEditor->bufypos)
if (actualEditor->ypos != bufypos)
DisplayBeep(NULL);
}
/*#ENDFD*/
I*#FOLD: undoLine
*1
1******************************************
* undoLine:
* Makes changes in actual line
* before cursor leaves line
******************************************1
void undoLine ()
{

if (z = actualEditor->actual)
{
actualEditor->buflen = 0;
z->flags &= -ZLF_USED;
1* Line output: *1
printLine(z,actualEditor->yoff
}
I*#ENDFD*I
I*#FOLD: getBufferPointer * I
1***************************************************
* getBufferPointer(pptr,pinc,prest):
592
ABACUS
ApPENDIX A VERSION
6 EDIT.C
Initialize pointer to actuaIEditor->buffer.

The biffer is initialized.
FALSE is returned if the buffer
cannot be initialized.
pptr A Pointer in buffer at cursor position.
pine A Offset to buffer start.
prestA buflen - inc.
*
*
*
*
*
*
***************************************************1
BOOL getBufferPointer(pptr,pinc, prest)
UBYTE
**pptr;
UWORD
*pinc,*prest;
{

register WORD inc, rest;
/* Buffer initialized: */
if (actuaIEditor->actual == NULL)
{
register UWORD wdy;

if (ZLINESPTR(wdy = actuaIEditor->wdy))
getLineForEdit(ZLINESPTR(wdy),ZLINESNR(wdy));
else
if (actuaIEditor->num_lines == 0)
{
/* create first line: */

if (z = newZlineUWORD) 1))
{
* UBYTE *) (z + 1)) = LF;

Add Head (&actuaIEditor->zlines,z);
ZLINESPTR(O) = Z;
ZLlNESNR(O) = (actuaIEditor->num_lines = 1):
getLineForEdit(ZLINESPTR(O),ZLINESNR(O));
printNumLines();
else
return (FALSE);
else
/* Cursor is not in line! */
return (FALSE);
/* Calculate pointer to cursor position: */
inc = actuaIEditor->xpos - 1;
ptr = actuaIEditor->buffer + inc:
rest= actuaIEditor->buflen - inc;
/* What if inc> buflen? */
if (inc < MAXWIDTH)
{
i f (rest < 0)
{
register UBYTE *p;

p = actuaIEditor->buffer + actuaIEditor->buflen:
actualEditor->buflen -= rest;
while (rest)
{
593
*p
AMIGA
= , ';
pH;
rest++;
}
else
return (FALSE);
*pptr z ptr;
*pinc = inc;
*prest= rest;
return (TRUE);
}
I*#ENDFD*I
I*#FOLD: deleteChar *1
1************************************
*
*
*
*
deleteChar:
Delet echaracter under the cursor.
False is returned if no more
characters.
************************************1
BOOL deleteChar()
{
UBYTE *ptr;
UWORD inc, rest;
register UWORD I:
return (FALSE);
if (rest)
{
p1 = ptr + 1;
p2 = ptr;
1 = --rest;
while (1)
(
*p2
*p1;
p1++;
p2++;
1--:
*p2
= actualEditor->buflastchar;
1* Lenght - 1 *1
actualEditor->buflen--;
printLine(actualEditor->actua1,actualEditor->yoff
return (TRUE);
}
I*#ENDFD*I
I*#FOLD: backspaceChar *1
1******************************
* backspaceChar:
* Delete character before cursor.
* False is returned when
594
A PPENDIX A VERSION 6 EDIT.('
ABACUS
cursor is at start of line.
******************************1
BOOL backspaceChar ()
{
UBYTE *ptr;
UWORD inc, rest;
register UWORD 1;
return (FALSE);
i f (inc)
{
{
pl
p2
ptr;
--ptr;
1 = rest;
while (1)
{
=
=
*p2 = *pl;
pl++;
p2++;
1--;
*p2
= actualEditor->buflastchar;
1* Lenght - 1 *1
actualEditor->buflen--;
inc--;
}
else
{
*--ptr = ';
inc--;
rest++;
cursorLeft () ;
printLine(actualEditor->actual,actualEditor->yoff
return (TRUE):
}
I*#ENDFD*I
I*#FOLD: insertLine *1
1**************************
* insertLine(c):
* Inserts new line behind
* actual line.
* FALSE returned if
* error encountered.
* c = line end (CR/LF).
**************************1
BOOL insertLine(c)
UBYTE
c;
{
static UBYTE buf[MAXWIDTH + l};
595
AMIGA
UBYTE *ptr;
WORD inc, rest, len;
UWORD autoindent;
register UWORD 1;
/* Insert new line only if actual line

has no fold markers!!! */
if z = actualEditor->actual) == NULL)
z = ZLINESPTR(actualEditor->wdy);
if (Iz I I ! (z->flags & ZLF FSE
{
/* get line in buffer: */
return (FALSE);
/* calculate auto indent: */
if (actualEditor->autoindent)
(
pI
actualEditor->buffer;
= inc;
while (1 && *pl
x=
')
II (*pl
TAB)
pl++;
1--;
}
i f (1)
auto indent = pI - actualEditor->buffer + 1;

else
/* line consisits only of spaces: */
autoindent = 1;
else
autoindent
= 1;
p2 = buf + (len = deconvertLine(buf,actualEditor->buffer,inc;
i f (*p2 = c)
len++;
= newZline(len
z->flags = (actualEditor->actual->flags
if (z
(
& -ZLF_USED);
/* save line that the cursor is in: */

pI = (UBYTE *) (z + 1);
p2 = buf;
1 = len;
while (1)
(
*pl. = *p2;
pl++;
p2++;
1-;
/* bind line: */
Insert (&actualEditor->zlines, z,actualEditor->actual->pred) ;
596
ABACUS
ApPENDIX A VERSION (; EDIT.C
/* move rest of buffer line in front: */
/* remember auto indent! */
pI = actualEditor->buffer;
1
= autoindent:
while (--1)
pl++;
/* Old contents remain! */
p2 = ptr;
1 = rest;
if
(actualEditor->autoindent)
1* Spaces at start cot off:
while (l && (*p2 = , '))
*/
p2++;
1-:
1* The write rest of the buffer start:

while (1)
*/
*pl
*p2;
pl++;
p2++;
1-;
pI - actualEditor->buffer;
/* rest filled with lastchar: */

1 = inc + 1 - autoindent:
while (1)
{
*pl = actualEditor->buflastchar:
pl++:
1--:
/* determine if new line is a fold marker: */
if (1 = getFoldInc(z))
{
register struct Zline *old:

register UWORD fold:
z->flags 1= ZLF_FSE:
/* change fold level of following lines: */
if old = z->succ)->succ)
while (old->succ)
{

I old->flags + 1) & ZLF_FOLD):
old = old->succ:
/* fit minfold and maxfold: */
if 1 > 0) II z->flags & ZLFYOLD) > 0))
{
fold = (z->flags & ZLF FOLD) + 1;

if (actualEditor->minf~ld > fold)
597
actualEditor->minfold = fold;
printFold () ;
}
if (actualEditor->maxfold < fold)

{
actualEditor->maxfold = fold;
printFold ( ) ;
ZLINESPTR(actualEditor->wdy) = Zi
actualEditor->wdy =
recalcTopOfWindow(actualEditor->wdy);
1* If cursor ontop line => zlinesptr[O]:

i f (actuaIEditor->wdy == 0)
*1
ZLINESPTR(O) = Z;
restoreZlinenptr();
1* redisplay window: *1
actuaIEditor->xpos = autoindent;
if (actuaIEditor->xpos <= actuaIEditor->leftpos)
{
1* If scrolling necessary: printAII *1

actualEditor->leftpos = actuaIEditor->xpos - 1;
,
ZLINESPTR(O) = ZLINESPTR(l);
restoreZlinenptr();
}
actuaIEditor->ypos = ZLINESNR(actualEditor->wdy);
printAll () ;
else
i f (1)
{
1* If fold marker: then redisplay window: *1

i~
(++actuaIEditor->wdy >= actuaIEditor->wch)

ZLINESPTR(O) = ZLINESPTR(I);
restoreZlinenptr();
printAlIO;
else
{
1* it was scrolled: */
{
1* Cursor was on last line: */

register UWORD nr;
598
ABACUS
APPENDIX A VERSION
6 EDIT.C
scrollUpUWORD) 1);
nr = (--actualEditor->wdy) - 1;
actualEditor->ypos = ZLINESNR(actualEditor->wdy);
printLine(ZLlNESPTR(nr),actualEditor->yoff
+ actualEditor->ch*nr);
else
{
/* Cursor near middle of window: */

register UWORD nr;
if (ZLlNESPTRnr
actualEditor->wdy) +
/* move rest of window down: *1

+ actualEditor->wdy*actualEditor->ch,
(LONG)actualEditor->yscr);
actualEditor->ypos
nr--;
printXpos () ;
printypos () ;
printNumLines();
printFlags () ;
return (TRUE);
else
return (FALSE);
else
{
cursorHome () ;
cursorDown () ;
return (FALSE);
}
/*#ENDFD*/
/*#FOLD: getLastWord */
/**************************************
* getLastWord(str,len)
* Returns pointer to last
* word in a string.
* str A String.
* len = buffer length.
**************************************/
UBYTE *getLastWord(str,len)
register UBYTE
*str;
599
register UWORD
AMIGA
len;
register UBYTE c,*lwb,*lwn,*end;

lwb
lwn
end
for
1* Last Word is space *1

= NULL;
1* Last Word normal */
= NULL;
/* pointer to string end */
= str + len - 1;
(str += len; len; len-)
if c
= *--str) == .)
lwb = str;
break;
else i f ( ! c >= 'A')
II (c >= 'a') &&
II (c >= '0') &&
II (c >= 192) II
i f (lwn = NULL)
lwn = str;
&& (c <= 'Z')

(c <= 'z' )
(c <= '9' )
(c == ,-
' )
i f (lwb == NULL)
i f (lwn)
lwb = lwn;
else
lwb = end;
else
i f (lwn)
if end - lwb) > MAXWIDTH I 5)
if end - lwn + 5) < (end - lwb
lwb = lwn;
return (lwb + 1);
}
/*#ENDFD*/
/*#FOLD: insertChar */
1******************************************
* insertChar (c)
* Inserts character in buffer.
* FALSE returned if character
* cannot be inserted.
* c= character.
******************************************/
BOOL insertChar(c)
register UBYTE
c;
UBYTE *ptr;
WORD inc, rest;
WORD xadd = 1;
return (FALSE);
if c == TAB) && (actualEditor->tabs
{
{
/* Insert spaces: */
register UWORD num,l;
600
ABACUS
p1 = actuaIEditor->tabstring + actuaIEditor->xpos - 1;
1
= MAXWIDTH - actuaIEditor->xpos;
num = 1;
while (1 && (*++pl
{
xadd++;
1--;
num++;

if (actuaIEditor->buflen + num > MAXWIDTH)
{
xadd -= num + actualEditor->buflen - MAXWIDTH;

num = MAXWIDTH - actuaIEditor->buflen;
1* move rest of line: */

pI = ptr + rest;
p2 = pI + num;
1
= rest;
while (1)
(
*--p2
*--pl;
1-;
)
pI = ptr;
1 = num;
while (1)
(
*pl =
pl++;
";
1--;
actuaIEditor->buflen += num;
rest += num;
)
else
{
1* move cursor only: *1

register UWORD max;
tab = actuaIEditor->tabstring + actuaIEditor->xpos - 1;
max = MAXWIDTH - actuaIEditor->xpos;
while (max && (*++tab
{
xadd++;
max--;
}
else
{

if
&&
actuaIEditor->skipblanks)
(actuaIEditor->xpos >= actuaIEditor->rm
601
1* Word-Wrap when SkipBlanks == 0 *1

actualEditor->xpos = getLastWord(actualEditor->buffer,
actualEditor->buflen)
- actualEditor->buffer + 1;
i f (! insertLine UBYTE) 0
return (FALSE);
if (! saveLine(actualEditor->actua1
DisplayBeep(NULL);
cursorEnd () ;
return (FALSE);
if (actualEditor->buflen < MAXWIDTH)
{

register UWORD 1;
p1
p2
= ptr + rest;
= pI;
1 = rest;
while (1)
{
*p2 = *--p1;
p2--;
1--;
*ptr
c;
/* Lenght + 1 */
rest++;
else
return (FALSE);
else
{
*ptr = c;
i f (rest =
0)
rest = 1;
1* move cursor: */
while (xadd)
{
i f (! cursorRight (
xadd--;
break;
1* Line output: *1
printLine(actualEditor->actual,
actualEditor->yoff + actualEditor->ch*actualEditor->wdy);
return (TRUE);
602
ABACUS
ApPENDIX A VERSION
EDIT.C
I*#ENDFD*I
deleteLine *1
1****************************
I*#FOLD:
*
*
*
*
deleteLine:
delete the line,
that the cusor is on.
FALSE returned if
cannot be deleted
****************************1
BOOL deleteLine()
{
register struct Zline *line,*next,*previ

register UWORD 1 = 0:
UWORD nextnr,prevnr:
if (line = ZLlNESPTR(actualEditor->wdy
{
1* erase reference to lien: *1

if (line == actualEditor->actual)
{
actualEditor->actual = NULL:
actualEditor->bufypos
0:
actualEditor->buflen = Oi
actuaIEditor->changed = Ii
actuaIEditor->num_lines--:
1* Reconstruct folding: *1
1* change fold level of following lines: *1

if z = line->succ)->succ)
{

1* line was fold start: *1
1
(z->flags & ZLF_FOLD
-Ii
else if line->flags & ZLF FOLD)

1* line was fold end: *1 -
z->flags+l)
&
ZLF]OLD
1 = 1;
else
1
= 0;
i f (l)
while (z->succ)
{
z->flags = (z->flags & -ZLF_FOLD)

I z->flags + 1) & ZLF_FOLD)i
z = z->succ;
1* Note
prevnr=
prev '"'
next '"'
nextnr=
next and previous line: */

ZLINESNR(actualEditor->wdY)i
prevLine (line, &prevnr) i
ZLINESNR(actuaIEditor->wdy):
603
1* Security check: *1
while prev == NULL) && (next
{
==
NULL) && (actualEditor->minfold
1 = 1;
1* Window completely output *1
actualEditor->minfold--;
printFold 0 ;
prevnr=
prey =
next =
nextnr=
ZLlNESNR(actualEditor->wdy);
prevLine (line, &prevnr);
ZLlNESNR(actualEditor->wdy):
if (prev II next) break:

if (actualEditor->minfold) continue:
if next = actualEditor->zlines.head)->succ
next = NULL:
else
nextnr = 1;
~=
NULL)
break;
1* delete line: *1
deleteZline(line);
1* pointer to actual line reset for recalc: *1
i f (next)
{
ZLlNESNR(actualEditor->wdy)
= next;
= nextnr;
else
(
ZLINESNR(actualEditor->wdy)
prey;
= prevnr;
1* Restore window contents: *1

i f (1)
(
1* Redisplay window in each case: *1

restoreZlinenptr();
printAllO;
cursorOnText 0 ;
}
else
(
1* with scrolling: *1
register UWORD oldwdy;
oldwdy = actualEditor->wdy;
restoreZlinenptr();
i f (next)
{
1* push up rest of window: *1

OL, (LONG)actualEditor->ch,
+ actualEditor->wdy*actualEditor->ch
604
ApPENDIX A VERSION 6 EDIT.C
ABACUS
(LONG)actualEditor->yscr):
printLine(ZLINESPTR(actualEditor->wch - 1),
actualEditor->yoff
+ actualEditor->ch*(actualEditor->wch - 1:
else
if (oldwdy == actualEditor->wdy)
{
/* scroll top section of window down: *1

(LONG)actualEditor->yoff,
+ (actualEditor->wdy + 1)
*actualEditor->ch - 1);
printLine(ZLlNESPTR(O),actualEditor->yoff);
}
else
printLine(ZLINESPTR(oldwdy),actualEditor->yoff
+ actualEditor->ch*oldwdy):
cursorOnText () :
}
printXpos () ;
printYpos 0 ;
printNumLines():
printFlags 0 ;
return (TRUE);
}
else
return (FALSE);
}
/*#ENDFD*/
/*#FOLD: initFolding */
/**********************************
* initFolding:
* calculate the fold level of all
* lines from text start.
**********************************/
void initFolding()
{

register UWORD level,l;
z = actualEditor->zlines.head;
level = 0:
while (z->succ)
(
z->flags = (z->flags & -ZLF_FOLD) I (level & ZLF_FOLD):

{
z->flags 1= ZLF_FSE;
level += 1;
z =
Z->SllCC;
/*#ENDFD*/
605
AMIGA
Version 6 Cursor.c
/**************************
* Module: Cursor.c
* Moving the cursors.
***************************/
/*#FOLD: Includes */
/************
**
Includes:
*************/
#inc1ude <exec/types.h>
#inc1ude <intuition/intuition.h>
#inc1ude "src/Editor.h"
/*#ENDFD*/
/*#FOLD: Defines */
/***********
* Defines:
***********/
#define
#define
#define
#define
#define
#define
#define
#define
#define
CSI Ox9B
CUU 'A'
CUD 'B'
CUF 'C'
CUB 'D'
SU 's'
SD 'T'
CHOME 'A'
CEND '@'
#define ESC 27
#define UNDO '?'
#define CFOLD 6
#define CENDF 5
#define DELLINE 2
#define DEL 127
#define BS 8
#define TAB 9
#define LF 10
#define CR 13
#define REPCOM 7
/*#ENDFD*/
/**********************
**********************/
void SetDrMd(),RectFill(),printAl1(),Disp1ayBeep(),undoLine();
606
ApPENDIX A VERSION
ABACUS
6 CURSOR.C
void ScroIIRaster(),cursorOnText(),printXpos(),printYpos();
void printFold(),printFlags(),cursorOnText();
struct Zline *nextLine(),*prevLine();
BooL insertChar(),deleteChar(),backspaceChar(),insertLine(),deleteLine();
BOOL ActivateGadget();
UBYTE *executeCommand();
/*#ENDFD*/
/*********************
**********************/
extern awORD SplitDec;
extern struct Gadget G Command;
/*#ENDFD*/
/***************
**
*
Functions: *
***************/
/*1IFOLD: restoreZlinenptr */
/************************************
* restoreZlinenptr:
*
*
*
*
*
Restore the zlinesptr/nr field

for all lines in window plus
the next line.
The first line must be
correct!
*************************************/
void restoreZlinenptr()
{
register UWORD n;
register struct Zline **zptr,*z;
register awORD *pnr;
awORD znr;
for (n
actuaIEditor->wch, zptr = actuaIEditor->zlinesptr,

pnr = actuaIEditor->zlinesnr,
z
*zptr++, znr = *pnr++; n; n--)
*zptr++
*pnr++
(z = nextLine(z,&znr;
znr;
/*1IENDFD*/
/*#FOLD: Cursor */
1**********************************
* Cursor:
* Set or erase the cursor

* at the actual position.
*
607
AMIGA
**********************************1
void Cursor 0
{
registerstruct RastPort *rp;

register LONG x,y;
register LONG cw,ch;
SetDrMd(rp = actualEditor->rp,COMPLEMENT);
x: (actualEditor->xpos - actualEditor->leftpos - 1)
*(cw = actualEditor->cw) + actualEditor->xoff;
y = actualEditor->wdy
*(ch = actualEditor->ch) + actualEditor->yoff;
RectFill(rp,x,y,x + cw-1,y + ch-1);
SetDrMd(rp,JAM2);
}
I*#ENDFD*I
I*#FOLD: scrollRight *1
1*******************************
* scrollRight(nurn):

* num characters to the right.
**
nurn = number of characters.
*******************************1
void scrollRight(nurn)
register UWORD
num;
{

actualEditor->leftpos -= num;
if (nurn >= actualEditor->wcw)
printAll () ;
else
{
1* Scrolling: *1
-num * (LONG)actualEditor->cw,OL,
1* now list printAll(): *1
wcw = actualEditor->wcw;
actualEditor->wcw = nurn + 1;
SplitDec = 1;
y
actualEditor->yoff;
db
actualEditor->ch;
1* Output line: *1
608
ABACUS
ApPENDIX A VERSION
CURSOR.C
z++, Y += chI
printLine(*z,y);
SplitDec = 0;
1* Display right border beside last character *1

xoff
= actualEditor->xoff;
actualEditor->wcw
= 2;
actualEditor->leftpos += (y = wcw - 2);
actualEditor->xoff
+= actualEditor->cw * y;
= actualEditor->yoff;
1* Output line: *1
z++, y += chI
printLine(*z,y);
actualEditor->wcw
= wcw,
actualEditor->leftpos= leftpos;
}
l*lIENDFD*1
l*lIFOLD: scrollLeft *1
1*******************************
* scrollLeft(num):
**

* num characters left.
**
num = number of characters.
*******************************1
void scrollLeft(num)
register UWORD num;
{

actuaIEditor->leftpos += num;
if (num >= actualEditor->wcw)
printAll () ;
else
{
1* Scrolling: */
num * (LONG)actualEditor->cw,OL,
1* now list printAll(): *1
wcw
actualEditor->wcw;
xoff
= actualEditor->xoff;
actuaIEditor->wcw
= num + 1;
609
AMIGA
aetuaIEditor->leftpos +~ (y ~ wew - nurn - 1);

aetuaIEditor->xoff
+~ aetuaIEditor->cw * y;
= aetualEditor->yoff;
aetuaIEditor->eh;
count = aetuaIEditor->wch;
ch
1* Output line: *1
for (z ~ actuaIEditor->zlinesptr; count-- && (*z 1= NULL);
z++, y += chI
printLine(*z,y);
actualEditor->wcw
= wcw;
actuaIEditor->leftpos= leftpos;
}
I*#ENDFD*I
I*#FOLD: scrollUp *1
1*******************************
* serollUp(nurn):
*
*
*
*

nurn lines up.
Nurn lines must
exist!
**
nurn = Number of the line.

********************************1
void scroll Up (nurn)
register UWORD num;
(
register struet Zline *z,**zptr,**zold;

register UWORD n,weh = actuaIEditor->weh,y;
i f (nurn >= weh)
{
1* Display window without Scrolling *1

1* First look for top line: *1
n = weh;
i f (nurn > wch)

(
z = ZLlNESPTR(n);
znr = ZLINESNR(n);
while (++n < nurn)
z = nextLine(z,&znr);
else
(
z = ZLINESPTR(n - 1);
znr = ZLINESNR(n - 1);
1* z points before the top line of the window *1

for (n = 0, zptr = aetuaIEditor->zlinesptr,
pnr = actualEditor->zlinesnr; n <= weh; n++)
*zptr++
610
= (z
= nextLine(z,&znr;
ABACUS
ApPENDIX A VERSION
*pnr++
CURSOR.C
znr;
1* Now redisplay the window: *1

printAll () ;
else
{
1* Scrolling: *1
OL,hum * (LONG)actualEditor->eh,
(LONG)actualEditor->xoff, (LONG)aetualEditor->yoff,
(LONG)actualEditor->xser, (LONG)aetualEditor->yscr);
1* zlinesptr/nr recalculated: *1
zptr = actualEditor->zlinesptr;
zold = &(ZLINESPTR(num;
pnr = aetualEditor->zlinesnr;
oldnr = &(ZLINESNR(num;
for (n = 0; n <= weh - num; n++)
{
*zptr++
*pnr++
*zold++;
*oldnr++;
z = *--zold;
zold = zptr - 1;
znr = *--oldnr;
oldnr = pnr - 1;
while (n <= wch)
{
1* mark for output! *1
*pnr++ = znr;
nt+;
1* Output new line: *1

for (n = num, y = aetualEditor->yoff+(wch-num)*actualEditor->ch;
(n && (*zold != NULL; n--, zold++, y += actualEditor->eh)
printLine(*zold,y);
}
1*ItENDFD*1
I*#FOLD: scrollDown *1
1*******************************
* scrollDown(num):

* nurn lines down.
* Nurn lines must
* exist!
* nurn
number of lines.
*******************************1
void scrollDown (nurn)

register UWORD num;
{

register UWORD n,wch = actualEditor->wch,y;
611
AMIGA
if (nurn >= wch)
{
/* Display window without Scrolling */

/* search backwards fo rtop line: */
for (z = ZLINESPTR(O), znr = ZLINESNR(O), n = num - wch; n; n--)
/* z points to first line after bottom window edge. */
for (n = 0, zptr = &(ZLINESPTR(wch)), pnr = &(ZLINESNR(wch));
n <= wch; n++)
*zptr-- = z;
*pnr-- = znr;
/* Now redisplay the window: */
printAll () ;
else
{
/* Scrolling: */
OL,-nurn * (LONG)actualEditor->ch,
/* zlinesptr recalculated: */
zptr = &(ZLINESPTR(wch));
zold = &(ZLINESPTR(wch - num));
pnr = &(ZLINESNR(wch));
oldnr = &(ZLINESNR(wch - num));
for (n = 0; n <= wch - num; n++)
{
*zptr-*pnr--
*zold--;
*oldnr--;
for (z = *++zold, znr
= *++oldnr,
num; n; n--)
*zptr-- = (z = prevLine(z,&znr));
*pnr-- = znr;
/* OUtput new line: */
for (n = nurn, y = actualEditor->yoff; (n && (*zold != NULL));
n--, zold++, y += actuaIEditor->ch)
printLine(*zold,y);
}
/*#ENDFD*/
/*#FOLD: cursorLeft */
/****************************************
* cursor1eft:
* moce cursor one position

* left, False returned if cursor,
* is in first column.
612
ApPENDIX A VERSION
ABACUS
CURSOR.C
****************************************1
BOOL cursorLeft()
{
if (actuaIEditor->xpos > 1)
{
actuaIEditor->xpos--;
if (actuaIEditor->xpos <z actualEditor->leftpos)
scroIIRightUWORO)l);
printXpos () ;
return (TRUE);
else
retum (FALSE);
}
I*I!ENDFD*I
I*I!FOLD: cursorRight *1
1*****************************************
** cursorRight:
** moce cursor one position
* right, False returned if
*
cursor,
is in last column.
******************************************1
BOOL cursorRight ()
{
if (actuaIEditor->xpos < MAXWIDTH)

{
actuaIEditor->xpos++;
if (actuaIEditor->xpos
>= (actuaIEditor->leftpos + actuaIEditor->wcw
scrollLeft ( (UWORD) 1) ;
printXpos () ;
return (TRUE);
}
else
retum (FALSE);
}
I*I!ENDFD*I
I*I!FOLD: cursorUp *1
1*****************************************
**
cursorUp:
**
*
*
Move cursor one line up.

False retumed when
on first line.
******************************************1
BOOL cursorUp ()
{
UWORO help;
if (actualEditor->wdy)
else
613
AMIGA
if (prevLine(ZLlNESPTR(O),&help
scroIIDownUWORD)I);
else
return (FALSE);
actuaIEditor->ypos
printYpos 0 ;
= ZLlNESNR(actuaIEditor->wdy);
retum (TRUE);
}
/*#ENDFD*/
/*#FOLD: cursorDown *1
/******************************************
* cursorDown:
**
Move cursor one line down.

* on last line.
******************************************/
BOOL cursorDown()
{
if (ZLINESPTR(actuaIEditor->wdy + 1
{

{
scroIIUpUWORD)I);
}
actuaIEditor->ypos
printYpos () ;
= ZLINESNR(actualEditor->wdy);
return (TRUE);
else
return (FALSE);
}
/*#ENDFD*I
I*#FOLD: halfPageUp *1
/**********************************
* halfPageUp:
**
Move cursor on half page

* up. The text is scrolled
* half page down
**
False retumed when

* in first line.
***********************************/
BOOL halfPageUp()
{

UWORD help;
max = actuaIEditor->wch 1 2;
for (n = 0, z = ZLlNESPTR(O); n < max; n++)
614
ABACUS
ApPENDIX A VERSION
CURSOR.C
if (! (z = prevLine(z,&help)
break;
i f (n)
{
scroIIDown(n);
actuaIEditor->ypos
printYpos () ;
return (TRUE);
}
else
return (FALSE);
}
l*flENDFD*1
l*flFOLD: halfPageDown *1
1**********************************
**
halfpageDown:
**
Move cursor on half page

text is scrolled
down.
* down. The
* half page

* on last line.
**********************************1
BOOL halfPageDown ()
{

UWORD help;
max = actuaIEditor->wch 1 2;
for (n = 0, z = ZLlNESPTR(actuaIEditor->wch - 1); n < max; n++)
if (! (z - nextLine(z,&help)
break;
i f (n)
{
scroll Up (n) ;
actuaIEditor->ypos - ZLlNESNR(actuaIEditor->wdy);
printYpos();
return (TRUE);
}
else
return (FALSE);
}
I*#ENDFD*I
I*#FOLD: cursorHome *1
1*************************************
*
**
cursorHome:
move cursor to first line.
*************************************1
BOOL cursorHome()
615
AMIGA
actuaIEditor->xpos z 1;
if (actuaIEditor->leftpos)
scroIIRight(actuaIEditor->leftpos);
printXpos 0 ;
return (TRUE);
}
l*flENDFD*1
I*#FOLD: cursorEnd *1
/***************************************
* cursorEnd:
** Set the cursor on the last line.

****************************************1
BOOL cursorEnd()
{

register awORD len,pos;
if (actuaIEditor->bufypos)
pos = actuaIEditor->buflen + 1;
else
if (z = ZLlNESPTR(actuaIEditor->wdy))
{
/* remove CR from lenght: *1

len = z->len;
ptr = UBYTE *) (z + 1)) + len - 1;
i f (len)
i f (*ptr = CR)
1en--;
else if (*ptr = : LF)
i f (--len)
if (*--ptr == CR)
len--;
if (actuaIEditor->tabs)
{
ptr
tab
pos
= UBYTE *) (z + 1));
actualEditor->tabstring;
= 1;
=
while (len--) && (pos < MAXWIDTH))

if (*ptr++ == TAB)
<b
{
tab++;
pos++;
} while *tab) && (pos < MAXWIDTH));
else
{
tab++;
pos++;
}
else
pos
616
len + 1;
ABACUS
ApPENDIX A VERSION
6 CURSOR.C
else
return (FALSE);
actualEditor->xpos = pos;
if (pos >= (actualEditor->leftpos + actualEditor->wcw
scrollLeft(pos + 1 - (actualEditor->leftpos
+ actualEditor->wcw;
else if (pos <= actualEditor->leftpos)
scrollRight(actualEditor->leftpos + 1 - pos);
printXpos () ;
return (TRUE);
}
/*#ENDFD*/
/*#FOLD: recalcTopOfWindow */
/***********************************
**
recalcTopOfWindow(y)
* Calculate the cursor position

* in line y of
* line in window.
* Get new Y-Position.
**
y = Number of line [O,wdy[

************************************/
UWORD recalcTopOfWindow(y)
register UWORD
y;
{
register struct Zline *z,*zp;

register UWORD znrp;
UWORD znr,oldy = y;
= ZLlNESPTR(y);
znr = ZLINESNR(y);
while (y)
{
2P
= z;
znrp = znr;
if z = prevLine (z, &znr
break;
== NULL)
y--;
i f (y)
{
ZLINESPTR(O)
ZLINESNR(O)
= zp;
= znrp;
else
{
ZLlNESPTR (0) = z;
ZLlNESNR(O) = znr;
return (oldy - y);
}
/*#ENDFD*/
617
I*#FOLD: eursorTop *1
1*****************************
*
* eursorTop:
*
* Set Cursor to to of
* first line.
Text
*****************************1
BOOL cursorTop()
{
register struet Zline *z;

UWORD n;
register UWORD cnt;
z = ZLINESPTR(O);
n = ZLINESNR(O);
ent = 0;
while (z = prevLine(z,&n
ent++;
i f (ent)
serollDown(ent);
aetualEditor->ypos
printypos 0 ;
ZLINESNR(aetualEditor->wdy
0);
retum (TRUE);
}
I*#ENDFD*I
I*#FOLD: cursorBottom *1
1*****************************
*
*
eursorBottom:
*
* Set Cursor to bottom of text
* last line.
*****************************1
BOOL cursorBottom()
{
register struet Zline *z;

UWORD n;
register UWORD weh,ent;
if
(z
:.
ZLlNESPTR (weh
aetualEditor->weh
- 1
n
ZLlNESNR(weh);
ent = 0;
while (z = nextLine(z,&n
ent++;
i f (ent)
scrollUp(ent);
aetualEditor->ypos
ZLlNESNR(aetualEditor->wdy
= weh);
ZLlNESNR(aetualEditor->wdy
= weh);
else
{
aetualEditor->ypos
cursorOnText 0 ;
618
ABACUS
ApPENDIX A VERSION
CURSOR.C
printYpos () ;
return (TRUE);
}
/*#ENDFD*/
/*#FOLD: enterFold */
/***************************
**
enterFold:
**
Step one fold in.
***************************/
void enterFold()
{

if (actuaIEditor->maxfold < ZLF_FOLD)
{
if (z = ZLINESPTR(actuaIEditor->wdy
if z = z->succ)->succ)
/* If there is a following line: */
if (ZLINESPTR(actuaIEditor->wdy)->flags & ZLF_FOLD)
&& z->flags & ZLF_FOLD) == actuaIEditor->maxfold
/* When this begins a new fold: */
actuaIEditor->minfold = actuaIEditor->maxfold;
ZLINESPTR(O) = z;
ZLINESNR(O)
actuaIEditor->wdy = 0;
actuaIEditor->wdy = recalcTopOfWindow(actuaIEditor->wdy);
restoreZlinenptr();
printAlI () ;
actualEditor->ypos = ZLINESNR(actuaIEditor->wdy);
printYpos () ;
printFold () ;
}
/*#ENDFD*/
/*#FOLD: exitFold */
/*****************************
* exitFold:
**
Step one fold out.
*****************************/
void exitFoldO
{
register UWORD wdy;

if (actualEditor->maxfold)
{
actuaIEditor->maxfold--;
619
AMIGA
if (actualEditor->minfold > actualEditor->maxfold)

actualEditor->minfold = actualEditor->maxfold;
wdy = actualEditor->wdy;
if (ZLINESPTR(wdy) ~ NULL)
actualEditor->wdy = wdy = 0;
if (ZLINESPTR(wdy
if (ZLINESPTR(wdy)->flags & ZLF FOLD) > actualEditor->maxfold)
I I ZLlNESPTR(wdy)->flags & ZLF FOLD) < actualEditor->minfold
ZLINESPTR(wdy) = prevLine(ZLlNESPTR(wdy),&(ZLINESNR(wdy);
actualEditor->wdy = recalcTopOfWindow(wdy);
restoreZlinenptr();
printAll () ;
cursorOnText () ;
printYpos () ;
printFold () ;
}
/*#ENDFD*/
/*#FOLD: handleKeys */
/*******************************
* handleKeys(buf,len):
*
*
*
*
Handles keyboard messages.

Give FALSE, if
the Editor commands
elset TRUE.
* buf
* len
buffer with character.

number of characters.
*******************************1
BOOL handleKeys(buf,len)
register WORD
len;
{

while (len> 0)
{
first = *buf;
buf++i
len--;
if first == CSI) && (len> 0
{
len-;
switch (*buf++)
{
case CUU:
cursorUp () ;
break;
case CUD:
cursorDown 0 ;
break;
case CUF:
cursorRight () ;
break;
case CUB:
620
ABACUS
ApPENDIX A VERSION
6 CURSOR.C
cursorLeft 0 ;
break;
case sU:
halfpageDown 0 ;
break;
case SD:
halfPageUpO;
break;
case ':
i f (len> 0)
{
len--;
switch (*buf++)
{
case CHOME:
cursorHome 0 ;
break;
case CEND:
cursorEnd 0 ;
break;
default:
i f (! insertChar ( (UBYTE) CSI) )
DisplayBeep(NULL);
buf -= 2;
len += 2;
}
break;
}
else
goto noCSI;
case UNDO:
if len> 0) && (*buf
==
._.
buf++;
len--;
undoLineO;
break;
}
/* Else: default */
default:
noCSI:
/* No command-Sequence! */
DisplayBeep(NULL);
buf-;
len++;
break;
/* switch (first)
*1
1
else
switch (first)
{
case ESC:
1* Rest of buf. transferred:
*1
register UBYTE *sibuf;

register UWORD cnt;
sibuf = actualEditor->gc SIBuffer;
while (len)
{
621
AMIGA
*sibuf++
len--;
= *buf++;
*sibuf = 0;
cnt = (sibuf - actuaIEditor->gc SIBuffer);
actuaIEditor->gc GadgetSI.Bufferpos = cnt:
actuaIEditor->gc-GadgetSI.NumChars = cnt:
actuaIEditor->gc=GadgetSI.DispPos
= 0;
if (!ActivateGadget(&(actuaIEditor->G_Command),
actuaIEditor->window,NULL
DisplayBeep(NULL):
break;
case REPCOM:
{

i f (ptr == -IL)
return (FALSE);
else
{
if
(ptr >= Ox80000000)

ptr = NULL - ptr:
1* Cursor in StringGadget in error set *1

pos = ptr - actuaIEditor->gc SIBuffer:
actualEditor->gc GadgetSI.BufferPos = pos;
if (pos < actuaIEditor->gc GadgetSI.DispCount)
actualEditor->gc GadgetSr.Oisppos = 0:
else
actuaIEditor->gc GadgetSI.DispPos = os
- actuaIEditor->gc_GadgetSI DispCount I 2:
ActivateGadget(&(actuaIEditor->G
C~mmand),
actuaIEditor->winciow~NULL);
DisplayBeep(NULL);
/* return now: */
return (TRUE):
break;
case CFOLD:
enterFold 0 ;
break:
case CENDF:
exi tFold () :
break:
case DELLlNE:
if (! deleteLine(
DisplayBeep(NULL);
break:
case DEL:
if (! deleteChar(
DisplayBeep(NULL);
622
ABACUS
ApPENDIX A VERSION
6 CURSOR.C
break;
case BS:
if (1 backspaceChar(
DisplayBeep(NULL);
break;
case LF:
case CR:
if (1 insertLine((UBYTE) LF
DisplayBeep(NULL);
break;
default:
if (1 insertChar(first
DisplayBeep(NULL);
/* switch (first) *1
/* while (len) *1
return (TRUE);
}
/*#ENDFD*I
623
AMIGA
Version 6 Output.c
/***********************************
** Module: Output
** Output text in a window.
************************************/
/************
**
Includes:
*************/
/*#ENDFD*/
/*#FOLD: Defines */
/***********
* Defines:
***********1
#define CR 13
#define LF 10
#define TAB 9
/*#ENDFD*/
/**********************
**********************1
void SetAPen(),RectFill(),Text(),Move(),SetFont();
/*#ENDFD*/
/*#FOLD: External Variables *1
/*********************
*********************1

extern struct TextFont *infoFont;
/*#ENDFD*/
/*#FOLD: Global Variables */
1*********************
**
*
Global Variables:
*********************1
624
ABACUS
ApPENDIX A VERSION
UWORD Split Dec

/*#ENDFD*/
OUTPUT.C
0;
/***************
**
Functions: *
***************/
/*#FOLD: initWindowSize *1
1**********************************
**
*
initWindowSize(ed)
* Initializes the width/height

* statement of the actual windows.
* Restores the zlineptr field
*
*
* ed
Editor structure.
**********************************/
void initWindowSize
(ed)
{
register struct Window *w;

register UWORD n,newh;
w
rp
ed->window;
ed->rp:
~
~
ed->xoff
ed->yoff
(UWORD)w->BorderLeft:
(UWORD)w->BorderTop;
ed->cw
ed->ch
(rp->TxWidth)? rp->TxWidth
8:
(rp->TxHeight)? rp->TxHeight : 8;
ed->wcw
newh
(w->Width - ed->xoff - w->BorderRight) 1 ed->cw;

(w->Height- ed->yoff - w->BorderBottom) 1 ed->ch;
ed->xrl = ed->xoff + ed->wcw*ed->cw;

ed->xrr ~ w->Width - w->BorderRight - 1;
ed->yru = w->Height - w->BorderBottom - 1;
ed->hl ~ rp->TxBaseline:
/* Position for Infoline: */
register ULONG xinc,xsize;
xinc
xsize~
w->Width + ed->G_Info.LeftEdge + ed->gi_IText.LeftEdge - 1;

infoFont->tf_XSize;
ed->ip xpos
= xinc + INFO XPOS INC * xsize;
ed->ip=ypos
= xinc + INFO-YPOS-INC * xsize;
ed->ip numzlines = xinc + INFO NUMOFLINES INC * xsize;
ed->ip-minfold
= xinc + INFO-MINFOLD INC * xsize;
ed->ip-maxfold
~ xinc + INFO-MAXFOLD-INC * xsize;
ed->ip-flags
= xinc + INFO FLAGS INC-* xsize;
ed->ip=topedge
= w->Height-+ ed->G_Info.TopEdge
625
AMIGA
C FOR
+ ed->gi IText.TopEdge
+ (ULONG)infoFont->tf_Baseline - 1;
/* Now restore the zlineptr field: */
{

UWORD znr,*pnr;
if (newh < ed->wch)
{
/* Pointer to Null set: */

*zptr++ = NULL;
else
{

z
*zptr++;
pnr = &(ed->zlinesnr[ed->wch]);
znr
= *pnr++;
{
*zptr++ - (z - nextLine(z,&znr;
*pnr++ = znr;
ed->wch
newh;
/* Top/bottom border for scrolling: */

ed->xscr = ed->xoff + ed->wcw*ed->cw - 1;
ed->yscr = ed->yoff + ed->wch*ed->ch - 1;
}
/*#ENDFD*/
/*#FOLD: convertLineForPrint */
/******************************************
**
convertLineForPrint(line,len,w,buf,lc)
* Convert line so that this can be

* displayed simply.
* Tabs converted into spaces
**
*
*
*
*
*
line ,.. Zline.

len = the length.
w = maximum wiidth.
buf Abuffer, in which the converted line
is stored.
lc ,.. to Variable.
******************************************/
UWORD convertLineForPrint(line,len,w,buf,lc)
UBYTE
*line;
regi ster WORD
len;
w;
register UWORD
register UBYTE
*buf;
626
ABACUS
ApPENDIX A VERSION
UBYTE
OUTPUT.C
*lc;

register UWORD 1 = 1, skip
UBYTE last char;
UWORD real Len;
= actualEditor->leftpos;
1* determine if ends with CR or LF: *1

i f (len)
{
tab
= line
+ len - 1;
CR)
i f (*tab =
{
len-;
lastchar
, ,.,
else i f (*tab =
LF)
len--;
lastchar ~ , ';
if len) && (*--tab
len--;
else
lastchar
CR
= 183;
else
last char
183;
{
1* line convert: *1
while len--) && (1 < wI)
if *buf ~ *line++) = TAB)
<b
{
tab++;
(skip)
skip--;
else
if
1++;
*buf++ =
, ,.,
while *tab) && (1 < wI);

else
{
tab++;
i f (skip)
skip--;
else
{
1++;
buf++;
else
{
1* line copied with no tabs: *1

while (len--) && (l < w
627
i f (skip)
skip--;
else
{
1++;
*buf++
*line;
line++;
1* determine if the line is not completely converted: *1

i f (len >= 0)
1astchar = 183;
1* save lenght of line *1

realLen = 1 - 1;
1* line filled from last character to end: *1
while (1++ <= w)
*buf++ = last char;
*lc = lastchar;
return (realLen);
}
l*ilENDFD*1
l*ilFOLD: printAt *1
1************************************
** printAt(buf,len,x,y,fgpen)
**Prive character string buf a
position
* (x,y) in a window. The character* string is changed!
**
*
*
*
*
buf A character sring.

len = the length.
x = Pixel-X-Position.
y = Pixel-Y-Position.
fgpen = Foreground color (Folding)
************************************1
void printAt (buf,len,x,y,fgpen)

WORD
len;
register UWORD
x,y;
ULONG
fgpen;
{
struct RastPort *rp = actualEditor->rpi

register UWORD cw = actualEditor->cw;
register UWORD x2;
UWORD y2 = y + actualEditor->ch - 1;
UBYTE *ptr;
ULONG 1;
while (len > 0)

if (*buf = , ')
{
1* output spaoes: *1
x2
628
x - 1;
ABACUS
ApPENDIX A VERSION
while *buf
== , ')
OUTPUT.C
&& len--)
buf++;
x2 += cw;
SetAPen(rp,BGPEN);
RectFill(rp, (ULONG)x, (ULONG)y, (ULONG)x2, (ULONG)y2);
SetAPen (rp, fgpen);
x
++x2;
else
{
/* Also output text: */

Move (rp, (ULONG)x, (ULONG)y + actualEditor->bl);
ptr = buf;
if (CONTROLCODE(*buf
(
SetAPen(rp,CTRLPEN);
while (CONTROLCODE(*buf) && len--)
*buf++ 1= 64;
1 = buf - ptr;
Text (rp,ptr, l);
SetAPen(rp,fgpen);
}
else
{
while (!CONTROLCODE(*buf) && (*buf != , ') && len--)

buf++;
1 = buf - ptr;
Text (rp,ptr, l);
x += cw*l;
)
/*iENDFD*/
/*#FOLD: printLine */
/************************************
* printLine(line,y)
Displays Zline at pixel position

*y on the screen.
* line A Zline structure.

* y = pixel positon.
************************************/
void printLine
(line,y)
register UWORD
y;
{
static UBYTE buf[MAXWIDTHj;

i f (line)
{
629

register UWORD w;
register ULONG pen;
UBYTE lc;
1* Test, if line even edited: *1

if (line->flags & ZLF USED)
{
1* Buffer copies to buf,since printAt changed this! *1
if (MAXWIDTH > actuaIEditor->leftpos)
{
w = MAXWIDTH - actuaIEditor->leftpos;
p2 = actuaIEditor->buffer + actuaIEditor->leftpos;
pl = buf;
while (w)
{
*pl = *p2;
p1++;
p2++;
w--;
}
else
pl = buf;
1* fill rest from lastchar : *1

p2 = buf + (w = actuaIEditor->wcw);
lc = actuaIEditor->buflastchar;
while (pl < p2)
{
*pl = lc;
p1++;
else
convertLineForPrint(line+l,line->len,
w = actuaIEditor->wcw,buf,&lc);
1* If start/end marker of fold => FOLDPEN *1

if z = line->succ)->succ)
{
if z->flags & ZLF FOLD) <= actuaIEditor->maxfold)
pen = FOLDPEN;
else
pen = FGPEN;
else
{
pen = FOLDPEN;
1* old write color first again: *1
630
ABACUS
ApPENDIX A VERSION
OUTPUT.C
if (pen != FGPEN)
else
printAt(buf,w - SplitDec,actualEditor->xoff,y,FGPEN);
else
{
1* if no line
~> erase in window: *1

struct RastPort *rp = actuaIEditor->rp;
SetAPen (rp, BGPEN) ;

RectFill(rp, (ULONG)actuaIEditor->xoff.
(ULONG)y.
(ULONG)actuaIEditor->xscr.
(ULONG)y + actuaIEditor->ch - I);
SetAPen(rp,FGPEN);
}
I*#ENDFD*I
I*#FOLD: printAlI *1
1**********************************
**
**
printAll
print new screen.
**********************************1
void printAll ()
{

y
ch
count
If>
actuaIEditor->yoff;
actuaIEditor->ch;
actualEditor->wch;
actualEditor->rp;
1* Output the line: *1

for (z = actuaIEditor->zlinesptr; count-- && (*z
z++, y += ch)
printLine(*z,y);
.=
NULL);
1* erase right and bottom border: *1

SetAPen(rp,BGPEN);
if (actuaIEditor->yru >= y)
RectFill(rp, (ULONG)actuaIEditor->xoff, (ULONG)y.
(ULONG)actuaIEditor->xrr, (ULONG)actualEditor->yru);
if actualEditor->xrr >= actuaIEditor->xrl)
&& (y > actualEditor->yoff
RectFill(rp, (ULONG)actuaIEditor->xrl, (ULONG)actualEditor->yoff.
(ULONG)actuaIEditor->xrr. (ULONG)y-l);
SetAPen(rp,FGPEN);
}
/*#ENDFD*I
I*#FOLD: cn utoa *1
/******************************
**
cn_utoa(buf,val,len):
*
631
AMIGA
C FOR
* Convert UWORD in ASCII.
* buf
* val
* len
=
=
buffer for ASCII.

Value
Maximum Lenght.
*******************************/
void cn_utoa();
#asm
cseg
. public _cn_utoa
; 4(sp)
; 8 (sp)
;lO(sp)
lea
move. I
moveq
rnove.w
rnove.w
lea
bra.s
4(sp),aO
(aO)+,al
#O,dO
(aO)+,dO
; = Value
(aD) ,d1
= maximum Lenght
o(a1,d1.w), aD ; ~ End of Buffers
.in
.lp:
divu
swap
add.b
move.b
clr.w
swap
.in:
dbeq
beq.s
bra.s
.bl:
rnove.b
.ib:
dbra
.r:
rts
=
=
Buffer,
Value,
maximum Lenght.
#lO,dO
dl
#'0' ,dO
dO,-(aO)
dO.w
VAlue MOD 10
dl
dl
Zero-Flag is 1, when dO.1
dl, .lp
.ib
.r
; solange bis Wert = 0 oder String voll

Value = 0 => Rest is full of blanks
String full!
#'
,,- (aO)
dl, .bl
#endasm
/*#ENDFD*/
/*#FOLD: printXpos */
/*****************************
* printXpos:
* Give Xpos of cursor.
*****************************/
void printXpos ()
{
632
ABACUS
ApPENDIX A VERSION
OUTPUT.C
register struct Editor *ae = actualEditor
Move (ae->rp,ae->ip xpos,ae->ip topedge);
Text (ae->rp, ae->gi-Text '. INFO- XPOS INC, (ULONG) INFO XPOS LEN);
SetFont (ae->rp, oldFont: ;
}
I*#ENDFD*/
I*#FOLD: printYpos */
/*****************************
*
*
*
printYpos:
Give Ypos of cursor.
*****************************/
void printypos()
{

Move(ae->rp,ae->ip-ypos,ae->ip_topedge);
Text(ae->rp,ae->gi Text + INFO YPOS INC, (ULONG) INFO YPOS LEN);
SetFont (ae->rp, oldFont) ;
}
I*#ENDFD*/
/*#FOLD: printNumLines */
/*****************************
**
printNumLines:
**
Give number of lines

* of text
*****************************/
void printNumLines()
{

en utoa(ae->gi Text + INFO NUMOFLINES INC,ae->num lines,
(UWORD)INFO_NUMOFLINES_LEN);
Move(ae->rp,ae->ip_numzlines,ae->ip_topedge);
Text(ae->rp,ae->gi Text + INFO NUMOFLINES INC,
(ULONG)INFO_NUMOFLINES_LEN);
SetFont(ae->rp,oldFont);
}
/*#ENDFD*I
/*#FOLD: printFold */
/********************************
633
AMIGA
* printFold:
* Gives minfold and maxfold display.
*********************************1
*
void printFold()
{

en utoa(ae->gi Text + INFO MINFOLD INC,ae->minfold,
(UWORD)INFO MINFOLD LEN);
cn_utoa(ae->gi_Text + INFO_MAXFOLD_INc,ae->maxfold,
(UWORD)INFO_MAXFOLD_LEN);
Move(ae->rp,ae->ip minfold,ae->ip topedge);
Text(ae->rp,ae->gi-Text + INFO MINFOLD INC, (ULONG) INFO MINFOLD LEN);
Move(ae->rp,ae->ip-maxfold,ae->ip topedge);
-Text(ae->rp,ae->gi-Text + INFO MAXFOLD INC, (ULONG) INFO MAXFOLD LEN);
Set Font (ae->rp, oldFont);
-}
I*#ENDFD*I
I*#FOLD: printFlags *1
1*****************************
**
printFlags:
**
At Ypos of cursor.
******************************1
void printFlags()
{

register UBYTE *fs = ae->gi_Text + INFO_FLAGS_INC;
i f (ae->insert)
*fs++
'I';
else
*fs++
'0';
if (ae->ehanged)
'C';
*fs++
else
*fs++
'c';
if (ae->autoindent)
*fs++
'A';
else
*fs++
'a';
i f (ae->tabs)
*fs++
'T';
else
*fs++
't';
if (ae->skipblanks)
*fs = 'B';
else
*fs = 'b';
634
ABACUS
ApPENDIX A VERSION
6 OUTPUT.C
Move (ae->rp,ae->ip_flags,ae->ip_topedge);
Text (ae->rp,ae->gi Text + INFO FLAGS INC, (ULONG)INFO FLAGS LEN);
SetFont (ae->rp, oldFont) r
-}
I*#ENDFD*I
I*#FOLD: printInfo *1
1****************************
*
* printInfo:
*
* Gives new Infoline.
*
****************************1
void printInfo()
{
printXpos () ;
printypos () ;
printNumLines();
printFold () ;
printFlags () ;
}
/*#ENDFD*I
635
AMIGA
C FOR
Version 6 Memory.c
/***************************************
* Module: Memory
* Organize memory handling.
***************************************/
/************
** Includes:
*************/
/*#ENDFD*/
/*#FOLD: Defines *(
/***********
**
Defines:
***********/
#define BIGBLOCK (10240L-sizeof(struct Memoryblock

#define BLOCKSIZE (5120L-sizeof(struct Memoryblock
/*#ENDFD*/
/**********************
**********************/
struct Memoryblock *AllocMem();

void NewList(),AddTail(),Remove(),FreeMem();
/*#ENDFD*/
/*********************
**********************/
/*#ENDFD*/
/***************
* Functions: *
***************/
63b
ABACUS
ApPENDIX A VERSION
MEMORY.C
/*#FOLD: freeMemoryblock */
/*******************************************
* freeMemoryblock(blk)
**
frees the memory of

* Memoryblocks, without regard for
* lines that lie in it.
* blk . r1emorybloc....
********************************************/
void freeMemoryblock(blk)
{
Remove(blk);
FreeMem(blk,blk->length + sizeof(struct Memoryblock;
}
/*#ENDFD*/
/*#FOLD: newerBlock */
/**********************************************
* newerBlock(len):
**
used memory for new memory block with len length

* and returns the pointer
* in case of error => NULL.
* len = Length of memory block.
**********************************************/
struct Memoryblock *newerBlock(len)

ULONG
len;
{

register struct Memorysection *spst;
/* Allocate memory: */
i f (blk = AllocMem(len+sizeof(struct Memoryblock),MEMF_CLEAR
{
/*Initialize memory block structure: */

blk->length = len;
blk->free = len - sizeof(struct Memorysection);
NewList(&(blk->freeliste;
spst = (struct Memorysection *) (blk + 1);
AddTail(&(blk->freeliste),spst);
spst->len = (UWORD)blk->free;
return (blk);
}
/*#ENDFD*/
/*#FOLD: searchMemorysection */
1***********************************************
**
searchMemorysection(block,len)
* Searches for memory piece with len length in the block block
* and returns the pointer or zero in case
* of no room.
637
**
block ~ Memoryblock.
* len = Length of line (even).
***********************************************1
struct Memorysection *searchMemorysection(block,len)

struct Memoryblock
*block;
UWORD
len;
{
register UWORD minl,l;

register struct Memorysection *st,*fnd = NULL;
1* Now searches through if enough memory: *1

if ULONG)len <= block->free)
{
1* Minimal length in case found Length <> len *1

minI = len + sizeof (struct Memorysection) + 2;
1* Searches through list: *1
for (st = block->freeliste.head; st->succ; st = st->succ)
if 1 = st->len) == len)
{
1* Optimal size found: *1

fnd = st;
break;
}
else
if 1 >= minI) && (1 > fnd)? fnd->len
1* Search for largest block: *1
fnd = st;
0)
return (fnd);
}
l*lIENDFD*1
l*lIFOLD: ConvertSpstToZline *1
1********************************************
* ConvertSpstToZline(blk,st,len)
**
Converts memory piece in line.

* If the piece was longer than the line,
* the piece is shortened, otherwise the piece is removed
* from the list.
** blk
~ memory block of memory piece.

* st ~ memory piece.
* len = Length of the line
*********************************************1
struct Memoryblock
*blk;
*st;
register UWORD
len;
{
register UWORD 1;
Z
638
= (struct
Zline *)st;
ABACUS
ApPENDIX A VERSION
MEMORY.C
if (st->len == (1 = EVENLEN(len)))
Remove (st):
else
{
/* distribute piece: */
UBYTE *)st) += (1 += sizeof (struct Zline)):
( st->succ = struct Memorysection *)z)->succ )->pred
( st->pred = struct Memorysection *)z)->pred )->succ
st->len = struct Memorysection *)z)->len - 1;
st:
st;
blk->free -- 1;
z->flags = 0;
z->len = len;
return (z);
}
I*#ENDFD*I
I*#FOLD: getZline *1
1***********************************************
* getZline (len)
*
*
*
*
Tries to get line with len length and,

searches through all of the memory blocks.
returns pointer to line or
zero if no room.
**
len = Length of the line.
***********************************************1

UWORD
len;
{
register UWORD 1:
register struct Memorysection *st,*fst
=
=
NULL:
NULL:
/* make even length: */

1 = EVENLEN(len);
1* run through list of blocks: */

for (blk = actualEditor->block.head; blk->succ; blk = blk->succ)
if (st->len == 1)
{
1* found optimal memory piece: *1

fblk = blk;
fst = st;
break;
else
if (st->len > fst)? fst->len : 0))
{
1* searches for largest memory piece: *1

fblk = blk;
fst = st;
i f (fst)
639
AMIGA
return (ConvertSpstToZline(fblk,fst,len:
else
1* no piece found: *1
return (NULL);
}
I*IIENDFD*I
I*IIFOLD: optimalBlock *1
1*********************************************
* optimalBlock(block)
* Searches through all of the memory pieces of the block

* and makes one out of all of the memory pieces
* that lie one after another.
** block ~ Memoryblock.
**********************************************f
void optimalBlock
(blk)
{
register struct Memorysection *stl,*st2,*stle;

for (stl
= blk->freeliste.head;
stl->succ; stl
stl->succ)
f* Certain pointer directly behind stl: *1

(UBYTE *)stl) + stl->len + sizeof(struct Memorysection;
for (st2 = blk->freeliste.head; st2->succ; st2
if (stle =" st2)
= st2->succ)
1* memory piece st2 lies behind stl: *f

stl->len +~ st2->len + sizeof(struct Memorysection);
blk->free +z sizeof(struct Memorysection);
Remove (st2);
(UBYTE *)stl)+stl->len+sizeof(struct Memorysection;
1* begin at the beginning of the list: *1

st2
(struct Memorysection *) &(blk->freeliste.head);
I*IIENDFD*I
I*IIFOLD: garbageCollection *f
/****************************************
* garbageCollection(len)
*
*
*
*
*
cleans the memory until at least

len bytes are free. returns pointer.
to memory block where there is room
or zero if no room.
zero if no room.
* len
= number
of bytes that should be accomodated
****************************************1
640
ABACUS
ApPENDIX A VERSION
6 MEMORY.C

OWORD
len;
{

struct Memorysection *st,spst;
register UBYTE ~ptr,*end;
register struct Zline *z,*fz;
register OWORD n;
1* Only garbage collection if more than one memory piece! *1
if st = blk->freeliste.head->succ)? st->succ : FALSE)
{
1* test for beginning and end of block memory: *1

1* lines lie at the beginning: *1
while (ptr < end)
{
1* searches for lines that lie at next ptr: *1

fz = NULL;
i f z < fz) II (fz == NULL
fz = z;
z->succ)
i f (fz)
i f (fz != ptr)
{
1* note old pointer: *1

z = fz;
spst = *st;
1* writes line after ptr: *1

for (n = (EVENLEN(fz->len) + sizeo(struct Zline 1;
*) fz)++ )
n; n--, OWORD *)ptr)++, UWORD
* OWORD *)ptr) = * OWORD
*) fz);
1* pointer correction: *1
fz = (struct Zline *)st;

fz->succ->pred = fz;
fz->pred->succ = fz;
if (actualEditor->lineptr == z)
actuaIEditor->lineptr =
fz;
if (*zptr == z)
{
*zptr = fz;
break;
st->succ = spst.succ;
st->pred = spst.pred;
st->len = spst.len;
spst.succ->pred = st;
641
AMIGA
spst.pred->succ = st;
1* push memory pieces together: *1

optimalBlock(blk);
}
else
1* set ptr high: *1

ptr += EVENLEN(fz->len) + sizeof(struct Zline);
else
1* no more lines in block! *1

break;
1* of while *1
1* Test if there is enough memory: *1
if (searchMemorysection(blk,len))
return (blk);
1* of if (at least 2 blocks) *1
return (NULL);
}
l*lIENDFD*1
l*lIFOLD: newZline *1
1*****************************************
**
newZline (len)
**
get room for new line,

* fits characters in len.
* returns pointer to new line,
* or zero if error.
* len = number of characters.
******************************************1
struct Zline *newZline(len)
UWORD
len;
{
register
register
register
register
struct Zline *z;

UWORD 1;
if (actualEditor->block.head->succ)
1* blocks already exist: *1
if (z = getZline(len))
return (z);
else
1* Garbage-Collection:line not directly' usable; *1
if (blk = garbageCollection(l = EVENLEN(len)))
return (ConvertSpstToZline(blk,st,len));
else
1* may not actually be encountered I!! *1
return (NULL);
else
1* no reults for Garbage-Collection ->pick new block *1
if (blk = newerBlock(BLOCKSIZE))
{
642
ABACUS
ApPENDIX A VERSION
MEMORY.C
else
return (NULL);
else
1* no block in the block list: *1

if (blk = newerBlock(BIGBLOCK
(
return (getZline(len;
}
else
return (NULL);
}
I*#ENDFD*I
I*#FOLD: deleteZline *1
1*********************************************
* deleteZline(line)
*
*
*
*
Delete line from the list of lines.

The occupied memory word in the memory
section is changed and the corresponding
memory block optimized.
* line " line.
*********************************************1
void deleteZline
(line)
{
register
register
register
register
struct Memoryblock *blk,*fblk

struct Zline **zptr;
UWORD n;
= NULL;
1* remove line from list: *1

Remove (line) ;
1* determine in which block the line lies: *1
if (line> blk) && (line < (UBYTE *) (blk + 1 + blk->length)
{
fblk = blk;
break;
i f (fblk)
{
1* convert line into memory peices, insert in list: *1

fblk->free += ( struct Memorysection *)line)->len
= EVENLEN(line->len) );
AddTail(&(fblk->freeliste),line);
optimalBlock(fblk);
if (fblk->length == (fblk->free + sizeof(struct Memorysection)
. 1* Block contains no more lines: Delete! *1
freeMemoryblock(fblk);
}
I*#ENDFD*I
643
AMIGA
Version 6 Command.c
See Section 4.3.9 for a complete listing of the comand. c module.
644
Index
+L compiler option
15
absolute instructions
64
absolute memory locations
77
ACTIVATE
96,428
ACTIVEWINDOW
263
ALERT_TYPE
258
Alerts
251
AIIocEntryO
372
AIIocMemO
370,372
AIIocRememberO
375,376
application
381
argc (argument count)
14,543
argv
14,543
Anays
41,382
ASCII
21,25,384,365,399
assembler
35,63
AUTO
234
Auto Indent
384,514
auto requester
229,232
AUTOBACKPEN
234
AUTODRAWMODE
234
AUTOFRONIPEN
234
AUTOINDENT
549
AUI'OITEXTFONT
234
AUTOLEFIEDGE
234
AUI'ONEXTIEXT
234
AUTOTOPEDGE
234
Aztec (Manx) C 4, 186, 381, 408, 446,
516
back gadget
BackDrop
Backspace
BCPL
BGPEN
BIGBLOCK
bit fields
bit-maps
bit-planes
BlockPen
BLOCKSIZE
Boolean gadgets
BOOLGADGET
BORDERLESS
BOITOMBORDER
178
87,95
476
3,62
433
422
41,43
126
158
85
422
177
182
86,95
181
breakpoint
buffer
BuildSysRequestO
75
407
234
Capsable
363,368
CD_ASKDEFAULTKEYMAP
365
CD_ASKKEYMAP
363
CD_SETDEFAULTKEYMAP
365
CFOLD
551
CHECKED
292
CHECKIT
292
Chip RAM
369
chip memory
162
ClearMenuStrip
289
CLI
407,543
CLI command access
385
CLlwindow
427
Close gadget
178
close gadget
88,428
CLOSEWINDO
264
CLOSEWINDOW
404
CLOSEWINDOW
428
code segment
64
command line
512
COMMSEQ
292,299
compiler
35,381
compiler errors
36
compiler options
36
144,449
COMPLEMENT mode
console.device
337
console.device flags
266
ConsoleReadMsg
338
ConsoleWriteMsg
338
Control characters 343,384,433,474
341
control sequence
385
control structures
control structures.
54
433
CONlROLCODE
coordinate
153
CPTR pointers
62
341,462,504
CSI
C1RLPEN
433
Cursor display
449
Cursor movement
513
445
cursor
cursor movement
449
645
INDEX
CURSORHOME
custom gadgets
custom requester
custom screen
AMIGA
551
177
234
85, 123
data segment
64
DEADEND_ALERT
252,255,258
debug post mortem
75
debugger
74,471,473
Delete character
476
Delete commands
513
Delete line
476
Deleting Lines
422
DELTAMOVE
264,278
DetailPen
85
development stage
381
DEVICE STATUS REPORT
346
devices
262
DIGIT
535
disk flags
281
DISKINSERTED
265
DISKREMOVED
265,281
double linked list
385,386
drag bar
88
dynamic arrays
78
editor commands
environment variables
ERROR_CLASS
Exchange commands
ExecMessage
505
7
259
513
268
Fast RAM
369
FGPEN
468
file attribute
17
Find commands
513
flags
266
FOLD
432
Fold levels
447
fold
496
fold end mark
445
fold mark
468
fold starting mark
445
Folding
384,432,445,446
FOLDPEN
468
FOLLOWMOUSE
264
font preference flags
146
fonts
145
FreeRememberQ
375,376
646
FreeSysRequestO
front gadget
FSE_LEN
FSE_SIG
functions
234
178
495
495
59
GADGDISABLED
180
gadget border
171
gadget flags 94, 180,264,266
gadget IDCMP flag
271
gadget manipulation
263
GADGETDOWN
263
gadgets
83,177,271
GADGETUP
264,515,541,544
GADGHBOX
179
GADGHCOMP
179
GADGHIGHBITS
179
GADGHIMAGE
179
GADGIMAGE
180
GADGIMMEDIATE
181
GADGIMMEDIATE
263
GAGDHNONE
179
garbage collection
409,535
GetMsgO
269
GIMMEZEROZERO 87,95, 104, 114,
181,428
global symbol
48
goto
59
graphics
158
graphics chip
162
GRELBOTIOM
179
GRELHEIGHT
180
GRELRIGHT
179
GRELWIDTH
179
Guru Meditation
258,385
Guru number
258
GZZGADGET
182
HAM
handleKeys
HIGHBOX
HIGHCOMP
HIGHFLAG
HIGHFLAGS
HIGHIMAGE
HIGHITEM
HIGHNONE
hunks
125
449
292,301
292
301
292
292,300
292
292,301
71
ABACUS
INDEX
IDCMP 96, 233, 262, 263, 266, 276,

321,337,349
IDCMPflag
263,271,429
IDCMP variables
268
IF command
58,513
INACTIVEWINOOW
263
indentation
384
initializing a variable
46
input.device flags
266
Insert character
476
Insert line
476
Insert mode
474
IntuiMessage structure
268,272
IntuiText
147
IntuiText structure
300,429,518
INTUmCKS
265
Intuition
83
Intuition
177
Intuition Direct Communication
Message Port
96,262
Intuition functions
429
Intuition structure
267
INVERSEVID
144
IOStdReq
338,349
IOStdRequest structure
338
ISDRAWN
292
ITEMENABLE
292,301
ITEMTEXT
292
JAM1
JAM2
jump table
144
144
70
keyboard input
keyboard menu item
keyboard table
KeyMap
Key Map Types
Kickstart ROM
338
299
361
363
367
369
large code
large data 3
Lattice C
LEFTBORDER
libraries
Line commands
line output
link
37, 70
7, 70
4
181
6,89,404,543
514
152
408
linked list
linker
linker options
Local variables
local symbol table
long variables
383,385
35,67
71
60,509
48
15
514
Macro command
12
macro definition
macros
385
main loop
404
make utility
9
9,393
makefile
137
MakeScreenO
MEMF_CHIP
370
MEMF_CLEAR
370
MEMF_FAST
370
MEMF_PUBLIC
370
Memory Management
408
memory
383
memory classes
53
memory management
369
Menu flags
266
menu reading
264
menu strip
283,287
menu structure
288
276
MENUOOWN
MENUENABLED
288
290,300
Menultem structure
MENUPICK
264,274
MENUPICK
321
Menus
274,283
MENUTOGGLE
305
276
MENUUP
MENUVERIFY
264
MessageKey
267
MessagePort
267,398
288
MIDRAWN
missing semicolon error
36
266
ModifyIDCMPO
266
Mouse flags
276
mouse
264,277
mouse status
MOUSEBUTIONS 264,277,466,543
MOUSEMOVE
264
107
MoveWindowO
268
MsgPort structure
MutualExclude
305
647
INDEX
AMIGA
NEWPREFS
265
NEWSIZE
263,429
NewWindow structure
90
No Intuition/IDCMP message flags 266
NOCAREREFRESH
95
numeric keypad
279
object me
octal code
Operating system programming
output
Overwrite mode
394
383
381
143
474
PC-relative instructions
65, 70
pointer assignments
62
Pointers
47
PowerWindows
325
Preferences change flags
266
PrintIText
147,429,518
program translation
40
programmability
384
PROPGADGET
182
178
proportional character sets
430
proportional gadget
189
puts function
543
Quit command
514
RAM disk
408,544
RAWKEY 265, 278, 337, 361, 399,
506,537
RAWKEY diagram
362
recoverable Alert
233
RECOVERY_ALERT
252,258
Redirection
35
REFRESHWINDOW 263, 429, 420,
436
register storage class
53
register variables
512
registers
53
relative instructions
64
relocatible code
63
RELVERIFY
181, 186, 264
RELVERIFY flag
514
REMOUSE
95
REPEAT
506
648
Repeat command
Repeatable
REPORTMOUSE
REQCLEAR
REQSET
Requester flags
requester structure
Requesters
REQVERIFY
resolution
retumvalue
RIGHTBORDER
RMB1RAP
RMB1RAP
ROM fonts
screen refreshing
screen types
screens
SCRGADGET
script file
Scrolling
SELECTOOWN
SELEC1ED
SELECTUP
Set commands
SetMenuStrip
SIMPLE_REFRESH
single-stepping
size gadget
SIZEBOTTOM
SIZEBRIGHT .
SIZEVERIFY
SizeWindowO
sizing gadget
SKIPBLANKS
small code
small data
SMART_REFRESH
source code
source code utilities
src/command.c
src/Cursor.c
src/Editor.c
src/Editor.h
src/Memory.c
src/OutpulC
src/feslc
stack
stack overflow
514
368
264
264,273
264,273
266
235
229,251,273
264,274
125
61
181
96
276
145
428
126
83, 123
182
9
445,446,497
276
180
276
514
289
86,94,263,428
75
428
94
94
263,273
108
88
535
70
70
86,95,263,428
381
325
522
451
400
399
410
436
419
49
49
INDEX
ABACUS
StandardRequest
339
start of block
512
startup code
68
startup-sequence
6
static variables
53
STRGADGET
182
String gadget
178,195,514
submenu
302
SUBSYSlEM_CODE
258
super-high-resolution screen
138
SUPER_BITMAP
86, 88, 95, 428
switch case
55, 505, 535
symbol table 37,41,44,61,64,68, 74
SYSGADGET
182
system gadgets
116, 177
system requester
229
WINDOWDEPTH
WINOOWDRAG
WindowLimitsO
WindowPort
WINOOWSIZING
WindowToBackO
WindowToFrontO
word processors
Workbench message flags
WRI1EMODE
Zeditor
94
94
109
267
94
110
110
466
266
549
384,446
Tab commands
TABMODE
Tabs
Text
Text output
text entry
timer.device flags
TOGGLESELECT
topaz.font
514
549
384,446,474,514
429
143,497
474
266
180
146
TOPBORDER
181
266
tIackdisk.device flags
type conversions
45
Undo command
Undo function
union
UNIX
user-friendly
UserPort
video chip
514
383,476
43
3
383
267,268,269
265,278,337,399
40
125
WaitO
while loop
Window flags
window
window flags
window title
window types
WINDOWCLOSE
270
54
266,272
383,398,427,445
93
93
95
94
VANILLAKEY
variable types
649
Optional Diskette
Amlga C for
Advanced
Programmers
Optional diskette
For your convenience, the program listings contained in this book are
available on an Amiga formatted floppy diskette. You should order the
diskette if you want to use the programs, but don't want to type them in from
the listings in the book.
All programs on the diskette have been fully tested. You can change the
programs for your particular needs. The diskette is available for $14.95 plus
$2.00 ($5.00 foreign) for postage and handling.
When ordering, please give your name and shipping address. Enclose a
check, money order or credit card information. Mail your order to:
Abacus
5370 52nd Street SE
Or for fast service, call 616-698-0330.
Credit Card orders only 1-800-451-4319.
New Software
All Abacl/S software runs on the Amiga 500, Amiga

}OOO or Amiga 2000. Each package isfully compatible
wilh OllT olher products in the Amiga liM
The Ideal AMloA wordprocessoc
TextPro
More than word processing .
AMIGA
TextPro AMIGA upholds the true spirit of !he AMlOA:
it's powerful, it has a surprising number of "extra"
features, but it's also very easy to use. TextPro
AMIGA-the Ideal AMlGA word processor that proves
just how easy word processing can be. You can write
your fmt documents immediately, with a minimum of
learning-without even reading the manual. But
TextPro AMIGA is much more than a beginner's
package. Ultra-fast onscreen formatting, graphic merge
capabilities, automatic hyphenation and many more
features make TextPro AMIGA ideal for !he
professional user as well. TextPro AMIGA features:
High-speed text input and editing

Functions accessible through menus or shortcut keys
Fast onscreen formatting
Automatic hyphenation
Versatile function key assignment
Save any section of an AMlOA screen & print as text
Loading and saving through the RS-232 interface
Multiple tab settings
Accepts IFF format graphics in texts
Extremely flexible printer adaptations. Printer drivers
for most popular dot-matrix printers included
Incl udes thorough manual
Not copy proteCted
TextPro AMIGA
sets a new standard
for word processing
packages in its price
range. So easy to
use and modestly
priced that any
AMlGA owner can
use it-so packed
with advanced
features, you can't
pass it up.
Suggested retail price:
'--------------=----'
7extPro
AM/GA'
$79.95
BeckerText
AMIGA
This is one program for BIiwI1 AMloA owners.
BeckerText Amiga is more than a word processor. It
has all the features ofTextPro AMIGA, butit also has
features that you might not expect:
FastWYSIWYG formatting
Calculations within a text-like having a spreadsheet
program anytime you want it
Templates for calculations in columns
Line spacing options
Auto-hyphenation and Auto-indexing
Multiple~olumn printing, up to 5 colunms on a single
page
Online dictionary checks spelling in text as it's written
Spell checker for interactive proofing of documents
Up to 999 characters per line (with scrolling)
Many more features for the professional
BeckerText AMIGA
is a vital addition for
C programmers-it's
an extremely flexible
C editor. Whether
you're deleting,
adding or duplicating
AMIGA"
a block of C sourcecode, BeckerText
AMIGA does it all,
automatically. And
the online dictionary
acts as a C syntax
checker and finds
syntax errors in a
flash.
BeckerText AMIGA. When you need more from your
word processor than just word processing.
8edferText
--
$150.00
Imagine the ~ database
DataRetrieve
AMIGA
Imagine, for a moment, what the perfect database for
your AMIOA would have. You'd want power and speed,
for quick access to your information. An unlimited
amount of storage space. And you'd want it easy to
use-no baming commands or file structures-with a
graphic interface that does your AMIoAjustice.
Enter DataRetrleve AIIIIGA. It's unlike any other
database you can buy. Powerful, feature-packed, with
the capacity for any business or personal applicationmailing lists, inventory, billing, etc. Yet it's so simple to
use, it's startling. DataRetrleve AIIIIGA'S drop-down
menus help you to define files quickly. Then you conveniently enter information using on-screen templates.
DataRetrieve AIIIIGA takes advantage of the Amiga's
multi-tasking capability for optimum processing speed.
DataRetrleve AIIIIGA features:
Open eight files simultaneously
Password protection
Edit files in memory
Maximum of 80 index fields with variable preciSion
(1-999 characters)
Convenient search/select criteria (range, AND/OR
compari sons)
Text, date, time, numeric and selection fields, IFF file
reading capability
Exchange data with other software packages (for form
letters, mailing lists, etc.)
Control operations with keyboard or mouse
Adjustable screen masks, up to 5000 x 5000 pixels
Insert graphic elements into screen masks (e.g.,
rectangles, circles, lines, patterns, etc.)
Screen masks support different text styles and sizes
Multiple text fields with word make-up and
formatting capabilities
Integrated printer masks and list editor.
Maximum filesize 2 billion characters
Maximum data record size 64,000 characters
Maximum data set 2 billion characters
Unlimited number of data fields
Maximum field size 32,000 characters
DataRetrieve AIIIIGA -it'll handle your data with the
speed and easy operation that you've come to expect
from Abacus products for the AMIOA.
$79.95
Notjnst for the experts
AssemPro
AMIGA
AssemPro AIIIIGA lets every Amigaownerenjoy the
benefits offast machine language programming.
Because machine language programming isn't just for
68000 experts. Assem Pro AIIIIGA is easily learned and
user-friendly-it uses Amiga menus for simplicity. But
AssemPro AIIIIGA boasts a long list of professional
features that eliminate the tedium and repetition of MIL
programming. Assem Pro AIIIIGA is the complete
developer's package for writing of 68000 machine
language on the Amiga, complete with editor, debugger,
disassembler and reassembler. AssemPro AIIIIGA is the
perfect introduction to machine langage development
and programming. And it's even got what you 68000
experts need.
AssemPro AIIIIGA features:
Written completely in machine language, for ultra-fast
operation
Integrated editor, debugger, disassembler, reassembler
Large operating system library
Runs under CLI and Workbench
Produces either PC-relocatable or absolute code
Macros possible for nearly any parameter (of different
types)
Error search function
Cross-reference list
Menu-controlled conditional and repeated aSfo~mbly
Full 32-bit arithmetic
Debugger with 68020 single-step emulation
Runs on any AMIGA with SI2K and Kickstart 1.2.
$99.95
Abacus Products for
Amiga
computers
Professional DataRetrieve
The Professional Level
Database Management System
Professional DatJIRetrieve, for the Amiga 500/1000/2000,
isa friendly easy-to-opcrate professional level data management package with the features most wanted in a relational
data bft3e syslem_
Professional DataRetrieve has complete relational data
mangagement capabilities. Define relationships between
different files (one to one, one to many, many to many).
Ch!lllge relations withOllt file reorgani7A1tion.
Professional DatJIRetrieve includes an extensive programming laguage which includes more than 200 BASIC-like
commands and functions and integrated program editor.
Design custom u~r interfaces with pulldown menus, icon
selection, window activation and more.
Professional DataRetrleve can perform calculations and
searches using complex mathematical comparisons using
over 80 functions and constants.
Professional DataRetrieve is a friendly, easy to operate
programmable RELATIONAL data base system. PDR includes PROFlL, a programming language similar to BASIC.
You can open and edit up to 8 files simultaneously and the
size ofYOIIrdata rlClds, records and files are limited only by
your memory and disk storage. You have complete interrelation between files which can include IFF graphics. NOT
COpy PROJECTED. ISBN 1-55755..0484
MORE features of Professional DataRdrleve
Easily import data from other databoscs ....file compatible
with standanl Data Retrieve....supPOrlS multitasking...design
your own custom forms with the completely integrated
printer mask editor... .includes PROFIL programming language that allows the programmer to custom tailor his database requirements.
MORE features ofPROFlL include:
Open Amiga devices including the console, printer,
serial and the CLI.
Create your own programmable requestors
Complete error trapping.
Built-in compiler and much, much more.
$295.00
---
Titt
PmfmimrJ/
lml
Features
Up to 8 files can be ed~ed simuHaneously
Maximum size of a data field 32,000 characters
(text fields only)
Maximum number of data fields limited by RAM
Maximum record size 01 64,000 characters
Maximum number of records diSk dependent
(2,000,000,000 maximum)
Up to 80 index fields per file
Up to 6 field types - Text, Date, Time, Numeric,
IFF, Choice
Unlimijed number 01 searches and subrange
cr~eria
Integrated liSt edHor and ful~page printer mtlsk

editor
Index accuracy selectable from 1-999 characters
MuHiple file masks on-screen
Easily create/edij on-screen masks for one or
many files
User-programmable pulidown menus
Operate the program from the mouse or the key
board
Calculation fields, Data Fields
IFF Graphics supported
Mass-storage-oriented file organlzation
Not Copy Protected, NO DONGLE; can be in
stalled on your hard drive
....DOS
.. "{fA
Abacus Amlga
...
:
.
:
~Toa
wX :
.. .. .. .. . . .. .. . ... .. ..
Essential software tools
for all Amiga users.
A collection of essential, powerful

and easy-to-use tools.
The Abacus AmigaDos ToolBox Includes 11 new fonts, new
handy eLI comm~,,!ds, a disk copier, floppy disk speeder,
screen capture utility and more.
from a company you can count on
AbacusllHill
5370 52nd Sireel SE Grand Rapids, M149508 Phone (616) 6980330 Telex 709101 Facsimile (616) 6980325
Companion Diskette
Amlga C for
Advanced
Programmers
Companion diskette
For your convenience, the program listings contained in this book are
available on an Amiga fonnatted floppy diskette. You should order the
diskette if you want to use the programs, but don't want to type them in from
the listings in the book.
All programs on the diskette have been fully tested. You can change the
programs for your particular needs. The diskette is available for $14.95 plus
$2.00 ($5.00 foreign) for postage and handling.
When ordering, please give your name and shipping address. Enclose a
check, money order or credit card infonnation. Mail your order to:
Abacus
5370 52nd Street SE
Or for fast service, call 6166980330.

Credit Card orders only 18004514319.
Books for the AMIGA

Amiga
3-~
Graphics Programming in BASIC
Shows you how to use the powerful graphics capabilities 01 the

Amiga. Details the techniques and algorithm lor wr~ing threedimensional graphics programs: ray tracing in all resolutions, light
sources and shading, saving graphics in IFF lorr:nat and more.
Topics include:
Basics 01 ray tracing
Using an object editor to enter three-dimensional objects
Material ed~or lor setting up materials
Automatic computation in different resolutions
USing any Amiga resolution (low-res, high-res, interface.
HAM)
Dillerentlight sources and any active pixel

Save graphics in IFF lormat lor later recall into any IFF
compatible drawing program
Mathematical baSics lor the non-mathematition
Volume 3
Suggested retail price $19.95
Amiga Machine Language

Amlga MachIne Language introduces you to 68000 machine
language programming presented in clear, easy to understand terms. If
you're a beginner, the introduction eases you into programming right
away. ~ you're an advance programmer, you'll discover the hidden
powers 01 your Amiga. Learn how to access the hardware registers, use
the Amiga libraries, create gadgels, work with Intu~ion and much more.
68000 address modes and instruction set
Accessing RAM, operating system and multitasking
capabilities
Details the powerful Amiga libraries lor using AmlgaDOS
Speech and sound lacilities from machine language
Simple number base conversions
Text input and output
Checking lor speCial keys
Opening CON: RAW: SER: and PRT: devices
New directory program that doesnl access the eLI
Menu programming explained
Complete Intuition demonstration program including
Proportional, Boolean and String gadgets.
Volume 4
Suggested retail price $19.95 ISBN 1-55755-025-5
Save Time and Money!-Optional program disks are available lor all our Amiga relerence
books (except Amiga lor Beginners). AH programs listed in the book are on each respective
disk and will save you countless hours 01 typing! $14.95
Books for the AMIGA

Amiga Tricks & Tips
Amlga Tricks & Tips follows our tradition of other Tricks and Tips
books for CBM users. Presents dozens of tips on accessing libraries
from BASfC, custom character sets, AmigaDOS, sound, important
68000 memory locations, and much morel
Topics include:
Diverse and useful programming techniques

Displaying 64 colors on screen simultaneously
Accessing libraries from BASIC
Creating custom character sets
Using Amiga DOS and graphics
Dozens of tips on windows
Programming aids
Covers important 68000 memory locations
Volume 5
Suggested retail price $19,95
ISBN 0-916439-88-7
Amiga System Programmer's Guide

Amfga System Programmer's Guide is a comprehensive guide to
what goes on inside the Amiga in a single volume. Explains in detail the
Amiga chips (68000, CIA. Agnus, Denise, Paula) and how to access
them. All the Amiga's powerful interfaces and features are explained
and documented in a clear precise manner.
Topics include:
EXEC Structure
Multijasking functions
I/O management through devices and I/O request
Interrupts and resource management
RESET and ijs operation
DOS libraries
Disk Management
Detailed infonmation about the CLI and Hs commands
Volume 6
Save TIme and MoneylOptional program disks are available for all our Amiga reference
books (except Amiga for Beginners). All programs listed in the book are on each respective
disk and will save you countless hours of typing! $14.95
Books for the AMIGA

AmigaDOS: Inside & Out
AmlgaDOS: Inside & Out-covers the insides of AmigaDOS from
the internal design up to practical applications. Includes detailed
reference section, tasks and handling, DOS editors ED and EDIT, how
to create and use batch files, multitasking, and much more.
Topics include:
68000 microprocessor architecture

AmigaDOS - Tasks and handling
Detailed explanations of CLI commands and their functions
DOS ed~ors ED and EDIT
Operating notes about the CLI
(Wildcards, shortening input and output)
Amiga devices and how the CLI uses them
Batch files - what they are and how to write them
Changing the startup sequence
AmigaDOS and multitasking
Writing your own CLI commands
Reference to the CLI, ED and EDIT commands
Resetting priorities - the TaskPri command
Protecting your Amiga from unauthorized use
Volume 8
Amiga Disk Drives: Inside & Out

Amlga Disk Drives: Inside & Out is the most in-depth reference
available covering the Amiga's disk drives. Learn how to speed up data
transfer, how copy protection works, computer viruses, Workbench
and the CLI DOS functions, loading, saving sequential, relative file
organization, more.
Topics include:
Floppy disk operation from the Workbench and ClI
BASIC: Loading, saving, sequential and relative files DOS
functions
File management: Block types, boot blocks, checksums, file
headers, hashmarks and protection methods
Viruses: Protecting your boot block
Trackdisk.device: Commands, structures
Trackdisk-task: Function and design
Diskette access with DOS
MFM, GCR, Track design, blockheader, data blocks,
checksums, coding and decoding data, hardware registers,
SYNC, and interrupts
Volume 9
Suggested retail price $29.95
ISBN 1-55755-042-5
Save Time and Money!-Optional program disks are available for all our Amiga reference
books (except Amiga for Beginners). All programs listed in the book are on each respective
disk and will save you countless hours of typing! $14.95
Books for the AMIGA

Amiga C for Beginners
An introduction to learning the popular C language. Explains the
language elements using examples specifically geared to the Arniga.
Describes C library routines, how the compiler woills and more.
Topics Include:
Particulars of C
How a compiler woills
Wr~ing your first program
The scope of the language (loops, cond~ions, functions,
structures)
Special features of C
Important routines in the C libraries
Inpul/Output
Tricks and TIps for finding errors
Introduction to direct programming of the operating system
(windows, screens, direct text output, DOS functions)
Using the LATTICE and AZTEC C compilers
Volume 10
Suggested retail price $19.95 ISBN 1-55755-045-X
Amiga C for Advanced Programmers

Amlga C for Advanced Programmers- contains a wealth of
information from the pros: how compilers, assemblers and linkers work,
designing and programming user friendly interfaces using Intuition,
managing large programming projects, using jump tables and dynamic
arrays, coming assembly language and C codes, and more. Includes
complete source code for text editor.
Topics Include:
Using INCLUDE, DEFINE and CASTS
Debugging and optimizing assembler sources
All about Intu~ion programming (windows, screens, pulldown
menus, requesters, gadgets)
Programming the console devices
A professional edito~s view of problems with developing
larger programs
Using MAKE correctly
Debugging C programs with dffferent utilities
Folding (formatting text lines and functions for readabilijy)
Volume 11
Save Time and Money!Optional program disks are available for all our Amiga reference
books (except Amiga for Beginners). AH programs listed in the book are on each respective
disk and will save you countless hours of typingl $14.95
Books for the AMIGA

Amiga Graphics Inside & Out
The Amiga Graphics tnslde & Out book will show you simply and In plain English the super graphic leatures and functions of the Amiga In detail. You will
learn the graphic features that can be accessed from AmigaBASIC or C.
The advanced user will lear" graphic programming In C with examples of
points, lines, rectangles, polygons, colors and more. Amlga GraphicS
Inside & Out contains a complete description of the Amlga graphic
system - View, ViewPort, RastPort, bitmap mapping, screens, and
windows.
Topics include:
Accessing fonts and type styles In AmlgaBASIC

CAD on a 1024 x 1024 super bHmap, Using graphic
library routines
New ways to access libraries and chips from BASIC - 4096
colors at once, color pattems, screen and window dumps to printer
Graphic programming In C - points, lines, rectangles, polygons,colors
Amiga animation explained inctuding sprHes, bobs and AnimObs, Copper and bliller programming
Volume 13 SUggested retail price $34.95 ISBN 1.s57~52-2

lOpllonal Diskette
$14.95
'727
Amiga Desktop Video Guide

The Amlga Desktop Video Guide is the most complete and useful guide to desktop video on the Amiga.
Amlga Desktop Video Guide covers all the basicS - defining video terms,
selecting genlocks, digHlzers, scanners, VCRs, camera and connecting
them to the Amlga.
Just a few of the topicS you'n find described In this excellent book:
The Basics of Video

Genlocks
Digitizers and Scanners
Frame Grabbers/Frame Buffers
How to connect VCRs, VTRs, and Cameras to the Amlga
Animation
Video THling
Music and Videos
Home Video
Advanced Techniques
Using the Amiga to add or Incorporate Special Effects to a video
Tips on Paint, Ray Tracing, and 3-D Rendering in Commercial Applications
Votume 14 Suggested Retail Price $19_95 ISBN 1-55755-057-3

Save Time and Moneyl-Optional program disks are available for all our Amlga reference
books (except Amiga for Beginners and AmlgasDOS Quick Reference). Programs listed in
the book are on each respective disk and saves countless hours of typing! $14.95
Presenting ...
Protect your Amiga computer

system with this collection of
essential and valuable tools!
The Virus Protection Toolbox

describes how computer viruses
work; what problems viruses
cause; how viruses invade the
Libraries, Handler and Devices
of the operating system;
preventive maintenance; how to
cure infected programs and disks.
Works with Workbench 1.2 and 1.31
5370 52nd Street S. E.

Grand Rapids, MI49512
Available at your local dealer or
Order Toll Free 1-800-451-4319

Amlga Is a registered trademarll 01 CommodoreAmlga Inc.
Some of our best tools included are:
Boot Checkto prevent startup viruses.

Recoverto restore the system information
to disk.
Change Control Checkerto record modifications to
important files.
Check Newto identify new program and data files.
Order now or call for your Free
pamphlet "What you should know
about Computer Viruses"
(while supplies last)
Books for the AMIGA
UiHHIH~1
AmigaDOS Quick Reference Guide

IImigaDos Quick Reference Guide Is an easylouse reference 1001 for
beginners and advanced programmers alike. You can quickly lind commands
lor your Amiga by using the three handy indexes designed with the user In
mind. All commands are in alphabetical order lor easy reference. The most
uselullnformalion you need fasl can be found including:
An AmigaDOS commands described, Including Workbench 1.3
Command syntax and arguments described with examples

CLI shortcuts
CTRL sequences
ESCape sequences
Amiga ASCII table
Guru Meditation Codes
Error messages with their corresponding IlIlmbers
IIlreil. Indexes lor quick Information at your fingertips! The AmlgaDOS

Ouick Reference Guide Is an Indispensable 1001 you'lI wanlto keep close
to your Amiga.
Suggested retail prIce US $9.95
ISBN 155755-049-2
Abacus Amiga Books

Vol. 1
Amlaa lor Bealnners
1-55755-021-2
Vol. 2
AmlaaBASIC Inside & Out
0-916439-87-9
$16.95
$24.95
Vol. 3
Amlaa 3D GraDhlc Proarammlna In BASIC
1-55755-044-1
119.95
Vol. 4
Amlaa Machine Lanauaae
1-55755-025-5
$19.95
Vol. 5
Amlaa Tricks & TIDS
0-916439-88-7
119.95
Vol. 6
Amlaa Svstem Proarammers Guide
1-55755-034-4
$34.95
Vol. 7
Advanced Svstem Proarammers Guide
1-55755-047-6
$34.95
Vol. 8
AmlaaOOS Inside & Out
1-55755-041-7
119.95
Vol. 9
Amlaa Disk Drives Inside & Out
1-55755-042-5
$29.95
Vol. 10 Amlaa C lor Bealnners
1-55755-045-X
$19 .95
Vol. 11 Amlaa C lor Advanced Proarammers
1-55755-046-8
$34.95
Vol. 12 More Tricks & Tips lor the Amlaa
1-55755-051-4
$19 . 95
Vol. 13 Amlaa GraDhlcs Inside & Oul
1-55755-052-2
$34.95
Vol. 14 Arnica Desktop Video Guide
1-55755-057-3
119.95
AmlgaDOS Quick Reference Guide
1-55755-049-2
$ 9.95
ISBN 1-55755-046-8

Amiga C For Advanced Programmers

Uploaded by

Copyright:

Available Formats

Amiga C For Advanced Programmers

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Amiga C For Advanced Programmers

Uploaded by

Copyright:

Available Formats

,[

A Data Becker Book

Second Printing, December 1989

Copyright 1987, 1988

Two Amiga C compilers

How a C compiler works

The screen functions

Operating System Programming

Planning the Editor

Editor Command Line Commands

Appendix A Editor Listings

AMIGA C FOR ADVANCED PROGRAMMERS

Two Amiga C compilers

The Aztec compiler

1.1 THE AZTEC COMPILER

C FOR ADVANCED PROGRAMMERS

(Assembler Include Files)

Disk 1 is enough to compile short programs. You may have to write a

;stack should be at least 10K

copy dfO:lib/c.lib to ram:

;where are the include files?

;where to put the temporary files?

Then the startup-sequence copies the linker library (which must be

1.1 THE A ZTEC COMPILER

But back to the startup-sequence. Next the environment variables of the

which directory will be searched to find the libraries for linking.

; stack should be at least 10K

;where are the include files?

C FOR ADVANCED PROGRAMMERS

The professional version

Another method for interrupting a script file is pressing <Ctrl><D>.

The developer's version

(FFP functions for 6881 and

C FOR ADVANCED PROGRAMMERS

printf ("Function 1 calls function 3!\n");

printf ("Function 2 is called by function 3!\n");

printf ("Function 3 calls function 2!'\n");

printf ("MAIN calls function l!\n");

1.1 THE AZTEC COMPILER

If modl.o is older than modl.e (i.e., if a change was made to modl.e

These are make's simplest functions. If you modify the makefile so

C FOR ADVANCED PROGRAMMERS

Another option of make is the macro defInition. In your makefile

1.1 THE AZTEC COMPILER

The commercial version

;get parameter from stack

AMIGA C FOR ADVANCED PROGRAMMERS

The compiler in action

argc (argument count) contains the number of arguments. It should be

This structure represents a pointer to a pointer of type cha r. A pointer

1.2 THE COMPILER IN ACTION

printf ("USAGE: compare FILE1 FILE2 \n");

It would be a better programming style to react to a "1" as an argument,

fopenO requires a second argument in addition to the filename to be

printf ("%s cannot be opened!! !\n",argv[l]);

printf ("%s cannot be opened !!!\n",argv[2]);

FOR ADVANCED PROGRAMMERS

$12 <> $3a

1.2 THE COMPILER IN ACTION