Advanced Programming C++

Advanced Programming with C++
Tim Love,
Cambridge University,
C++ notes for engineers,
adapted by Hugh Murrell,
for Computer Science UKZN.
February 12, 2011
Abstract
This document aims to provide people who have done an introductory
programming course the skills required to write C++ programs.
Contents
1 Introduction 3
1.1 A second programming language . . . . . . . . . . . . . . . . . . . . 3
1.2 Web pages and Text books . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 The UKZN Teaching labs . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 Microsoft Windows . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Major project for 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 C++ programming constructs 5
2.1 Keywords . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Built-in Types and Enumerations . . . . . . . . . . . . . . . . . . . . 5
2.3 Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4.1 if . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4.2 switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5.1 while, do . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5.2 for . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.6 Aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.6.1 arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.6.2 structures and classes . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.8 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.9 Declarations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.10 Memory Allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.11 Scope and Namespaces . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.12 Exceptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1
CONTENTS CONTENTS
3 Object-orientated programming 15
3.1 Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2 Class member privacy . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Friend classes and functions . . . . . . . . . . . . . . . . . . . . . . . 17
3.4 A complete example . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5 Derived classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.6 Static members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.7 Const members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.8 Static Const members . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.9 Overriding behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.10 Templates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4 The Standard Library 20
4.1 Iterators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.1 Iterators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.2 Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.2.3 Routines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3 vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.4 Queues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.5 list . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.6 map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.7 bitset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.8 valarray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.9 Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.10 Set algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.11 Using member functions . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.12 Predicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.12.1 Creating predicates . . . . . . . . . . . . . . . . . . . . . . . . 32
4.12.2 Adapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5 More on Classes 32
5.1 Virtual members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.2 Abstract Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.3 Redening operators . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.4 A class denition example . . . . . . . . . . . . . . . . . . . . . . . . 34
5.5 Redening [ ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.6 Redening () . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.7 Redening -> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6 Input/Output 38
6.1 Simple I/O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.2 Formatting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.3 Stream Iterators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
6.4 Output of User-Dened types . . . . . . . . . . . . . . . . . . . . . . 41
6.5 Input of User-Dened types . . . . . . . . . . . . . . . . . . . . . . . 41
6.6 String streams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7 Odds and Ends 42
7.1 Casts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7.2 Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.3 Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.4 Maths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.5 Hardware Interfacing: bit operations and explicit addresses . . . . . 46
2
Advanced Programming with C++ 1 INTRODUCTION
7.6 Using routines from other languages . . . . . . . . . . . . . . . . . . 46
7.6.1 C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
7.6.2 Fortran . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
7.7 bad habits to give up . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
7.8 Non-essentials (Things you probably dont need to know yet) . . . . 48
7.8.1 virtual inheritance . . . . . . . . . . . . . . . . . . . . . . . . 48
7.8.2 typeid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7.8.3 virtual destructors . . . . . . . . . . . . . . . . . . . . . . . . 49
7.8.4 Auto pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7.8.5 operator functions . . . . . . . . . . . . . . . . . . . . . . . . 49
7.8.6 union . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7.8.7 Non-type template parameters . . . . . . . . . . . . . . . . . 50
8 Program Development 50
8.1 Style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
8.2 Makeles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
8.3 Debugging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
8.4 Common Diculties and Mistakes . . . . . . . . . . . . . . . . . . . 53
9 CImg, an image processing library for C++ 54
9.1 library structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
9.2 CImg, Hello world . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
9.3 compiling a CImg application . . . . . . . . . . . . . . . . . . . . . . 56
10 Glossary 57
11 References 58
12 Program listings 59
Listings
1 intTree.cpp, A binary search tree for integers. . . . . . . . 59
2 genTree.cpp, A template binary search tree. . . . . . . . . 61
1 Introduction
1.1 A second programming language
This document is aimed at those who have done two years of Java programming at
UKZN. These Java courses introduced all loop constructions and many operators.
Arrays and data structures were also introduced. All of this was done in an object
oriented setting using Java classes.
However some issues such as software housekeeping (e.g. Makeles) were glossed
over and hidden behind IDEs. Others, such as templates and many things were
not touched upon at all. Some of these newer things are easy conceptually, and
fundamental to modern C++ programming, though implementation details can be
complicated. The pay-o comes when you use the Standard Library - a library
which exploits advanced C++ features to make life easy for the programmer.
3
Advanced Programming with C++ 1 INTRODUCTION
C++ can be used in various ways: procedural (like Fortran or C) or Object-
orientated (like Java). This document covers the techniques to support each of
these programming paradigms, dealing with
Namespaces
Using classes to build new classes - inheritance
Abstract Classes and Templates
Using the Standard Library
Techniques useful when writing big, project-sized programs.
1.2 Web pages and Text books
To aid readability, the examples in the text are mostly code fragments. Full exam-
ples will be posted online at the modules homepage:
http://www.cs.ukzn.ac.za/ murrellh/ap
Some complete program examples are given in the in the appendix to this text.
Most of these have been adapted from examples on the Ira Pohl website:
http://users.soe.ucsc.edu/ pohl/
Many more complete examples are posted there.
A few programming exercises are suggested in this document. Any computing book
will oer more.
Unless otherwise stated, the page numbers in the margin refer to:
Bjarne Stroustrup, The C++ Programming Language (third edition), AT&T,
1997.
You can get this book from the library or buy your own copy from amazon.
There are also many free books on C++ on the web. For extra reading material
you can download from the advanced programming webpage:
Frank Brokken, C++ Annotations (version 6.5.0)
1.3 The UKZN Teaching labs
Advanced Programming students will either use G21 or the honours lab. Students
can code under Microsoft Windows or Linux.
1.3.1 Microsoft Windows
Here your compiler will be MinGw and your IDE will be Code::Blocks. The CodeBlocksManual
can be downloaded from the course website.
4
Advanced Programming with C++ 2 C++ PROGRAMMING CONSTRUCTS
1.3.2 Linux
Here your compiler is g++ and your IDE is Eclipse
1.4 Major project for 2010
At the end of this module you will construct a program that will be written written
using a large library of C++ classes.
In this major project you will try to simulate water owing over a terrain. As C++
doesnt have any graphics routines for this project you will make use of the image
processing library CImg to do our graphics work.
your program must initialise a terrain and then dump water on it. At each iteration
the water will ow downhill and the terrain with its current water prole will be
re-rendered.
2 C++ programming constructs
2.1 Keywords
You havent been introduced to all of these keywords before. Many youll never need.
Theyre included here for completeness because you cant use them for variable
names, etc.
and and eq asm auto bitand bitor
bool break case catch char class
compl const const cast continue default delete
do double dynamic cast else enum explicit
export extern false float for friend
goto if inline int long mutable
namespace new not not eq operator or
or eq private protected public register reinterpret cast
return short signed sizeof static static cast
struct switch template this throw true
try typedef typeid typename union unsigned
using virtual void volatile wchar t while
xor xor eq
2.2 Built-in Types and Enumerations
There are integral types (char, short, int, long) which can be intended to hold
signed (the default) or unsigned values. So for example unsigned char c creates
a one-byte variable big enough to hold any integer from 0 to 255. There are also
oating point types (float, double, long double).
5
If you want an integer variable to be restricted to containing just a few specic
values you can create an enumerated type
enum user_type {ugrad, pgrad, staff, visitor};
user_type u;
creates an enumerated type called user_type. u can only be set to the values 0, 1,
2 or 3 (or equivalently - and preferably - ugrad, pgrad, staff or visitor). The
values can be set explicitly; for example
enum user_type {ugrad=2, pgrad=7, staff=9, visitor=3};
user_type u;
Enumerated types can be used wherever integer types can be.
2.3 Operators
The lines of the table are in order of precedence, so a * b + 6 is interpreted as
(a * b) + 6. When in doubt put brackets in!
The Associativity column shows how the operators group. E.g. < groups left to
right, meaning that a < b < c is equivalent to (a < b) < c rather than a < (b < c).
Both are pretty useless expressions - (a < b) evaluates to 1 or 0 depending on
whether its true or not.
Associativity Operator
left to right ::
left to right () [], ->, ., typeid, casts,
right to left ! (negation), ~ (bit-not)
new, delete ++, --, - (unary) , * (unary), & (unary), sizeof
right to left (type)
left to right *, /, % (modulus)
left to right - +
left to right <<, >>
left to right <, <=, >, >=
left to right ==, !=
left to right & (bit-and), | (bit-or)
left to right ^ (bit-xor)
left to right && (logical and)
left to right || (logical or)
right to left ?:
right to left =, +=, -=, /=, %=, >>=, &=
left to right throw
left to right ,
2.4 Selection
2.4.1 if
...
6
if (i==3) // checking for equality; != tests for inequality
// no braces needed for a single statement
j=4;
else{
// the braces are necessary if the
// clause has more than one statement
j=5;
k=6;
}
...
2.4.2 switch
...
// switch is like a multiple if.
// The values that the switching variable is compared with
// have to be constants, or default.
switch(i){
case 1: printf("i is one\n");
break; // if break wasnt here, this case will
// fall through into the next.
case 2: printf("i is two\n");
break;
default: printf("i is neither one nor two\n");
break;
}
...
2.5 Loops
Note that if you use the Standard Librarys containers and vectors, explicit loops
might not be needed very much.
2.5.1 while, do
...
while(i<30){ // test at top of loop
something();
...
}
...
do {
something();
} while (i<30); // test at bottom of loop
...
7
2.5.2 for
The for construction in C++ is very general. In its most common form its much
like for in other languages. The following loop starts with i set to 0 and carries on
while i<5 is true, adding 1 to i each time round.
...
for(int i=0; i<5; i=i+1){
something();
}
...
The general form of for is
for ([expression1]; [expression2]; [expression3])
something();
where all the expressions are optional. The default value for expression2 (the
while condition) is 1 (true). Essentially, the for loop is a while loop. The above
for loop is equivalent to
...
expression1; // initialisation
while (expression2){ // condition
something();
expression3; // code done each iteration
};
...
E.g. the 2 fragments below are equivalent. i is set to 3, the loop is run once for
i=3 and once for i=4, then iteration nishes when i=5.
for (int i = 3; i < 5; i=i+1)
total = total + i;
int i = 3;
while(i < 5){
total = total + i;
i=i+1;
}
Within any of the above loop constructions, continue stops the current iteration
and goes to the next and break stops the iterations altogether. E.g. in the following
fragment 0 and 2 will be printed out.
...
int i=0;
while (i<5){
if (i==1){
i = i+1;
continue;
}
if (i==3)
break;
printf("i = %d\n", i);
8
i=i+1;
}
...
If you want a loop which only ends when break is done, you can use while(true)
or for(;;).
2.6 Aggregation
Variables of the same or dierent types can be bundled together into a single obect.
2.6.1 arrays
Variables of the same type can be put into arrays.
char letter[50];
denes an array of 50 characters, letter[0] being the 1st and letter[49] being
the last character. C++ arrays have no subscript checking; if you go o the end of
an array C++ wont warn you.
Multidimensional arrays can be dened too. E.g.
char values[50][30][10];
denes a 3D array. Note that you cant access an element using values[3,6,1];
you have to type values[3][6][1].
The Standard Library oers various alternatives (vector, for example) which are
often preferable to the basic arrays described above because
They can be made to grow on demand
Many routines for copying and modifying them exist.
2.6.2 structures and classes
Variables of dierent types can be grouped into a structure or a class.
class person {
int age;
int height;
string surname;
};
person fred, jane;
9
denes 2 objects of type person each of 3 elds. Fields are accessed using the .
operator. For example, fred.age is an integer which can be used in assignments
just as a simple variable can.
Structure and Class objects may be assigned, passed to functions and returned,
but they cannot compared, so continuing from the above fragment fred = jane;
is possible (the elds of jane being copied into fred) but you cant then go on to
do
if (fred == jane)
fprintf(stderr, "The copying worked ok\n");
- you have to compare eld by eld.
Classes can contain member functions as well as data. The data can be made
private so that other objects cant access it directly.
2.7 Pointers
Even if you dont use pointers yourself, the code youll learn from is likely to have
them. Suppose i is an integer. To nd the address of i the & operator is used
(&i). Setting a pointer to this value lets you refer indirectly to the variable i. If
you have the address of a variable and want to nd the variables value, then the
dereferencing operator * is used.
...
int i=0;
cout << i << " is at memory location " << &i << endl;
// The next statement declares i_ptr to be a pointer at
// an integer. The declaration says that if i_ptr is
// dereferenced, one gets an int.
int *i_ptr;
i_ptr = &i; // initialise i_ptr to point to i
// The following 2 lines each set i to 5
i = 5;
*iptr = 5; // i.e. set to 5 the int that iptr points to
Pointers arent just memory addresses; they have types. A pointer-to-an-int is of
type int*, a dierent type to a pointer-to-a-char (which is char*). The dierence
matters especially when the pointer is being incremented; the value of the pointer
is increased by the size of the object it points to. So if we added
iptr=iptr+1;
in the above example, then iptr wouldnt be incremented by 1 (which would make
it point somewhere in the middle of i) but by the length of an int, so that it would
10
point to the memory location just beyond i. This is useful if i is part of an array.
In the following fragment, the pointer steps through an array.
...
int numbers[10];
numbers[0] = 1;
numbers[1] = 2;
numbers[2] = 3;
int *iptr = &numbers[0]; // Point iptr to the first element in numbers[].
// A more common shorthand is iptr = numbers;
// now increment iptr to point to successive elements
for (int i=0; i<3; i++){
printf("*iptr is %d\n", *iptr);
iptr= iptr+1;
}
...
Pointers are especially useful when functions operate on structures. Using a pointer
avoids copies of potentially big structures being made.
typedef struct {
int age;
int height;
string surname;
} person;
person fred, jane;
int sum_of_ages(person *person1, person *person2){
int sum; // a variable local to this function
// Dereference the pointers, then use the . operator to get the
// fields
sum = (*person1).age + (*person2).age;
return sum;
}
Operations like (*person1).age are so common that theres a special, more natural
notation for it: person1->age.
2.8 Functions
The form of a function denition is
<function return type> <function name> ( <formal argument list> )
{
<local variables>
11
<body>
}
E.g.
int mean(int x, int y)
{
int tmp;
tmp = (x + y)/2;
return tmp;
}
or, if the function is a member of class stats, ::, the scope resolution operator can
be used to specify which particular function of that name to access.
int stats::mean(int x, int y)
{
int tmp;
tmp = (x + y)/2;
return tmp;
}
You can provide default values for the trailing arguments. E.g.
int mean(int x, int y=5)
{
int tmp;
tmp = (x + y)/2;
return tmp;
}
...
int foo=mean(7)
is legal, setting foo to 6.
You can have more than function of a particular name provided that the compiler
knows unambiguously which one is required for a function call, so the functions
with identical names have to take dierent arguments or must only be visible one
at a time. Its not enough for functions to dier only in the return type.
By default, the variables given to a function wont have their values changed when
the function returns. In the following fragment number wont be increased to 6.
void add (int x)
{
x = x+1;
}
int number = 5;
add(number);
12
When add(number) is called, the x variable is set to the current value of number,
then incremented, but number is unchanged. The following slight variation that
uses call by reference will increase number
void add(int& x)
{
x = x+1;
}
int number = 5;
add(number);
2.9 Declarations
First, a note on terminology. A variable is dened when it is created, and space
is made for it. A variable is declared when it already exists but needs to be re-
described to the compiler (perhaps because it was dened in another source le).
Think of declaring in C++ like declaring at customs admitting to the existence
of something. In C++ you can only dene a variable once but you can declare it
many times.
C++ declarations are not easy to read. You shouldnt need to use complicated
declarations so dont worry too much if you cant decode them. Keep a cribsheet
of useful ones The following examples show common declarations.
int *p pointer to an int
int x[10] an array of 10 ints
const int *p pointer to a constant int
int * const p a constant pointer to an int
int (*x)[10] a pointer to an array of 10 ints
int *x[10] array of 10 pointers to ints
int (*f)(int) pointer to a function taking and returning an int
void (*f)(void) pointer to a function taking no args and returning nothing
int (*f[])(int) An array of pointers to a functions taking and returning an int
Note the importance of the brackets in these declarations. On the Teaching System
the c++decl program can help. E.g.
c++decl explain "char *&x"
prints
declare x as reference to pointer to char
typedef can be used to create a shorthand notation for a verbose type.
typedef int (*PFI)(int) // declare PFI as pointer to function that
// takes and returns an int.
PFI f[]; // declare f as an array of pointers of type PFI.
13
2.10 Memory Allocation
Space is automatically set aside for variables when they are dened, but sometimes
you dont know beforehand how many variables youll need or just how long an
array might need to be. The Standard Library copes with such situations but you
may need to deal with things yourself.
The new command creates space on the y. new[] creates arrays. There are
corresponding delete and delete[] calls.
int i; cout <<"Type an integer "; cin >> i; int *ia = new int[i];
creates an integer array of the required size that can be destroyed by
delete[] ia;
If theres not enough memory for new to provide what you want, it will cause a
bad_alloc exception (see section 2.12 for details).
2.11 Scope and Namespaces
A namespace is like a class that has no functionality. Stroustrup uses them a lot.
Theyre a way of controlling scope - the part of the program where the entity has
a meaning. The more localised variables are the better. Theres
le scope - entities can be made visible only to entities in the same le.
function scope - entities can be made visible only to the function theyre
created in.
block scope - delimited by { and }.
class scope - entities can be made visible only to the class theyre created in.
namespaces - You can put an entity into a namespace by doing something
like
namespace test {
int i;
}
then using test::i to access the variable. The command using namespace test
will make all the entities in the namespace available for the rest of the unit
that the statement is in, without the test:: being necessary. Its tempt-
ing to put using namespace std at the top of each le to access all the
standard routines, but this pollutes the global namespace with many rou-
tine names youll never use, so consider using more specic commands like
using std:string
Its possible for a local variable to mask a global variable of the same name. If, in a
function that has a local variable i you want to access a global variable i, you can
use ::i, but its better to avoid such nameclashes in the rst place.
14
Advanced Programming with C++3 OBJECT-ORIENTATED PROGRAMMING
2.12 Exceptions
p.356
In C every call had to be checked for error values - which could double the code
size. C++ exceptions are an alternative to traditional techniques when they are
insucient, inelegant, and error-prone. Three keywords are involved
try - species an area of code where exceptions will operate
catch - deals with the exceptional situation produced in the previous try clause.
throw - causes an exceptional situation
When an exception is thrown it will be caught by the local catch clause if one
exists, otherwise it will be passed up through the call hierarchy until a suitable
catch clause is found. The default response to an exception is to terminate.
There are many types of exception - they form a class hierarchy of their own. Here
just a few will be introduced.
try {
// code
}
// catch a standard exception
catch(std:exception& e){
// order of the catch clauses matters; theyre tried
// in the given order
}
// catch all the other types
catch(...) {
// cleanup
throw; // throw the exception up the hierarchy
}
Heres an example that sooner or later will produce an exception. p.128
try{
for(;;)
new char[10000];
} catch(bad_alloc) {
cerr << "No memory left";
}
A useful exception is out_of_range (header <stdexcept>), which is thrown by at()
and by bitset<>::operator[]().
Exception handling code can slow a program down even when no exceptions happen.
The code size will also increase.
3 Object-orientated programming
Object-orientated languages support to a greater or lesser extent the following con-
cepts
15
Encapsulation (including in an object everything it needs, hiding elements
that other objects neednt know about). This keeps data and related routines
together and unclutters the large-scale organisation of the program. Each
object has a public set of routines that can be called, and these routines are
all that other objects need to know.
Inheritance (creating new types of objects from existing ones). Rather than
having many seemingly unrelated objects, objects can be organised hierarchi-
cally, inheriting behaviour. Again, this simplies the large-scale organisation.
Polymorphism (dierent objects responding to the same message in dierent
ways). Rather than having a dierent routine to do the same thing to each of
many dierent types of objects, a single routine does the job. An example of
this is how the + operator can be overloaded in C++ so that it can be used
with new classes.
Using an Object-orientated language often means that
program entities can more closely model real-world entities. As Stroustrup
wrote, For small to medium projects there often is no distinction made be-
tween analysis and design: These two phases have been merged into one.
Similarly, in small projects there often is no distinction made between design
and programming.
complexity is more localised
code re-use is easier
3.1 Classes
Classes are at the heart of C++s object-orientation. Classes are created with data
in them, and constructors (routines that are run when an object is created) and
member functions are associated with a class.
3.2 Class member privacy
Class members can have 3 types of privacy
private - can be used only by member functions and friends
protected - like private except that derived classes have access to them too. Try
not to use this - it can help performance but it breaks encapsulation.
public - available to any function.
By default class members are private. You often want the data to be private (so
that they cant be tampered with from the outside) but the functions public (so
that they can be called by other objects). If you want other objects to be able to
change the data, write a member function that the objects can call and make sure
that the member function does validity checking. Then other member function that
use the data wont have to check for validity rst.
16
A private function is often called a helper or utility function because its for the
benet of other member functions. Note that a member function p.278
1. can access private data
2. is in scope of class
3. must be invoked on an object
If you have the choice of writing a friend or member function, choose a member
function. p.280
Note that structs have no private or protected parts.
3.3 Friend classes and functions
It is sometimes useful to let the functions of one class have access to the components
of another class without making the components public and without the overhead
of having to call member functions to get private data.
The following shows how one class (in this case Another) can let another class (in
this case Base) be a friend class
class Another { friend class Base; int value; ... }
For the sake of clarity its a good idea to have the friend lines at the start of the
class denition. Note that friendship is granted by the class being accessed, not
claimed by the class wanting access.
A friend may be a non-member function, a member function of another class, or an
entire class. Friends and derived classes dier in that derived classes might not be
able to access all the members of the base class.
3.4 A complete example
The intTree class, given in the appendix, is an example of a class built from scratch.
The intTree class makes use of a binary node class, bnode to hold and integer and
a count and pointers to left and right branches of a binary tree. intTree has a
constructor, intTree, and three member functions, find, insert and print.
The code shown here has no destructor dened. As an exercise try to nd out
what happens to memory after repeated use of intTree and insert the appropriate
destructors to rectify matters.
3.5 Derived classes
The more you use C++ the more youll be developing classes which require greater
sophistication. Often youll need to add extra functionality to an existing class.
C++ provides a mechanism to build new classes from old ones
17
class More : public Base {
int value;
...
};
Here More inherits the members of Base. The public keyword means that the
members of Base are as accessible to More as Base allows. By default derivation
is private (which means that even Bases public members cant be accessed by
More), but usually derivation should be public, privacy being control by the base Coplien, p.103
class.
When an object of type More is created, rst the Base constructor is called, then
Mores constructor. When the object is destroyed, the destructors are called in
reverse order. The constructors/destructors are not inherited, but you can say
More::More(int sz):Base(sz){}
passing arguments to a base member class constructor (or to a number of construc-
tors using a comma-separated list).
A derived class can be assigned to any of its public base classes, so the following is
possible
class More : public Base {
int value;
...
};
Base b;
More m;
b=m;
It will ll the elds of b with the corresponding values in m. However, m=b isnt
possible.
You can derive new classes from a derived class, and you can derive many dierent
classes from a base class so you can develop a family tree of related classes. As
well see soon, the parent classes can keep parts of themselves private, not only
to unrelated classes but to derived classes too.
Big classes dont necessarily imply big objects: all objects of a class share member
functions, for instance.
Its possible for a class to be derived from 2 or more other classes. Its called Multiple
Inheritance but it introduces complications, so dont use it unless you know what
youre doing.
3.6 Static members
p.228
A static member is part of a class, but not part of an object of a class, so theres
only one instance of them however many objects of that class are created. This is
useful if, for instance, you want to keep a count of how many objects are created.
A static member function
18
1. can access private data
2. is in scope of class
3. but cant be invoked on an object - objectname.staticmemberfunction() is im-
possible.
3.7 Const members
p.229
The value of a const data member cant be changed. A const member function
cant change the data of the class.
int day() const { return d};
doesnt modify state of the class.
3.8 Static Const members
Details regarding how to initialise static const members changed late in the C++
specication. A static const data member of integral or enum type can be initialized
in the class denition, but you still need to provide another denition (without an
initializer) for it. So
class C {
const static int csi = 5;
};
const int C::csi;
is legal, but you cant do the same for oats, non-statics or non-consts. With some
old compilers you have to initialise outside the class.
3.9 Overriding behaviour
As well as adding members a derived class can override behaviour simply by reden-
ing a member function. For instance, if you are developing objects for a graphics
editor, youd like each kind of object to have a draw function, but youd like each
kind of object to have its own version.
Later (section 5) more details of class derivation will be described.
3.10 Templates
A class template is a type generator: a way to fabricate many similar classes or
functions from one piece of code. This is how the Standard Library can support
many types of vector, etc. C++s Templates are parameterized types. They
support generic programming by obviating the need for explicit switch statements
to deal with various data types. Lets look at an example. The following code swaps
2 integers in the usual C++ way
19
Advanced Programming with C++ 4 THE STANDARD LIBRARY
void swap (int& a, int& b) {
int tmp = a;
a = b;
b = tmp;
}
If you wanted to swap other types of variables (including those youve dened) you
could copy this code, replacing int by (say) float. But in situations like this,
where the code is potentially generic, templates can be used.
template <class T>
void swap (T& a, T& b) {
T tmp = a;
a = b;
b = tmp;
}
Here the T name is arbitrary (like a formal parameter in a function denition)
and the class keyword can be replaced by the newer typename keyword. Note
that the function is written much as before. When the compiler is now given the
following code
int a = 3, b = 5;
float f=7.5, g=9.5;
swap (a, b);
swap (f, g);
it will create a version (instantiation) of swap for each type required. These
versions have nothing in common except that they were created from the same
template - they wont share members, not even static ones. Templates are an
example of source code re-use not object code re-use.
You can have template functions, but also template classes and derived classes.
In the appendix we give a full listing, genTree.cpp, of the template version of
the binary search tree. The programmer can now create a binary search tree of
integers, characters, strings or any other data type provided a comparator function
is available.
As an exercise, make use of the genTree class to count the occurrences of ascii
characters in any text le. Then adapt your program to count word usage in any
text.
4 The Standard Library
The Standard Library includes container classes: classes whose purpose is to contain
other objects. There are classes for vectors, lists, etc. Each of these classes can
contain any type of object (as long as the object can be copied safely - if it cant, use
pointers to elements instead). You can, for example, use a vector<int> (meaning p.466
a vector of ints) in much the same way as you would use an ordinary C array,
except that vector eliminates the chore of managing dynamic memory allocation
by hand.
20
vector<int> v(3); // Declare a vector (of size 3 initially) of ints
v[0] = 7;
v[1] = v[0] + 3;
v[2] = v[0] + v[1]; // v[0] == 7, v[1] == 10, v[2] == 17
v.push_back(13); // add another element - the vector has to expand
Note the use of the push_back function - adding elements this way will make the
container expand.
The Standard Library also includes about 70 algorithms that manipulate the data
stored in containers - reverse, insert, unique, transform etc. Note that these opera-
tions act on the elements without explicit use of loops.
To use these features you need to include the corresponding header le. For example,
#include <vector> is needed for vectors.
Read the Standard C++ Library Class Reference .
4.1 Iterators
Iterators are like pointers. They save algorithm writers the worry of having to
deal with the internals of containers. They come in various types: random access,
bidirectional, forward, reverse, input, output. They can be const. Some containers
support only some Iterator types (you cant for example randomly access a list).
See the documentation for details.
4.2 Strings
In older C++ books and in C books, strings are arrays of characters - C-style
strings. Except when speed is vital, C++ strings are preferable to these. For
some operators (concatenation, for example) strings may be a lot faster than Cs
character arrays. Also
They grow on demand
The Standard Library algorithms work on them
Other containers use them
Natural constructions like s=s1+s2; are possible
String objects have many features in common with other Standard Library objects.
Since strings are familiar to you, its useful to look at the facilities in more detail
to prepare you for other Standard Library components.
4.2.1 Iterators
The following routines provide iterators (pointers) of various kinds, depending on
whether you want to go forwards or backwards through the string and whether you
want to change the string. The same kind of routines exist for vector etc.
21
// Iterators - note the various types
iterator begin();
const_iterator begin() const;
iterator end();
const_iterator end() const;
reverse_iterator rbegin();
const_reverse_iterator rbegin() const;
reverse_iterator rend();
const_reverse_iterator rend() const;
4.2.2 Size
These routines (similar ones exist for other containers) give you size information. p.456
reserve() doesnt change size or initialize elements - resize() does. capacity()
size() = reserved size.
// Capacity
size_type size() const;
size_type length() const;
size_type max_size() const;
void resize(size_type, charT);
void resize(size_type);
size_type capacity() const;
void reserve(size_type);
bool empty() const;
4.2.3 Routines
These string-specic routines (only a selection is shown below) provide the func-
tionality that C people are used to. The names arent too hard to remember, and
the default arguments are reasonable. The routines are overloaded, so find can be
used to look for strings, characters or even C-strings.
Note also that the c_str routine returns a C string.
const charT* c_str() const;
size_type find(const basic_string&,
size_type = 0) const;
size_type find(charT, size_type = 0) const;
size_type rfind(const basic_string&,
size_type = npos) const;
size_type rfind(charT, size_type = npos) const;
size_type find_first_of(const basic_string&,
size_type find_last_of(const basic_string&,
size_type find_first_not_of(const basic_string&,
size_type find_last_not_of(const basic_string&,
basic_string substr(size_type = 0, size_type = npos) const;
int compare(const basic_string&) const;
int compare(size_type, size_type, const basic_string&) const;
22
The case history (section ??) has many examples of string usage. Here are 2 short
programs -
#include <fstream.h>
#include <string>
int main()
{
string s1= "pine";
string s2= "apple";
cout << "length of s1 is " << s1.length() << endl;
string s3= s1+s2;
cout << "s1 + s2 = " << s3 << endl;
string s4= s3.substr(2,4);
cout << "s3.substr(2,4)=" << s4 << endl;
cout << "s3.find(\"neap\")=" << s3.find("neap") << endl;
cout << "s3.rfind(p)=" << s3.rfind(p) << endl;
return 0;
}
To get a line of text (that may include spaces) from the user, do p.30
#include <fstream.h>
#include <string>
int main()
{
string str;
cout <<"Type a string\";
getline(cin,str);
cout <<"you typed" << str << endl;
return 0;
}
4.3 vector
p.447
Standard Library Vectors can have various element types and like other containers
can be initialised in various ways:
vector <int> v; // an empty integer vector
vector <int> v(5); // 5 elements, initialised to the default
// value of the elements, in this case 0.
// Vector resizing is expensive, so set to
// the max size if you can.
vector <int> v(5,13); // 5 elements initialised to 13
vector <int> w(v.begin(),v.end()); // initialised from another container.
Many routines are available for vectors. For example, if you have a vector of strings, p.452
you can remove all the strings beginning with p by doing the following
23
sort(fruit.begin(), fruit.end());
// create an iterator p1 for use with a vector of strings
// and set it to point to the first string to remove
vector<string>::iterator p1=
find_if(fruit.begin(), fruit.end(), initial(p));
// create and set p2 to point to the first string after
// those that begin with p
find_if(p1, fruit.end(), initial_not(p));
// Now erase
fruit.erase(p1, p2);
Note that you cant do the following (searching from each end of the vector) because
forward and reverse iterators are dierent types, and you cant use erase with mixed
types.
sort(fruit.begin(), fruit.end());
find_if(fruit.begin(), fruit.end(), initial(p));
vector<string>::reverse_iterator p2=
find_if(fruit.rbegin(), fruit.rend(), initial(p));
fruit.erase(p1, p2);
These above examples use the standard algorithms, which are always useful as a
fallback option, but the vector class has a more specialised routine which is likely
to be faster
fruit.remove_if(initial(p));
p.454
Remember that
vector elements can move! insert, for instance, changes its 1st argument.
vector doesnt have arithmetic operations: valarray (which is described
later) does. p.51
vector doesnt have range-checking by default.
4.4 Queues
Variations include p.478??
deque - (pronounced deck) double-ended queue. One of the basic containers.
queues - items are pushed into the back of the container and removed from
the front. The underlying container can be vector or a deque. The items
can be whatever you like.
// create a message structure
struct Message{ ... }
24
// wait until an item appears on the queue, then process it.
void server(queue<Message> &q)
{
while(!q.empty()) {
Message& m = q.front();
m.service();q.pop();
}
}
priority queues - items are pushed into the back of the container and removed
from the front in order of priority. The underlying container can be vector,
amongst other things. In the following example the priority is determined by
comparing the value of the integers in the queue using less.
#include <queue>
#include <vector>
#include <iostream.h>
int main()
{
// Make a priority queue of int using a vector container
priority_queue<int, vector<int>, less<int> > pq;
// Push a couple of values
pq.push(1);
pq.push(7);
pq.push(2);
pq.push(2);
pq.push(3);
// Pop all the values off
while(!pq.empty()) {
cout << pq.top() << endl;
pq.pop();
}
return 0;
}
4.5 list
Lists are useful when items are likely to be inserted/deleted from the middle of long
sequences. This example creates a vector then copies it into a list, reversing as it
does so.
#include <vector>
#include <list>
#include <algorithm>
int main()
{
vector<int> V;
V.push_back(0);
V.push_back(1);
V.push_back(2);
25
list<int> L(V.size());
reverse_copy(V.begin(), V.end(), L.begin());
exit(0);
}
4.6 map
A map stores a mapping from one set of data (e.g. Names) to another (e.g. Phone
numbers). The following example shows how to set up an associative array an
array where you can use a string as an index. It records how many times each word
appears on cin, the standard input.
int main()
{
string buf;
map<string, int> m;
while (cin>>buf) m[buf]++;
// write out result (see next example)
}
A map keeps items in order and has no duplicate keys. Variations include p.478??
hash_map - an unsorted map.
multi_map - can have duplicate keys
set - a map with ignored values
p.483
// read in lines of the form
// red 7
// blue 3
// red 4
void readitems(map<string, int>& m)
{
string word;
int val=0;
while(cin>>word>>val) m[word]+=val;
}
int main()
{
map<string, int>tbl;
readitems(tbl);
int total=0;
// We want to create an iterator p for use with this type of map.
// Since were not going to change the map values using it,
// well make a const_iterator. We could do this using
// map<string, int>::const_iterator p
// but if we wanted to create other iterators of this type
26
// it would be tedious. So we use typedef to make CI a
// shorthand for the long expression.
typedef map<string, int>::const_iterator CI;
for (CI p=tbl.begin(); p!=tbl.end(); ++p) {
total+=p->second;
cout<<p->first << \t << p->second << \n;
}
cout << "total\t" << total << endl;
return !cin;
}
Heres a fragment using multimap showing how to print out all the values associated
a particular key (in this case the string Gold). It uses typedef as in the above
example, and the useful equal_range function which returns an iterator to the rst
and last elements with a particular key value. Pairing up variables is so common
that a facility called pair declared in the <utility> le is available to assist in their
creation. The pair used here holds 2 iterators, but it can hold 2 items of dierent
types too.
void f(multimap<string,int>&m)
{
typedef multimap<string, int>::iterator MI;
pair <MI,MI> g = m.equal_range("Gold");
for (MI p=g.first; p!=g.second; ++p) {
// print the value out
}
}
In maps, m["blah"]=1; overwrites, m.insert(val) doesnt.
4.7 bitset
This provides ecient operators on bits.
#include <stdio.h>
#include <bitset>
int main()
{
const unsigned int width = 32;
bitset<width> b1("10111");
bitset<width> b2 = 1023;
bitset<width> b3;
cout << "Type in a binary number (type a non-binary digit to end) ";
cin >> b3;
cout << "You typed " << b3 << endl;
cout << "b2 flipped=" << b2.flip() << endl;
cout << "b1=" << b1 << endl;
return 0;
}
27
4.8 valarray
p.663
valarrays are designed for doing fast vector calculations. They can be initialised
in various ways
// v1 is a vector of 2000 integer elements each initialised to 7
valarray<int> v1(7,2000);
// v2 is initialised to the first 4 elements of d
const double d[]= {1.0, 1.0, 2.0, 3.0, 5.0};
valarray<double> v2(d,4);
// v3 is a copy of v2
valarray<double> v3=v2;
Operations include
v2 = 5; // sets each element to 5
v3 *= .5; // halve each element
v2 = v3.shift(2); // copy the elements of v3 shifted 2 to the left, into v2
// filling spaces with 0.0
v3 = v2.cshift(-2) //copy the elements of v2 shifted 2 to the right, into v3
// filling the spaces with elements that have fallen off
// the other end
v2= v2*cos(v3);
A slice is what you get when you take every nth element of a valarray. This is p.671
especially useful if the valarray represents the values in a 2D array - slices are the
rows or columns. A slice needs to know the index of the rst element, the number
of elements and the stride - the step-size.
slice_array<double>& v_even= v2[slice(0,v2.size()/2,2);
slice_array<double>& v_odd= v2[slice(1,v2.size()/2,2);
v_odd *=2 ; // double each odd element of v2
A gslice lets you extract subarrays from a valarray that represent the values in a p.677
2D array.
A mask array lets you pick out arbitrary elements. It needs to contain bool values p.678
and shouldnt be bigger than the valarray its used with. When used as a subscript,
the valarray elements corresponding to the true elements of the mask array will be
chosen.
bool b[]= { true, false, true, true};
valarray<bool>mask(b,4);
valarray<double>v4= v[mask];
Theres also an indirect array. p.679
valarrays are built for speed - they arent intended to grow, and operations arent
error-checked - if you multiply together 2 valarrays of dierent lengths the be-
haviour is undened.
28
4.9 Algorithms
p.57
Algorithms (in <algorithm>) operate on containers, including strings. They may
not always be the fastest option but theyre often the easiest. They fall into 6
groups
Search algorithms - search(), count_if(), etc.
Sorting algorithms - sort(), merge(), etc.
Deletion algorithms - remove(), unique(), etc.
Numeric algorithms - partial_sum(), inner_product(), etc.
Generation algorithms - generate(), for_each(), etc.
Relational algorithms - equal(), min(), etc.
Algorithms usually take as their rst 2 arguments iterators to mark the range of
elements to be operated on. For example, to replace A by B in a string str one
could do
replace(str.begin(), str.end(), A, B);
Heres one way to count how many instances of a particular character there are in
a string using the find algorithm.
int count (const string& s, char c)
{
string::const_iterator i=find(s.begin(), s.end(),c);
int n=0;
while(i!=s.end()){
++n;
i=find(i+1, s.end(),c);
}
return n;
}
Heres a way to count the number of z characters in a C-style string using the
count algorithm. Note how the character array can be used like other containers. p.59
void g(char cs[], int sz)
{
int i1 = count(&cs[0], &cs[sz], z);
}
Heres the count algorithm again, this time being used to nd a particular complex
number in a list.
void f(list<complex>&lc)
{
int i1 = count(lc.begin(), lc.end(), complex(1,3));
}
29
Other algorithms include next_permutation and prev_permutation
int main()
{
char v[]="abcd";
cout << v << \t;
while (next_permutation(v, v+4))
cout << v << \t;
}
random_shuffle,
char v[]="abcd";
cout << v << \t;
for (int i=5;i;i--) {
random_shuffle(v, v+4);
cout << v << \t;
}
and a routine to cycle elements
rotate (first, middle, last)
rotate swaps the segment that contains elements from rst through middle-1
with the segment that contains the elements from middle through last.
4.10 Set algorithms
Set algorithms include
includes()
set_union()
set_intersection()
set_difference()
set_symmetric_difference()
The following, while demonstrating some of these routines, uses 2 ideas that are
often useful
an output stream can be thought of as a container, so you can copy other
containers onto it. The line
copy(s1.begin(), s1.end(), ostream_iterator<int>(cout," "));
also puts spaces between the integers being printed. For now, dont worry if
you dont understand the mechanics - its useful so just use it!
30
Sometimes you wont know how many resulting elements will be produced
by an operation, you just want them all added to a container. inserter
(which calls the containers insert() routine), front inserter (which calls
the containers push_front() routine - which vectors for example dont have)
or back inserter (which calls push_back()) do that for you.
#include <algorithm>
#include <set>
#include <iterator> // needed for ostream_iterator
int main()
{
//Initialize some sets
int a1[10] = {1,2,3,4,5,6,7,8,9,10};
int a2[6] = {2,4,6,8,10,12};
set<int> s1(a1, a1+10), s2(a2, a2+6), answer ;
cout << "s1=";
copy(s1.begin(), s1.end(),
ostream_iterator<int>(cout," "));
cout << "\ns2=";
copy(s2.begin(), s2.end(),
//Demonstrate set_difference
set_difference(s1.begin(), s1.end(),
s2.begin(), s2.end(), inserter(answer,answer.begin()));
cout << "\nThe set-difference of s1 and s2 =";
copy(answer.begin(), answer.end(),
cout << endl;
return 0;
}
4.11 Using member functions
p.311
Algorithms can use the member functions of objects theyre manipulating.
void print_list(set<Employee*>& s)
{
for_each(s.begin(), s.end(), mem_fun(&Employee::print));
}
4.12 Predicates
Some algorithms can be given a function to modify their behaviour. For example,
the find_if algorithm can nd items that conform to a particular condition as
specied by a function. A predicate is a function object (an object that behaves
like a function) that returns a bool showing whether or not an item matches the
condition. p.515
31
Advanced Programming with C++ 5 MORE ON CLASSES
4.12.1 Creating predicates
<functional> supplies common conditions that can operate on various types: e.g.,
greater_equal<int>(), plus<double>(), logical_or<int>() etc. The following
does a vector multiplication of a and b, putting the answer in res to which elements
are added one at a time.
void discount(vector<double>& a, vector<double>& b, vector<double>& res)
{
transform(a.begin(), a.end(), b.begin(), back_inserter(res), multiplies<double>());
// vector multiplication
}
4.12.2 Adapters
p.518
Note: These are sometimes useful but look complicated, so dont worry if you dont
understand them..
Adapters are ways to adapt existing predicates - they take and return a function
that we can then use with Standard Library algorithms. There are 4 types -
binder - to use a 2-arg function object as a 1-arg function. less is a binary predi-
cate. If we want make a unary predicate from it we can do bind2nd(less<int>(),7)).
member function adapter - to use a member function p.520
pointer to function adapter - to use a pointer to a member function
negator - to negate an existing predicate
These can be combined -
... find_if ( ....... not1(bind2nd(ptr_fun(strcmp), "funny")));
5 More on Classes
Concrete Classes are the simple classes that behave just like int, char, etc once p.247
theyre set up.
To make best use of Classes its sometimes useful to redene the operators as well
as dene member functions. For instance, if you were going to invent a class to deal
with strings, it would be nice for + to join strings together. Or if you wanted to
add range-checking to arrays, youd have to modify the [ ] behaviour. Below are
some examples of overloading, but rst more on the mechanism of inheritance.
5.1 Virtual members
As mentioned earlier, its easy to redene a base classs function in a derived class,
but to fully exploit polymorphism (the ability to deal with many types transpar-
ently), a new concept needs to be introduced.
32
Suppose that we were designing a graphics editor. We might have a base class
Shape and various derived classes Triangle, Rectangle, etc, each with their own
print member function. Wed like to collect these objects together into groups
sometimes. An array of pointers to the objects would be appropriate, but what sort
of pointers could we use? Fortunately, its possible to assign a derived class pointer
to a base class pointer, so the following is possible, creating an array of pointers to
the base class.
Triangle t1, t2;
Rectangle r1, r2;
typedef ShapePtr Shape*;
vector<ShapePtr> shapes(4);
shapes[0]=&t1;
shapes[1]=&r1;
shapes[2]=&t2;
shapes[3]=&r2;
Then we could draw all the shapes in the group by calling the print function of
each element. But theres a problem - because the elements are pointers to Shape,
its Shapes print function which is called, rather than the appropriate derived
objects function.
The solution to this is to dene print in the base class as a virtual function by
having something like
virtual void print();
which lets us deal with the shapes array without having to worry about the real
type of each component; the derived objects print function will always be called.
5.2 Abstract Classes
p.313
Some classes represent abstract concepts for which objects cannot exist. The func-
tions have to be dened in the derived classes. Such functions are called pure virtual
functions and are denoted by the following notation
virtual void print() =0;
in the base class. A class with one or more pure virtual functions is an Abstract
Class.
5.3 Redening operators
If you need to redene operators remember to redene then consistently (if you
redene + dont forget +=) and remember that you cant change precedence order. p.198
The default assignment operator does a simple copy of members, which might not
be what you want.
33
Examples of redening operators are in most books. Some things to remember Deitel, p.474,
Lippman, p.737 when redening assignment operators are
that reference parameters help to overload operators eciently (objects arent
copied) while keeping their use intuitive (pointers would require the user to
supply addresses of objects).
to deal with situations where a variable is assigned to itself. A pointer/refer-
ence to the object on the right of the = will be given to the function explicitly.
A pointer to the object that the function is a member of is supplied implicitly
to all non-static member functions. Its called this, so if this equals the
supplied pointer, youd usually want to do nothing.
to return a const. This catches errors like (i+j)=k
to return *this so that assignments can be chained (i.e. i=j=k becomes
possible)
5.4 A class denition example
The following example highlights some of the more subtle issues associated with
Constructors and Destructors.
Its the job of Constructors to allocate the resources required when new objects are
created. If these are created using new then the Destructor should free the resources
to stop memory being wasted. For example, if we want to store information about
criminals we might dene an object as follows
class criminal {
string name;
char* fingerprint;
int fingerprint_size;
}
where fingerprint points to an image. The memory (if any) allocated for the
image would be freed by the destructor.
assignment operator
That works ne until the following kind of situation arises
void routine (criminal a)
{
criminal b;
b=a;
...
}
The assignment b=a copies the bytes of a to b, so the fingerprint eld of both
objects will be the same, pointing to the same memory. At the end of this rou-
tine criminals destructor will be called for b, which will delete the memory that
34
b.fingerprint points to, which unfortunately is the same memory as a.fingerprint
points to.
To cure this we need to redene the assignment operator so that the new object
has its own copy of the resources.
void criminal::operator=(criminal const &b)
{
name=b.name;
fingerprint_size=b.fingerprint_size;
delete fingerprint;
fingerprint = new char[fingerprint_size];
for (int i=0;i<fingerprint_size;i++)
fingerprint[i]=b.fingerprint[i];
}
Note that space used by an existing ngerprint is freed rst before a new ngerprint
image is created - memory would be wasted otherwise.
this
But problems remain. If the programmer writes a=a, the delete command above
frees a.fingerprint, which we dont want to free. We can get round this by making
use of the this variable, which is automatically set to be the address of the current
object. If the code above is bracketted by if (this != &b) { ... } then its
safe.
Theres another improvement that we can make. So that assignments like a=b=c
can work, its better not to return void.
criminal const& criminal::operator=(criminal const &b)
{
if (this != &b) {
name=b.name;
delete fingerprint;
}
return *this;
}
copy constructor
But theres still a problem! Suppose the programmer writes criminal a=b;. a is
being created, so a constructor a called - not an assignment. Thus we need to write
another constructor routine using similar ideas to those used in assignment. Note
that in this case resources dont need deleting, and that this neednt be checked
(because criminal a=a; is illegal).
Putting all these ideas together, and sharing code where possible we get
35
void criminal::copy (criminal const &b)
{
name=b.name;
}
void criminal::destroy()
{
delete fingerprint;
}
// Assignment Operator
criminal const& criminal::operator=(criminal const &b)
{
if (this != &b) {
destroy(this);
copy(b);
}
}
// Default Constructor
criminal()
{
fringerprint=0;
fingerprint_size=0;
}
// Copy Constructor
criminal(criminal const &b)
{
copy(b);
}
// Destructor
void ~criminal()
{
destroy(this);
}
This seems like a lot of work, but its necessary if objects use pointers to resources.
You might think that by avoiding the use of commands like a=a in your own code,
you can save yourself the trouble of writing these extra routines, but sometimes the
problem situations arent obvious. For instance, the copy constructor is called when
an object is returned or used as an argument to a routine. Better safe than sorry.
5.5 Redening [ ]
p.286
vector doesnt have range-checking by default, but if you use the member func-
tion vector::at(), out_of_range exceptions can be trapped, so you can add
36
out_of_range checking by redening [ ] in a class derived from vector, and using
vectors constructors -
template<class T> class Vec: public vector<T> {
public:
Vec() : vector<T>() {}
Vec(int s) : vector<T>(s) {}
T& operator[] (int i) {return at(i);}
const T& operator[] (int i) const {return at(i);}
}
The following more complicated example creates a mapping (relating a string to a
number) without using the Standard Librarys map facility. It does a word-frequency
count of the text given to it.
#include <stdio.h>
#include <string>
#include <vector>
// The Assoc class contains a vector of string-int Pairs.
class Assoc {
struct Pair {
string name;
int val;
// The following line is a constructor for Pair
Pair (string n="", int v=0): name(n), val(v) {}
};
// create a vector of the Pairs weve just created
vector <Pair> vec;
public:
// redefine []
int& operator[] (const string&);
// a member function to print the vector out
void print_all() const;
};
// This redefines [] so that a ref to the value corresponding to the
// string is returned if it exists. If it doesnt exist, an entry
// is created and the value returned.
int& Assoc::operator[] (const string& s)
{
// The next lines creating an appropriate iterator
// for the vector of Pairs
for (vector<Pair>::iterator p=vec.begin(); p!=vec.end(); ++p)
if (s == p->name)
return p->val;
else{
vec.push_back(Pair(s,0));
return vec.back().val;
}
}
37
Advanced Programming with C++ 6 INPUT/OUTPUT
void Assoc::print_all() const
{
for(vector<Pair>::const_iterator p=vec.begin(); p!=vec.end(); ++p)
cout << p->name<<": " << p->val << endl;
}
int main()
{
string buf;
Assoc vec;
while(cin>>buf) vec[buf]++;
vec.print_all();
}
5.6 Redening ()
p.287
Coplien p.165
This can be used to provide the usual function call syntax for objects that in some
way behave like functions (function objects or functors). These are used by the
Standard Library.
5.7 Redening ->
p.289
Smart pointers are pointers that do something extra (e.g. increment a counter)
when they are used.
6 Input/Output
As with C, Input/Output is not part of the language, but support is provided by a
library. Most programs need to use #include <iostream>. p.612
You may also need #include <fstream> for le I/O, #include <iomanip> for
greater control over formatting, and #include <strstream> for I/O to and from
strings.
6.1 Simple I/O
The following illustrates the use of putback() for an input stream. istreams also
have an ignore function (is.ignore(5) would ignore the next 5 characters) and
a peek() function which looks at the next character without removing it from the
stream.
istream& eatwhite(istream& is) { char c;
while(is.get(c)) {
if (!isspace(c)) {
is.putback(c);
break;
38
}
return is;
}
}
The following illustrates the use of I/O to and from les. Given the name of 2 les
on the command line (which are available to main as C-style character arrays), it
does a le copy.
void error(string p, string p2="") {
cerr<<p << << p2 << endl;
std::exit(1);
}
int main (int argc, char* argv[]) {
if (argc != 3) error("wrong number of arguments");
std::ifstream from(argv[1]);
if (!from) error("cannot open input file", string(argv[1]));
std::ofstream to(argv[2]);
if (!to) error("cannot open output file", string(argv[2]));
char ch;
while (from.get(ch)) to.put(ch);
if (!from || !to) error("something strange happened");
}
You can force the screen/le output to be up-to-date by using
cout.flush() or cout << flush;
but this isnt usually necessary.
If you want to write or read raw (binary) data (rather than strings that represent
numbers) you can use write, read and casts (section 7.1) to keep the compiler
happy. The following program writes 100 random integers into a le then reads
them back.
#include <iostream.h> #include <stdlib.h> // to use rand #include
<fstream.h>
int main() {
ofstream outfile("myresults",ios::out|ios::binary);
if (outfile.good() == false) {
cerr << "Cannot write to myresults" << endl;
exit(1);
}
int num;
for (int i=0; i<100; i++){
num=rand();
39
outfile.write(reinterpret_cast<const char*>(&num), sizeof(num));
}
outfile.close();
ifstream infile("myresults");
if (infile.good() == false) {
cerr << "Cannot open myresults" << endl;
exit(1);
}
int count=0;
while(infile.read(reinterpret_cast<char*>(&num), sizeof(num))) {
cout << num << endl;
count++;
}
cout << count << " numbers read in" << endl;
infile.close();
return 0;
}
6.2 Formatting
The way that text and numbers are output can be controlled. p.626
#include <iostream.h> #include <iomanip.h> #include <cmath> int
main() {
int i=10;
cout << "i = " << i << " (default)\n" ;
cout << hex << "i = " << i << " (hex)\n";
double d = sqrt(7.0);
cout << "d=sqrt(7.0)=" << d << endl;
cout.precision(3);
cout << "After cout.precision(3), d=" << d << endl;
return 0;
}
There many other routines too, amongst them
cout.setf(ios_base::oct,ios_base::basefield); // set base to 8
const ios_base::fmtflags myopt = ios_base::left|ios_base::oct;
ios_base::fmtflags old_options = cout.flags(myopt); //set base to 8
// and alignment to left.
cout.precision(8); // set precision to 8
cout.setf(ios_base::scientific, ios_base::floatfield);
cout.width(4); // output at least four characters cout.fill(*);
// fill gaps with * cin.noskipws(); // dont skip white space
40
6.3 Stream Iterators
Iterators can be used on streams, providing an elegant way to integrate I/O with p.60
container classes. Youve seen them already in the Standard Library section.
ostream_iterator<string> oo (cout); int main()
*oo = "Hello";
++oo;
*oo = "World";
6.4 Output of User-Dened types
One way to do this is by overloading the stream insertion operator. The following
denes how complex numbers should be printed out -
ostream& operator<<(ostream&s, complex z) // returns ostream& so can
be chained {
return s << ( << z.real() << , << z.imag() << );
}
6.5 Input of User-Dened types
The following code reads in a complex number thats provided in the form (a,b) or
(a) p.621
istream& operator>>(istream&s, complex &a) // returns istream& so
can be chained {
double re=0, im=0;
char c=0;
s>>c;
if (c== () {
s>>re >> c;
if (c == ,) s >> im >> c;
if (c != )) s.clear(ios_base::badbit);
}
else {
s.putback(c);
s>> re;
}
if (s) a = complex(re,im);
return s;
}
6.6 String streams
stringstreams are streams that are attached to a string rather than a le and
letting you use the same syntax as le I/O. They are dened in <sstream>. p.640
41
Advanced Programming with C++ 7 ODDS AND ENDS
#include <string>
#include <sstream>
int main() {
int i = 7;
string s1 = "He is ";
string s2 = " years old";
ostringstream ostring;
ostring << s1 << i << s2;
cout << "ostring =" << ostring;
return 0;
}
7 Odds and Ends
7.1 Casts
C++ is strict about converting from one type to another - much stricter that C is.
You can use C-style casting but its not recommended - instances are hard to nd
in the code, and its less type-safe. Try to use the least dangerous (most specic)
casting alternative from those below
const cast - used to remove the const qualier.
dynamic cast - does careful, type-sensitive casting at run time. Cant deal with
void*.
reinterpret cast - produces something that has the same bit pattern as the orig-
inal.
static cast - doesnt examine the object it casts from at run time. This is fast
if its safe - and the compiler should tell you if it isnt safe. Can deal with
void*. static_cast can be used to reverse any of the implicit conversions
(like int to oat).
To convert from a const void * to a File_entry** (a common kind of thing to
do in C) I had to do
File_entry**fe2=
reinterpret_cast<File_entry**>(const_cast<void*>(entry2));
42
7.2 Limits
Use <climits> to access information on the systems limits. E.g. to check if a long
variable can t into a short you could try
long i; if (i<numeric_limits<short>::min() ||
numeric_limits<short>::max() <i)
cout << i << "cannot fit into a short"
7.3 Performance
The quality of the compiler and libraries can greatly aect the speed of the resulting
code, but theres a lot you can do to help.
Read about the available optimiser options. On HPs for example, using +O4
instead of no optimisation can speed up some code by orders of magnitude.
Try not to use call-by-value for big objects. Unless you need to change the
object, use a reference to a const value
Some C++ features have compile time implications, others have run time,
program size or debugging implications. For example, making a function
virtual can add 15% to the cost of a function call whereas templates have
no runtime implications. They do, however, make things more dicult for
debuggers and compilers. The Standard Template Library, for example, avoids
virtual functions for performance reasons.
Inlined routines are faster, but theyll bloat the code. Note that a member
function dened in the class declaration is automatically inline. In particular,
note that an inline destructor can be inserted at each exit point of a function.
Though Standard Library algorithms like reverse work on both list and
vector containers, the choice of container is still important for performance.
Though vectors can grow, its more ecient to create them with their maxi-
mum size.
Try to create and initialise a variable in one operation.
7.4 Maths
If youre using any of the maths routines (sqrt, sin, etc) remember that youll need
to include cmath to declare the routines.
Before you start writing much maths-related code, check to see that it hasnt all been
done before. Many maths routines, including routines that oer arbitrary precision
are available from netlib, http://www.hensa.ac.uk/netlib/master/readme.html.
Before you do any heavy computation, especially with real numbers, I suggest that
you browse through a Numerical Analysis book. Things to avoid are
43
Finding the dierence between very similar numbers (if youre summating an
alternate sign series, add all the positive terms together and all the negative
terms together, then combine the two).
Dividing by a very small number (try to change the order of operations so
that this doesnt happen).
Multiplying by a very big number (use logs).
Common problems that you might face are :-
Testing for equality :- Real numbers are handled in ways that dont guarantee
expressions to yield exact results. Just as in base 10 to 2 decimal places,
3 (10/3) doesnt equal 10, so computer arithmetic has its limitations. Its
especially risky to test for exact equality. Better is to use something like
d = max(1.0, fabs(a), fabs(b))
and then test fabs(a - b) / d against a relative error margin. Useful con-
stants in <climits> are FLT_EPSILON, DBL_EPSILON, and LDBL_EPSILON, de-
ned to be the smallest numbers such that
1.0f + FLT_EPSILON != 1.0f
1.0 + DBL_EPSILON != 1.0
1.0L + LDBL_EPSILON != 1.0L
respectively.
Avoiding over- and underow :- You can test the operands before performing
an operation in order to check whether the operation would work. You should
always avoid dividing by zero. For other checks, split up the numbers into
fractional and exponent part using the frexp() and ldexp() library functions
and compare the resulting values against HUGE (all in <cmath>).
Floats and Doubles :- You can use the information in <limits> or use sizeof(double),
etc to compare the size of oat, double and long double.
Keep in mind that the double representation does not necessarily in-
crease the precision over float. Actually, in most implementations the
worst-case precision decreases but the range increases.
Do not use double or long double unnecessarily since there may a large
performance penalty. Furthermore, there is no point in using higher pre-
cision if the additional bits which will be computed are garbage anyway.
The precision one needs depends mostly on the precision of the input
data and the numerical method used.
#include <iostream.h> #include <stdio.h>
int main(int argc, char* argv[]) { float f1; double d1; long double
ld1;
f1=d1=ld1=123.45678987654321; cout.precision(50); cout << "f1=" <<
f1 << ", f1**2=" << f1*f1 << endl; cout << "d1=" << d1 << ", d1**2="
<< d1*d1 << endl;
//The following doesnt work on HPs!
44
//cout << "ld1=" << ld1 << "ld1**2=" << ld1*ld1 << endl;
printf("ld1=%25.25Lf, ld1*ld1=%25.25Lf\n", ld1, ld1*ld1);
return 0; }
Innity :- The IEEE standard for oating-point maths recommends a set of func-
tions to be made available. Among these are functions to classify a value as
NaN, Infinity, Zero, Denormalized, Normalized, and so on. Most imple-
mentations provide this functionality, although there are no standard names
for the functions. Such implementations often provide predened identiers
(such as _NaN, _Infinity, etc) to allow you to generate these values.
If x is a oating point variable, then (x != x) will be TRUE if and only
if x has the value NaN. Some C++ implementations claim to be IEEE 748
conformant, but if you try the (x!=x) test above with x being a NaN, youll
nd that they arent.
The following sythesises a NaN on HP machines, then uses isnan().
#include <iostream.h> #include <cmath>
double make_a_NaN () { double ret; char* pointer =
reinterpret_cast<char*>(&ret); int i;
// IEEE spec for doubles on HPs
// 0 : sign bit
// 1-11: exponent
// 12-63: fraction
// For QNaNs on HPs
// set 1-11 to 1, 12=0, rest to 1
for (i=0; i<8; i++)
*(pointer+i) = 255;
*(pointer+1)= 247;
return ret; }
int main(int argc, char* argv[]) { double d;
d=make_a_NaN (); if (isnan(d))
cout << d << " is a nan" << endl;
else
cout << d << " is not a nan" << endl;
return 0;
}
An increasing amount of maths work is being done with C++. One development is
the Template Numerical Toolkit built around the Standard Library.
http://math.nist.gov/tnt/index.html
45
7.5 Hardware Interfacing: bit operations and explicit ad-
dresses
If youre interfacing with hardware and need to operate on bits you can use bitset
but you may prefer to use C++s low-level operations.
Setting a bit :- Suppose you wanted to set bit 6 of i (a long, say) to 1. First
you need to create a mask that has a 1 in the 6th bit and 0 elsewhere by doing
1L<<6 which shifts all the bits of the long 1 left 6 bits. Then you need to
do a bit-wise OR using i = i | (1L<<6).
Unsetting a bit :- Suppose you wanted to set bit 6 of i (a long, say) to 0. First
you need to create a mask that has a 0 in the 6th bit and 1 elsewhere by doing
1L<<6 then inverting the bits using the ~ operator. Then you need to do a
bit-wise AND using the & operator.
The whole operation is i =i & ~(1<<6) which can be contracted to i &= ~(1<<6).
Creating a mask :- In X graphics, masks are often created each of whose bits
represent a option that is to be selected in some way. Each bit can be referred
to using an alias that has been set up in an include le. For example, a mask
which could be used in a call to make a window sensitive to key presses and
buttonpresses could be set up by doing
unsigned int mask = KeyPressMask | ButtonPressMask ;
Suppose because of hardware considerations the i variable needed to be the value
of memory location 2000. The following code will do what you want.
long *memory_pointer = reinterpret_cast<long*>(2000); long i =
*memory_pointer;
7.6 Using routines from other languages
7.6.1 C
When your C++ code is presented with the prototypes of C routines, the prototypes
need to be bracketed by the following
#ifdef __cplusplus extern "C" { #endif ... #ifdef __cplusplus }
#endif
so that the code compiles.
7.6.2 Fortran
This is a rather machine-specic issue. Rudi Vankemmel ([email protected]) men-
tions some general points that are worth noting :-
46
1. Fortran uses a column wise storage of matrices while C++ stores them row
wise. This means that when you want to parse a matrix from your C++
program to the fortran routine you must transpose the matrix in your program
before entering the routine. Of course, any output from such a routine must
be transposed again.
If you omit this step, then probably your program will run (because it has
data to compute on) but it will generate wrong answers.
If you have the Fortran source code (of any routine) then on some platforms
you may use compiler directives specifying that the Fortran compiler must use
row wise storage. Some platforms support these directives. However watch out
with this if you call the same routine from another Fortran routine/program.
2. Your Fortran compiler may add an underscore to the routine name in the
symbol table. Hence in the calling C++ program you must add a trailing
underscore! Otherwise the loader will complain about an undened symbol.
3. Fortran passes its variables by reference. This means that you MUST give
adresses in your calling C++ program.
4. Watch out especially with floats and doubles. Make sure that the size of the
variable in the calling program is identical to the size in the Fortran routine.
7.7 bad habits to give up
C++ is a exible language. You can write C++ that a C compiler could compile.
You can even write in a style that isnt too distant from Fortran 77. But if youre a
C programmer who wants to make the most of modern C++, the following points
are worth bearing in mind
Think in terms of objects rather than functions. Groups of objects are easier
to organise (using inheritance) than are groups of functions.
Use namespaces: functions can have shorter names; information hiding is eas-
ier; and its easier to switch between dierent implementations. Use names-
pace rather than les to control scoping.
Use references rather than pointers when you can.
Use strings. You can convert between C and C++ strings
char cstring[]="a test";
string a_str;
a_str=string(cstring);
strcpy(cstring, a_str.c_str());
Dont use NULL - use 0
Create variables just before rst use. In particular, create for loop index
variables within the loop construction.
Use exceptions when the code that has to deal with a problem is far from the
code that discovers the problem.
Use const as much as possible. Note that in C++, const variables have
internal linkage by default whereas in C, they have external linkage.
47
Use enum
Dont use type-identifer elds in structures
Avoid pre-processor macros. const, inline, template and namespace replace p.161
most of #define usage, leading to stronger typing and better control of scope.
For example
const int answer = 42; template<class T> inline T min(T a, T b)
{return (a<b)?a:b;}
Avoid static except within functions.
// better than a global
int & use_count() { static int uc =0; return uc; }
or use nameless namespaces
namespace { class X{}; void f(); // like a static int i; }
7.8 Non-essentials (Things you probably dont need to know
yet)
7.8.1 virtual inheritance
Inheritance can be private (where even the public elds of the base class are
treated as private), public (where the base class determines visibility), or virtual.
With virtual inheritance there is only one copy of each object even if (because
of multiple inheritance) the object appears more than once in the hierarchy. In
the following example, derived1a and derived1b each contain a base object, but
derived2 contains just one i eld.
class base{ public:
int i;
};
class derived1a: virtual public base { public:
int j;
};
class derivedb: virtual public base { public:
int k;
};
class derived2: public derived1a, public derived1b{ public:
int product() {return i*j*k;}
};
main() { derived2 obj; obj.i = 10; obj.j = 3; obj.k = 5;
cout << "product is" << obj.product() << \n; return 0; }
48
7.8.2 typeid
p.414
In C++ you can identify the type of an object dynamically using RTTI (run-time
type information). Given a type name, typeid returns a reference to a type_info
object (in <type-info>). Can also give it an expression
#include <typeinfo> void g(Component* p) { cout <<
typeid(*p).name();
}
7.8.3 virtual destructors
p.422
You cant have virtual constructors, but you can have virtual destructors. Theyre Coplien p.140
not quite ordinary functions - no pointer to them is possible, for example).
7.8.4 Auto pointers
p.367, Deitel
p.688, Lippman
p.409
auto_ptr in <memory> - object pointed to will be deleted when pointer goes out of
scope
7.8.5 operator functions
An operator function can be a member or non-member. The assignment, subscript,
call and member selection operators are required to be class member functions.
7.8.6 union
A union is like a struct except that the elds occupy the same memory location
with enough memory allocated to hold the largest item. The programmer has to
keep a note of what the union is being used for. What would the following code
print out?
... union person {
int age;
int height;
char surname[20];
} fred;
fred.age = 23;
fred.height = 163;
cout << "Fred is " << fred.age << "years old\n"; ...
If fred started at memory location 2000, then fred.age, fred.height and fred.surname
would all begin at memory location 2000 too, whereas in a struct the elds wouldnt
overlap. So setting fred.height to 163 overwrites fred.age (and the 1st 4 char-
acters of fred.surname) making fred 163 years old.
49
Advanced Programming with C++ 8 PROGRAM DEVELOPMENT
7.8.7 Non-type template parameters
Values as well as types can be used as template parameters.
template < class T, int size> class MyArray { ...
This lets the template create arrays of a xed size (good for performance reason)
but it means that a new piece of code will be created for each dierent class/size
combination used.
8 Program Development
Stroustrup writes that the last section of his book aims p.691
to bridge the gap between would-be language-independent design and pro-
gramming that is myopically focussed on details. Both ends of this spec-
trum have their place in a large project, but to avoid disaster and exces-
sive cost, they must be part of a continuum of concerns and techniques.
Programs vary considerably in their composition. Ive seen these gures quoted
Coding takes 10% of the total time. Debugging takes 50%.
The Graphical User Interface is 80% of the code
Error handling can be 50% of the code
Complexity is the enemy, so
Divide and conquer.
Use modules - namespaces or les (helps the optimiser too).
Dont re-invent the wheel
Copy models
Adapt existing parts
When making new parts design them for re-use
8.1 Style
It helps if you decide upon a uniform style for writing code. Its common to suggest
that for all except the most trivial classes its wise to dene Coplien p.38-45
a void constructor -
50
a copy constructor - to create a new object using values from an existing one.
Use call-by-reference to avoid innite recursion. Deitel p.483
the assignment operator - that returns a reference.
And then theres the the rule of three - if you write any one of a copy constructor,
copy assignment operator or destructor, then you will almost certainly need to write
all three for your class to function properly.
Stanley Lippman in Dr.Dobbs Journal, October 99, (p.40) noted that unnecessary
denition of these functions is likely in practise to make the code bigger and/or
slower - he got a 40% speed improvement by removing all 3 from a piece of code,
the compilers default alternatives being more ecient.
When deriving classes Coplien p.192
if using public inheritance, most base class member functions should be de-
clared virtual to ensure that the derived class customisations will override
the base class behaviour even in a context where a base class object is ex-
pected.
use virtual inheritance if you are not especially concerned about performance
or if you might use multiple inheritance later.
8.2 Makeles
If you have many source les you dont need to recompile them all if you only change
one of them. By writing a makefile that describes how the executable is produced
from the source les, the make command will do all the work for you. In the 1A/1B
courses makeles were created by the xcc
1
program, but you may well need to write
your own eventually.
The following makele says that pgm depends on two les a.o and b.o, and that
they in turn depend on their corresponding source les (a.cc and b.cc) and a
common le incl.h:
pgm: a.o b.o
aCC a.o b.o -o pgm
a.o: incl.h a.cc
aCC -Aa -c a.cc
b.o: incl.h b.cc
aCC -Aa -c b.cc
Lines with a : are of the form
target : dependencies
make updates a target only if its older than a le it depends on. The way that
the target should be updated is described on the line following the dependency line
(Note: this line needs to begin with a TAB character).
1
http://www-h.eng.cam.ac.uk/help/tpl/languages/xccmake.html
51
Heres a more complex example of a makefile for a program called dtree. First
some variables are created and assigned. In this case typing make will attempt to
recompile the dtree program (because the default target is the rst target men-
tioned). If any of the object les it depends on are older than their corresponding
source le, then these object les are recreated.
The targets neednt be programs. In this example, typing make clean will remove
any les created during the compilation process.
# Makefile for dtree DEFS = -O2 -DSYSV
CFLAGS = $(DEFS) -O
LDFLAGS = CC = aCC
LIBS = -lmalloc -lXm -lXt -lX11 -lm
BINDIR = /usr/local/bin/X11 MANDIR = /usr/local/man/man1
OBJECTS_A = dtree.o Arc.o Graph.o #using XmGraph
ARCH_FILES = dtree.1 dtree.cc Makefile Dtree Tree.h TreeP.h \
dtree-i.h Tree.cc Arc.cc Arc.h ArcP.h Graph.cc Graph.h GraphP.h
dtree: $(OBJECTS_A)
$(CC) -o dtree $(LDFLAGS) $(OBJECTS_A) $(LIBS)
Arc.o: Arc.cc
$(CC) -c $(CFLAGS) Arc.cc
Graph.o: Graph.cc
$(CC) -c $(CFLAGS) Graph.cc
dtree.o: dtree.cc
$(CC) -o dtree.o -c $(CFLAGS) -DTREE dtree.cc
install: dtree dtree.1
cp dtree $(BINDIR)
cp dtree.1 $(MANDIR)
clean:
rm -f dtree *.o core tags a.out
8.3 Debugging
Code Complete by Steve McConnell (Microsoft Press, 1993) is the book to read. It
has very useful sections on testing and debugging, and quotes results from investi-
gations of software quality assurance. On p.567 it notes that the industry average
for code production is 8-20 lines of correct code per day. On p.610 it notes that
industry average experience suggests that there are 15-50 errors per 1000 lines of
delivered code. The recommendation is to design and code defensively - it saves
time in the end.
C++ is a strongly-typed language with many features to help write safe code, but
you still need to be cautious
Let the compiler help you as much as possible! Many compilers generate
52
warning messages if you ask them to.
Assume when youre writing your program that it wont be bug-free. Write it
so that you (and others) can easily debug it: make routines small and control
ow simple.
Concrete examples are easier to debug. You can make them into a template
once theyre working.
Write a test suite (for you and future users)
Dont get stuck at one level - zoom in and pan out. Your code should be un-
derstandable locally (parts should be as isolated from other parts as possible)
and at the top level.
Write defensive code. Always assume the worst.
Its not uncommon for beginners to prove to themselves on paper that their
broken code is correct and the compiler is wrong. Add cout statements to
print out intermediate values and test assertions.
If youre using oating point arithmetic, check divisions to see if youre divid-
ing by zero.
Check that youre not going o the end of arrays.
To comment-out large chunks of code bracket it with
#if 0 ... #endif
See the Debugging handout
2
for details of the wdb debugger
3
.
8.4 Common Diculties and Mistakes
Obscure Error Messages - small errors in a simple-looking line can produce
complicated messages like the following
Error 226: "set1.cc", line 16 # No appropriate function found for
call of
operator <<. Last viable candidate was "ostream &ostream::operator
<<(char)" ["/opt/aCC/include/iostream/iostream.h", line 509]. Argument of
type class set<int,less<int>,allocator> could not be converted to
char.
cout << "s1 = " << s1;
^^^^^^^^^^^^^^^^^^^^^
The ^^^^ symbols show which part of the line is at fault. In this case the
compiler cant narrow down the problem, so next read the rst (easiest) part
of the error message. Theres something wrong with the use of <<. Remember
that in C++ many functions can be invoked by an operator - it all depends
on the number and type of operands. In this case s1 (a set) isnt something
that theres a matching function for - hence the error message. The compiler,
trying to be helpful, gives the next best option - a function tied to << that
can deal with a char.
2
http://www-h.eng.cam.ac.uk/help/tpl/languages/debug/debug.html
3
http://www-h.eng.cam.ac.uk/help/tpl/languages/debug/node5.html
53
Advanced Programming with C++ 9 CIMG, AN IMAGE PROCESSING LIBRARY FOR C++
Another example: the line double d = sqrt(4); produces the following error
because the integer given to sqrt has to be promoted to a real, but the
compiler doesnt know what kind of real.
Error 225: "foo.cc", line 6 # Ambiguous overloaded function call;
more than
one acceptable function found. Two such functions that matched were "long
double sqrt(long double)" ["/opt/aCC/include/cmath", line 107] and "float
sqrt(float)" ["/opt/aCC/include/cmath", line 57].
Obscure declarations and denitions - C++ has fewer keywords than many
languages but it makes up for this by having many meaning attached to a
symbol. For example < is used in #include lines, cout, in template de-
nitions as well as meaning less-than. This makes deciphering declarations
hard. c++decl can sometimes help.
Deriving from Container Classes - the container classes werent designed to
act as base classes: they dont have virtual functions. Use them as members
of other classes instead.
Object creation - The rst 2 lines below create a string variable s. The 3rd
doesnt create an empty string - it declares a function s that returns a string.
String s; String s("initial"); String s();
Input and newline characters - If you try to get a character then a string from
the user with
cout << "Type a character: "; cin >> ch; cout << "Type a string: ";
cin.getline(str,20);
youre likely to get an empty string because the >> operator leaves the typed
newline in the input stream. One way round this is to use istream::ignore(INT_MAX, \n);.
9 CImg, an image processing library for C++
The CImg library is an image processing library, designed for C++ programmers.
It provides useful classes and functions to load/save, display and process various
types of images.
9.1 library structure
The CImg Library consists in a single header le CImg.h providing a set of C++
template classes that can be used in your own sources, to load/save, process and
display images or list of images.
Very portable (Unix/X11, Windows, MacOS X, FreeBSD,..), ecient, simple to
use, its a pleasant toolkit for coding image processing applications in C++.
The header le CImg.h contains all the classes and functions that compose the
library itself. This particularly means that :
54
No pre-compilation of the library is needed, since the linking of the CImg
functions is done at the same time as the linking of your own C++ code.
No complex dependencies have to be handled : Just include the CImg.h le,
and you get a working C++ image processing toolkit.
The compilation is done on the y : only CImg functionalities actually used
by your program are compiled and appear in the executable. This leads to
very compact code, without any unused code.
Class members and functions are inlined, leading to better performance
during the program execution.
The CImg Library is structured as follows:
All library classes and functions are dened in the namespace cimg library.
This namespace encapsulates the library functionalities and avoid any class
name collision that could happen with other includes. Generally, one uses this
namespace as a default namespace :
#include "CImg.h"
using namespace cimg_library;
...
The namespace cimg library::cimg denes a set of low-level functions and
variables used by the library. Documented functions in this namespace can
be safely used in your own program. But, dont use the cimg library::cimg
namespace as a default namespace, since it contains functions whose names
are already dened in the standard C/C++ library.
The class cimg library::CImg<T> represents images up to 4-dimensions wide,
containing pixels of type T (template parameter). This is actually the main
class of the library.
The class cimg library::CImgList<T> represents lists of cimg library::CImg<T>
images. It can be used for instance to store dierent frames of an image se-
quence.
The class cimg library::CImgDisplay is able to display images or image
lists into graphical display windows. As you may guess, the code of this class
is highly system-dependent but this is transparent for the programmer, as
environment variables are automatically set by the CImg library
The class cimg library::CImgException (and its subclasses) are used by
the library to throw exceptions when errors occur. Those exceptions can
be catched with a try { ..} catch (CImgException) { .. } construct.
Subclasses dene precisely the type of encountered errors.
9.2 CImg, Hello world
.
Below is a very simple code that creates a Hello World image.
55
#include "CImg.h"
using namespace cimg_library;
int main() {
CImg<unsigned char> img(640,400,1,3);
// Define a 640x400 color image with 8 bits per color component.
img.fill(0);
// Set pixel values to 0 (color : black)
unsigned char purple[] = { 255,0,255 };
// Define a purple color
img.draw_text(100,100,"Hello World",purple);
// Draw a purple "Hello world" at coordinates (100,100).
img.display("My first CImg code");
// Display the image in a display window.
return 0;
}
9.3 compiling a CImg application
The CImg library is a very light and user-friendly library. Only standard system
libraries are used. It avoids complex dependencies and problems with library com-
patibility. The only thing you need is a (quite modern) C++ compiler:
g++ (Linux version): Use the following command to compile and link a CImg
application with g++, on Linux:
g++ -o hello_word.exe hello_world.cpp -O2 -L/usr/X11R6/lib -lm
-lpthread -lX11
g++ (MingW windows version): Use the following command to compile and
link a CImg application with g++, on Windows:
g++ -o hello_word.exe hello_word.cpp -O2 -lgdi32
56
Advanced Programming with C++ 10 GLOSSARY
10 Glossary
Abstract Class - a class with at least one pure virtual function.
Abstract Data Type (ADT ) - An abstract class.
Adapters - ways to adapt existing predicates.
Algorithm - general-purpose routines for use with container classes.
Bag - a multiset.
Cast - explicit conversion of an object type.
Container - A class that can hold other objects.
Iterator - Something that can point to an object in a container.
Multiple Inheritance - When a class has more than one base class.
Namespace - a scoping mechanism that can scan les.
Overloading - When calling a function runs dierent code . according to the
types and number of arguments provided.
Polymorphism - not having to worry about the type of object youre dealing
with.
Predicate - a function object that returns a bool.
Pure Virtual function - a function that must be over-ridden in a derived class.
The Standard Library - classes, algorithms, and function objects providing
support for strings, vectors, lists, maps, strings and random numbers.
Template - A class that can be lled in with various types.
Virtual function - a function that can be over-ridden in a derived class.
Volatile - a variable is volatile if it might change beyond control of the program.
57
Advanced Programming with C++ 11 REFERENCES
11 References
Stroustrup B., The C++ Programming Language (third edition), AT&T,
1997. Also known as the bible. See his homepage:
http://www.research.att.com/ bs/homepage.html
Deitel and Deitel,C++: How to Program, (second edition), Prentice Hall,
1998. Used by undergraduates at Cambridge.
Coplien J.O., Advanced C++ Programming Styles and Idioms, Addison-
Wesley, 1992.
Lippman S.B. and Lajoie J.,C++ Primer (3rd Edition), Addison-Wesley,
1997. Despite the title, this contains some advanced material not covered by
Stroustrup or Deitel.
Lippman S.B.,Inside the C++ Object Model, Addison-Wesley, 1996.
Coplien J.O.,Multiparadigm Design for C++, Addison-Wesley, 1998.
Cambridge Universitys C++ help page:
http://www-h.eng.cam.ac.uk/help/tpl/languages/C++.html
Brokken F.B.C++ Annotations, (version 6.5.0), available for download from
the modules webpage but comprises 675 pages so print at your own risk.
Ezust A. and Ezust P.Design Patterns in C++ with Qt4, Prentice Hall,
available for download but also huge, Makes use of the Qt programming en-
vironment which will not be used in this course.
58
Advanced Programming with C++ 12 PROGRAM LISTINGS
12 Program listings
Listing 1: intTree.cpp, A binary search tree for integers.
// a t r e e f or s t or i ng i nt e g e r s
#include <i ostream>
#include <c s t dl i b >
// c l a s s i n t t r e e ;
cl ass bnode {
private :
friend cl ass i n t t r e e ;
bnode l e f t ;
bnode r i ght ;
i nt data ;
i nt count ;
bnode ( i nt d , bnode l , bnode r ) :
data ( d) , l e f t ( l ) , r i ght ( r ) , count ( 1) { }
void pr i nt ( ) const
{ std : : cout << data << : << count << \ t ; }
};
cl ass i n t t r e e {
public :
i n t t r e e ( ) { r oot = 0; }
void i ns e r t ( i nt d ) ;
i nt f i nd ( i nt d) const { return ( f i nd ( root , d ) ) ; }
void pr i nt ( ) const { pr i nt ( r oot ) ; }
private :
bnode r oot ;
i nt f i nd ( bnode r , i nt d) const ;
void pr i nt ( bnode r ) const ;
};
void i n t t r e e : : i ns e r t ( i nt d)
{
bnode temp = r oot ;
bnode ol d ;
i f ( r oot == 0) {
r oot = new bnode ( d , 0 , 0 ) ;
return ;
}
while ( temp != 0) {
ol d = temp ;
i f ( temp > data == d) {
( temp > count )++;
return ;
}
i f ( temp > data > d)
temp = temp > l e f t ;
el se
59
temp = temp > r i ght ;
}
i f ( ol d > data > d)
ol d > l e f t = new bnode ( d , 0 , 0 ) ;
el se
ol d > r i ght = new bnode ( d , 0 , 0 ) ;
}
i nt i n t t r e e : : f i nd ( bnode r , i nt d) const
{
i f ( r == 0)
return 0;
el se i f ( r > data == d)
return ( r > data ) ;
el se i f ( r > data > d)
return ( f i nd ( r > l e f t , d ) ) ;
el se
return ( f i nd ( r > r i ght , d ) ) ;
}
void i n t t r e e : : pr i nt ( bnode r ) const
{
i f ( r != 0) {
pr i nt ( r > l e f t ) ;
r > bnode : : pr i nt ( ) ;
pr i nt ( r > r i ght ) ;
}
}
i nt main ( )
{
i n t t r e e t ;
i nt p ;
while ( std : : ci n>>p){
i f ( p==0)break ;
t . i ns e r t ( p ) ;
}
t . pr i nt ( ) ;
std : : cout << \n ;
std : : cout<< ent er another zer o to cont i nue \n ;
std : : ci n>>p ;
i n t t r e e i t r e e ;
for ( i nt i = 150; i > 0; i )
i t r e e . i ns e r t ( i nt ( rand ( ) ) ) ;
i t r e e . pr i nt ( ) ;
std : : cout<< ent er another zer o to t ermi nat e \n ;
std : : ci n>>p ;
return 0;
}
60
Listing 2: genTree.cpp, A template binary search tree.
//Templ ate Versi on of genTree .
#include <i ostream>
#include <s t r i ng >
template <cl ass T> cl ass genTree ; // f orward de c l
template <cl ass T> cl ass bnode {
private :
friend cl ass genTree<T>;
bnode<T> l e f t ;
bnode<T> r i ght ;
T data ;
i nt count ;
bnode (T d , bnode<T> l , bnode<T> r ) :
data ( d) , l e f t ( l ) , r i ght ( r ) , count ( 1) { }
void pr i nt ( ) const
{ std : : cout << data << : << count << \ t ; }
};
template <cl ass T>
cl ass genTree {
public :
genTree ( ) { r oot = 0; }
void i ns e r t (T d ) ;
T f i nd (T d) const { return ( f i nd ( root , d ) ) ; }
void pr i nt ( ) const { pr i nt ( r oot ) ; }
private :
bnode<T> r oot ;
T f i nd ( bnode<T> r , T d) const ;
void pr i nt ( bnode<T> r ) const ;
};
template <cl ass T>
void genTree<T>: : i ns e r t (T d)
{
bnode<T> temp = r oot ;
bnode<T> ol d ;
i f ( r oot == 0) {
r oot = new bnode<T>(d , 0 , 0 ) ;
return ;
}
while ( temp != 0) {
ol d = temp ;
i f ( comp( temp > data , d) == 0) {
( temp > count )++;
return ;
}
i f ( comp( temp > data , d) > 0)
temp = temp > l e f t ;
el se
temp = temp > r i ght ;
61
}
i f ( comp( ol d > data , d) > 0)
ol d > l e f t = new bnode<T>(d , 0 , 0 ) ;
el se
ol d > r i ght = new bnode<T>(d , 0 , 0 ) ;
}
template <cl ass T>
T genTree<T>: : f i nd ( bnode<T> r , T d) const
{
i f ( r == 0)
return 0;
el se i f ( comp( r > data , d) == 0)
return ( r > data ) ;
el se i f ( comp( r > data , d) > 0)
return ( f i nd ( r > l e f t , d ) ) ;
el se
return ( f i nd ( r > r i ght , d ) ) ;
}
template <cl ass T>
void genTree<T>: : pr i nt ( bnode<T> r ) const
{
i f ( r != 0) {
pr i nt ( r > l e f t ) ;
r > bnode<T>: : pr i nt ( ) ;
pr i nt ( r > r i ght ) ;
}
}
template <cl ass T> // gener al case
i nt comp(T i , T j )
{
i f ( i == j ) //assumes == < def i ned f or T
return 0;
el se
return ( ( i < j ) ? 1 : 1 ) ;
}
i nt main ( )
{
genTree<char> t ;
char p ;
std : : cout<< ent er any char s to gener at e a charTree ( t er mi nt e wi th z char ) \n ;
while ( std : : ci n>>p){
i f ( p== z ) break ;
t . i ns e r t ( p ) ;
}
t . pr i nt ( ) ;
std : : cout << EOF << \n ;
62
std : : cout<< ent er any char to gener at e an i nt Tr ee \n ;
std : : ci n>>p ;
genTree<int> i Tr ee ;
for ( i nt i = 15; i > 5; i )
i Tr ee . i ns e r t ( i ) , i Tr ee . i ns e r t ( i ) ;
i Tr ee . pr i nt ( ) ;
std : : cout<< ent er any char to t ermi nat e \n ;
std : : ci n>>p ;
return 0;
}
63

Advanced Programming C++

Uploaded by

Copyright:

Available Formats

Advanced Programming C++

Uploaded by

Document Information

Original Description:

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

Advanced Programming C++

Uploaded by

Copyright:

Available Formats

Advanced Programming with C++

You might also like