From Wikibooks, the open-content textbooks collection

Authors

The following people are authors to this book

Panic

This work is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license. In short: you are free to share and to make derivatives of this work under the conditions that you appropriately attribute it, and that you only distribute it under the same, similar or a compatible license. Any of the above conditions can be waived if you get permission from the copyright holder.

Acknowledgment is given for using some contents from other works like Wikipedia, the Wikibooks Java Programming, C Programming and C++ Exercises for beginners, the C/C++ Reference Web Site, and from Wikisource, as from the authors Scott Wheeler, Stephen Ferg and Ivor Horton.

There are many other contributors/editors to the book; a verifiable list of all contributions exist as History Logs at Wikibooks (http://en.wikibooks.org/).

Structures

A simple implementation of the object paradigm from (OOP) that holds collections of data records (also known as compound values or set). A struct is like a class except for the default access (class has default access of private, struct has default access of public). C++ also guarantees that a struct that only contains C types is equivalent to the same C struct thus allowing access to legacy C functions, it can (but may not) also have constructors (and must have them, if a templated class is used inside a struct), as with Classes the compiler implicitly-declares a destructor if the struct doesn’t have a user-declared destructor.

A struct is defined by:

struct myStructType /*: inheritances */ {
public: 
 // public members
protected:
 // protected members
private:
 // private members
} myStructName;

Because it is not supported in C, it is uncommon to have structs in C++ using inheritances even though they are supported just like in classes. The more distinctive aspect is that structs can have two identities one is in reference to the type and another to the specific object. The public access label can sometimes be ignored since the default state of struct for member functions and fields is public.

An object of type myStruct (case-sensitive) is declared using:

 myStruct obj1;

Why should you Use Structs, Not Classes?

Older programmer languages used a similar type called Record (ie: COBOL, FORTRAN) this was implemented in C as the struct keyword. And so C++ uses structs to comply with this C's heritage (the code and the programmers). Structs are simpler to be managed by the programmer and the compiler. One should use a struct for POD (PlainOldData) types that have no methods and whose data members are all public. struct may be used more efficiently in situations that default to public inheritance (which is the most common kind) and where public access (which is what you want if you list the public interface first) is the intended effect. Using a class, you typically have to insert the keyword public in two places, for no real advantage. In the end it's just a matter of convention, which programmers should be able to get used to.

Point objects 

As a simple example of a compound structure, consider the concept of a mathematical point. At one level, a point is two numbers (coordinates) that we treat collectively as a single object. In mathematical notation, points are often written in parentheses, with a comma separating the coordinates. For example, (0, 0) indicates the origin, and (x, y) indicates the point x units to the right and y units up from the origin.

The natural way to represent a point is using two doubles. The structure or struct is one of the solutions to group these two values into a compound object.

// A struct definition:
 struct Point { 
   double x, y; };

This definition indicates that this structure contains two members, named x and y. These members are also called instance variables, for reasons I will explain a little later.

It is a common error to leave off the semi-colon at the end of a structure definition. It might seem odd to put a semi-colon after a squiggly-brace, but you'll get used to it. This syntax is in place to allow the programmer the facility to create an instance[s] of the struct when it is defined.

Once you have defined the new structure, you can create variables with that type:

Point blank; 
blank.x = 3.0; 
blank.y = 4.0;

The first line is a conventional variable declaration: blank has type Point. The next two lines initialize the instance variables of the structure. The "dot notation" used here is similar to the syntax for invoking a function on an object, as in fruit.length(). Of course, one difference is that function names are always followed by an argument list, even if it is empty.

As usual, the name of the variable blank appears outside the box and its value appears inside the box. In this case, that value is a compound object with two named instance variables.

Accessing instance variables 

You can read the values of an instance variable using the same syntax we used to write them:

int x = blank.x;

The expression blank.x means "go to the object named blank and get the value of the member named x." In this case we assign that value to a local variable named x. Notice that there is no conflict between the local variable named x and the instance variable named x. The purpose of dot notation is to identify which variable you are referring to unambiguously.

You can use dot notation as part of any expression, so the following are legal.

cout << blank.x << ", " << blank.y << endl; 
double distance = blank.x * blank.x + blank.y * blank.y;

The first line outputs 3, 4; the second line calculates the value 25.

Operations on structures 

Most of the operators we have been using on other types, like mathematical operators ( +, %, etc.) and comparison operators (==, >, etc.), do not work on structures. Actually, it is possible to define the meaning of these operators for the new type, but we won't do that in this book.

On the other hand, the assignment operator does work for structures. It can be used in two ways: to initialize the instance variables of a structure or to copy the instance variables from one structure to another. An initialization looks like this:

Point blank = { 3.0, 4.0 };

The values in squiggly braces get assigned to the instance variables of the structure one by one, in order. So in this case, x gets the first value and y gets the second.

Unfortunately, this syntax can be used only in an initialization, not in an assignment statement. Therefore, the following is illegal.

Point blank; 
blank = { 3.0, 4.0 }; // WRONG !!

You might wonder why this perfectly reasonable statement should be illegal, and there is no good answer. (Note, however, that a similar syntax is legal in C since 1999, and is under consideration for possible inclusion in C++ in the future.)

On the other hand, it is legal to assign one structure to another. For example:

Point p1 = { 3.0, 4.0 }; 
Point p2 = p1; 
cout << p2.x << ", " <<  p2.y << endl;

The output of this program is 3, 4.

Structures as return types 

You can write functions that return structures. For example, findCenter takes a Rectangle as an argument and returns a Point that contains the coordinates of the center of the Rectangle:

Point findCenter (Rectangle& box) 
{ 
  double x = box.corner.x + box.width/2; 
  double y = box.corner.y + box.height/2; 
  Point result = {x, y}; 
  return result; 
}

To call this function, we have to pass a box as an argument (notice that it is being passed by reference), and assign the return value to a Point variable:

Rectangle box = { {0.0, 0.0}, 100, 200 }; 
Point center = findCenter (box); 
printPoint (center);

The output of this program is (50, 100).

Passing other types by reference 

It's not just structures that can be passed by reference. All the other types we've seen can, too. For example, to swap two integers, we could write something like:

void swap (int& x, int& y) 
{ 
  int temp = x; 
  x = y; 
  y = temp; 
}

We would call this function in the usual way:

int i = 7; 
int j = 9; 
swap (i, j); 
cout << i << j << endl;

The output of this program is 97. Draw a stack diagram for this program to convince yourself this is true. If the parameters x and y were declared as regular parameters (without the &s), swap would not work. It would modify x and y and have no effect on i and j.

When people start passing things like integers by reference, they often try to use an expression as a reference argument. For example:

int i = 7; 
int j = 9; 
swap (i, j+1); // WRONG!!

This is not legal because the expression j+1 is not a variable — it does not occupy a location that the reference can refer to. It is a little tricky to figure out exactly what kinds of expressions can be passed by reference. For now, a good rule of thumb is that reference arguments have to be variables.

Pointers and structures

Structures can also be pointed by pointers and store pointers. The rules are the same as for any fundamental data type. The pointer must be declared as a pointer to the structure.

Nesting structures

Structures can also be nested so that a valid element of a structure can also be another structure.

this

The this keyword is a implicitly created pointer that is only accessible within nonstatic member functions of a struct (or a union or class) and points to the object for which the member function is called. The this pointer is not available in static member functions. This will be restated again on when introducing unions a more in depth analysis is provided in the Section about classes.

Unions

Unions are similar to structs, but they differ in one aspect: the fields of a union share the same position in memory. The size of the union is the size of its largest field (or larger if alignment so requires, for example on a SPARC machine a union contains a double and a char [17] so its size is likely to be 24 because it needs 64-bit alignment). Unions cannot have a destructor.

What is the point of this? Unions provide multiple ways of viewing the same memory location, allowing for more efficient use of memory. Most of the uses of unions are covered by object-oriented features of C++, so it is more common in C. However, sometimes it is convenient to avoid the formalities of object-oriented programming when performance is important or when one knows that the item in question will not be extended.

Writing to Different Bytes

Unions are mostly useful for low-level programming tasks that involve writing to the same memory area but at different portions of the allocated memory space, for instance:

union item {
  // The item is 16-bits
  short theItem;
  // In little-endian lo accesses the low 8-bits -
  // hi, the upper 8-bits
  struct { char lo; char hi; } portions;
};

  item tItem;
  tItem.theItem = 0xBEAD;
  tItem.portions.lo = 0xEF; // The item now equals 0xBEEF

Using this union we can modify the low-order or high-order bytes of theItem without disturbing any other bytes.

Example in Practice: SDL Events

One real-life example of unions is the event system of SDL, a graphics library in C. In graphical programming, an event is an action triggered by the user, such as a mouse move or keyboard press. One of the SDL's responsibilities is to handle events and provide a mechanism for the programmer to listen for and react to them.

// primary event structure in SDL
 
typedef union {
  Uint8 type;
  SDL_ActiveEvent active;
  SDL_KeyboardEvent key;
  SDL_MouseMotionEvent motion;
  SDL_MouseButtonEvent button;
  SDL_JoyAxisEvent jaxis;
  SDL_JoyBallEvent jball;
  SDL_JoyHatEvent jhat;
  SDL_JoyButtonEvent jbutton;
  SDL_ResizeEvent resize;
  SDL_ExposeEvent expose;
  SDL_QuitEvent quit;
  SDL_UserEvent user;
  SDL_SysWMEvent syswm;
} SDL_Event;

Each of the types other than Uint8 (an 8-bit unsigned integer) is a struct with details for that particular event.

// SDL_MouseButtonEvent
 
typedef struct{
  Uint8 type;
  Uint8 button;
  Uint8 state;
  Uint16 x, y;
} SDL_MouseButtonEvent;

When the programmer receives an event from SDL, he first checks the type value. This tells him what kind of an event it is. Based on this value, he either ignores the event or gets more information by getting the appropriate part of the union.

For example, if the programmer received an event in SDL_Event ev, he could react to mouse clicks with the following code.

if (ev.type == SDL_MOUSEBUTTONUP && ev.button.button == SDL_BUTTON_RIGHT) {
  cout << "You have right-clicked at coordinates (" << ev.button.x << ", "
       << ev.button.y << ")." << endl;
}

While identical functionality can be provided with a struct rather than a union, the union is far more space efficient; the struct would use memory for each of the different event types, whereas the union only uses memory for one. As only one entry has meaning per instance, it is reasonable to use a union in this case.

This scheme could also be constructed with polymorphism and inheritance features of object-oriented C++, however the setup would be involved and less efficient than this one. Use of unions loses type safety, however it gains in performance.

this

The this keyword is a implicitly created pointer that is only accessible within nonstatic member functions of a union (or a struct or class ) and points to the object for which the member function is called. The this pointer is not available in static member functions. This will be restated again on when introducing unions a more in depth analysis is provided in the Section about classes.

Classes

Classes are used to create user defined types. An instance of a class is called an object and programs can contain any number of classes. As with other types, object types are case-sensitive.

Classes provide encapsulation as defined in the Object Oriented Programming (OOP) paradigm. A class can have both data members and functions members associated with it. Unlike the built-in types, the class can contain several variables and functions, those are called members.

Classes also provide flexibility in the "divide and conquer" scheme in program writing. In other words, one programmer can write a class and guarantee an interface. Another programmer can write the main program with that expected interface. The two pieces are put together and compiled for usage.

Declaration

A class is defined by:

class myClass {
 /* public, protected and private
 variables, constants, and functions */
};

An object of type myClass (case-sensitive) is declared using:

myClass Object;

• by default, all class members are initially private.
• keywords public and protected allow access to class members.
• classes contain not only data members, but also functions to manipulate that data.
• a class is used as the basic building block of OOP (this is a distinction of convention, not of language-enforced semantics).

A class can be created

• before main() is called.
• when a function is called in which the object is declared.
• when the "new" operator is used.

Class Names

• Name the class after what it is. If you can't determine a name, then you have not designed the system well enough.
• Compound names of over three words are a clue your design may be confusing various entities in your system. Revisit your design. Try a CRC card session to see if your objects have more responsibilities than they should.
• Avoid the temptation of naming a class something similar to the class it is derived from. A class should stand on its own. Declaring an object with a class type doesn't depend on where that class is derived from.
• Suffixes or prefixes are sometimes helpful. For example, if your system uses agents then naming something DownloadAgent conveys real information.

Data Abstraction

A fundamental concept of Object Oriented (OO) recommends an object should not expose any of its implementation details. This way, you can change the implementation without changing the code that uses the object. The class, by design, allows its programmer to hide (and also prevents changes as to) how the class is implemented. This powerful tool allows the programmer to build in a 'preventive' measure. Variables within the class often have a very significant role in what the class does, therefore variables can be secured within the private section of the class.

Access labels

The access labels Public, Protected and Private are used within classes to set access permissions for the members in that section of the class. All class members are initially private by default. The labels can be in any order. These labels can be used multiple times in a class declaration for cases where it is logical to have multiple groups of these types. An access label will remain active until another access label is used to change the permissions.

We have already mentioned that a class can have member functions "inside" it; we will see more about them later. Those member functions can access and modify all the data and member function that are inside the class. Therefore, permission labels are to restrict access to member function that reside outside the class and for other classes.

For example, a class "Bottle" could have a private variable fill, indicating a liquid level 0-3 dl. fill cannot be modified directly (compiler error), but instead Bottle provides the member function sip() to reduce the liquid level by 1. Mywaterbottle could be an instance of that class, an object.

/* Bottle - Class and Object Example */
#include <iostream>
#include <iomanip>
 
using namespace std;
 
class Bottle
{
  private:  // variables are modified by member functions of class
  int fill; // dl of liquid
 
  public:
    Bottle()  // Default Constructor
    : fill(3) // They start with 3 dl of liquid
      {
        // More constructor code would go here if needed.
      }
 
     bool sip() // return true if liquid was available
     {
 
       if (fill>0)
       {
         --fill;
         return true;
       }
       else
       {
         return false;
       }
 
     }
 
     int level() const  // return level of liquid dl
     {
         return fill;
     }
 };  // Class declaration has a trailing semicolon
 
int main()
{
  // terosbottle object is an instance of class Bottle
  Bottle terosbottle;
  cout << "In the beginning, mybottle has "
       << terosbottle.level()
       << "  dl of liquid"
       << endl;
 
  while (terosbottle.sip())
  {
     cout << "Mybottle has "
          << terosbottle.level()
          << " dl of liquid"
          << endl;
  }
 
  return 0;
}

These keywords, private, public, and protected, affect the permissions of the members -- whether functions or variables.

public

This label indicates any members within the 'public' section can accessed freely anywhere a declared object is in scope.

private

Members defined as private are only accessible within the class defining them, or friend classes. Usually the domain of member variables and helper functions. It's often useful to begin putting functions here and then moving them to the higher access levels as needed so to reduce complexity.

(This is an example where the default copy constructor will do the same thing.)

class Foo
{
 public:
   Foo( const Foo &f )
   {
     m_value = f.m_value; // perfectly legal
   }
 
 private:
   int m_value;
}

protected

The protected label has a special meaning to inheritance, protected members are accessible in the class that defines them and in classes that inherit from that base class, or friends of it. In the section on inheritance we will see more about it.

Inheritance ( Derivation )

As we have seen early on in the Programming Paradigms Section, Inheritance is a property that describes a relationship between two (or more) types, or classes, of objects in OOP and so C++ classes share this property.

Derivation is the action of creating a new class using the inheritance property of the C++ Programming language. It is possible to derive one class from another or even several (Multiple inheritance), like a tree we can call base class to the root and child class to any leaf; in any other case the parent/child relation will exist for each class derived from another.

Base Class

A base class is a class that is created with the intention of deriving other classes from it.

Child Class

A child class is a class that was derived from another, that will now be the parent class to it.

Parent Class

A parent class is the closest class that we derived from to create the one we are referencing as the child class.

As an example, suppose you are creating a game, something using different cars, and you need specific type of car for the policemen and another type for the player(s). Both car types share similar properties. The major difference (on this example case) would be that the policemen type would have sirens on top of their cars and the players' cars will not.

One way of getting the cars for the policemen and the player ready is to create separate classes for policemen's car and for the player's car like this:

class PlayerCar {
private:
  int color;
 
public:
  void driveAtFullSpeed(int mph){
    // code for moving the car ahead
  }
 
};
 
class PoliceCar {
private:
  int color;
  bool sirenOn;  // identifies whether the siren is on or not
  bool inAction; // identifies whether the police is in action (following the player) or not
 
public:
  bool isInAction(){
    return this->inAction;
  }
 
  void driveAtFullSpeed(int mph){
    // code for moving the car ahead
  }
 
};

and then creating separate objects for the two cars like this:

PlayerCar player1;
PoliceCar policemen1;

Now all is fine and supe'cool at the moment, except for one thing that you can easily notice: there are certain chunks of code that are very similar (if not exactly the same) in the above two classes. So, in essence, you have to type in the same code at two different places! And when you update your code to include methods (functions) for handBrake() and pressHorn(), you'll have to do that in both the classes above.

Therefore, to escape this frustrating (and confusing) task of writing the same code at multiple locations in a single project, you use Inheritance.

Now that you know what kind of problems Inheritance solves in C++, let’s examine how to implement Inheritance in our programs. As its name suggests, Inheritance lets us create new classes which automatically have all the code from existing classes. It means that if there is a class called MyClass, a new class with the name MyNewClass can be created which will have all the code present inside the MyClass class. The following code segment shows it all:

class MyClass {
  protected:
         int age;
  public:
         void sayAge(){
             this->age = 20;
             cout << age;
         }
};
 
class MyNewClass : public MyClass {
 
};
 
int main() {
 
  MyNewClass *a = new MyNewClass();
  a->sayAge();
 
  return 0;
 
}

As you can see, using the colon ':' we can inherit a new class out of an existing one. It’s that simple! All the code inside the MyClass class is now available to the MyNewClass class. And if you are intelligent enough, you can already see the advantages it provides. If you are like me (i.e. not too intelligent), you can see the following code segment to know what I mean:

class Car {
  protected:
         int color;
         int currentSpeed;
         int maxSpeed;
  public:
         void applyHandBrake(){
             this->currentSpeed = 0;
         }
         void pressHorn(){
             cout << "Teeeeeeeeeeeeent"; // funny noise for a horn
         }
         void driveAtFullSpeed(int mph){
              // code for moving the car ahead;
         }
};
 
class PlayerCar : public Car {
 
};
 
class PoliceCar : public Car {
  private:
         bool sirenOn;  // identifies whether the siren is on or not
         bool inAction; // identifies whether the police is in action (following the player) or not
  public:
         bool isInAction(){
             return this->inAction;
         }
};

In the code above, the two newly created classes PlayerCar and PoliceCar have been inherited from the Car class. Therefore, all the methods and properties (variables) from the Car class are available to the newly created classes for the player's car and the policemen's car. Technically speaking, in C++, the Car class in this case is our "Base Class" since this is the class which the other two classes are based on (or inherit from).

Just one more thing to note here is the keyword protected instead of the usual private keyword. That’s no big deal: We use protected when we want to make sure that the variables we define in our base class should be available in the classes that inherit from that base class. If you use private in the class definition of the Car class, you will not be able to inherit those variables inside your inherited classes.

There are three types of class inheritance: public, private and protected. We use the keyword public to implement public inheritance. The classes who inherit with the keyword public from a base class, inherit all the public members as public members, the protected data is inherited as protected data and the private data is inherited but it cannot be accessed directly by the class.

The following example shows the class Circle that inherits "publically" from the base class Form:

class Form {
private:
  double area;
 
public:
  int color;
 
  double getArea(){
    return this->area;
  }
 
  void setArea(double area){
    this->area=area;
  }
 
};
 
class Circle: public Form {
public:
  double getRatio() {
    double a;
    a= getArea();
    return sqrt(a/2*3.14);
  }
 
  void setRatio(double diameter) {
    setArea( pow( diameter * 0.5, 2) * (3.14));
  }
 
  bool isDark() {
    return color>10;
  }
 
};

The new class Circle inherits the attribute area from the base class Form (the attribute area is implicitly an attribute of the class Circle), but it cannot access it directly. It does so through the functions getArea and setArea (that are public in the base class and remain public in the derived class). The color attribute, however, is inherited as a public attribute, and the class can access it directly.

The following table indicates how the attributes are inherited in the three different types of inheritance:

As the table above shows, protected members are inherited as protected methods in public inheritance. Therefore, we should use the protected label whenever we want to declare a method inaccessible outside the class and not to lose access to it in derived classes. However, losing accessibility can be useful sometimes, because we are encapsulating details in the base class.

Let’s imagine that we have a class with a very complex method "m" that invokes many auxiliary methods declared as private in the class. If we derive a class from it, we should not bother about those methods because they are inaccessible in the derived class. If a different programmer is in charge of the design of the derived class, allowing access to those methods could be the cause of errors and confusion. So, it is a good idea to avoid the protected label whenever we can have a design with the same result with the private label.

Multiple inheritance

Multiple inheritance is the construction in which objects inherit from more than one object type or class. This contrasts with single inheritance, where objects can only inherit from one type or class.

Multiple inheritance can cause some confusing situations, and is much more complex than single inheritance, so there is some debate over whether or not its benefits outweigh its risks. Multiple inheritance has been a touchy issue for many years, with opponents pointing to its increased complexity and ambiguity in situations such as the "diamond problem" Most modern OOP languages do not allow multiple inheritance.

Data Members

TODO Add missing info

this pointer

this is a C++ keyword that acts as a pointer to a class being referenced. The this pointer acts like any other pointer, although you can't change the pointer itself. Read the section concerning pointers and references to understand more about what a pointer does. The this pointer is only accessible within nonstatic member functions of a class (or a union or struct), and is not available in static member functions. It is not necessary to write code for the this pointer as the compiler does this implicitly. When you debug code in a GUI compiler with some way of viewing the current variables, you can see the this pointer in some variable list when the program steps into nonstatic class functions.

In the following example, the compiler inserts an implicit parameter this in the nonstatic member function int getData(). Additionally, the code initiating the call passes an implicit parameter (provided by the compiler).

class Foo
{
private:
    int x;
public:
    Foo(){  x=5;  };
    int getData() 
    {   
        return this->x;  // this is provided by the compiler at compile time
    }
};
 
int main()
{
    Foo Example;
    int temp;
 
    temp = Example.getData(&Example);  // compiler adds the &Example reference at compile time
    return 0;
}

There are certain times when a programmer should know about and use the this pointer. The this pointer should be used when overloading the assignment operator to prevent a catastrophe. For example, add in an assignment operator to the code above.

class Foo
{
private:
    int x;
public:
    Foo(){  x=5;  };
    int getData()          
    {                            
        return x;  
    }
    Foo& operator=( const Foo &RHS );
};
 
Foo& Foo::operator=( const Foo &RHS )
{
    if( this != &RHS ){    // the if this test prevents an object from copying to itself (ie. RHS = RHS;)
        this->x = RHS.x;        // this is suitable for this class, but can be more complex when
                                // copying an object in a different much larger class
    }
 
    return ( *this );       // returning an object allows chaining, like  a = b = c; statements
}

However little you may know about this, it is important in implementing any class.

Member Functions

Member functions can (and should) be used to interact with data contained within user defined types. User defined types provide flexibility in the "divide and conquer" scheme in program writing. In other words, one programmer can write a user defined type and guarantee an interface. Another programmer can write the main program with that expected interface. The two pieces are put together and compiled for usage. User defined types provide encapsulation defined in the Object Oriented Programming (OOP) paradigm.

Within classes, to protect the data members, the programmer can define functions to perform the operations on those data members. Member functions and functions are names used interchangeably in reference to classes. Function prototypes are declared within the class definition. These prototypes can take the form of non-class functions as well as class suitable prototypes. Functions can be declared and defined within the class definition. However, most functions can have very large definitions and make the class very unreadable. Therefore it is possible to define the function outside of the class definition using the scope resolution operator "::". This scope resolution operator allows a programmer to define the functions somewhere else. This can allow the programmer to provide a header file .h defining the class and a .obj file built from the compiled .cpp file which contains the function definitions. This can hide the implementation and prevent tampering. The user would have to define every function again to change the implementation. Functions within classes can access and modify (unless the function is constant) data members without declaring them, because the data members are already declared in the class.

Simple example:

file: Foo.h

// the header file named the same as the class helps locate classes within a project
// one class per header file makes it easier to keep the header file readable (some classes can become large)
// each programmer should determine what style works for them or what programming standards their
// teach/professor/employer has
 
#ifndef FOO_H
#define FOO_H
 
class Foo{
public:
  Foo();                  // function called the default constructor
  Foo( int a, int b );    // function called the overloaded constructor
  int Manipulate( int g, int h );
 
private:
  int x;
  int y;
};
 
#endif

file: Foo.cpp

#include "Foo.h"
 
/* these constructors should really show use of initialization lists
Foo::Foo() : x(5), y(10)
{
}
Foo:Foo(int a, int b) : x(a), y(b)
{
}
*/
Foo::Foo(){
  x = 5;
  y = 10;
}
Foo::Foo( int a, int b ){
  x = a;
  y = b;
}
 
int Foo::Manipulate( int g, int h ){
  x = h + g*x;
  y = g + h*y;
}

Accessors and Modifiers (Setter/Getter)

What is an accessor?

An accessor is a member function that does not modify the state of an object. The accessor functions should be declared as const.

Getter is another common definition of an accessor due to the naming ( GetSize() ) of that type of member functions.

What is a modifier?

A modifier, also called a modifying function, is a member function that changes the value of at least one data member. In other words, an operation that modifies the state of an object. Modifiers are also known as ‘mutators’.

Setter is another common definition of an modifier due to the naming ( SetSize( int a_Size ) ) of that type of member functions.

Overloading

Member functions can be overloaded. This means that multiple member functions can exist with the same name on the same scope, but must have different signatures. A member function's signature is comprised of the member function's name and the type and order of the member function's parameters.

Due to name hiding, if a member in the derived class shares the same name with members of the base class, they will be hidden to the compiler. To make those members visible, one can use declarations to introduce them from base class scopes.

Constructors and other class member functions, except the Destructor, can be overloaded.

Dynamic Polymorphism ( Overrides )

So far, we have learned that we can add new data and functions to a class through inheritance. But what about if we want our derived class to inherit a method from the base class, but to have a different implementation for it? That is when we are talking about polymorphism, a fundamental concept in OOP programming.

As seen previously in the Programming Paradigms Section, Polymorphism is subdivided in two concepts static polymorphism and dynamic polymorphism. This section concentrates on dynamic polymorphism, which applies in C++ when a derived class overrides a function declared in a base class.

We implement this concept redefining the method in the derived class. However, we need to have some considerations when we do this, so now we must introduce the concepts of dynamic binding, static binding and virtual methods.

Suppose that we have two classes, A and B. B derives from A and redefines the implementation of a method c() that resides in class A. Now suppose that we have an object b of class B. How should the instruction b.c() be interpreted?

If b is declared in the stack (not declared as a pointer or a reference) the compiler applies static binding, this means it interprets (at compile time) that we refer to the implementation of c() that resides in B.

However, if we declare b as a pointer or a reference of class A, the compiler could not know which method to call at compile time, because b can be of type A or B. If this is resolved at run time, the method that resides in B will be called. This is called dynamic binding. If this is resolved at compile time, the method that resides in A will be called. This is again, static binding.

virtual

The virtual member functions are an essential part of designing a class hierarchy and sub-classing classes from a toolkit. The concept is relatively simple, but often misunderstood. Specifically it determines the behavior of overridden methods in certain contexts.

Virtual member functions are class member functions, that can be overridden in any class derived from the one were they where declared. The member function body is then replaced with a new set of implementation in the derived class.

By placing the keyword virtual before a method declaration we are indicating that when the compiler has to decide between applying static binding or dynamic binding it will apply dynamic binding. Otherwise, static binding will be applied.

Again, this should be clearer with an example:

class Foo
{
public:
  void f()
  {
    std::cout << "Foo::f()" << std::endl;
  }
  virtual void g()
  {
    std::cout << "Foo::g()" << std::endl;
  }
};
 
class Bar : public Foo
{
public:
  void f()
  {
    std::cout << "Bar::f()" << std::endl;
  }
  virtual void g()
  {
    std::cout << "Bar::g()" << std::endl;
  }
};
 
int main()
{
  Foo foo;
  Bar bar;
 
  Foo *baz = &bar;
  Bar *quux = &bar;
 
  foo.f(); // "Foo::f()"
  foo.g(); // "Foo::g()"
 
  bar.f(); // "Bar::f()"
  bar.g(); // "Bar::g()"
 
  // So far everything we would expect...
 
  baz->f();  // "Foo::f()"
  baz->g();  // "Bar::g()"
 
  quux->f(); // "Bar::f()"
  quux->g(); // "Bar::g()"
 
  return 0;
}

Our first calls to f() and g() on the two objects are straightforward. However things get interesting with our baz pointer which is a pointer to the Foo type.

f() is not virtual and as such a call to f() will always invoke the implementation associated with the pointer type -- in this case the implementation from Foo.

Remember that the overloading and overriding are distinct concepts.

Pure virtual member function

There is one additional interesting possibility -- sometimes we don't want to provide an implementation of our function at all, but want to require people sub-classing our class to provide an implementation on their own. This is the case for pure virtuals.

To indicate a pure virtual function instead of an implementation we simply add an "= 0" after the function declaration.

Again -- an example:

class Widget
{
public:
   virtual void paint() = 0;
};
 
class Button : public Widget
{
public:
   void paint() // is virtual because it is an override
   {
       // do some stuff to draw a button
   }
};

Because paint() is a pure virtual function in the Widget class we are required to provide an implementation in all concrete subclasses. If we don't the compiler will give us an error at build time.

This is helpful for providing interfaces -- things that we expect from all of the objects based on a certain hierarchy, but when we want to ignore the implementation details.

So why is this useful?

Let's take our example from above where we had a pure virtual for painting. There are a lot of cases where we want to be able to do things with widgets without worrying about what kind of widget it is. Painting is an easy example.

Imagine that we have something in our application that repaints widgets when they become active. It would just work with pointers to widgets -- i.e. Widget *activeWidget() const might be a possible function signature. So we might do something like:

Widget *w = window->activeWidget();
w->paint();

We want to actually call the appropriate paint member function for the "real" widget type -- not Widget::paint() (which is a "pure" virtual and will cause the program to crash if called using virtual dispatch). By using a virtual function we insure that the member function implementation for our subclass -- Button::paint() in this case -- will be called.

TODO Mention interface classes

virtual Constructors

There is a hierarchy of classes with base class Foo. Given an object bar belonging in the hierarchy, it is desired to be able to do the following:

1. Create an object baz of the same class as bar (say, class Bar) initialized using the default constructor of the class. The syntax normally used is:

   Bar* baz = bar.create();

2. Create an object baz of the same class as bar which is a copy of bar. The syntax normally used is:

   Bar* baz = bar.clone();

In the class Foo, the methods Foo::create() and Foo::clone() are declared as follows:

class Foo
{
    // ...
 
    public:
        // Virtual default constructor
        virtual Foo* create() const;
 
        // Virtual copy constructor
        virtual Foo* clone() const;
};

If Foo is to be used as an abstract class, the functions may be made pure virtual:

class Foo
{
   // ...
 
    public:
        virtual Foo* create() const = 0;
        virtual Foo* clone() const = 0;
};

In order to support the creation of a default-initialized object, and the creation of a copy object, each class Bar in the hierarchy must have public default and copy constructors. The virtual constructors of Bar are defined as follows:

class Bar : ... // Bar is a descendant of Foo
{
    // ...
 
    public:
    // Non-virtual default constructor
    Bar ();
    // Non-virtual copy constructor
    Bar (const Bar&);
 
    // Virtual default constructor, inline implementation
    Bar* create() const { return new Foo (); }
    // Virtual copy constructor, inline implementation
    Bar* clone() const { return new Foo (*this); }
};

The above code uses Covariant return types. If your compiler doesn't support Bar* Bar::create(), use Foo* Bar::create() instead, and similarly for clone().

While using these virtual constructors, you must manually deallocate the object created by calling delete baz;. This hassle could be avoided if a smart pointer (e.g. std::auto_ptr<Foo>) is used in the return type instead of the plain old Foo*.

Remember that whether or not Foo uses dynamically allocated memory, you must define the destructor virtual ~Foo () and make it virtual to take care of deallocation of objects using pointers to an ancestral type.

virtual Destructor

It is of special importance to remember to define a virtual destructor even if empty in any base class, since failing to do so will create problems with the default compiler generated destructor that will not be virtual.

A virtual destructor is not overridden when redefined in a derived class, the definitions to each destructor are cumulative and they start from the last derivate class toward the first base class.

Pure virtual Destructor

Every abstract class should contain the declaration of a pure virtual destructor.

Pure virtual destructors are a special case of pure virtual functions (meant to be overridden in a derived class). They must always be defined and that definition should always be empty.

class Interface {
public:
  virtual ~Interface() = 0; //declaration of a pure virtual destructor 
};
 
Interface::~Interface(){} //pure virtual destructor definition (should always be empty)

Covariant return types

Covariant return types is the ability for a virtual function in a derived class to return a pointer or reference to an instance of itself if the version of the method in the base class does so. e.g.

class base
{
public:
  virtual base* create() const;
};
 
class derived : public base
{
public:
  virtual derived* create() const;
};

This allows casting to be avoided.

const

This type of member function cannot modify the member variables of a class. It's a hint both to the programmer and the compiler that a given member function doesn't change the internal state of a class; however, any variables declared as mutable can still be modified.

Take for example:

 class Foo
 {
  public:
    int value() const
    {
      return m_value;
    }
 
    void setValue( int i )
    {
      m_value = i;
    }
 
  private:
    int m_value;
 };

Here value() clearly does not change m_value and as such can and should be const. However setValue() does modify m_value and as such cannot be const.

Another subtlety often missed is a const member function cannot call a non-const member function (and the compiler will complain if you try). The const member function cannot change member variables and a non-const member functions can change member variables. Since we assume non-const member functions do change member variables, const member functions are assumed to never change member variables and can't call functions that do change member variables.

The following code example explains what const can do depending on where it is placed.

 class Foo
 {
 public:
    /*
     * Modifies m_widget and the user
     * may modify the returned widget.
     */
    Widget *widget();
 
    /*
     * Does not modify m_widget but the
     * user may modify the returned widget.
     */
    Widget *widget() const;
 
    /*
     * Modifies m_widget, but the user
     * may not modify the returned widget.
     */
    const Widget *cWidget();
 
    /*
     * Does not modify m_widget and the user
     * may not modify the returned widget.
     */
    const Widget *cWidget() const;
 
 private:
    Widget *m_widget;
 };

static

Member functions or variables declared static are shared between all instances of an object type. Meaning that only one copy of the member function or variable does exists for any object type.

static member function

Named constructors are a good example of using static member functions. Named constructors is the name given to functions used to create an object of a class C without (directly) using its constructors. This might be used for the following:

1. To circumvent the restriction that constructors can be overloaded only if their signatures differ.
2. Making the class non-inheritable by making the constructors private.
3. Preventing stack allocation by making constructors private

Declare a static member function that uses a private constructor to create the object and return it. (It could also return a pointer or a reference but this complication seems useless, and turns this into the factory pattern rather than a conventional named constructor.)

Here's an example for a class that stores a temperature that can be specified in any of the different temperature scales.

class Temperature
{
    public:
        static Temperature Fahrenheit (double f);
        static Temperature Celsius (double c);
        static Temperature Kelvin (double k);
    private:
        Temperature (double temp);
        double _temp;
};
 
Temperature::Temperature (double temp):_temp (temp) {}
 
Temperature Temperature::Fahrenheit (double f)
{
    return Temperature ((f + 459.67) / 1.8);
}
 
Temperature Temperature::Celsius (double c)
{
    return Temperature (c + 273.15);
}
 
Temperature Temperature::Kelvin (double k)
{
    return Temperature (k);
}

static data member

The use of the static specifier in a data member, it will make that member to be shared all instances of owner class and derived classes. To use static data members you must declare the data member as static and initialize it outside of the class declaration, at file scope.

class Temperatures
{
  private:
    static int Celsius_Degrees;
};
 
int Temperatures::Celsius_Degrees = 0; // Or any other needed initialization.

inline

Sharing most of the concepts we have seen before on the introduction to inline functions, when dealing with member function those concepts are extended, with a few additional considerations.

If the member functions definition is included inside the definition of the class. that function is by default made implicitly inline. Compiler options may also override this behavior.

Calls to virtual functions cannot be inlined if the object's type is not known at compile-time, because we don't know which function to inline.

Constructors

A constructor is a special member function which is called whenever a new instance of a class is created. The compiler calls the constructor after the new object has been allocated in memory, and converts that "raw" memory into a proper, typed object.

The constructor is declared as any normal member function but it will share the name of the class. Constructors are responsible for almost all of the run-time setup necessary for the class operation. Its main purpose becomes in general defining the data members upon object instantiation (when an object is declared), they can also also have arguments, if the programmer so chooses. If a constructor has arguments, then they should also be added to the declaration of any other object of that class when using the new operator. Constructors can also be overloaded.

Foo myTest;                 // essentially what happens is:  Foo myTest();
Foo myTest( 3, 54 );        // accessing the overloaded constructor
Foo myTest = Foo( 20, 45 ); // although a new object is created, there are some extra function calls involved
                            // with more complex classes, an assignment operator should
                            // be defined to ensure a proper copy (includes ''deep copy'')
                            // myTest would be constructed with the default constructor, and then the
                            // assignment operator copies the unnamed Foo( 20, 45 ) object to myTest

using new with a constructor

Foo* myTest = new Foo();           // this defines a pointer to a dynamically allocated object
Foo* myTest = new Foo( 40, 34 );   // constructed with Foo( 40, 34 )
// be sure to use delete to avoid memory leaks

A constructor can't delegate to another. It is also considered desirable to reduce the use of default arguments, if a maintainer has to write and maintain multiple constructors it can result in code duplication, which reduces maintainability because of the potential for introducing inconsistencies and even lead to code bloat.

Default Constructors

A default constructor is one which can be called with no arguments. Most commonly, a default constructor is declared without any parameters, but it is also possible for a constructor with parameters to be a default constructor if all of those parameters are given default values.

In order to create an array of objects of a class type, the class must have an accessible default constructor; C++ has no syntax to specify constructor arguments for array elements.

Overloaded Constructors

When an object of a class is instantiated, the class writer can provide various constructors each with a different purpose. A large class would have many data members, some of which may or may not be defined when an object is instantiated. Anyway, each project will vary, so a programmer should investigate various possibilities when providing constructors.

These are all constructors for a class myFoo.

 
myFoo(); // default constructor, the user has no control over initial values
         // overloaded constructors
 
myFoo( int a, int b=0 ); // allows construction with a certain 'a' value, but accepts 'b' as 0
                         // or allows the user to provide both 'a' and 'b' values
 // or
 
myFoo( int a, int b ); // overloaded constructor, the user must specify both values
 
class myFoo {
private:
  int Useful1;
  int Useful2;
 
public:
  myFoo(){                     // default constructor
           Useful1 = 5;
           Useful2 = 10;  
          };
 
  myFoo( int a, int b = 0 ) { // two possible cases when invoked
         Useful1 = a;
         Useful2 = b;
  };
 
};
 
myFoo Find;           // default constructor, private member values Useful1 = 5, Useful2 = 10
myFoo Find( 8 );      // overloaded constructor case 1, private member values Useful1 = 8, Useful2 = 0
myFoo Find( 8, 256 ); // overloaded constructor case 2, private member values Useful1 = 8, Useful2 = 256

Constructor initialization lists

Constructor initialization lists (or member initialization list) are the only way to initialize data members and base classes with a non-default constructor. Constructors for the members are included between the argument list and the body of the constructor (separated from the argument list by a colon). Using the initialization lists is not only better in terms of efficiency but also the simplest way to guarantee that all initialization of data members are done before entering the body of constructors.

// Using the initialization list for _myComplexMember 
MyClass::MyClass(int mySimpleMember, MyComplexClass myComplexMember)
: _myComplexMember(myComplexMember) // only 1 call, to the copy constructor
{
 _mySimpleMember=mySimpleMember; // uses 2 calls, one for the constructor of the mySimpleMember class
                                 // and a second for the assignment operator of the MyComplexClass class
}

This is more efficient than assigning value to the complex data member inside the body of the constructor because in that case the variable is initialized with its corresponding constructor.

Note that the arguments provided to the constructors of the members do not need to be arguments to the constructor of the class; they can also be constants. Therefore you can create a default constructor for a class containing a member with no default constructor.

Example:

MyClass::MyClass() : _myComplexMember(0) { }

It is useful to initialize your members in the constructor using this initialization lists. This makes it obvious for the reader that the constructor does not execute logic. The order the initialization is done should be the same as you defined your base-classes and members. Otherwise you can get warnings at compile-time. Once you start initializing your members make sure to keep all in the constructor(s) to avoid confusion and possible 0xbaadfood.

It is safe to use constructor parameters that are named like members.

Example:

class MyClass : public MyBaseClassA, public MyBaseClassB {
  private:
    int c;
    void *pointerMember;
  public:
    MyClass(int,int,int);
};
/*...*/
MyClass::MyClass(int a, int b, int c):
 MyBaseClassA(a)
,MyBaseClassB(b)
,c(c)
,pointerMember(NULL)
,referenceMember()
{
 //logic
}

Note that this technique was also possible for normal functions but it is now obsoleted and is classified as an error in such case.

Destructors

Destructors like the Constructors are declared as any normal member functions but will share the same name as the Class, what distinguishes them is that the Destructor's name is preceded with a "~", it can not have arguments and can't be overloaded.

Destructors are called whenever an Object of the Class is destroyed. Destructors are crucial in avoiding resource leaks (by deallocating memory), and in implementing the RAII idiom. Resources which are allocated in a Constructor of a Class are usually released in the Destructor of that Class as to return the system to some known or stable state after the Class ceases to exist.

The Destructor is invoked when Objects are destroyed, after the function they were declared in returns, when the delete operator is used or when the program is over. If an object of a derived type is destructed, first the Destructor of the most derived object is executed. Then member objects and base class subjects are destructed recursively, in the reverse order their corresponding Constructors completed. As with Structs the compiler implicitly-declares a Destructor as a inline public member of its class if the class doesn’t have a user-declared Destructor.

The dynamic type of the object will change from the most derived type as Destructors run, symmetrically to how it changes as Constructors execute. This affects the functions called by virtual calls during construction and destruction, and leads to the common (and reasonable) advice to avoid calling virtual functions of an object either directly or indirectly from its Constructors or Destructors.

Subsumption property

Subsumption is a property that all objects that reside in a class hierarchy must fulfill: an object of the base class can be substituted by an object that derives from it (directly or indirectly). All mammals are animals (they derive from them), and all cats are mammals. Therefore, because of the subsumption property we can "treat" any mammal as an animal and any cat as a mammal. This implies abstraction, because when we are "treating" a mammal as an animal, the only we should know about it is that it lives, it grows, etc, but nothing related to mammals.

This property is applied in C++, whenever we are using pointers or references to objects that reside in a class hierarchy. In other words, a pointer of class animal can point to an object of class animal, mammal or cat.

Let's continue with our example:

 //needs to be corrected
  enum animalType {
       Herbivore;
       Carnivore;
       Omnivore;
  }
 
  class animal {
        public:
               bool isAlive;
               animalType Type;
               int numberOfChildren;
  }
 
  class mammal : public animal{
        public:
               int numberOfTeats;
  }
 
  class cat: public mammal{
        public:
              bool likesFish; //probably true
  }
 
  int main {
      animal* a1= new animal;
      animal* a2= new mammal;
      animal* a3= new cat;
      mammal* m= new cat;
 
      a2->isAlive= True; //Correct
      a2->Type= Herbivore; //Correct
      m->numberOfTeats=2; //Correct
 
      a2->numberOfTeats=6; //Incorrect
      a3->likesFish=True; //Incorrect
 
      cat* c= (cat*)a3; //Downcast, correct (but very poor practice, see later)
      c->likesFish=False //Correct (although it is a very awkward cat)
   }

In the last lines of the example there is cast of a pointer to animal, to a pointer to cat. This is called "Downcast". Downcasts are useful and should be used, but first we must ensure that the object we are casting is really of the type we are casting to it. Downcasting a base class to an unrelated class is an error. To resolve this issue, the casting operators dynamic_cast<>, or static_cast<> should be used. These correctly cast an object from one class to another, and will throw an exception if the class types are not related. eg. If you try:

cat* c = new cat;
motorbike* m;
m = dynamic_cast<motorbike*>(c);

then the app will throw an exception as a cat is not a motorbike. Static_cast is very similar, only it will perform the type checking at compile time. If you have an object where you are not sure of its type then you should use dynamic_cast, and be prepared to handle errors when casting. If you are downcasting objects where you know the types, then you should use static_cast. Do not use old-style C casts as these will simply give you an access violation if the types cast are unrelated.

Ensuring objects of a class are never copied

This is required e.g. to prevent memory-related problems that would result in case the default copy-constructor or the default assignment operator is unintentionally applied to a class C which uses dynamically allocated memory, where a copy-constructor and an assignment operator are probably an overkill as they won't be used frequently.

Some style guidelines suggest making all classes non-copyable by default, and only enabling copying if it makes sense. Other (bad) guidelines say that you should always explicitly write the copy constructor and copy assignment operators; that's actually a bad idea, as it adds to the maintenance effort, adds to the work to read a class, is more likely to introduce errors than using the implicitly declared ones, and doesn't make sense for most object types. A sensible guideline is to think about whether copying makes sense for a type; if it does, then first prefer to arrange that the compiler-generated copy operations will do the right thing (e.g., by holding all resources via resource management classes rather than via raw pointers or handles), and if that's not reasonable then obey the law of three. If copying doesn't make sense, you can disallow it in either of two idiomatic ways as shown below.

Just declare the copy-constructor and assignment operator, and make them private. Do not define them. As they are not protected or public, they are inaccessible outside the class. Using them within the class would give a linker error since they are not defined.

class C
{
  ...
 
  private:
    // Not defined anywhere
    C (const C&);
    C& operator= (const C&);
};

Remember that if the class uses dynamically allocated memory for data members, you must define the memory release procedures in destructor ~C () to release the allocated memory.

A class which only declares these two functions can be used as a private base class, so that all classes which privately inherits such a class will disallow copying. One of the Boost libraries, contains the boost:noncopyable utility class which is a practical example. Below shows an example of using this class:

class C : boost::noncopyable
{
  ...
};

Container class

A class that is used to hold objects in memory or external storage is often called a container class. A container class acts as a generic holder and has a predefined behavior and a well-known interface. It is also a supporting class whose purpose is to hide the topology used for maintaining the list of objects in memory. When it contains a group of mixed objects, the container is called a heterogeneous container; when the container is holding a group of objects that are all the same, the container is called a homogeneous container.

Interface class

TODO Complete

Singleton class

A Singleton class is a class that can only be instantiated once (similar to the use of static variables or functions). It is one of the possible implementations of a creational pattern, which is fully covered in the Design Patterns Section of the book.

Abstract Classes

An abstract class is, conceptually, a class that cannot be instantiated and is usually implemented as a class that has one or more pure virtual (abstract) functions.

A pure virtual function is one which must be overridden by any concrete (i.e., non-abstract) derived class. This is indicated in the declaration with the syntax " = 0" in the member function's declaration.

Example

class AbstractClass {
public:
  virtual void AbstractMemberFunction() = 0; //pure virtual function makes this class Abstract class
  virtual void NonAbstractMemberFunction1(); //virtual function
 
  void NonAbstractMemberFunction2();
};

In general an abstract class is used to define an implementation and is intended to be inherited from by concrete classes. It's a way of forcing a contract between the class designer and the users of that class. If we wish to create a concrete class (a class that can be instantiated) from an abstract class we must declare and define a matching member function for each abstract member function of the base class. Otherwise we will create a new abstract class (this could be useful sometimes).

Sometimes we use the phrase "pure abstract class," meaning a class that exclusively has pure virtual functions (and no data). The concept of interface is mapped to pure abstract classes in C++, as there is no construction "interface" in C++ the same way that there is in Java.

Example

 class Vehicle {
 public:
     explicit
     Vehicle( int topSpeed )
     : m_topSpeed( topSpeed )
     {}
     int TopSpeed() const {
        return m_topSpeed;
     }
 
     virtual void Save( std::ostream& ) const = 0;
 
 private:
     int m_topSpeed;
 };
 
 class WheeledLandVehicle : public Vehicle {
 public:
     WheeledLandVehicle( int topSpeed, int numberOfWheels )
     : Vehicle( topSpeed ), m_numberOfWheels( numberOfWheels )
     {}
     int NumberOfWheels() const {
       return m_numberOfWheels;
     }
 
     void Save( std::ostream& ) const; // is implicitly virtual
 
 private:
     int m_numberOfWheels;
 };
 
 class TrackedLandVehicle : public Vehicle {
 public:
    int TrackedLandVehicle ( int topSpeed, int numberOfTracks )
    : Vehicle( topSpeed), m_numberOfTracks ( numberOfTracks )
    {}
    int NumberOfTracks() const {
       return m_numberOfTracks;
    }
    void Save( std::ostream& ) const; // is implicitly virtual
 
  private:
    int m_numberOfTracks;
  };

In this example the Vehicle is an abstract base class as it has an abstract member function. It is not a pure abstract class as it also has data and concrete member functions. The class WheeledLandVehicle is derived from the base class. It also holds data which is common to all wheeled land vehicles, namely the number of wheels. The class TrackedLandVehicle is another variation of the Vehicle class.

This is something of a contrived example but it does show how that you can share implementation details among a hierarchy of classes. Each class further refines a concept. This is not always the best way to implement an interface but in some cases it works very well. As a guideline, for ease of maintenance and understanding you should try to limit the inheritance to no more than 3 levels. Often the best set of classes to use is a pure virtual abstract base class to define a common interface. Then use an abstract class to further refine an implementation for a set of concrete classes and lastly define the set of concrete classes.

Pure Abstract Classes

An abstract class is one in which there is a declaration but no definition for a member function. The way this concept is expressed in C++ is to have the member function declaration assigned to zero.

Example

class PureAbstractClass 
{
public:
  virtual void AbstractMemberFunction() = 0;
};

A pure Abstract class has only abstract member functions and no data or concrete member functions. In general, an abstract class is used to define an interface and is intended to be inherited by concrete classes. It's a way of forcing a contract between the class designer and the users of that class. The users of this class must declare a matching member function for the class to compile.

Example of usage for a pure Abstract Class

 class DrawableObject 
 {
  public:
    virtual void Draw(GraphicalDrawingBoard&) const = 0; //draw to GraphicalDrawingBoard
 };
 
 class Triangle : public DrawableObject
 {
 public:
   void Draw(GraphicalDrawingBoard&) const; //draw a triangle
 };
 
 class Rectangle : public DrawableObject
 {
 public:
   void Draw(GraphicalDrawingBoard&) const; //draw a rectangle
 };
 
 class Circle : public DrawableObject
 {
 public:
   void Draw(GraphicalDrawingBoard&) const; //draw a circle
 };
 
 typedef std::list<DrawableObject*> DrawableList_t;
 
 DrawableList_t drawableList;
 GraphicalDrawingBoard gdrawb;
 
 drawableList.pushback(new Triangle());
 drawableList.pushback(new Rectangle());
 drawableList.pushback(new Circle());
 
 for(DrawableList::const_iterator iter = drawableList.begin(), 
    endIter = drawableList.end();
    iter != endIter;
    ++iter)
 { 
   DrawableObject *object = *iter;
   object->Draw(gdrawb);
 }

Note that this is a bit of a contrived example and that the drawable objects are not fully defined (no constructors or data) but it should give you the general idea of the power of defining an interface. Once the objects are constructed, the code that calls the interface does not know any of the implementation details of the called objects, only that of the interface. The object GraphicalDrawingBoard is a placeholder meant to represent the thing onto which the object will be drawn, i.e. the video memory, drawing buffer, printer.

Note that there is a great temptation to add concrete member functions and data to pure abstract base classes. This must be resisted and in general it is a sign that the interface is not well factored. Data and concrete member functions tend to imply a particular implementation and as such can inherit from the interface but should not be that interface. Instead if there is some commonality between concrete classes, creation of abstract class which inherits its interface from the pure abstract class and defines the common data and member functions of the concrete classes works well. Some care should be taken to decide whether inheritance or aggregation should be used. Too many layers of inheritance can make the maintenance and usage of a class difficult. Generally, the maximum accepted layers of inheritance is about 3, above that and refactoring of the classes is generally called for. A general test is the "is a" vs "has a", as in a Square is a Rectangle, but a Square has a set of sides.

What is a "nice" class?

A "nice" class takes into consideration the use of the following functions:

1. The copy constructor.
2. The assignment operator.
3. The equality operator.
4. The inequality operator.

Class Declaration

class Nice
{
public:
    Nice(const Nice &Copy);
    Nice &operator= (const Nice &Copy);
    bool operator== (const Nice &param) const;
    bool operator!= (const Nice &param) const;
};

Description

A "nice" class could also be called a container safe class. Many containers such as those in the Standard Template Library (STL), that we'll see later, use copy construction and the assignment operator when interacting with the objects of your class. The assignment operator and copy constructor only need to be declared and defined if the default behavior, which is a member-wise (not binary) copy, is undesirable or insufficient to properly copy/construct your object.

A general rule of thumb is that if the default, member-wise copy operations do not work for your objects then you should define a suitable copy constructor and assignment operator. They are both needed if either is defined.

Copy Constructor

The purpose of the copy constructor is to allow the programmer to perform the same instructions as the assignment operator with the special case of knowing that the caller is initializing/constructing rather than an copying.

It is also good practice to use the explicit keyword when using a copy constructor to prevent unintended implicit type conversion.

Example

class Nice
{
public:
  explicit Nice(int _a) : a(_a)
  {
    return;
  }
private:
  int a;
};
 
class NotNice
{
public:
  NotNice(int _a) :  a(_a)
  {
    return;
  }
private:
  int a;
};
 
int main()
{
  Nice proper = Nice(10); //this is ok
  Nice notproper = 10; //this will result in an error
  NotNice eg = 10; //this WILL compile, you may not have intended this conversion
  return 0;
}

Equality Operator

The equality operator says, "Is this object equal to that object?". What constitutes equal is up to the programmer. This is a requirement if you ever want to use the equality operator with objects of your class.

However, in most applications (e.g. mathematics), it is usually the case that coding the inequality is easier than coding the equality. In which case the following code can be written for the equality.

inline bool Nice::operator== (const Nice& param) const
{
    return !(*this != param);
}

Inequality Operator

The inequality operator says, "Is this object not equal to that object?". What constitutes not equal is up to the programmer. This is a requirement if you ever want to use the inequality operator with objects of your class.

However, in some applications, coding the equality is easier than coding the inequality. In which case the following code can be written for the inequality.

inline bool Nice::operator!= (const Nice& param) const
{
    return !(*this == param);
}

If the statement about the (in)equality operators having different efficiency (whatever kind) seems complete nonsense to you, consider that typically, all object attributes must match for two objects to be considered equal.
Typically, only one object attribute must differ for two objects to be considered unequal. For equality and inequality operators, that doesn't mean one is faster than the other.

Note, however, that using both the above equality and inequality functions as defined will result in an infinite recursive loop and care must be taken to use only one or the other. Also, there are some situations where neither applies and therefore neither of the above can be used.

Given two objects A and B (with class attributes x and y), an equality operator could be written as

  if (A.x != B.x) return false;
  if (A.y != B.y) return false;
  return true;

while an inequality operator could be written as

  if (A.x != B.x) return true;
  if (A.y != B.y) return true;
  return false;

So yes, the equality operator can certainly be written ...!(a!=b)..., but it isn't any faster. In fact, there's the additional overhead of a method call and a negation operation.

So the question becomes, is a little execution overhead worth the smaller code and improved maintainability? There is no simple answer to this it all depend on how the programmer is using them. If your class is composed of, say, an array of 1 billion elements, the overhead is negligible.

Operator overloading

Operator overloading (less commonly known as ad-hoc polymorphism) is a specific case of polymorphism (part of the OO nature of the language) in which some or all operators like +, = or == are treated as polymorphic functions and as such have different behaviors depending on the types of its arguments. Operator overloading is usually only syntactic sugar. It can easily be emulated using function calls:

:a + b × c

In a language that supports operator overloading is effectively a more concise way of writing:

operator_add (a, operator_multiply (b,c))

(Assuming the × operator has higher precedence than +.)

Operator overloading provides more than an aesthetic benefit when the language allows operators to be invoked implicitly in some circumstances.

Operator overloading has been criticized because it allows programmers to give operators completely different functionality depending on the types of their operands, usage of the << operator is an example of this problem.

// The expression
a << 1;

Will return twice the value of a if a is an integer variable, but if a is an output stream instead this will write "1" to it. Because operator overloading allows the programmer to change the usual semantics of an operator, it is usually considered good practice to use operator overloading with care.

To overload an operator is to provide it with a new meaning for user-defined types. This is done in the same fashion as defining a function. The basic syntax follows (where @ represents a valid operator):

return_type operator@(parameter_list)
{
    // ... definition
}

Not all operators may be overloaded, new operators cannot be created, and the precedence, associativity or arity of operators cannot be changed (for example ! cannot be overloaded as a binary operator). Most operators may be overloaded as either a member function or non-member function, some, however, must be defined as member functions. Operators should only be overloaded where their use would be natural and unambiguous, and they should perform as expected. For example, overloading + to add two complex numbers is a good use, whereas overloading * to push an object onto a vector would not be considered good style.

A simple Message Header

// sample of Operator Overloading
 
#include <string>
 
class PlMessageHeader
{
    std::string m_ThreadSender;
    std::string m_ThreadReceiver;
 
    //return true if the messages are equal, false otherwise
    inline bool operator == (const PlMessageHeader &b) const
    {
        return ( (b.m_ThreadSender==m_ThreadSender) &&
                (b.m_ThreadReceiver==m_ThreadReceiver) );
    }
 
    //return true if the message is for name
    inline bool isFor (const std::string &name) const
    {
        return (m_ThreadReceiver==name);
    }
 
    //return true if the message is for name
    inline bool isFor (const char *name) const
    {
        return (m_ThreadReceiver.compare(name)==0);
    }
};

Operators as member functions

Operators may be overloaded as member or non-member functions. Aside from the operators which must be members, the choice of whether or not to overload as a member is up to the programmer. Operators are generally overloaded as members when they:

1. change the left-hand operand, or
2. require direct access to the non-public parts of an object.

When an operator is defined as a member, the number of explicit parameters is reduced by one, as the calling object is implicitly supplied as an operand. Thus, binary operators take one explicit parameter and unary operators none. In the case of binary operators, the left hand operand is the calling object, and no type coercion will be done upon it. This is in contrast to non-member operators, where the left hand operand may be coerced.

    // binary operator as member function
    Vector2D Vector2D::operator+(const Vector2D& right) {...}
 
    // binary operator as non-member function
    Vector2D operator+(const Vector2D& left, const Vector2D& right) {...}
 
    // binary operator as member function with 2 arguments 
    friend Vector2D operator+(const Vector2D& left, const Vector2D& right) {...}
 
    // unary operator as member function
    Vector2D Vector2D::operator-() {...}
 
    // unary operator as non-member function
    Vector2D operator-(const Vector2D& vec) {...}

Overloadable operators

Arithmetic operators

• + (addition)
• - (subtraction)
• * (multiplication)
• / (division)
• % (modulus)

As binary operators, these involve two arguments which do not have to be the same type. These operators may be defined as member or non-member functions. An example illustrating overloading + for the addition of a 2D mathematical vector type follows.

Vector2D operator+(const Vector2D& left, const Vector2D& right)
{
    Vector2D result;
    result.set_x(left.x() + right.x());
    result.set_y(left.y() + right.y());
    return result;
}

It is good style to only overload these operators to perform their customary arithmetic operation.

Bitwise operators

• ^ (XOR)
• | (OR)
• & (AND)
• ~ (complement)
• << (shift left, insertion to stream)
• >> (shift right, extraction from stream)

All of the bitwise operators are binary, excepting complement, which is unary. It should be noted that these operators have a lower precedence than the arithmetic operators, so if ^ were to be overloaded for exponentiation, x ^ y + z may not work as intended. Of special mention are the shift operators, << and >>. These have been overloaded in the standard library for interaction with streams. When overloading these operators to work with streams the rules below should be followed:

1. overload << and >> as friends (so that it can access the private variables with the stream be passed in by references
2. (input/output modifies the stream, and copying is not allowed)
3. the operator should return a reference to the stream it receives (to allow chaining, cout << 3 << 4 << 5)

An example using a 2D vector

friend ostream& operator<<(ostream& out, const Vector2D& vec) // output
{
    out << "(" << vec.x() << ", " << vec.y() << ")";
    return out;
}
 
friend istream& operator>>(istream& in, Vector2D& vec) // input
{
    double x, y;
    in >> x >> y;
    vec.set_x(x);
    vec.set_y(y);
    return in;
}

Assignment operator

The assignment operator, =, must be a member function, and is given default behavior for user-defined classes by the compiler, performing an assignment of every member using its assignment operator. This behavior is generally acceptable for simple classes which only contain variables. However, where a class contains references or pointers to outside resources, the assignment operator should be overloaded (as general rule, whenever a destructor and copy constructor are needed so is the assignment operator), otherwise, for example, two strings would share the same buffer and changing one would change the other.

In this case, an assignment operator should perform two duties:

1. clean up the old contents of the object
2. copy the resources of the other object

For classes which contain raw pointers, before doing the assignment, the assignment operator should check for self-assignment, which generally will not work (as when the old contents of the object are erased, they cannot be copied to refill the object). Self assignment is generally a sign of a coding error, and thus for classes without raw pointers, this check is often omitted, as while the action is wasteful of cpu cycles, it has no other effect on the code.

Example

class WithRawPointer {
    T *m_ptr;
    public:
    WithRawPointer(T *ptr) : m_ptr(ptr) {}
    WithRawPointer& operator=(WithRawPointer const &rhs) {
        delete m_ptr; // free resource;
        m_ptr = 0;
        m_ptr = rhs.m_ptr;
        return *this;
    };
};
 
WithRawPointer x(new T);
x = x; // x.m_ptr == 0.
 
class WithRawPointer2 {
    T *m_ptr;
    public:
    WithRawPointer2(T *ptr) : m_ptr(ptr) {}
    WithRawPointer2& operator=(WithRawPointer2 const &rhs) {
        if (this != &rhs) {
            delete m_ptr; // free resource;
            m_ptr = 0;
            m_ptr = rhs.m_ptr;
        }
        return *this;
    };
};
 
WithRawPointer2 x2(new T);
x2 = x2; // x2.m_ptr unchanged.

Another common use of overloading the assignment operator is to declare the overload in the private part of the class and not define it. Thus any code which attempts to do an assignment will fail on two accounts, first by referencing a private member function and second fail to link by not having a valid definition. This is done for classes where copying is to be prevented, and generally done with the addition of a privately declared copy constructor

Example

class DoNotCopyOrAssign {
    public:
        DoNotCopyOrAssign() {};
    private:
        DoNotCopyOrAssign(DoNotCopyOrAssign const&);
        DoNotCopyOrAssign &operator=(DoNotCopyOrAssign const &);
};
 
class MyClass : public DoNotCopyOrAssign {
    public:
        MyClass();
};
 
MyClass x, y;
x = y; // Fails to compile due to private assignment operator;
MyClass z(x); // Fails to compile due to private copy constructor.

Relational operators

• == (equality)
• != (inequality)
• > (greater-than)
• < (less-than)
• >= (greater-than-or-equal-to)
• <= (less-than-or-equal-to)

All relational operators are binary, and should return either true or false. Generally, all six operators can be based off a comparison function, or each other, although this is never done automatically (e.g. overloading > will not automatically overload < to give the opposite). There are, however, some templates defined in the header <utility>; if this header is included, then it suffices to just overload operator== and operator<, and the other operators will be provided by the STL.

Logical operators

• ! (NOT)
• && (AND)
• || (OR)

The ! operator is unary, && and || are binary. It should be noted that in normal use, && and || have "short-circuit" behavior, where the right operand may not be evaluated, depending on the left operand. When overloaded, these operators get function call precedence, and this short circuit behavior is lost. It is best to leave these operators alone.

Example

bool Function1();
bool Function2();
 
Function1() && Function2();

If the result of Function1() is false, then Function2() is not called.

MyBool Function3();
MyBool Function4();
 
bool operator&&(MyBool const &, MyBool const &);
 
Function3() && Function4()

Both Function3() and Function4() will be called no matter what the result of the call is to Function3() This is a waste of CPU processing, and worse, it could have surprising unintended consequences compared to the expected "short-circuit" behavior of the default operators. Consider:

extern MyObject * ObjectPointer;
bool Function1() { return ObjectPointer != null; }
bool Function2() { return ObjectPointer->MyMethod(); }
MyBool Function3() { return ObjectPointer != null; }
MyBool Function4() { return ObjectPointer->MyMethod(); }
 
bool operator&&(MyBool const &, MyBool const &);
 
Function1() && Function2(); // Does not execute Function2() when pointer is null
Function3() && Function4(); // Executes Function4() when pointer is null

Compound assignment operators

• += (addition-assignment)
• -= (subtraction-assignment)
• *= (multiplication-assignment)
• /= (division-assignment)
• %= (modulus-assignment)
• &= (AND-assignment)
• |= (OR-assignment)
• ^= (XOR-assignment)
• >>= (shift-right-assignment)
• <<= (shift-left-assignment)

Compound assignment operators should be overloaded as member functions, as they change the left-hand operand. Like all other operators (except basic assignment), compound assignment operators must be explicitly defined, they will not be automatically (e.g. overloading = and + will not automatically overload +=). A compound assignment operator should work as expected: A @= B should be equivalent to A = A @ B. An example of += for a two-dimensional mathematical vector type follows.

Vector2D& Vector2D::operator+=(const Vector2D& right)
{
    this->x += right.x;
    this->y += right.y;
    return *this;
}

Increment and decrement operators

• ++ (increment)
• -- (decrement)

Increment and decrement have two forms, prefix (++i) and postfix (i++). To differentiate, the postfix version takes a dummy integer. Increment and decrement operators are most often member functions, as they generally need access to the private member data in the class. The prefix version in general should return a reference to the changed object. The postfix version should just return a copy of the original value. In a perfect world, A += 1, A = A + 1, A++, ++A should all leave A with the same value.

Example

SomeValue& SomeValue::operator++() // prefix
{
    ++data;
    return *this;
}
 
SomeValue SomeValue::operator++(int unused) // postfix
{
    SomeValue result = *this;
    ++data;
    return result;
}

Often one operator is defined in terms of the other for ease in maintenance, especially if the function call is complex.

SomeValue SomeValue::operator++(int unused) // postfix
{
    SomeValue result = *this;
    ++(*this); // call SomeValue::operator++()
    return result;
}

Subscript operator

The subscript operator, [ ], is a binary operator which must be a member function (hence it takes only one explicit parameter, the index). The subscript operator is not limited to taking an integral index. For instance, the index for the subscript operator for the std::map template is the same as the type of the key, so it may be a string etc. The subscript operator is generally overloaded twice; as a non-constant function (for when elements are altered), and as a constant function (for when elements are only accessed).

Function call operator

The function call operator, ( ), is generally overloaded to create objects which behave like functions, or for classes that have a primary operation. The function call operator must be a member function, but has no other restrictions - it may be overloaded with any number of parameters of any type, and may return any type. A class may also have several definitions for the function call operator.

Address of, Reference, and Pointer operators

These three operators, operator&(), operator*() and operator->() can be overloaded. In general these operators are only overloaded for smart pointers, or classes which attempt to mimic the behavior of a raw pointer. The pointer operator, operator->() has the additional requirement that the result of the call to that operator, must return a pointer, or a class with an overloaded operator->(). In general A == *&A should be true.

Example

class T {
    public:
        const memberFunction() const;
};
 
// forward declaration
class DullSmartReference;
 
class DullSmartPointer {
    private:
        T *m_ptr;
    public:
        DullSmartPointer(T *rhs) : m_ptr(rhs) {};
        DullSmartReference operator*() const {
            return DullSmartReference(*m_ptr);
        }
        T *operator->() const {
            return m_ptr;
        }
};
 
class DullSmartReference {
    private:
        T *m_ptr;
    public:
        DullSmartReference (T &rhs) : m_ptr(&rhs) {}
        DullSmartPointer operator&() const {
            return DullSmartPointer(m_ptr);
        }
        // conversion operator
        operator T { return *m_ptr; }
};
 
DullSmartPointer dsp(new T);
dsp->memberFunction(); // calls T::memberFunction
 
T t;
DullSmartReference dsr(t);
dsp = &dsr;
t = dsr; // calls the conversion operator

These are extremely simplified examples designed to show how the operators can be overloaded and not the full details of a SmartPointer or SmartReference class. In general you won't want to overload all three of these operators in the same class.

Comma operator

The comma operator,() , can be overloaded. The language comma operator has left to right precedence, the operator,() has function call precedence, so be aware that overloading the comma operator has many pitfalls.

Example

MyClass operator,(MyClass const &, MyClass const &);
 
MyClass Function1();
MyClass Function2();
 
MyClass x = Function1(), Function2();

For non overloaded comma operator, the order of execution will be Function1(), Function2(); With the overloaded comma operator, the compiler can call either Function1(), or Function2() first.

Member access operators

The two member access operators, operator->() and operator->*() can be overloaded. The most common use of overloading these operators is with defining expression template classes, which is not a common programming technique. Clearly by overloading these operators you can create some very unmaintainable code so overload these operators only with great care.

When the -> operator is applied to a pointer value of type (T *), the language dereferences the pointer and applies the . member access operator (so x->m is equivalent to (*x).m). However, when the -> operator is applied to a class instance, it is called as a unary postfix operator; it is expected to return a value to which the -> operator can again be applied. Typically, this will be a value of type (T *), as in the example under Address of, Reference, and Pointer operators above, but can also be a class instance with operator->() defined; the language will call operator->() as many times as necessary until it arrives at a value of type (T *).

Memory management operators

• new (allocate memory for object)
• new[ ] (allocate memory for array)
• delete (deallocate memory for object)
• delete[ ] (deallocate memory for array)

The memory management operators can be overloaded to customize allocation and deallocation (e.g. to insert pertinent memory headers). They should behave as expected, new should return a pointer to a newly allocated object on the heap, delete should deallocate memory, ignoring a NULL argument. To overload new, several rules must be followed:

• new must be a member function
• the return type must be void*
• the first explicit parameter must be a size_t value

To overload delete there are also conditions:

• delete must be a member function (and cannot be virtual)
• the return type must be void
• there are only two forms available for the parameter list, and only one of the forms may appear in a class:
  • void*
  • void*, size_t

Conversion operators

Conversion operators enable objects of a class to be either implicitly (coercion) or explicitly (casting) converted to another type. Conversion operators must be member functions, and should not change the object which is being converted, so should be flagged as constant functions. The basic syntax of a conversion operator declaration, and declaration for an int-conversion operator follows.

operator ''type''() const; // const is not necessary, but is good style
operator int() const;

Notice that the function is declared without a return-type, which can easily be inferred from the type of conversion. Including the return type in the function header for a conversion operator is a syntax error.

double operator double() const; // error - return type included

Operators which cannot be overloaded

• ?: (conditional)
• . (member selection)
• .* (member selection with pointer-to-member)
• :: (scope resolution)
• sizeof (object size information)
• typeid (object type information)

To know the reason why we cannot overload above operators refer to "Why can't I overload dot, ::, sizeof, etc.?" see www.research.att.com ( http://www.research.att.com/~bs/bs_faq2.html#overload-dot ).

Chapter Summary

1. Structures
2. Unions
3. Classes (Inheritance , Member Functions, Polymorphism and this pointer)
   1. Abstract Classes including Pure abstract classes (abstract types)
   2. Nice Class
4. Operator overloading