asked on

Virtual functions

#include <iostream.h>

class Animal
{
      public:
      virtual char* speak(int) {return "Silence";}
};
class Cat:public Animal
{
      public:
      virtual char* speak(int) {return "Meow";}
};
class Dog:public Animal
{
      public:
      virtual char* speak(short) {return "Bow-wow";}
};

int main()
{
      Animal *p = new Dog;
      cout << p->speak( 0 ) << endl;
      return 0;
}

The speech of dogs is supposed to be 'Bow-wow' but instead the program prints 'silence'.Why did the dog not bark?

ASKER CERTIFIED SOLUTION

alexo

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

alexo

In C, a function is known by its name. In C++, a function is known by its "signature", which is the name and the types of the arguments. Therefore f(int) and f(short) are two different functions.

Now, for your example:

Animal *p = new Dog;

Although p points to Dog, its static type is Animal*.

/* ... */ p->speak( 0 ) /* ... */

The compiler does not know what actual object p will point to at run-time. So, it generates code to call the virtual function whose signature is known at compile-time, based on the information it knows about p.

class Animal
{
public:
virtual char* speak(int) {return "Silence";}
};

Since p is a pointer to Animal, the compiler will (virtually) call the speak(int) function.

class Dog:public Animal
{
public:
virtual char* speak(short) {return "Bow-wow";}
};

p points to a Dog object. However, the Dog object does not define a speak(int) finction (the speak(short) function has a different signature and thus considered a different function). Therefore, the inherited speak(int) function of the superclass (Animal) is called.

nietod

By the way, A nice thing about the Borland C++ compiler (there aren't many) is that it warns you when you do this. Most compilers don't warn you since this is legal C++. However, almost no one every really wants to do this on purpose, so Borland has enough sense to warn you.

danny_pav

this looks like one of those PC-lint bug of the month ads.

sreejesh

ASKER

Alexo, You pointed out the correct answer. Congrats ! and Thank You. Same to nietod and danny_pav also.

alexo

>> this looks like one of those PC-lint bug of the month ads.
Bingo! No wonder the "Bow-wow" looked familiar...