Consider the following C++ code:
#include <iostream>
using std::cout;
class A
{
public:
int a;
A():a(0)
{
cout << "A constructor\n";
}
virtual void f()
{
cout << "f inside A\n";
}
};
class C : public A
{
public:
int c;
virtual void f()
{
cout << "f inside C\n";
}
C():c(0)
{
cout << "C constructor\n";
}
};
int main()
{
A varA = C();
cout << "Size of C class: " << sizeof(C) << "\n";
cout << "Size of varA object: " << sizeof(varA) << "\n";
C* varC = static_cast<C*>(&varA);
varC->f();
cout << "varC->a is " << varC->a << "\n";
cout << "varC->c is " << varC->c << "\n";
}
The output of this program is:
A constructor
C constructor
Size of C class: 16
Size of varA object: 8
f inside A
varC->a is 0
varC->c is 1726166356
I initialize the varA
object with the constructor of class C
. Constructors of A and C class are called, but the varA
is a simply a A
object. I cast the address of varA
to C*
type and I try to call its f()
function, but it prints the f()
function of class A
, so I deduce that it is use the early binding mechanism to call it.
I think if I call the constructor of derived class, like this case, I obtain the same object if I had called the base constructor.
I think the only difference is the other constructors are called. Is my supposition right or there are any other differences?
Classic example of slicing. A varA = C();
leaves you with object of static and dynamic type of A
. As a result, C* varC = static_cast<C*>(&varA);
exhibits undefined behavior.