Question

#define CLASS(ID) class ID{ \
    public: \
            ID(int) { cout<<#ID <<"\tconstructor"<<endl; } \
           ~ID(){cout<<#ID<<"\tDestroyed  "<<endl;} \
};
CLASS(Base);
CLASS(Member);

class Derived : public Base {
    public:
        Member *mem;
        Derived(int x) : Base(1) {
            cout<<"Derived constructor"<<endl;
            mem=new Member(2);
        }
        ~Derived()
        {
            cout<<"Derived Destroyed"<<endl;
            delete mem;
        }
};

int main(int argc, char** argv) {
    Derived * der=new Derived(1);
    cout<<"****"<<endl;
    delete der;
}

The output for this is:

Derived constructor
Member  constructor
****
Derived Destroyed
Member  Destroyed  
Base    Destroyed

In the second version :

class Derived : public Base {
    public:
        Member *mem;
        Derived(int x) : Base(1) {
            cout<<"Derived constructor"<<endl;
        }

        ~Derived()
        {
            cout<<"Derived Destroyed"<<endl;
            delete mem;
        }
};

Why in the first version the Base constructor don’t executes when an instance from the same class has been instantiated?

Answer 1

On GCC version 4.6 I got this

Base    constructor
Derived constructor
Member  constructor
****
Derived Destroyed
Member  Destroyed  
Base    Destroyed  

So it seems to be working

On a side note your base destructor should be virtual

 virtual ~ID(){}

this way when you only have a reference to the base you derived will still cleanly be removed.

On a side note, please don’t use defines like this unless you have too, also wrapping up mem in a smart point is also a good idea. Like auto_ptr in c++98 (ideally the boost ptrs) or unique_ptr in c++11.