Question

I am new to C++. Please consider the following code:

class foo 
{ 
 int f; 

 public: 
   foo(int f1=0) : f(f1) { 
   cout<<"In conversion ctor\n"; 
 } 

foo(const foo & rhs) : f(rhs.f) 
{ 
    cout<<" In copy ctor\n"; 
} 

 foo& operator=(const foo & that) 
{ 
    f=that.f; 
    cout<<"In = optor\n"; 
    return *this; 
} 

}; 

 foo rbv() 
  { 
    foo obj(9); 
    return obj;                        //named return by value [def. 1] 
  } 

   foo caller() 
   { 
      return rbv();              // return by value [def. 2] 
   } 

int main(void) 
{ 
   foo box=caller(); 
   return 0; 
 }

1. Are the definitions for RBV and NRBV correct as indicated in the
   comments?
2. Is it mandatory to have an accessible copy ctor defined though it
   is not called during RVO?

3. Without RVO, in the code blocks

      foo rbv() 
       { 
        foo obj(9); 
        return obj; 
       } 
   
       foo ret= rbv();
   

Are the following steps correct in creation of ‘ret’

(1) a temporary ( say obj_temp) is created using copy ctor from obj,
stack object ‘obj’ destroyed,

(2) ret is copy constructed from obj_temp, obj_temp destroyed later;

which implies there are three objects, ‘obj’ , ‘obj_temp’ and ‘ret’ and two copy ctors involved.

Answer 1

1. Looks right to me.
2. You have to have an accessible copy constructor defined if you are going to copy objects. Even if the copy can be optimized away.
3. That sounds right to me, though the precise ordering might be implementation defined. You would have to