How do references resolve the ambiguity in multiple inheritance?

Question

The following code shows the notorious diamond inheritance. But it seems that an appropriate reference designation avoids the ambiguity.

# include <iostream>
# include <stdio.h>

class B {
public:
  int m_a ;
} ;

class D1 : public B { } ;

class D2 : public B { } ;

class E : public D1, public D2 { } ;

  int main () {
  E e ;
  D1& c = e ;
  D2& d = e ; 

  e.D1::m_a = 2 ;
  e.D2::m_a = 2 ;

  std::cout << c.m_a << std::endl ;
  std::cout << d.m_a << std::endl ;

  c.m_a = 3 ;
  std::cout << c.m_a << std::endl ;
  std::cout << d.m_a << std::endl ;

  printf ( "%p\n", &c ) ;
  printf ( "%p\n", &d ) ;
  printf ( "%p\n", &e ) ;
  printf ( "\n" ) ;
  printf ("%p\n", (void*) &(c.m_a)) ;
  printf ("%p\n", (void*) &(d.m_a)) ;

  return 0 ;
}

The output is:

 2
 2
 3
 2
 0xffffcbf0
 0xffffcbf4
 0xffffcbf0

 0xffffcbf0
 0xffffcbf4

So it seems that a reference 'knows' where it should start in the memory layout of the object 'e' which contains duplicated D1::m_a and D2::m_a . I wonder how it is achieved in C++ implementation. Thanks!

Answer 1

The easiest way to discover what's happening is by printing (void*)(&c) and (void*)(&d) . The two references refer to distinct subobjects of e .

Also, the references to D1 and D2 do not know they're inside some E . A reference can refer to any object in memory, but it doesn't know about the surroundings of that object.