Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!sun-barr!cs.utexas.edu!uunet!mcvax!unido!iraun1!iravcl!s_heising
From: s_heising@iravcl.ira.uka.de
Newsgroups: comp.lang.c++
Subject: RE:Return types of virtual member functions
Message-ID: <297@iravcl.ira.uka.de>
Date: 1 Jun 89 15:02:02 GMT
Lines: 42
Organisation: Universitaet Karlsruhe, IRA, F.R. Germany


In his article 
<3403@hplabsz.HPL.HP.COM>  bs@hplabsz.HPL.HP.COM (Bob Shaw) 
writes about his problems with return types of virtual 
member functions.


void main()
{
  parent* presult;                                                                      
  child*  cresult;

  parent* p    = new parent;
  child*  c    = new child;

  presult = p->msg();           <-- expectation correct

  cresult = c->msg();           <-- expectation correct

  p = c;                        <-- misunderstanding!
                                    change in value, 
                                    not in type
  presult = p->msg();           <-- still calls parent::msg
                                    not child::msg as expected

}

I'm not an experienced C++ or C-programmer, 
my interest is theoretically rather. 
But I think the solution to the problem is as follows:

Bob is mixing up assigning values and declaring types.
to be more concrete, that part of c's value p can hold.
It doesn't makes p to become an object of class child, 
of class parent, not the one of class child. 

Bye!