Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!think.com!barmar
From: barmar@think.com (Barry Margolin)
Newsgroups: comp.lang.c++
Subject: Re: Comparison functions for qsort() and bsearch()
Keywords: Sun, C++
Message-ID: <1991Jan2.191630.18581@Think.COM>
Date: 2 Jan 91 19:16:30 GMT
References: <3069@lupine.NCD.COM> <277640DD.4A70@tct.uucp> <1990Dec26.160636.15566@clear.com>
Sender: news@Think.COM
Organization: Thinking Machines Corporation, Cambridge MA, USA
Lines: 40

In article <1990Dec26.160636.15566@clear.com> rmartin@clear.com (Bob Martin) writes:
>	I dissagree.  As Ron said in his article.  It is always possible
>	to cast a pointer to anything to a pointer to void.  But it is
>	not guaranteed that you can cast a pointer to void back to 
>	its original.  

This was a point I made in my article a week or two ago, and I was
corrected.  You can always cast a pointer to void back to its original
type.  However, if you cast T* to void*, you can't always cast the void* to
<parent-of-T>*.

>		       Case in point is a class derived from multiple bases:
>
>		class AB : public A, public B;
>
>	Now take the following variable:
>		AB *ab;
>
>	The following two casts work just fine:  
>		(A *)ab;  (B *)ab;
>
>	But the following, though it may compile, is likely not to work properly
>	because it miscalculates the address of at least one of the base classes.
>		(A *)(void *)ab; or (B *)(void *)ab;

But the following do work:

	(A *) (AB *) (void *) ab;
	(B *) (AB *) (void *) ab;

>	Yet if qsort is to sort objects of type AB, then this is precicely what
>	the void* implementation of compar would be asked to do!  

No, that is only true if you are using a compar function intended for
sorting As or Bs.  If it is expecting ABs then it will work fine.
--
Barry Margolin, Thinking Machines Corp.

barmar@think.com
{uunet,harvard}!think!barmar