Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!think.com!barmar
From: barmar@think.com (Barry Margolin)
Newsgroups: comp.lang.c++
Subject: Re: char* vs void*
Keywords: char* vs void* pointer arithmitic
Message-ID: <1991Jan4.063649.6464@Think.COM>
Date: 4 Jan 91 06:36:49 GMT
References: <277BAC53.F2A@tct.uucp> <1991Jan2.171429.11566@glinj.gli.com> <11223@cae780.csi.com>
Sender: news@Think.COM
Organization: Thinking Machines Corporation, Cambridge MA, USA
Lines: 36

In article <11223@cae780.csi.com> donald@cae780.UUCP (Donald Maffly) writes:
>Why did ANSI decide that one cannot do pointer arithmetic on a void* 
>pointer, you ask.
>
>Well, BY DEFINITION, void* means that 
>we are dealing with a pointer to an object of unknown type and size.  
>In pointer arithmetic, how could a C compiler possibly know how much 
>to decrement or increment a pointer by if it didn't know the size of 
>the object it was pointing to? 

Well, they could have defined it in terms of the units that sizeof measures
in.  For instance, if you have

	typedef ... foo_struct;
	foo_struct[N] foo;
	void *vp;
	foo_struct *fp;

	fp = &foo_struct[0];
	vp = (void *) fp;
	vp += sizeof foo_struct;
	fp = (foo_struct *) vp;

At this point, fp == &foo[1] .

This currently works if you replace "void" with "char", because sizeof char
is required to be 1.  They could have specified that sizeof void must
also be 1, and then all the uses of "char" as the generic basic memory
object could be abandoned.  This would pave the way for future revisions of
the C standard to allow sizeof char > 1 (e.g. to get rid of the char/w_char
distinction).
--
Barry Margolin, Thinking Machines Corp.

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