From 7135578897801561753
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: f78e5,e94695845965b249
X-Google-Attributes: gidf78e5,public
X-Google-ArrivalTime: 1992-09-10 03:23:16 PST
Newsgroups: comp.std.c++
Path: sparky!uunet!decwrl!borland.com!pete
From: pete@genghis.borland.com (Pete Becker)
Subject: Re: Zero-length structures and pointer comparisons
Message-ID: <1992Sep10.162906.14398@genghis.borland.com>
Originator: pete@genghis.borland.com
Sender: news@borland.com (News Admin)
Organization: Borland International
References: <9225302.22791@mulga.cs.mu.OZ.AU> <4945@holden.lulea.trab.se>
Date: Thu, 10 Sep 1992 16:29:06 GMT
Lines: 35

In article <4945@holden.lulea.trab.se> jbn@lulea.trab.se (Johan Bengtsson) writes:
>
>If sizeof(long) >= sizeof(void*), then you should be able to
>test like this:
>
>assert( (long)p1 != (long)p2 );
>
>since a pointer stored in an integer variable must keep all information
>needed to restore the pointer (if the integer type is large enough).
>

	Not necessarily. The requirement is only that the original pointers can
be restored, not that their representations as longs be identical.  For 
example:

	0x0000:0x0010
	0x0001:0x0000

These two addresses refer to the same memory location.  Converting them to
longs in the most obvious way produces these two values:

	0x00000010
	0x00010000

The trick on segmented architectures is to always normalize the pointers before
comparing. The first pointer above isn't normalized, because it has an offset
greater than 15.  The second one is normalized. Comparisons between normalized
pointers are always valid.
	-- Pete