From owner-freebsd-hackers  Sun Feb 23 21:50:44 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id VAA29724
          for hackers-outgoing; Sun, 23 Feb 1997 21:50:44 -0800 (PST)
Received: from dg-rtp.dg.com (dg-rtp.rtp.dg.com [128.222.1.2])
          by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id VAA29715
          for <FreeBSD-hackers@freebsd.org>; Sun, 23 Feb 1997 21:50:38 -0800 (PST)
Received: by dg-rtp.dg.com (5.4R3.10/dg-rtp-v02)
	id AA07910; Mon, 24 Feb 1997 00:50:06 -0500
Received: from ponds by dg-rtp.dg.com.rtp.dg.com; Mon, 24 Feb 1997 00:50 EST
Received: from lakes.water.net (lakes [10.0.0.3]) by ponds.water.net (8.8.3/8.7.3) with ESMTP id TAA08441; Sun, 23 Feb 1997 19:56:49 -0500 (EST)
Received: (from rivers@localhost) by lakes.water.net (8.8.3/8.6.9) id UAA11182; Sun, 23 Feb 1997 20:01:57 -0500 (EST)
Date: Sun, 23 Feb 1997 20:01:57 -0500 (EST)
From: Thomas David Rivers <ponds!rivers@dg-rtp.dg.com>
Message-Id: <199702240101.UAA11182@lakes.water.net>
To: ponds!lakes.water.net!phobos.illtel.denver.co.us, ponds!lambert.org!terry
Subject: Re: hmm
Cc: ponds!freebsd.org!audit-bin, ponds!best.net!dillon,
        ponds!freebsd.org!FreeBSD-hackers, ponds!gvr.win.tue.nl!guido,
        ponds!resnet.uoregon.edu!gurney_j, ponds!sonic.cris.net!top
Content-Type: text
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> 
> On Sat, 22 Feb 1997, Terry Lambert wrote:
> 
> > > with p=""  
> > > *p != '\0' && p[strlen(p) - 1] == '[' _still_ will read a byte p[-1].
> > 
> > What?
> > 
> > 	IF p = ""
> > 	THEN *p = '\0'
> > 
> > by definition of ""...
> 
> but p[strlen(p) - 1] will be p[-1]
> Optimization done by compiler may skip it, but depending on that will be
> rather dangerous.
> 
> --
> Alex
> 
> 

 It may not be too clear... but, and at the risk of repeating what
other's may have indicated - the C language, by definition, evaluates
this from left to right.  And, in the situation above (a logical-and)
the first 0-valued expression will cause the result of the logical-and
to be 0.  Furthermore, the a compiler, or optimizer, is forbidden
from evaluating the rest of the expression.

 That is, if *p  is equal to  '\0' - the remainder of the logical-and
will not be evaluated... if an optimizer does cause this to happen,
the compiler/optimizer is in error.

 So, if strlen(p) was 0, it wouldn't matter; p[-1] would never be
referenced.

 In this veign you'll also often see code like:
	
	if(p != NULL && !strcmp(p,"some value") ) 

 the left side of logical-and guarantees that the pointer 'p' is not
NULL. The right side of the logical-and can then safely dereference 'p'.

 Again; this isn't a coincedence, C is defined (from the early K&R 
days) to behave in this manner.  In compiler terms it's called
short-circuit evaluation of logical expressions...

	- Dave Rivers -