Path: utzoo!utgpu!water!watmath!clyde!cbosgd!ihnp4!ihlpf!nevin1
From: nevin1@ihlpf.ATT.COM (00704A-Liber)
Newsgroups: comp.lang.c
Subject: Re: noalias and vectors
Keywords: noalias, auto-vectorizing
Message-ID: <3458@ihlpf.ATT.COM>
Date: 21 Jan 88 01:43:50 GMT
References: <2942@hall.cray.com> <531@cresswell.quintus.UUCP> <7088@brl-smoke.ARPA> <3419@ihlpf.ATT.COM> <7142@brl-smoke.ARPA>
Reply-To: nevin1@ihlpf.UUCP (00704A-Liber,N.)
Organization: AT&T Bell Laboratories - Naperville, Illinois
Lines: 21

In article <7142@brl-smoke.ARPA> gwyn@brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>) writes:
>In article <3419@ihlpf.ATT.COM> nevin1@ihlpf.UUCP (00704A-Liber,N.) writes:
>-Actually, from what I understand, all that noalias would mean here is that s1
>-and s2 don't point to EXACTLY the same place ... Overlap would still
>-be allowed.
>
>Not the way I understand it.

Are you telling me that noaliasing a pointer to char means that I cannot have
overlap anywhere past the character that I am pointing to?  Do you think
that noalias believes that all uses of pointers to char are for null-terminated
strings (which is not the only way to define strings)?  Since string operations
are made by function calls, they are technically NOT part of the semantics of
the C language, and noalias cannot determine what is meant by overlap (except
for EXACTLY pointing to the same place).  I simply cannot believe that
noalias would not allow pointing to two different elements of the same array.
-- 
 _ __			NEVIN J. LIBER	..!ihnp4!ihlpf!nevin1	(312) 510-6194
' )  )				"The secret compartment of my ring I fill
 /  / _ , __o  ____		 with an Underdog super-energy pill."
/  (_</_\/ <__/ / <_	These are solely MY opinions, not AT&T's, blah blah blah