Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!njin!princeton!njsmu!mccc!pjh
From: pjh@mccc.UUCP (Pete Holsberg)
Newsgroups: comp.lang.c
Subject: Re: Another silly question
Message-ID: <759@mccc.UUCP>
Date: 22 May 89 21:36:24 GMT
References: <17812@cup.portal.com> <607@kl-cs.UUCP> <749@mccc.UUCP> <756@mccc.UUCP> <17657@mimsy.UUCP>
Reply-To: pjh@mccc.UUCP (Pete Holsberg)
Organization: The College On The Other Side of U. S. Route 1
Lines: 34

In article <17657@mimsy.UUCP> chris@mimsy.UUCP (Chris Torek) writes:
=>>In article <749@mccc.UUCP> pjh@mccc.UUCP (Pete Holsberg) asked
=>>>why compilers produce different code for "a[i]" and "*(a+i)"?
=
=>In article <17635@mimsy.UUCP> I noted that
=>>I have never observed one to do so.  There is no reason for a compiler
=>>to generate different code, as the expressions are semantically identical.
=
=I have none of these machines, and only gcc as a compiler.  The code
=produce by
=
=	GNU C version 1.35 (vax) compiled by GNU C version 1.35.
=
=for both loops in
=
=	int a[20];
=	main(){int i;
=	for(i=0;i<20;i++)a[i]=0;
=	f();
=	for(i=0;i<20;i++)*(a+i)=0;
=	f();
=	}
=
=was identical.  

=Differing code sequences is one of two obvious possibilities, the other
=being differing multi-user loads.  The latter seems less likely, especially
=if the results are repeatable.  Why not compile to assembly and compare?

Thanks, Chris.  I will try that on the 386 machine, as that assembly language
is not as much an unknown as those for the 680x0 and the WE320x0!
-- 
Pete Holsberg, Mercer County Community College, Trenton, NJ 08690  
{backbone}!rutgers!njin!princeton!njsmu!mccc!pjh