Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!snorkelwacker!husc6!m2c!wpi!jhallen
From: jhallen@wpi.wpi.edu (Joseph H Allen)
Newsgroups: comp.sys.ibm.pc.programmer
Subject: Re: Best C Compiler
Keywords: Compilers
Message-ID: <9954@wpi.wpi.edu>
Date: 21 Mar 90 06:21:09 GMT
References: <1722@geocub.greco-prog.fr> <1264@doitcr.doit.sub.org> <1990Mar19.175316.16898@Octopus.COM> <9873@wpi.wpi.edu>
Reply-To: jhallen@wpi.wpi.edu (Joseph H Allen)
Organization: Worcester Polytechnic Institute, Worcester ,MA
Lines: 30

In article <9873@wpi.wpi.edu> jhallen@wpi.wpi.edu (Joseph H Allen) writes:
>The only wierdness is the
>way it handles postincrement (and this weirdness is not incorrect.. it's just
>weird and slightly slower).

Oops... my apologies to zortech.  I was quite mistaken about what other
compilers do and my understanding of C.  In particular, this program:

main()
{
int a=0;
int *b= &a;
printf("%d\n", (*b) + (a++) );
}

Produces the result '1' on all compilers I've tried it on except GNU C, where
it produces the result '0'.

Most compilers assume 'a++' means "move a to tmp, increment a, use tmp for
rest of expression"

I (and I guess RMS too) thought it meant "use a for expression.  When
expresion is finished, increment a"   I think this definition is slightly
faster since with it one less register is free to use for other things in an
expression.  But K&R doesn't say which is to be used so I guess they're both
right.

-- 
            "Come on Duke, lets do those crimes" - Debbie
"Yeah... Yeah, lets go get sushi... and not pay" - Duke