Xref:	news2.mv.net comp.os.msdos.djgpp:4982
From:	broeker AT PROBLEM_WITH_INEWS_DOMAIN_FILE (Hans-Bernhard Broeker)
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: Is post-increment a pre-increment?
Date:	14 Jun 1996 11:46:28 GMT
Organization:	RWTH -Aachen / Rechnerbetrieb Informatik
Lines:	31
Message-ID:	<4prjek$8l7@news.rwth-aachen.de>
References:	<Pine DOT NEB DOT 3 DOT 93 DOT 960612212436 DOT 3438B-100000 AT ananke DOT amu DOT edu DOT pl>
NNTP-Posting-Host:	axpcl6.physik.rwth-aachen.de
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

Mark Habersack (grendel AT ananke DOT amu DOT edu DOT pl) wrote:
[...]
> len += CharWidth(*walker++, curFont); 

> Here, the walker variable has been everytime incremented TWICE, not
> ONCE as it was meant to be - what is wrong?

>  Later on I changed the offending line to
>    len += CharWidth(*walker, curFont);
>    walker++;

>  and everything worked just fine. Where lies the error - is it my
> ignorance or a bug or side effect?

This heavily looks like a case of macro parameters with
side-effects. If CharWidth is a macro that evaluated its first
parameter twice, this would be a perfect explanation of the behvaviour
you describe. Just imagine Charwidth were something like this:

#define CharWidth(char, font) ((is_alnum(char)? (font)[char].width : 0)

Now, if 'char' is '*walker++', it will get incremented twice if
is_alnum(char) is true. 

This is a well-known cause of problems with C, but can be hard to spot
sometimes. The fix is exactly the one you proposed (pass first, then
increment), and can be put as a rule like this:

  Don't use an expression with side effects as a macro parameter!

Hans-Bernhard Broeker (Aachen, Germany)

- Raw text -