Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uunet!munnari.oz.au!bruce!cechew
From: cechew@bruce.OZ (Earl Chew)
Newsgroups: comp.os.minix
Subject: Re: New wait.h
Message-ID: <1688@bruce.OZ>
Date: 13 Nov 89 01:50:57 GMT
References: <4507@ast.cs.vu.nl>
Organization: Monash Uni. Computer Science, Australia
Lines: 28

From article <4507@ast.cs.vu.nl>, by ast@cs.vu.nl (Andy Tanenbaum):
> In article <1684@bruce.OZ> cechew@bruce.OZ (Earl Chew) writes:
>>#define WIFSIGNALED(__s)	((unsigned int) ((__s) - 1 & 0xffff) < 0xff)
> 
> You may be right, but I find mine an awful lot clearer:
> 
> #define WIFSIGNALED(--s)        (_HIGH(__s) == 0 && __s != 0)
> 
> Does anyone know if mine is ok?

No it's not ok :-)

#define WIFSIGNALED(__s)        (_HIGH(__s) == 0 && (__s) != 0)

I think that the standard does not impose any rules about macros (it does about
functions implemented as macros) -- so you are ok.

However, I feel that where there is an alternative, you are better off coding
safe macros so that you won't get a surprise some time later. The point about
clarity is valid but I think that safety is more important. Furthermore, the
code is hidden within a macro and there's only 1 short line --- clarity is not
a big issue in this case.

Earl
-- 
Earl Chew, Dept of Computer Science, Monash University, Australia 3168
ARPA: cechew%bruce.cs.monash.oz.au@uunet.uu.net  ACS : cechew@bruce.oz
----------------------------------------------------------------------