Path: utzoo!mnetor!uunet!pwcmrd!skipnyc!atpal!tneff
From: tneff@atpal.UUCP (Tom Neff)
Newsgroups: comp.lang.c
Subject: Re: noalias comments to X3J11
Message-ID: <118@atpal.UUCP>
Date: 28 Mar 88 15:23:15 GMT
References: <12578@brl-adm.ARPA> <1988Mar25.172355.348@utzoo.uucp>
Reply-To: tneff@atpal.UUCP (Tom Neff)
Organization: Rational Technologies, Inc.
Lines: 26
Summary: /volatile/ isn't a frill, interrupt handlers aren't esoteric

In article <1988Mar25.172355.348@utzoo.uucp> henry@utzoo.uucp (Henry Spencer) writes:
>Interrupt routines are almost by definition esoteric, not to mention highly
>machine-specific.  Only on PCs do users commonly write their own interrupt
>routines...

It would be fairer to say "only on microcomputers" rather than "only on
PCs."  I write real-time systems for a living, and so do a lot of other
folks I know.  We do write interrupt routines, and plenty of 'em.  "C" is
a strong contender among languages for this work, because of its portability
and lack of overhead.  The basic concept of /volatile/ is very important to
me, and I'm glad XJ311 recognizes it.  Many's the time I have had to surround
a few statements with forests of labels or calls to dummy() just to get one
compiler or another to stop optimizing access to a location!

I will admit, though, that /vol/ is a grown-up keyword, which plenty of
programmers will have little use for.  But it needn't intrude in your lives
at all.  Those who need it, use it.  (Fortunately it's not too much of a pain
to implement -- effectively it just flags optimization off for certain
variables).  It does sound like /noalias/ is making a much noisier and messier
splash on its arrival in the standard.  I am not too keen on forcing a strange
new keyword into half the declarations, for instance, in dear old stdio.h. TMN


-- 

Tom Neff