Path: utzoo!mnetor!uunet!husc6!hao!oddjob!gargoyle!ihnp4!homxb!mhuxt!m10ux!mnc
From: mnc@m10ux.UUCP (Michael Condict)
Newsgroups: comp.lang.c
Subject: Re: (So-Called) ANSI C
Message-ID: <453@m10ux.UUCP>
Date: 22 Dec 87 21:17:33 GMT
References: <4668@pyr.gatech.EDU> <3873@uw-june.UUCP> <6896@brl-smoke.ARPA>
Organization: AT&T Bell Labs, Murray Hill
Lines: 43
Summary: reason for noalias not a pragma

In article <6896@brl-smoke.ARPA>, gwyn@brl-smoke.ARPA (Doug Gwyn ) writes:
> In article <3873@uw-june.UUCP> pardo@uw-june.UUCP (David Keppel) writes:
> >This is why I think pragmas are a better mechanism for doing things like
> >``noalias'', etc.  I'm now a little bothered that the debate over whether
> >to have ``noalias'' rages on yet nobody has talked about leaving the
> >language alone and using pragmas to get the optimizations...
> 
> There are (at least) two reasons for this.
> 	(1) X3J11 proposed "noalias".  They did not propose a specific
> 	pragma for this.
> 	(2) A #pragma cannot, according to the rules, alter program
> 	semantics.  Therefore it cannot be used to obtain the kind of
> 	optimization that "noalias" permits, because that IS
> 	semantically visible.

With all due respect, I think Doug is off-base one this one.
Addressing his points in turn:

	(1) This is begging the question, which might just as well have been
	stated: Why did X3J11 propose a new keyword instead of a pragma?

	(2) This is not an unavoidable problem with a "noalias" pragma.  All
	that needs to be done is to say that the noalias pragma may only
	appear at those places in the program where it is true (i.e., the
	variable truly has only one handle).  Then it does not alter the
	meaning of any correct C program.  (Of course, one would not
	require that the compiler or run-time system detect this error.)
	Surely it is considered okay to have pragmas whose use is
	restricted in this or other ways?  Or must every pragma be allowed
	to occur everywhere?  If so, why?

I strongly agree with Dennis Ritchie and others that adding keywords for the
sole purpose of optimization is a very big mistake, especially at this late
stage in the standardization process.  Isn't the committee aware that each new
reserved word may invalidate a substantial number of existing C programs that
previously used that word as a variable?  I though one of the prime directives
was to cause the maximum possible number of existing (correct) C programs to
conform to the standard.  This goal REQUIRES using a pragma instead of a new
reserved word wherever possible.
-- 
Michael Condict		{ihnp4|vax135|cuae2}!m10ux!mnc
AT&T Bell Labs		(201)582-5911    MH 3B-416
Murray Hill, NJ