Subject: Re: archive.c/arm32
To: None <current-users@NetBSD.ORG>
From: Greg A. Woods <woods@kuma.web.net>
List: current-users
Date: 03/24/1998 11:53:39
[ On Tue, March 24, 1998 at 14:48:31 (+0100), Wolfgang Solfrank wrote: ]
> Subject: Re: archive.c/arm32
>
> Yes, in a word, this GCC warning features (despite the fact that they might
> uncover some bugs) _is_ _brain damaged_.

Ditto.  Me too!  Amen!

I, for one, would be much happier if the NetBSD code were forced through
lint (and of course that lint were improved as much as possible) than
through GCC's -Wall, etc..  GCC with -Wall, plus all the warnings "all"
doesn't include, still doesn't catch some nasty things that SunOS-4's cc
does catch with no options!  People get a completely invalid sense of
security from it and sometimes don't realize they're getting bitten by
it until much later (i.e. when it bites someone else!).  The fact that
people are also forced to jump through stupid and often inefficient
hoops to avoid warnings is just about as bad.

Careful construction of makefile rules should even allow the use of lint
in places where it would do the most good, and eliminate the extra
overhead in places where it simply slows things down (I'd guess that
even gcc runs a wee bit faster without any -W's).  [What I really want
is a compiler that just compiles and a totally separate programming tool
for catching *all* of my errors (though if both generate code they
should use common code to do so, just as the original Unix lint does.]

-- 
							Greg A. Woods

+1 416 443-1734			VE3TCP			robohack!woods
Planix, Inc. <woods@planix.com>; Secrets Of The Weird <woods@weird.com>