X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-workers-bounces using -f
Date: Thu, 20 Dec 2001 01:27:46 -0500
From: AAganichev AT netscape DOT net (Alexander Aganichev)
To: djgpp-workers AT delorie DOT com
Subject: RE: Re: regcomp NLS fix
Message-ID: <2FDE018B.3AD4F419.09ACFA57@netscape.net>
X-Mailer: Atlas Mailer 1.0
Content-Type: text/plain; charset=iso-8859-1
Reply-To: djgpp-workers AT delorie DOT com

Hans-Bernhard Broeker <broeker AT physik DOT rwth-aachen DOT de> wrote:
>> I have redone the patch (regex.nls+warning.diff) by casting all
>> ctype.h function's arguments to unsigned char. Also I have fix the
>> warnings caused by the missed brackets and unused values calculated.
>> There's 4 more warnings in pedantic mode about possible use of
>> uninitialized variables, but they seems to be safe to ignore.
>Most of those changes are OK, but of one, I'm quite sure we don't want
>to do that:
>@@ -865,11 +865,11 @@
> static void
> ordinary(p, ch)
> register struct parse *p;
>-register int ch;
> {
>    register cat_t *cap = p->g->categories;

Sorry, I'm really won't do that.

>Compiling patched sources with the fully pedantic warning set (both the
>XGCC=$(XLGCC) and CFLAGS=-w overrides commented out in the regex
>makefile), I receive a lot more warnings than just those 4 "might be used
>uninitialized" ones from my old GCC-2.8.1 Alpha-to-DJGPP cross compiler
>here:
>
>   2 cast discards `const' from pointer target type
>   4  might be used uninitialized in this function
> 132 initialization discards `const' from pointer target type

Hmm... I've tried to add -pedantic to the makefile explicitly, so there's probably a problem in my gcc 2.95.2 with warnings :-(

>I extended the fix work based on yours to fix all those const-ness issues
>(they tend to avalanche quite a bit). I managed to get away with only one
>call to DJ's unconst() macro (in regcomp(), where it assigns the 'next'
>pointer), and an otherwise clean build in fully pedantic mode.  I think
>it's a useful achievement that this part now builds with full warning
>levels active, too.

Yesterday, I found that OpenBSD guys did more changes:
 - they fixed the regexp code for the 64-bit platforms,
 - they added more accurate handling of malloc and realloc return value.

Is it OK to peek these changes intact (with BSD copyright) or should we made something alike by ourselves?
-- 
alexander aganichev
url: http://aaganichev.narod.ru



__________________________________________________________________
Your favorite stores, helpful shopping tools and great gift ideas. Experience the convenience of buying online with Shop AT Netscape! http://shopnow.netscape.com/

Get your own FREE, personal Netscape Mail account today at http://webmail.netscape.com/