X-Authentication-Warning:	acaxp6.physik.rwth-aachen.de: broeker owned process doing -bs
Date:	Tue, 1 Jul 2003 14:38:06 +0200 (MET DST)
From:	Hans-Bernhard Broeker <broeker AT physik DOT rwth-aachen DOT de>
To:	Andris Pavenis <pavenis AT latnet DOT lv>
cc:	djgpp-workers AT delorie DOT com
Subject:	Re: conio.h bug?
In-Reply-To:	<200307011522.57814.pavenis@latnet.lv>
Message-ID:	<Pine.OSF.4.44.0307011428170.16922-100000@acaxp6.physik.rwth-aachen.de>
MIME-Version:	1.0
Reply-To:	djgpp-workers AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On Tue, 1 Jul 2003, Andris Pavenis wrote:

> No. I haven't done anything with it. Did tests under Linux with gcc-3.4
> 20030630 (experimental) and gcc-3.3.1 20030627 (prerelease) Results are
> below. So it's not a DJGPP only problem.

... except that DJGPP 2.03 would be probably be the only port of GCC to
have a version of gettext() in its libc that conflicts with the builtin
expectation of GCC.

> gettext appears in file gcc/builtin-attrs.def for CVS at least beginning
> with gcc-3.2 branch (haven't looked earlier).

Ouch.  What on earth did they think they'ld have to hardwire expectations
about a non-standard function like gettext() into GCC for?

This leaves us with two-and-a-half options, I think:

1) Patch the GCC port to DJGPP to remove this hardwired expectation (and
   try to get the GCC maintainers to accept that patch).

2) Add a patch for V2.03 conio.h to our GCC port's README that effectively
   brings it to the state of V2.04alpha (--> rename to _conio_gettext,
   and conditionally define gettext as a macro).

3) Take this issue as a reason to immediately release V2.04, and make all
   coming GCC ports require that.

I'll leave it to you to judge which of these only counts as half an option
;-)

-- 
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.

- Raw text -