DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :cc:content-type;
        b=wdzwH+3gq5DltHWw3e7BWd6jCgxLDs2xq919QfbBlShILXIl7vLavr+B4/b6ofY8dr
         xx/jpFO8YaWNxDEp9EJDAcgxNwgS/TwgzLCvVO/7YuZKdcUcTvBtiXt1KfdTl5UahAWI
         G50sjL3GYMhNKfSOCFFe9gMTVJE7wtiT3JoWI=
MIME-Version: 1.0
In-Reply-To: <BANLkTikX-_apoMDNg7n9EDq4WSM7xLg_PA@mail.gmail.com>
References: <4DDB35D2 DOT 7030600 AT iki DOT fi>
	<BANLkTikX-_apoMDNg7n9EDq4WSM7xLg_PA AT mail DOT gmail DOT com>
Date: Sat, 28 May 2011 16:17:09 -0500
Message-ID: <BANLkTikTAzR9844GVT16jfXeDdwpAwSMyw@mail.gmail.com>
Subject: Re: GCC-4.6.0
From: Rugxulo <rugxulo AT gmail DOT com>
To: djgpp-workers AT delorie DOT com
Cc: Gautier de Montmollin <esgm AT bluewin DOT ch>
Content-Type: text/plain; charset=ISO-8859-1
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
Precedence: bulk

Hi again,    :-(

On Fri, May 27, 2011 at 9:25 PM, Rugxulo <rugxulo AT gmail DOT com> wrote:
>
> On Mon, May 23, 2011 at 11:36 PM, Andris Pavenis <andris DOT pavenis AT iki DOT fi> wrote:
>>
>> The first test build of gcc-4.6.0 for DJGPP is available at
>>
>> http://ap1.pp.fi/djgpp/gcc/4.6.0/
>
> I'll have to test this with Gautier's 3D engine (Ada), esp. since
> 4.5.2 (IIRC) didn't compile it correctly anymore. I mean, I blindly
> figure 4.6.0 won't have the same bugs or regressions (hopefully).
> I guess I could always try 4.5.3 also if necessary.

4.5.3 (like previous versions) does compile it, but there must be some
bug in Zip_Ada because it crashes (again, like older versions) in
Zip.Load before anything actually runs. I'll have to cc Gautier and/or
test newer Zip_Ada file routines to see if that would fix it.

Worse, 4.6.0 doesn't even halfway compile it at all, and I'm not sure
if it's due to a broken gnatmake (probably, it whines saying it can't
handle *.gpr or esp. whines not finding in path searching for *.gpr
files ... seems ';' is missing as path separator) and/or missing
system files. (Can't find system.ads though it's there. A quick diff
of ada453b.mft and ada460b.mft does show some changes but only like
two or three files that are missing [eh?] in 4.6.0. Either way, it
could just be the new way upstream organizes things now, I have no
idea.)

Apparently there is (was?) a Linux version fork by Duncan Sands, but
it's not integrated, so I can't even pretend to try compiling that
with latest Ada (which isn't installed here anyways, another thing
I've have to do, heh). Perhaps Gautier can share those files. At least
then we'd know if it's buggy there too or not (I'm using F14 w/ DOSEMU
on persistent 4 GB liveUSB stick).

N.B. This compiled and worked fine in 4.3.2, so it's either an
undiscovered bug (totally possible) or regression and not even then
necessarily a fault with the DJGPP port proper.
http://sites.google.com/site/rugxulo/eng3d018.zip?attredirects=0