From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM autolearn=ham autolearn_force=no version=3.4.4 X-Google-Thread: 103376,2c7b0b777188b7c4 X-Google-Attributes: gid103376,public X-Google-Language: ENGLISH,ASCII Path: g2news1.google.com!news1.google.com!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail From: "Alex R. Mosteo" Newsgroups: comp.lang.ada Subject: Re: GNAT GPL Edition Maintenance and Upgrades Date: Wed, 05 Oct 2005 16:06:30 +0200 Message-ID: <4343DDE6.2020506@mailinator.com> References: <1128499462.850353.146890@g44g2000cwa.googlegroups.com> <9070id.mp6.ln@hunter.axlog.fr> <1128510619.707554.152420@f14g2000cwb.googlegroups.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-Trace: individual.net v3PAaNs5KegFJ5bRM+oT/gi/eE0fFYguIxG8hVFvviDB8XbVo= User-Agent: Mozilla Thunderbird 1.0.6 (X11/20050912) X-Accept-Language: en-us, en In-Reply-To: <1128510619.707554.152420@f14g2000cwb.googlegroups.com> Xref: g2news1.google.com comp.lang.ada:5404 Date: 2005-10-05T16:06:30+02:00 List-Id: Ludovic Brenta wrote: > Jean-Pierre Rosen a �crit : > > >> Ludovic Brenta a �crit : >> >>> [1] Quite to the contrary; the more I hear the "selfish >>> argument", the more I feel inclined to package GNAT GPL Edition, >>> just to teach selfish people that if they refuse to give, then >>> they cannot take, as Georg Bauhaus said so nicely. >>> >> >> Not a formal vote, but... >> >> I think that this argument goes *against* the spirit of free >> software. As far as I can understand, the basis of free software is >> "you can do anything with this software, *except* deny to others >> the rights you have received". Anything, including proprietary >> software. > > > Precisely. With the GMGPL, you receive the right to see and modify > the source code of the GNAT run-time; but you can also deny others > this same right. The GPL is more "free" than the GMGPL, since with > it you cannot deny others this right anymore. This is the "free" > spirit as defined by Stallman and the FSF. The BSD license does > allow you to deny rights to others, and has a different definition of > "free". > > >> The GPL edition is a big mistake made by AdaCore, both from a >> marketing and a popularity point of view. If the community clearly >> refuses this edition (and not having it in Debian is quite a strong >> indication), maybe AdaCore will revise its policy. > > > Yes, this is a possibility. > > >> Moreover, I find it quite logical for Debian to just use the >> standard GCC tree. > > > Well, latest /= greatest, that's why I kept gnat 3.15p in Debian for > so long instead of moving to GCC. Also, GCC is not "standard" by any > measure (Ada is not a release criterion for GCC). Indeed I think this is an important point. I don't use myself gcc-gnat but follow the gcc developers list for Ada related news. My (admitely perhaps innacurate) impression is that gcc-ada is in a state of great flow and sometimes brokeness that make it not that desirable as a free compiler at the moment. (it's known that one of the 3.3/3.4 versions was strongly not recommended --perhaps directly broken--, but I'm not sure which one of the two, and this was (IIRC) as a result of Ada not being a release criterion for gcc). > In contrast, GNAT GPL is "standard" by two measures: it has been > blessed by AdaCore, and is known to build ASIS, GLADE, GPS etc. > correctly. And so, as a hobbyist GPL developer, I will cast my late vote for the GPL version ;) followed by the latest gcc one. We must consider, and this I think is related with the vote to "wait" used by something else, that these gnat versions at the moment are changing to include the 2005 features. By experience I can say that if you use them [the features], you're going to see a lot more ICEs than when doing regular Ada 95 programming. That is, there must still pass quite some time until gnat become "stable" and the only changes are ironing out of rare bugs. (Is even the 0Y standard closed for that matter?) That's a reason not to include a version who's not actively receiving the changes (gcc.3.4? gcc.4.0?) if we're interested in a 0Y compiler instead of a 95 one. In any case, I'm afraid that any of these versions may require a lot of work for maintenance and patching. Ideally, I think we should aim to have the GPL edition and a GMGPL one, but I understand that it is a) more work and b) still more work to maintain the gcc one who's under heavy changing. Indeed I'm worried that the most work-effective stance, apart from the GPL edition who has all the toolkit ready, is to wait to see if a better replacement for 3.15p (and maybe the GPL) arises in the gcc branch in the future. In short, if I were Ludovic and were faced with a maintenance task, I'd leave 3.15p as it is, as the Ada95 choice; the GPL version as experimental 0Y with full toolkits, and would aim to replace it in the future with the GMGPL gcc one when it starts to settle and the 0Y feature set is complete. >> The whole issue is about packaging, not the compiler. I even think >> that it would be OK to take the GPL edition, replace the offending >> packages from the version in the GCC tree, and release the whole >> stuff under GMGPL. But then, someone has to do the packaging... > > > I contemplated this idea, but when I saw the size of the diff, I > backed out. I did "diff -I^-- gcc/gcc/ada gnat-gpl-2005-src/src/ada" > (note: ignoring comments and therefore the change of license) and > found: > > GCC 3.4.4 to GPL: 16.0 megabytes GCC 4.0.1 to GPL: 11.0 megabytes GCC > HEAD to GPL: 7.1 megabytes > > Even with 4.1, the difference is huge. And note that this is only > the Ada part of GCC. >