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 autolearn=ham autolearn_force=no version=3.4.4 X-Google-Thread: 103376,d581a4c04b0d7daf X-Google-Attributes: gid103376,public X-Google-Language: ENGLISH,ASCII-7-bit Newsgroups: comp.lang.ada Subject: Re: SAL, Auto_Text_IO release References: <1150721215.234108.92870@g10g2000cwb.googlegroups.com> From: M E Leypold Date: 19 Jun 2006 15:11:10 +0200 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii User-Agent: Some cool user agent (SCUG) NNTP-Posting-Host: 88.72.235.113 X-Trace: news.arcor-ip.de 1150722328 88.72.235.113 (19 Jun 2006 15:05:28 +0200) X-Complaints-To: abuse@arcor-ip.de Path: g2news2.google.com!news2.google.com!news.germany.com!news.unit0.net!newsfeed.arcor-ip.de!news.arcor-ip.de!not-for-mail Xref: g2news2.google.com comp.lang.ada:4832 Date: 2006-06-19T15:11:10+02:00 List-Id: "Ludovic Brenta" writes: > M E Leypold declaimed in a spectral voice: > > Some of us are still stuck with 3.15p. We can't GPL our customers code > > and being so small we can't pay the ACT support. So we use 3.15p. We > > cannot use Ada 2005. We are stuck. "This will take longer" we say to > > our customers. "We don't have 'not null'". "This will not look like > > OO, this must be Ada 95". And our projects are riddled with old bugs > > of 3.15p and ICEs and everything takes a million years and all our > > programs crash. We all die. Oh the embarrasment. > > But the upgrade path from GNAT 3.15p is quite clear: if you need the > GMGPL but can't or won't pay for GNAT Pro, then migrate to GCC 4.1, > then 4.2 when it is released. Debian implements that strategy: Sarge > since June 2005 has GNAT 3.15p with several bug fixes; Etch in December > 2006 will have GCC 4.1 (already has it, in fact, in testing). (This release cycles will be my death one day: Another release after 1.5 years? Hell, I worked happily with Turbo Pascal 5+6 for years (or it only seemed so, I'm not sure any more :-)). Well, I know that. I'v only been asking to keep the old versions. Two points to think about: - When we started the project there was no GCC 4.1 and GCC 3.x Ada support was rumored to be bad. So we decided to stick with 3.15p at that time. Call me conservative. I prefer known bugs and weaknesses to new ones. - Project is cross-platform: 3.15p came readily packaged and self contained. It will take some time to shift the Windows build machine to gcc 4.1 (if that is even possible yet). That will have to use MINGW (I think?) so this is a larger change from what we had, since that needs to coexists with cygwin. - I'm somewhat disinclined to change horses in midrace as it is. So migration to gcc 4.1 (for all builds) will be something which has do be done of a course of months if not years. If a build fails with gccada we'll have to build under 3.15p anyway so retaining 3.15p as an insurance (also against wild license changes, you understand ...) is a matter of policy. But thanks for your explanation anyway. > If you need all of AdaCore's libraries in GMGPL binary packages, that's > one more reason to use Debian or, alternatively, one of GNU Ada's > packages. Or roll your own binary packages. Exactly. Why do you think I _am_ using Debian? :-) I'd like to thank you and the Debian team here for the really solid piece of work you're doing. BTW: Is there a bug tracking for gnat 3.15p somewhere? I've found various bugs in the runtime and have not reproduced them with newer gnats but suppose that they are still here (at least some of them ...). Regards -- Markus