The default Ada compiler for Debian 8 "Jessie"
Hello.
GCC 4.9.0 should be released in a few months from now, possibly in March
2014. It is already available in experimental and Matthias Klose is
actively working on it. Apparently[1] he intends to make GCC 4.9 the
default compiler for C, C++ and other languages as soon as it reaches
unstable, on as many architectures as possible.
The maintainer of Ada in FreeBSD and Dragonlace has stated at FOSDEM[2]
that he intends to skip GCC 4.8 altogether for Ada and package GCC 4.9
instead (but note that the default C and C++ compiler on FreeBSD is now
clang/LLVM, not GCC).
Debian 8 "Jessie" will be frozen on November 5, 2014 [3], which leaves
us 9 months to transition all Ada packages to the next default Ada
compiler.
gnat-4.8 has been in Jessie (testing) since November 2013 but gnat-4.9
does not exist at all yet.
We are faced with a tough choice for the next default Ada compiler. If
we choose gnat-4.8, then the transition of all packages can start
immediately but Jessie ends up with an "old" compiler (4.8.0: March
2013) which is not the default for other languages and which is
different from the one in FreeBSD. If we choose gnat-4.9, this will
allow better support for Ada 2012 (e.g. contracts and other aspects) and
probably a more recent version of PolyORB too.
I have just created the branch org.debian.gnat-4.9 in monotone and I
propose the following plan:
- starting right now, everyone interested (and in particular the
maintainers I talked to at FOSDEM: you know who you are!) works hard
on updating all the Debian patches for gnat-4.9; this is the top
priority.
- at the end of March 2013 (two months from now), we review the state of
gnat-4.9: is upstream GCC 4.9.0 released? Is it in unstable? Are we
satisfied with the quality and stability of gnat-4.9? and we make the
final decision as for the Ada compiler for Jessie.
- Immediately after this decision is made, we update all other packages
to the chosen new compiler, starting with ASIS and PolyORB.
The obvious risk with this plan is that, if gnat-4.9 turns out not to be
viable, we'll have wasted two precious months for the big transition.
Objections? Commitments? Exuberant enthusiasm? Lukewarm support?
Please tell me...
[1] https://lists.debian.org/debian-gcc/2013/12/msg00034.html
[2] http://people.cs.kuleuven.be/~dirk.craeynest/ada-belgium/events/14/140201-fosdem/10-ada-bsd.pdf
[3] https://lists.debian.org/debian-devel-announce/2013/10/msg00004.html
--
Ludovic Brenta.
Reply to: