Debian's Ada metapackage gnat depends on gcc-14,
whereas the current gcc version is 16.

Impacted packages;

   adasockets
   gcc-doc-defaults
   libgmpada
   music123  (rebuild only)
   topal     (rebuild only)


===========================================================
Ben file:

title = "gnat-16";
is_affected = .depends ~ /libgnat/ & !.source ~ /gcc/;
is_good = .depends ~ /libgnat-16/;
is_bad = .depends ~ /libgnat-\d+/ & !.depends ~ /libgnat-16/;

===========================================================

Packages not in Testing depending on gnat-14

ada-bar-codes
adacgi
ahven
anet
dbusada
ghdl
gprbuild
libflorist
libgnatcoll
libgnatcoll-bindings
libgtkada
libtemplates-parser
libxmlada
libxmlezout

Packages not in Testing depending on unversioned gnat
alire
dh-ada-library


I understand there is an issue with 32bit architectures,
that is very unlikely to be resolved any time soon.
(Maybe never, as there has been no progress since last year.)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114065
https://bugs.debian.org/1121759
https://bugs.debian.org/1121232

To avoid users installing a broken package,
I'm thinking of adding a preinst to gnat,
to prevent installing on 32bit arches.
(attached)

Control: tags -1 moreinfo

No, this is wrong. If gnat-16 is broken on 32 bit architectures and
unfixable, then the package and its reverse dependencies need to be
removed from the 32 bit architectures first. Please remove the moreinfo
tag once this is resolved -- either by gnat-16 being fixed or the binary
packages removed.

Cheers

Some comments:

  - Please could you coordinate/announce your work on the
    debian-ada ML (CCed here)?

  - Usually all depending packages need a new upstream version
    for a major new gnat version. At least for all past gnat
    transitions these were prepared in experimental, and then
    brought to unstable together

  - Please ignore the comment of that rather unfriendly release
    team member.  According to Nicolas this is work in progress,
    and should be addressed for gnat 16.

  - There are also some disabled patches in the gcc-16 package,
    which would need a review and probably an update.
    See #1121232

  - If you run into #1136069, then that should be addressed as well.

Thanks for working on this

Matthias