Bug#1149705: Bug#1146496: dh-fortran: follow-up regression in get_fc_default() when FC_DEFAULT=gfortran (see #1149705)

[Viktor Pashaiev]

Package: dh-fortran
Followup-For: Bug #1146496

Control: block 1146496 by 1149705

Hi,

This is a follow-up to bug #1146496 and links with bug #1149705.

The fix for resolving 'gfortran' via its symlink in get_fc_flavor_arch() works, but get_fc_default() checks os.environ["FC_DEFAULT"] first and calls canon(), which still maps "gfortran" to hardcoded default_compilers["gfortran"] = "gfortran-16". Packages setting FC_DEFAULT=gfortran (like fortran-assert) still hit this issue.

Attached is an updated patch against dh-fortran 0.85 that:
1. Resolves symlinks in canon() and default_compilers["gfortran"] via _detect_default_compiler("gfortran", "gfortran-16").
2. Keeps the generic upstream fallback "gfortran-16" if unversioned gfortran is absent.
3. Adds a test in dhfortran/tests/compilers.py for FC_DEFAULT=gfortran.

Please consider applying this together with the fix for #1146496.

Thanks,
Viktor Pashaiev <w.paszajew at gmail.com>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dh-fortran_0.85ubuntu2.debdiff
Type: text/x-diff
Size: 4015 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/debian-science-maintainers/attachments/20261005/2eb9e611/attachment-0003.diff>