Re: Octave 10 transition

[Rafael Laboissière <rafael@debian.org>]

* Sébastien Villemot <sebastien@debian.org> [2025-09-19 09:37]:

> Le jeudi 18 septembre 2025 à 19:52 +0200, Rafael Laboissière a écrit :
> > * Rafael Laboissière <rafael@debian.org> [2025-09-18 08:44]:
> >
> > > * Sébastien Villemot <sebastien@debian.org> [2025-09-17 10:38]:
> > >
> > > > Any chance that you have a look at the swig issue (#1114586), since 
> > > > you are listed as uploader of the package?
> > >
> > > The issue in Bug#1114586 can indeed be fixed by either upgrading the 
> > > swig package to the new upstream version 4.3.1 or by applying the 
> > > patch that you indicated [*]. However, the plplot package would still 
> > > FTBFS on unstable, because the gnat stack is not yet updated to gcc 15 
> > > (see Bug#1097622).
> > >
> > > I think we could ignore the plplot package for the Octave 10 transition.
> >
> > I could find a solution for the gcc-15-releated issue with plplot. I just 
> > uploaded version 4.3.0-2 of the swig package, including the patch 
> > indicated by Sébatien. I will upload a new version of the plplot package 
> > which buidl-depends on swig >= 4.3.0-2.
>
> Thanks for fixing the bug in swig.
>
> I don’t think you need to tighten the build dependency on swig in 
> plplot, because fundamentally plplot can build with older versions on 
> swig. If I were to add a new version constraint, I would maybe add a 
> breaks relationship in octave, saying that it breaks swig << 4.3.0-2.

Your analysis sounds correct. Thanks for sharing your thoughts! However, 
it is generally not true that plplot can be built with older versions of 
swig. In a system where swig 4.3.0-1 and octave 10 are present, plplot 
will FTBFS. This is why I tightened the build dependency.

In any case, I agree that the optimal solution would be for octave to 
break swig 4.3.0-1. Could you please add that? Once you have done so, I 
will remove the versioned dependency on swig from the plplot 
Build-Depends list.

Best,

Rafael