Bug#1080921: rhythmbox: Missing Build-Depends on python3-setuptools

[Alexandre Detiste]

Le mar. 21 janv. 2025 à 12:11, Simon McVittie <smcv at debian.org> a écrit :
>
> Control: severity -1 important
> Control: tags -1 + moreinfo unreproducible
>
> On Thu, 05 Sep 2024 at 16:58:05 +0200, stefanor at debian.org wrote:
> > This package failed build from source when test-built against a version of
> > dh-python without a python3-setuptools dependency.
>
> How can this be reproduced? Please could you share a concrete proposed
> version of dh-python, or a patch or merge request with the proposed change,
> or a failing build log?
>
> It would be helpful if changes like this, that are expected to cause some
> build failures, went via a version of dh-python in experimental that
> maintainers could easily test against.
>
> It would also be helpful for reports of build failures to be accompanied
> by a (link to a) build log, so that if the maintainer cannot reproduce the
> failure themselves, there is at least some information available.
>
> I tried building rhythmbox in a schroot against a locally-built version
> of dh-python with the attached change (is this what you meant is going
> to happen?) but it built successfully (for _amd64 + _all, separately,
> in unstable).
>
> > Please add a Build-Depends on python3-setuptools to your package, or migrate
> > the package's build system away from setuptools/distutils.
>
> I cannot find any references to setuptools or distutils in rhythmbox,
> so I think there is nothing to be migrated, and I think it would be
> wrong to add a Build-Depends on python3-setuptools.
>
> Are you sure that the build failure had anything to do with dh-python
> dropping its dependency on python3-setuptools? rhythmbox build-depends on
> libgirepository1.0-dev and meson, both of which pull in python3-setuptools
> (and already did that at the time this bug was opened, as far as I can
> see), so I don't see how dh-python dropping its equivalent dependency
> would make any difference?
>
> Looking at recent reproducible-builds results, rhythmbox's upstream test
> suite does not seem to be completely stable - is it possible that the
> build just failed its tests by bad luck, for reasons that are orthogonal
> to setuptools? A typical symptom seems to be that "test-rhythmdb" fails.
> I've reported a separate bug for that.
>
> Thanks,
>     smcv