[DRE-maint] Bug#1104697: ruby-dbf: FTBFS caused by failure/Error: expect(record.name.encode('UTF-8').unpack1('H4')).to eq 'd0b0'

[Leandro Cunha]

On Tue, May 20, 2025 at 2:16 PM Santiago Vila <sanvila at debian.org> wrote:
>
> El 20/5/25 a las 18:19, Leandro Cunha escribió:
> > The build may fail on any build that does not have unicode as UTF-8 in
> > its environment. I don't know if this would be a possible scenario
> > given our reality living in Brazil and speaking Portuguese (or English
> > or Spanish), as the tests fail, both reproducible builds and Salsa CI.
> > Since there is a new version pending and trixie has already advanced
> > to hard freeze, I think it is more interesting to wait to pass this
> > stage. This is according to what is stated in the log:
>
> Hello. I can build this package just fine using sbuild, so I agree this is not RC.
>
> But the builds which fails in Salsa CI are the normal builds (amd64 and i386),
> not the one by reprotest. Salsa CI is supposed to be as similar as possible to
> a "normal" build (made by sbuild), so I believe there is something
> subtle going on here (which I would like to know if somebody figures out).
>
> Thanks.

This could be resolved normally in GitLab CI with:
variables:
LANG: en_US.UTF-8
LC_ALL: en_US.UTF-8
And I'm checking in the Salsa CI if this would be possible, but I'm
giving priority to reproducible builds in which the test fails for
unstable and trixie. But I confess that this bug left me with doubts
about which would be more appropriate, serious or important. But due
to the automatic removal already scheduled, it was better to down to
important due to the failure in reprotest (FTBFS flag) that I'm giving
more emphasis to.

-- 
Cheers,
Leandro Cunha