Re: Bug#1054657: transition: r-bioc-biocgenerics
Hi Sebastian,
Am Tue, Nov 07, 2023 at 03:12:39PM +0100 schrieb Sebastian Ramacher:
> > Charles and I tried to explain in different ways: We do not have simple
> > means to answer this question.
>
> Picking a random r-bioc-* package:
> https://salsa.debian.org/r-pkg-team/r-bioc-aroma.light/-/blob/master/DESCRIPTION
> has an "Imports" field. Those are mapped to dependencies in the package.
> So I presume that when importing those packages into the packaging
> repository, changes to this field can be identified and checked. I would
> excpect this information to be enough to identify any currently missing
> packages.
Well, I'm one of the authors of dh-r. I know where to find the
dependencies since we are parsing this file. We even go further: When
trying to build an R package and find a missing dependency this will be
nearly ready packaged automatically. This is NOT the problem. The
point is that this workflow is completely automated. Parsing 170
DESCRIPTION files of not yet packaged versions is not automated and thus
quite some manual work. And I would really like to hear better reasons
why you want us to do this manual work in addition to something which is
done automatically.
> > But I had a different question; What
> > exactly is the problem of a transition taking about 1 month due to some
> > delay by waiting for packages in new?
> >
> > I somehow have the feeling that this transition is currently delayed by
> > some bug-mail / tagging ping-pong which is demotivating for both sides.
> > You make a request to some volunteers to do some extra work that was not
> > requested before and we volunteers explained that it is really hard
> > work. I think it is fair to ask for the reasons you want us to do some
> > work which is definitely hard to do and for us painful and unproductive.
>
> We should have requested this information for all transitions in the
> past. We did not and thus had the same problems for the last couple of
> transitions including missing packages and a significant number of
> autopkgtest regressions.
The autopkgtest regressions are not caused due to missing new packages.
They are mostly due to issues on some architectures. We can stop those
by restricting Bioconductor packages to those tested by upstream by
loosing the advantages Debian provides to the community. In fact
dealing with the regressions has taken us usually longer than the
missing packages. I'm not proud on it that it takes so long to deal
with the regressions but cutting from our time constraints even more
will not help at all.
> The r-bioc-* transition is special in the sense that it requires all
> involved packages to be ready to migrate at the same time. This is where
> delays become an issue. It essentially blocks all other transitions that
> could potentially overlap (e.g., auto-hdf5) from being started or
> progressing.
We can wait until auto-hdf5 transition is done.
> All of that binds resources on our side to track down the remaining bits
> and pieces to make everything migrate at the same time. This is usually
> not an issue with a typical shared library transition. Hence we are
> asking you to identify possible NEW packages that will be required to
> complete the transition.
You did not yet answered the question I asked twice whether we can find
a compromise by simply removing packages with missing new dependencies
from testing. I consider this a compromise which I would really love to
discuss honestly.
I try to repeat: Please trust me that finding those packages is not as
simple as picking a random DESCRIPTION. We need time for it, time were
other packages (RC bugs, autopkgtest regressions etc. are suffering).
I'm not yet convinced that we should do this extra work for a couple of
days win. If you insist on your position I will escalate the issue to
the technical committee.
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: