the things I worked on.
retrying autopktest regressions
===============================
At the start of my shift, the autopkgtest queues were empty so I
retried all autopkgtest regressions (876 tests triggered). Once these
had completed and the results reflected in the update_excuses
report[1], I triggered migration-reference/0 tests (--no-proposed) for
those that still failed (739 tests triggered).
scikit-learn / matplotlib
=========================
scikit-learn was missing a build on amd64 due to a missing
build-dependency on python-matplotlib-doc. This new package was
introduced in a new version of matplotlib in Debian, but had not
auto-sync'd because of the Ubuntu delta. I checked whether a merge
was needed or whether matplotlib could be sync'd, and decided on the
latter.
various syncs
=============
I looked at the Merge-o-Matic report for packages in universe [2], and
sync'd several packages from Debian where our delta could be dropped:
mercurial, pytest, dm-writeboost, flowblade, hamster-time-tracker,
apt-cacher-ng, poetry,
virtnbdbackup and r-cran-glmmtmb
I checked whether the delta for abyss (linking with -latomic on
riscv64) could be dropped, but for some reason it is still needed in
Ubuntu, and not in Debian.
more retrying autopktest regressions
====================================
I noticed many autopkgtest regressions with '/unknown' versions shown
in the update_excuses report[1]. I retried these (--only-unknown) and
214 tests were triggered.
silver-platter
==============
silver-platter was unable to migrate for 53 days due to tests stuck in
"Tests in progress" on architectures other than amd64. silver-platter
only builds binaries on amd64, so I added a hint:
https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=7eb30f973d69ec04b81239f5a935bbcb065692c2
r-cran-dimred
=============
r-cran-dimred is missing a build on armhf, but this was intentional,
so I filed a bug requesting removal of the armhf binaries:
https://bugs.launchpad.net/ubuntu/+source/r-cran-dimred/+bug/2076327
bowtie2
=======
bowtie2 was unable to migrate due to a regression on s390x, but it is
not built for that architecture, so I triggered a
migration-reference/0 test which failed successfully.
r-cran-swagger
==============
r-cran-swagger's autopkgtests were failing due to differing MD5 sums
of some images. I suspected that this was due to the PNG optimization
step in pkgbinarymangler, so I tried exporting NO_PKG_MANGLE=1, in
debian/rules. This worked, but then I found that the reference images
were being downloaded during the autopkgtests, so I decided to skip
those tests altogether:
https://launchpad.net/ubuntu/+source/r-cran-swagger/5.17.14-1ubuntu1
matplotlib / gammapy
====================
My earlier sync of matplotplib became blocked by an autopkgtest
regression in gammapy on s390x, which turned out to be due to a
missing build of a new version of gammapy on s390x. I uploaded fix:
https://launchpad.net/ubuntu/+source/gammapy/1.2-1ubuntu1
[1] https://ubuntu-archive-team.ubuntu.com/proposed-migration/update_excuses.html
[2] https://merges.ubuntu.com/universe.html
--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
