release week, as time permitted. Below are the packages I
investigated.
python-sparse/hyperspy/sfepy/python-cogent
==========================================
Numba and several of its reverse dependencies were removed in LP:
#1951814. Thanks to the work by Matthias and Erich, we got new
versions of onetbb and numba in time for jammy.
I looked at getting the reverse dependencies back in. I uploaded
python-sparse fixing the FTBFS with the new numba, and proposed a MR
hinting the previously hinted autopkgtest failure on s390x. This
unblocked hyperspy. I sync'd a new version of sfepy from Debian,
which fixed the FTBFS, but still needed another upload to fix the
autopkgtests (not tested in Debian due to missing numba). The
autopkgtest failures in python-cogent were previously hinted, but I
proposed a MR adding it to big_packages instead, and then only
required hints on i386 and s390x.
fpc/lazarus/view3dscene
=======================
While merging fpc from Debian, I found it FTBFS with a '-f may not be
used without -shared' link error. This was caused by the
-ffat-lto-objects flag now injected in jammy, but not in impish.
Filtering it out fixed fpc, and I found similar issues in the related
packages lazarus and view3dscene.
mercurial/hg-git/git-remote-hg/totoisehg
=========================================
Mercurial 6.1 was FTBFS due to test failures with Python 3.10. I
sync'd the 6.1.1 bugfix release from Debian, which solved all but one
of the test failures. After discussing with Simon, we found Debian
had enabled a previously disabled test which was trying to test TLS
1.0/1.1 which are disabled in Ubuntu. Disabling this test again fixed
the build, and I followed up with syncs of tortoisehg, git-remote-hg
and hg-git to allow migration.
r-cran-relsurv/r-cran-pec/r-cran-riskregression/r-cran-rms
==========================================================
I found r-cran-relsurv had not built on riscv64 due to a missing
build-dependency. I followed the chain of build-dependencies and was
able to build them all locally. Eventually, I found that r-cram-rms
had not built on riscv64 in impish, and there was no new version in
jammy, so I uploaded a no-change rebuild and checked that r-cran-rms,
r-cran-riskregression, r-cran-pec and r-cran-relsurv all built
correctly on riscv64.
abyss/asymptote
===============
I found the latest version of abyss was running out of memory during
its autopkgtest on ppc64el and the latest version of asymptote was
doing the same on all architectures. I proposed MRs adding both to
big_packages.
paraview
========
Paraview had been removed in November 2021 for blocking the libglew
transition. The fixed version was unable to migrate, so I proposed a
MR hinting the previously hinted autopkgtest failures on armhf and
s390x.
sagemath/sagetex
================
Debian officially dropped armhf from the architecture list in
sagemath. I sync'd this version and proposed a MR hinting sagetex's
autopkgtest failure on armhf.
various syncs
=============
I sync'd several packages from Debian to fix FTBFS and/or autopkgtest
failures; breathe, deepdiff, dwarves, hydrogen, pyqi,
python-django-celery-beat and sparse.
various uploads
===============
I uploaded some packages with minor changes to fix FTBFS and/or
autopkgtest failures and filed bugs in Debian where applicable;
aspcud, ganeti, lnav, macs, mlpack, node-marked and openmw.
various sponsored uploads
=========================
I reviewed and sponsored scilab (LP: #1969227), pagekite (LP:
#1963719), ignition-physics (LP: #1965064), ignition-launch and
ignition-gazebo (both in LP: #1960940).
--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
