I was on +1 maintenance last week and below are the packages I investigated.
numpy ===== numba was the last blocker preventing the migration of numpy 1:1.20.2-1ubuntu1. It takes upstream some time to add support for new major python versions, and in the course of the python 3.9 transition it was temporarily removed from the archive. I filed LP: #1951814 requesting removal, which sil2100 took care of (thanks!). Once numpy had migrated, I retried all failed autopkgtests and another 127 packages followed. I then sponsored dbungert's sync request of 1:1.21.4-2 in LP: #1951860. gsl === libmath-gsl-perl's autopkgtest failure is caused by an ABI break in gsl. I filed LP: #1951828 to track this and linked to the bug in Debian. Per Debian bug #998192, it should be resolved shortly. netcdf ====== Previously, I had uploaded no-change rebuilds for the netcdf transition, but libpdl-netcdf-perl, ferret-vis and vtk6 still FTBFS (thanks cpaelzer for pointing these out). libpdl-netcdf-perl was missing a build dependency on pdl, pdl was missing a build dependency on libmoosex-getopt-perl, and libmoosex-getopt-perl was missing a build dependency on... itself. I tried a 'nocheck' build in a PPA in an attempt to break the cycle, which was successful, and I was about to upload and follow up with a normal upload, but locutusofborg suggested it may be possible to build against -release in bileto, which he then did and saved me the two uploads. With libmoosex-getopt-perl built, I was able to successfully retry autopkgtests of libmarkdent-perl, libmoosex-app-cmd-perl and libmoosex-getopt-perl itself on behalf of xypron. ferret-vis only needed a merge, which I did. vtk6 FTBFS against GCC 11, and I found Debian intend to remove older versions of vtk, in favour of the current vtk9. I filed LP: #1952422 requesting removal, and doko took care of that (thanks!). Unfortunately, netcdf remains entangled with proj, glew and vtk9. r-api-bioc-3.14 =============== There was another r-bioc* transition incoming from debian. I set up a tracker and retried builds when they were attempted in the incorrect order when autosync-ing from Debian. https://people.canonical.com/~ubuntu-archive/transitions/html/r-api-bioc-3.14.html Once the builds were complete, I triggered autopkgtests of all the r-bioc* packages with all-proposed=1. There are still two regressions; r-bioc-biocparallel and r-bioc-gviz for which we should get fixes from Debian soon enough. python-skbio ============ python-skbio needed re-cythonizing and some fixes from upstream for python 3.10. I uploaded 0.5.6-5ubuntu1 and forwarded my changes to Debian. segyio ====== There was an incoming NMU from Debian which fixed the FTBFS with python 3.10. Once that had been autosync'd, I applied a patch to fix the test failure with glibc 2.34 and filed a bug in Debian for when they eventually upgrade glibc. -- ubuntu-devel mailing list [email protected] Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
