Foundations Team Updates - Thursday 2024/03/21


  • Re-bootstrapped rustc 1.75


Some time_t transition work

Transitioned the following packages (armhf time_t 64-bit transition, with help from @vpa1977 @vorlon @mwhudson, thank you!):

  • gdal
  • leptonlib
  • ffmpeg
  • ogre-1.12
  • nodejs
1 Like


  • added an option to build hotspot disassembler
  • backported -testsupport package to JDK 21
  • added specification draft to disable unattended upgrades of openjdk (work in progress)

libheif MIR:

  • enabled pacakge tests at build time for aom


  • filed removal for dogtag pki to unblock tomcat-9 migration

spring boot snap:

  • pivoted to package Spring Boot CLI instead of trying to create offline Spring Boot installation. Work in progress.

time_t transition:

  • retried autopkgtests
  • no change rebuild of openjdk-21
  • fixed yapet build failure
1 Like
1 Like




  • hiring
  • networking meeting with Michael
1 Like


  • MIR libtraceevent (see LP #2051916)
    • patch refinement


  • Backlog grooming
  • Reviewed and merged # 194 - Consolidated docs for contributors
  • Reviewed and merged # 163 - Make snap more self-contained
  • Improved SnapStore page of ubuntu-image (with @rkratky)
  • LP: #2057885
    • Investigated, but I am unable to reproduce. Laider Lai will try to reproduce on their build server.
  • LP: # 2058629
    • Investigation in progress. This bug appears locally but this is working fine in CI. This is breaking the build.
  • PR #195 - Fix mountpoint management
    • merged
  • PR # 188 - Improve the manual “execute” customization
    • Reviewed by @sil2100
    • Need some improvements.


  • Use the package-repositories stanza in imagecraft YAML
    • Investigation in progress.
    • PR # 23 - Use “standard” package-repositories key in imagecraft.yaml
      • Exploring ideas
  • Setup RTD webhook
  • PR #22 - Use new ubuntu-image feature to build a rootfs easily
    • merged


  • Setting up a RPi5 to have a local arm64 builder
  • Reading about SSDLC and what we plan to do around it
1 Like


  • Worked on the new .NET snap:
    • Worked on setting up the snap hooks to install the latest SDK on first install and update components on subsequent refreshes using the installer tool shipped in the snap (PR #5).
    • Set up the install command so that only one version of the runtime/SDK is kept per major/feature band (PR #6).
    • Started adding unit tests to the installer project to get some coverage going (PR #7).
    • Set up a CI pipeline to run these tests and publish coverage information to a GitHub Pages website.
    • Refactored some components of the installer project into dependency-injected services to achieve better code testability (PR #8)
  • Experimented building current snapcraft.yaml from Microsoft for the dotnet-sdk snap.
  • Attended weekly .NET partners sync meeting
1 Like



  • Final changes before openssl 3.0.13 was uploaded; thanks @tobhe and @schopin in particular
  • Triggering tests again dozen of times, targetting tests with only armhf failures as soon as logs say “badpkg” or something like that
  • Doing the same for arm64 and ppc64el because queues are empty

armhf Y2038 support (a.k.a armhf 64-bit time_t)

  • Went over a list of packages needing merges in Ubuntu and dumped/diffed the ABI for more than a dozen in order to know if we can skip merging them (around 10 could be skipped)
1 Like
1 Like


  • Investigated a bug related to the integration with network manager LP#2055148
  • Preparing SRUs of 0.107.1 for Jammy and Mantic LP#2058051 LP#2058031
  • Investigated what is needed to support Mellanox Scalable Functions on Netplan. Will work on an specification next
  • Investigating how we could add support for fuzzing for libnetplan in our CI
1 Like




proposed migration

  • freetype: only failing armhf autopkgtests. I checked some of the failures. All of those were caused by badpkg. So these test needs to run against all-proposed (due to the time_t transition).

64-bit time_t transition

  • Sync zbar 0.23.93-4 to get a rebuild against libv4l-0t
  • Retried mjpegtools 1:2.1.0+debian-8.1
  • gr-fosphor, gr-iqbal, and gr-radar → python3-pyqt6 (pyqt6) → retried since libqt6widgets6 (qt6-base) → waits on qt6-webengine rebuild
  • Uploaded ucommon 7.0.1-0.2ubuntu3 to fix symbols
  • Please remove minizip
  • Uploaded mapnik 3.1.0+ds-7ubuntu1 to use scons from the Debian package (for Python 3.12 support)
  • libapache2-mod-tile → mapnik
  • Uploaded poppler 24.02.0-1ubuntu8 to rename libraries for 64-bit time_t transition
  • sndobj 2.6.7+ds1-4 for Python 3.12 fix
  • Uploaded polylib 5.22.5-4.2+dfsg to Debian
  • Synced pacman-package-manager 6.0.2-6
  • owncloud-client: FFe: Sync owncloud-client (universe) from Debian unstable (main) and synced
  • Sponsored gvmd 23.1.0-1ubuntu1 for @hyask
  • Sponsored Sync rocr-runtime 5.7.1-2 (universe) from Debian unstable (main)
  • Retried rsyslog build after mongo-c-driver build failure was fixed, but it fails to build on armhf (needs investigation)


1 Like


  • I’ve been working on GRUB update for Mantic, Noble, and Debian Sid
  • ubuntu-boot-test
    • BIOS netboot tests
    • 3-rd party distro chainloading
    • [WIP] mokutil+mokmanager tests


  • dnspython vs samba proposed migration passed with enough retries
  • bpfcc and bpftrace autopkgtests now working for MIR
1 Like
  • Pemmican MIR passed security review \o/ (LP: #2056553)
  • nbd-client needed more attention
    • Enhanced the existing autopkgtests to add read/write testing of nbd devices
    • Wrote an initrd netboot test
    • Tidied up a bunch of lintian warnings
    • Added some tentative systemd tightening
    • Proposed MR upstream
  • Worked on linux-firmware-raspi; still more to do here after discussion with kernel team re: bluez-firmware (LP: #2056588, LP: #2058234)
  • Worked on autopkgtest spec and implementation for isolation of tests (LP: #2058040)
  • Pi meetings
1 Like


Ubuntu Packaging Guide


  • +1 Maintenance (Week 12)


  • paperwork
1 Like


  • Further investigations into creating a testable environment for MOK Enrollment and re-drafting the spec after review
  • PR #1945 - Refactor Subiquity’s Journal based reporting and add new features to the context logger
  • PR #1946 - Support new context logging events in the progress view
  • PR #1947 - Adds a warning that usage of unknown top level keys in autoinstall configurations will be a fatal error in later version (Originally opened as PR #1899 but was closed in favor of the new PR. )
  • Subiquity bug triage shift this week


  • Investigating Vim’s reverse-depends autopkgtest failures. This has been tangled in the time_t transition and has mostly been re queuing tests will all-proposed as needed.
1 Like



  • Rewrote the TCK driver script in Python
  • Did repository reorganization for easier maintenance.


  • Investigated (wip) an openjdk-lts build freeze on armhf, reproduced with upstream build, isolated into a Java program.
  • openjdk-8 security update for Jan 2024 published by the Security team

FIPS prototype


Proposed migration - ccache vs elfutils - reran autopkgtests, fixed
Reviews - jruby-ftbfs

1 Like


  • cryptsetup & jq migration. Retry the tests and they are waiting the results and openssl to be migrated first.
  • Rename the bamf library name for time_t transition. Ubuntu version is newer than Debian, so the time_t change is not synced.
  • Rename libcupsfilter library name for time_t transition. The cups package is split in Ubuntu, so the time_t change is not synced.
  • Continue debugging kmod on armhf after time_t changes. The test failures still exist and are suspicious.
1 Like

auto upgrade testing

  • PR to support double LTS upgrades
  • Opened bug regarding apt-xapian-index crash.


  • Investigated an issue with noble images being old on LXD armhf remote, and starting to behave strangely. As they’re not updated regularly, we split the work in the team to update all the armhf LXD remotes with both a new autopkgtest branch supporting deb822, and updated the noble image, in the hope of having things more stable over the week-end
  • Lots of issue during the week-end and beginning of the week, but everything was stabilized by Tuesday afternoon.
  • Added 14 new armhf workers in bos03, along with all the issues that needed to be addressed, like disk device block name that changed, network device changed, LXD images building issues, etc… Throughput for armhf has now increased consequently, and is being much more stable.
  • Couple of MP opened and reviewed


  • Investigated gvmd dependency issue related to time_t transition. Thanks @bdrung for the upload :slight_smile:
1 Like
  • Fixed startup of software-properties
  • Some grub review stuff
  • Seriously extended my dose-distcheck human report script so that you see conflicting chains, have results for armhf, as well as binary uninstallability.
  • Used report output to fix some time_t issues on armhf by uploading rebuilds :slight_smile:
1 Like


  • AppArmor vs glibc:
    • patch sent upstream to mark tests as unsupported if running when apparmor userns mitigation is enabled
    • tweaked autopkgtests to disable userns mitigation entirely. We want those tests to run!
  • filed a bug to track a ppc64el test regression
  • Conformance tests in t64


  • gtk+3.0: fixed FTBFS due to implicit function declaration (t64 transition)
  • qtwebengine-opensource-src: fix Py 3.12 FTBFS (t64 transition)
  • GH work
  • openssl upload.
1 Like


  • Worked with the team to update the armhf test runners to the latest version of autopkgtest and rebuild their noble images.
  • Reported a bug regarding the amd64 worker behavior when running i386 tests.
  • Reported a bug regarding the web interface at autopkgtest.u.c and non-existent packages.
  • Thought deeply about an MP fixing the above bug and commented on the MP.
  • Submitted an MP updating the modification of apt sources for Noble but it doesn’t work for other releases. :frowning:
  • Reported a bug regarding a test with an “invalid PPA” that managed to get queued.
  • Reviewed an MP from @andersson123 regarding database backup and checksums. This was later approved and merged.
  • Commented on an MP regarding how to clean the queues of obsolete package versions.
  • Reported a bug regarding the yaml file produced by a script of ours not being in the correct format for juju. Then some investigation into what the actual correct format is.
  • Queue management…


  • Submitted an RT regarding archiving Lunar to old-releases.u.c.
  • Some binary package removals!


  • More stuff I failed to write down.
1 Like