Desktop Team Updates - Monday 17th May 2021

Hi everyone, below you will find the updates from the Desktop team from the last week. If you’re interested in discussing a topic please start a thread in the Desktop area of Discourse .

Last week’s notes are here: Desktop Team Updates - Monday 10th May 2021





  • Added doc command to adsysctl
  • Added documentation to project.
  • Continued backport to 20.04.3
  • Work on CI to:
    • Add automated source to wiki and wiki to source update for documentation
    • CLA check, with automated comment on MP
    • Move scrips in yaml directly and clean them up.
  • Augment shell completion to dynamically file up documentation chapters, users, files for all ADSys commands.
  • Code reorganization to have policies available in top directory
  • Cleanups (golint), small fixes and debug infos for tests
  • GNOME updates, gnome-boxes, gnome-autoar, gnome-online-accounts, gnome-initial-setup, dconf-editor
  • debugged the desktop daily iso not getting promoted, proposed a fix to a script used by the CI jobs which got merged, we have current images now
  • Debian merges, libinih, libvncserver, geoclue, webkitgtk, extractpdfmark, freedombox, libao, fonts-khmeros, gnome-settings-daemon, gnome-control-center, gnome-online-accounts, gnome-initial-setup, pulseaudio
  • Syncs from experimental, since Debian is frozen updates are often staged there instead which means no autosyncs
  • did some desktop SRU verifications
  • dashboards
    • did some extra tweaks to the versions collector, created a ‘current’ symlinks in stats that point to the current records, got merged, thanks Laney! Added it to the grafana views
  • installer
    • spent a day debugging the new installer snap segfault in dri drivers, turned out that patchelf is doing changes to the binary that lead to a bug and snapcraft is calling it to set an rpath in classic … fixed now by using no-patchelf for the dri directory
    • joined the virtual hackfest
  • sponsoring
    • uploaded the wslu update from Patrick
  • Updated the libsoup tests depends for the php8 transition
  • Other
    • spent some hours trying to understand why multipass and lxc were getting stucked starting a new instance, turned out they don’t handle well connectivity issues to the images server (which was failing to Canonical VPN problems)


  • Preparing gnome-40 for impish
  • Screen shield mutter work
  • Various gjs cleanups and upstream maintainance work
  • Short week
  • Finished off some more impish opening tasks; there are a bunch of things to do “in the weeks” following opening, mostly uploading things with a new release name.
  • Couple of updates orca glib2.0
  • Took a look at the proposed patch for LP: #1921862. I don’t think it solves the identified problem (although is a good idea anyway), so investigated a bit, proposed a MR to lp:ubuntu-archive-publishing and suggested some ways to unblock the issue until the MRs get reviewed. All linked from / on the bug.
  • autopkgtest was mostly calm, except one issue where some of the new ARM servers don’t boot bionic :slightly_frowning_face: so looked into that a bit, asked IS to help, and have installed a temporary change to make sure these servers don’t run bionic tests

Ubuntu Core GDM Experiment:

  • I made a number of improvements to the prototype image to make testing with additional snaps easier:
    • Built a 6GB image rather than 4GB. The gadget snap reserves almost 2GB for the ESP and boot partitions, so space could get a bit tight after installing a number of desktop snaps.
    • Updated the wrapper script that launches the confined desktop session to (a) set the DISPLAY environment variable in the dbus and systemd activation environments, and (b) wait for gnome-shell to create its Xauthority file, and copy it to a location snapd’s x11 plug will let snaps read.
    • Provided a helper script that will create launcher .desktop files to launch snap applications via the desktop-launch D-Bus interface.
  • All combined, we were able to launch and use complex desktop apps like Chromium and LibreOffice from the confined desktop session (after installing them and connecting their x11 plug from an unconfined shell).
  • There’s still some work to snap apps communicating via Wayland working though.

snapd PRs:

  • worked on improving testing of the polkit interface PRs, and ran into some trouble trying to write an in-tree test a test snap that talked to polkitd: it turns out dbus-send is not expressive enough to encode the arguments to polkit’s D-Bus interface. After discussing it on IRC, I put together core20 PR #103 to add libglib2.0-bin to the base snap to make the gdbus tool available.

outside of work: