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 the Community Hub (this site).
We also have our weekly meeting on IRC. We meet on Tuesday at 13:30 UTC in #ubuntu-desktop on Freenode. There will be an “Any Other Business” section at the end where you are welcome to raise topics. These topics might be discussed during the meeting, or afterwards depending on the time, depth of conversation, topic and so on.
Tried to backport the missing commits required by the Nvidia driver to bionic, only to find it’s already been done upstream. We just need to wait for mutter version 3.28.4 which contains them.
FYI, note that both mutter and Xorg will need to be rebuilt in disco after we get egl-wayland into main. The latter is required to get acceleration in (GL)X clients (they only get llvmpipe software rendering at the moment).
Wayland issues are not a priority for Ubuntu at the moment. But at the same time I would expect most of the above to be resolved in time for 19.04 anyway.
64.0 was released as a stable update to all supported series (trusty, xenial, bionic, cosmic, disco)
chromium
updated dev to 72.0.3626.14
updated beta to 72.0.3626.17
prepared stable update to 71.0.3578.98
libreoffice
fixed s390x autopkgtest failures (bug #1808147) and backported to cosmic
rls-bb-tracking bugs
bug #1754671 in network-manager: @willcooke’s testing is conclusive, we agreed to upload to bionic-proposed and leave it there for (at least) the duration of the Xmas break to give it wider and in-depth testing
bug #1765304 in gnome-shell: verified (by myself and others)
bug #1802208 in libreoffice: no progress this week
rls-cc-tracking bugs
bug #1765304 in gnome-shell: in unapproved queue, but the bug doesn’t really affect cosmic anyway
bug #1803142 in libreoffice: was accepted in cosmic-proposed, but I had to do another upload to fix autopkgtests on s390x, so back in the unapproved queue
internal: arranged travel for a couple of upcoming trips
rls: I do have bug #1805857 assigned, which I didn’t work on this week. Looks like 1.12.6 didn’t fix it, so still something to do here. That’ll be one of my tasks before xmas.
Still progressing on PR #6258. It is mostly done, but needs review.
While updating the tests to use a confined client app so they could function on Ubuntu Core, I discovered activation triggered by confined apps was broken on 18.04+. I initially tried fixing this with a change to the AppArmor rules, but @jdstrand instead pointed me at the AssumedAppArmor key that can be placed in D-Bus service files.
I came up with some ideas to fix up the incompatibilities with 14.04, which I’m testing at the moment. If those are acceptable, it should get rid of the question about Trusty support.
snapd desktop portal testing
At the moment, all of snapd’s tests for portal support run against a fake version of xdg-desktop-portal, or a filesystem hierarchy that resembles that provided by xdg-document-portal. It doesn’t test behaviour against the real thing.
I started PR #6313 to address this. The idea is to test against the real xdg-desktop-portal and xdg-document-portal with a fake UI service. This allows testing of things like the portal file chooser API without trying to drive a GUI from the test harness.
The plan is to only run the test on a few systems where we know portals should work. As we SRU portals to older Ubuntu releases, we can expand the list of systems to verify that they work with snaps.
snap application polkit support
A while back Ken asked me to look at what would be needed to allow snapped daemons to make use of polkit authorisation (e.g. for things like fwupd).
I started a thread on the forum outlining what I think is needed. The main open question is whether namespacing polkit action IDs to a snap is acceptable. Without namespacing, it is not clear we can safely let snaps install new .policy files.
At present, the forum thread is the extent of my work on this: it was a task to look at after sorting out dbus activation / shell search providers.
Sponsored about a dozen uploads so that we could have the newest fontmake and friends.
Cherry-picked Unicode/Emoji 11 support from GNOME 3.32’s version of pango down to Debian Buster and Ubuntu 19.04 (Ubuntu 18.10 SRU is in the queue). At this time, I don’t plan to SRU that to 18.04 LTS (it has an older pango so it would be more complicated to backport there).
Packaged gtk 3.24.2. This version had some regressions so I’m planning to wait for 3.24.3 before SRUing to 18.10. The known regressions are fixed, but this way we won’t be immediately out of date there.
Identified some emoji fixes from 3.24.2 I am backporting to 18.04 LTS: LP: #1807719 (some emoji in chooser are black & white) and LP: #1807721 (crash when trying to use variant selector in emoji chooser).
Prepared the gnome-desktop SRUs for 18.04 and 18.10 to fix the 32-bit thumbnail issue that Laney fixed upstream. Security is handling 18.04.
Packaged devhelp 3.30. It needed amtk packaged first. devhelp has a soname bump so it’s also in the NEW queue. I demoted devhelp to universe last cycle so that we wouldn’t need to do a MIR for amtk.
Fixed the build tests for tepl (required by latest GNOME Latex release [previously latexila]).
Encouraged Debian to update to poppler 0.71 (it’s in experimental now). Sponsored popplerkit.framework and NMU’d kitinerary to make the transition easier.
Discussed the Epiphany browser upstream maintainer’s concerns about the browser being included in Debian and Ubuntu LTS releases. He would prefer users use a Flatpak or Snap version. (A Snap version doesn’t exist yet.), His concerns aren’t mostly about epiphany itself but about its dependencies: webkit2gtk, gstreamer, glib-networking, which need to be updated for websites like YouTube to keep working.
Other
I’m going to work on the sane-backends issue this week.
I had some hardware issues with my computer this week that interrupted my productivity.