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.
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.
Geometric (OpenGL-less) picking: This week completed full 3D transformation support, reaching full feature parity with the code it replaces. Still using much less CPU than the code it replaces.
Started tracking OpenGL pipeline stalls which account for a lot of frame skips and limited frame rates. The good news is that fixes are already proposed.
[nvidia] Corrupted wallpaper after resuming from suspend or hibernation: Fixed upstream in mutter 3.33.4. Just waiting for that to trickle into eoan. Because I noticed different behaviour between 3.33 and 3.32, I am not game to declare the same fix a complete solution for 3.32. So I would like to let the fix reach 3.34 first and mature.
tested and confirmed bug #1837437 (“disk content permanently lost when changing LUKS password”), formatted the bug following the SRU template and attached a debdiff that @laney kindly sponsored
Cover boot commit by a complete testsuite and various fixes for those.
Add “zsys boot prepare” and “zsys boot commit” command line arguments (the first will be hooked into the zfs mount generator and the second in a systemd service to commit boot)
Start looking at hooking up in zfs mount generator + various experiments and fixes on real double machine systems + appropriate fixes.
Found a zfs 0.8 regression in mount ordering, debug, find a reproducer and then backport a corresponding an upstream fix. Test + upload in the canary-image ppa (with other changes right now)
Debugged and found that the grub merge erased some changes in master (like our awk -> shell changes) and has flatten all our patches + drop some other parts. Redid the whole work and upload grub+ grub-signed.
PR #6954 (the skeleton of the session agent daemon) was finally merged last week.
Following up on that has been a few PRs:
PR #7162 – some code reorganisation brought up in the first PR. (mered)
PR #7194 – adjust how new systemd units are enabled on RPM systems to match their policy. (draft)
PR #7197 – have the session agent exit after a period of inactivity. (draft)
Still todo is write some client code to talk to all available session agents, and implement the user unit management API calls. Once those are done, it should unblock the user daemons PR.
snapd interfaces for Snap Store:
The packagekit-control interface is live on the edge channel of the core snap, so available for testing.
PR #7042 (add the appstream-metadata interface) is progressing, having received more review comments. Depending on Jamie’s decision, the interface may need an allow-installation: false restriction on the plug side so only authorised snaps can use it.
snapd icon theme support:
More review comments on PR #6767: there is some concern about the potential for problems if a snap provides overly large icons. This might get pushed off to a follow-up PR though.
Other:
While going through some of my old PRs, I fixed some conflicts in snapcraft PR #2229 and pinged Sergio to see why it hadn’t been reviewed yet. This was a small PR that extracts a bit more info from ELF files as a first step towards extracting debug information during a snap build. Not much has come of that project, but it seems a waste not to get these clean ups merged.
cups-filters: Based on a bug report from my GSoC student I did several improvements on cups-browsed. More to come. It is mainly about handling emulations of IPP printers running locally (Printer Applications, ippusbxd, …) and coping with the case of network connections be taken down mid-session (like Wi-Fi).
network-manager: More testing on the autopkgtest script ending up finding a solution to make the tests pass reliably. The test connections got actually all reliably established, the solution was a slight change in the test method (Pull request).
network-manager: Succeeded to have more dialog with the reporters of the Bionic SRU regressions. Bug 1829566 closed, it was user error. Got more input on bug 1829838. Reporter of bug 1829913 gave up helping.
Google Summer of Code 2019: Second round of evaluations (after the second month of coding). All 5 OpenPrinting Students passed again. 2 of the other Linux Foundation students failed. I had to take care that all mentors turn in their evaluations to not get penalizations from Google. Continued mentoring my student on Printer Applications.
Avahi: Still waiting for continuation on the localhost problem by Trent.