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.
Spent a few days auditing graphics drivers: i915, nouveau, nvidia-390, nvidia-340 to examine their advertised vs working vsync features. Because confusion between what is advertised and what actually works was holding up progress in !73, !191 and !334.
FYI, some generally exciting news: Upstream some people have tracked down two causes of disk IO impacting graphics smoothness. Something I knew was a problem but had never investigated yet. It’s encouraging there are fixes proposed already (1, 2).
Waited a few days for the “required” dependencies to get published but they’re still not done. Built them myself and then found they’re not required to get as far as logging in and using a Wayland session with Nvidia. Still, there are actually multiple separate problems that I’ve now tried to outline in the bug.
Users of 19.04 can indeed install nvidia-driver-410 and enable Wayland with nvidia-drm.modeset=1 right now. You can log in and use the desktop but:
it’s very heavy on the CPU (likely because EGLStreams support is missing, slow buffer copies?); and
GLX apps (via Xwayland) are not accelerated at all (they use LLVMpipe software rendering).
PR #6258 is still progressing. There is still some work to be done on the tests and I need to get it reviewed.
Under testing, I discovered that directly activated (as opposed to systemd activated) system bus services were failing. As Ubuntu 14.04 doesn’t support systemd activation, it basically meant system bus service activation didn’t work at all there. Further more, changing things to work with 14.04 would make supporting Ubuntu Core systems more difficult. I’m following up with the snapd guys to decide how to proceed.
With bus activated services the most appropriate systemd service type is dbus, where the service is considered started when a given bus name is claimed. Snapd can generate unit files of this type, but Snapcraft and review-tools both consider it a syntax error. I started a thread to see what we can do about this. On the bright side, the fact you can’t currently publish a snap using daemon: dbus means we’re free to change the behaviour if appropriate.
Packaged some GTK emoji fixes for Disco & started SRUs for 18.04 and 18.10
Released GNOME Tweaks 3.30.2 and prepared the Cosmic SRU
Did GNOME 3.31.3 releases for gedit, gnome-menus, Mahjongg, Tali, Tetravex, Tweaks, Quadrapassel
Proposed a Night Light color slider for GNOME Settings
Cleaned up deprecations in GNOME Settings app’s Mouse & Power panels
Added Keyboard Shortcut overlays to Chess, Mahjongg, Mines, Tali, and Tetravex
Updated these apps for the new GNOME 3.32 menu recommendations: Chess, D-Feet, gtk-widget-factory, Deja Dup, Dictionary (proposed), Iagno, Klotski, Nibbles, Quadrapassel, Robots, Tali, Tetravex, Yelp (proposed)
Switched Tali and Tetravex from autotools to meson
Started discussion about how keyboard accessibility for the new GNOME “primary menus” should work
Reverted our Tracker packaging to use autotools because of autopkgtest regressions seen with Meson
Helped with the ongoing poppler/hunspell/libzip/glew transition
Sponsored appstream-glib and snapd-glib disco updates for Robert
Segmentation fault on gnome-shell debugt exit: mutter is in main, gjs is in cosmic-proposed
Ubuntu dock/launcher is shown on the lock screen: it looks like that the problem has not be completely fixed. I took another look last week but without being able to reproduce it’s a bit hard.
gnome session: Must ask twice to lock the screen: same as above
PulseAudio:
Gnome Shell performance (
Spent a few days auditing graphics drivers: i915, nouveau, nvidia-390, nvidia-340 to examine their advertised vs working vsync features. Because confusion between what is advertised and what actually works was holding up progress in 
Again almost completed 
Refreshed, fixed conflicts and retested:
FYI, some generally exciting news: Upstream some people have tracked down two causes of disk IO impacting graphics smoothness. Something I knew was a problem but had never investigated yet. It’s encouraging there are fixes proposed already (
Investigated 
Housekeeping:
Finally migrated my development machine to disco.
Bug management:
No significant changes this week, which itself is unusual.
Release bugs:
firefox
chromium
libreoffice