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.
(Briefly) resolved all discussions that were blocking 10-bit deep colour support. Just awaiting more testing now.
Major improvements to the Nvidia latency fix to resolve outstanding discussions, become compatible with triple buffering and generally perform better than before.
fixed avahi shipping an invalid ifdown entry and changed it to clean obsolete conffiles on upgrade
sponsoring
OEM fixes for audio on AMD Ryzen acp devices (F/G/H)
helped syncing the bluez vcs after direct archive uploads, commented about the changes since review was not asked before
reviewed and merged extras versions-script changes from Marco
pango fixes backported by Gunnar
sorted out the details of a wpa patch submitted for review and SRUed the fix to bionic
SRU
backported a fix for gdm segfaulting in some cases when using xfvb to focal
did some bug triaging on the snap store
thunderbird snap
updated to 78.4.3
debugged opening files from the composer not working
restored the removable-media interface, some users rely on it to share profiles between partitions
versions script
fixed a bug in listing upgrade-software-version entries
spent some more times trying to figure out how to build a liveCD ISO locally, using ubuntu-old-fashioned was alright but figuring out how to combine the artifacts in an ISO using ubuntu-cdimage is no fun, finally got an iso locally built!
Again babysat the clouds for autopkgtest, couple of outages including our quota getting out of sync meaning we couldn’t spawn any new instances for a bit, yay
We got a new perl and a new glibc thousands more tests each, will the queues ever get
to zero?
The previous perl tests weren’t finished before it got superseded, so I cleaned those away and requeued with new perl.
Reviewed some changes to various autopkgtest things from others.
Worked with Canonical IS a bit on metrics, should be able to push from autopkgtest now, will try to get that up and running soon-ish
Worked on a change to proposed-migration to avoid queueing tests too early for transitions - no point running tests for a thing which is uninstallable - will propose that tomorrow.
Chugged through more outstanding actions from groovy release and hirsute opening with some others on the release team.
Discussed smartcard authentication with @3v1n0 and others
the fix to backport isn’t in the 2.50 series yet, so I asked Federico what his plans for a 2.50.2 release are, and he plans on making it before the end of the week
Continued to worked on fixing an LO snap bug regarding the libreoffice.desktop actions… tried several things but then came across this bug where the startcenter and math icons were removed from the deb back in 2018. I believe the snap should match the deb but there’s pushback already so want to hear others opinion.
Read through the new netplan spec… Basically the plan is to add to libnetplan so that it can write YAML and to netplan to read/translate NM keyfile data so that there is feature parity of all settings from NetworkManager, to be representable in netplan YAML format
LAS
Deep dived into flutter snaps, put together presentation for LAS, and learned a lot! Found the VS code snap bug that Marcus proposed a fix for
Spent half of Thursday and all of Friday on LAS. Ken and I gave a talk/demo of flutter snaps that went really well, and coordinated the canonical office hour.
PostScript Printer Application: Continued the development. The user of the PostScript Printer Application does not need to weed through a list of ~4000 printers to find his model (it is sorted at least) as now the first entry of this list reads “Automatic Selection” and this picks the correct model. I have also added support for all options in the PPD file, also the ones which do not have an equivalent standard IPP attribute, at least one can set defaults for these option via the Printer Application’s web interface now.
PAPPL: The PostScript Printer Application is exercising PAPPL very well and reveals a lot of bugs, missing features and other shortcomings. During the last 7 days I have reported a total of 13 bugs and feature requests, 8 still open with all of them already triaged and accepted by Michael Sweet, 5 closed with 2 fixed/implemented and 3 turned into questions/discussions. All the 5 issues I reported the week before Michael has fixed/implemented/merged.
cups-filters: The development of the PostScript Printer Application also caused some changes and fixes on cups-filters: Moved some functios for PPD option name conversion to the public API and improved them somewhat, fixed also some minor bugs.
CUPS Snap: Still waiting for @jamesh’s API extension of snapd to pass Canonical’s security team review so that the CUPS Snap makes it into the Snap Store.
CUPS Snap and Printer Applications: Helped @laney on using these projects for an Ubuntu appliance with printing functionality.
sane-airscan: Waiting for the MIR to get reviewed by the security team.
system-config-printer: Hugh Chao’s SRU for Focal and Groovy made it into the updates.
Google Season of Docs 2020: Mentoring our technical writer, doing corrections. Follow his work growing on our web site.
Seb asked me to look into a problem reported with the Thunderbird snap, where it could not launch the associated application for an attachment in the compose window.
The trigger was calling the OpenURI.OpenFile portal API using a document portal file path (i.e. $XDG_RUNTIME_DIR/doc/...) as returned by a portal file chooser. I was able to reproduce this with a simplified example that showed it to be caused by the way these paths have different inode values inside and outside of the snap sandbox.
I filed xdg-desktop-portal issue #545 with the details, since it looked to be something that could be reproduced with Flatpak confined applications too.
I had a quick look to see how easy it would be to fix, but the document portal code is a lot more complex than last time I looked, having grown the ability to share directory trees rather than just files. This is important, since while treating leaf files as hard links, sharing inodes for directories could allow sandboxed apps to escape from their subtree of the doc portal file system.
snapd PRs:
I’ve been working to update snapd PR #9497 after mvo gave it a review.
While the D-Bus activation branch received the security sign-off, but still needs a snapd team review. I’m following up to make sure we get security reviews of the remaining blocked branches.
Gnome Shell and friends:
Other:
Bug tracking:
thousands more tests each, will the queues ever get
firefox
chromium
flutter
other