#1032674 /usr/bin/plasmashell: when killing a window via the grouped task bar, the task bar disappears #1032674
- Package:
- plasma-workspace
- Source:
- plasma-workspace
- Description:
- Plasma Workspace for KF5
- Submitter:
- Paul Gevers
- Date:
- 2023-03-10 21:30:07 UTC
- Severity:
- normal
- Tags:
Dear maintainers, Thank you for maintaining KDE in Debian, it must be a hell of a job. I want to share the following annoyance which is probably due to a crash and auto restart (so an annoyance, but probably it really should be fixed). When I have multiple windows of the same program (qgit in my case) I have their entries grouped in the task bar. If I hover over their entry in the task bar, multiple minified windows appear with a close button. When I press the close button of the minified window I want to close, the task bar dissapears for several seconds. When the bar returns I can see that `plasmashell` got started in the same minute as the disappearance. I traced it a bit with the one-liner below. `plasmashell` actually seems to stay running until the task bar reappears, at which moment the start time and ID get bumped. So maybe the problem lies elsewhere, please tell me if you need more from me trace back the culprit. Paul paul@mulciber ~ $ while true ; do ps aux | grep /usr/bin/plasmashell | grep -v grep ; echo . ; sleep 1 ; done paul 220257 1.4 8.3 3211068 309192 ? Rsl 20:47 0:10 /usr/bin/plasmashell --no-respawn . paul 220257 1.4 8.6 3211068 319060 ? Ssl 20:47 0:10 /usr/bin/plasmashell --no-respawn . paul 220257 1.4 8.2 3211068 305960 ? Ssl 20:47 0:10 /usr/bin/plasmashell --no-respawn . paul 220257 1.4 7.5 3211068 279624 ? Ssl 20:47 0:10 /usr/bin/plasmashell --no-respawn . paul 220257 1.4 7.4 3211068 275072 ? Ssl 20:47 0:10 /usr/bin/plasmashell --no-respawn . paul 220858 54.0 1.7 378372 64852 ? Rsl 20:59 0:00 /usr/bin/plasmashell --no-respawn . paul 220858 42.2 3.5 1234588 130732 ? Dsl 20:59 0:00 /usr/bin/plasmashell --no-respawn . paul 220858 43.6 4.6 1505996 173336 ? Rsl 20:59 0:01 /usr/bin/plasmashell --no-respawn . paul 220858 48.8 5.3 2043572 198924 ? Rsl 20:59 0:01 /usr/bin/plasmashell --no-respawn . paul 220858 43.4 5.7 2060256 214048 ? Ssl 20:59 0:01 /usr/bin/plasmashell --no-respawn
control: forwarded -1 https://bugs.kde.org/show_bug.cgi?id=464186 Le 10 mars 2023 21:45:57 GMT+01:00, Paul Gevers <elbrus@debian.org> a écrit : Dear Paul, Thank you for coordinating Debian releases, it must be a hell of a job. 🙂 This is tracked upstream as bug 464186 (now linked). The fix is in KDE Frameworks 5.104 which is expected to be released tomorrow (March 11.). I've not seen the release note yet but KDE Frameworks is a tightly coupled set of 83 source packages, and I'd prefer avoiding having to backport individual fixes as much as possible. If I could trick the Release Team into accepting 5.104 into bookworm as a whole, it would fix this issue and probably several others in a safe way. Frameworks 5.x is now in maintenance mode and upstream is supposed to include only bugfixes there. This particular fix is a oneliner so not an issue to backport in itself, but it would mean I'm going down the long hard path of patching on top of the 83 frameworks packages version 5.103, hoping not to (cross-)break more things. So patching is definitely not my preferred option. Let's see what the complete Frameworks 5.104 release note brings us and I'll file a proper unblock pre-approval bug to discuss this. Happy hacking, -- Aurélien
control: forwarded -1 https://bugs.kde.org/show_bug.cgi?id=464186 Le 10 mars 2023 21:45:57 GMT+01:00, Paul Gevers <elbrus@debian.org> a écrit : Dear Paul, Thank you for coordinating Debian releases, it must be a hell of a job. 🙂 This is tracked upstream as bug 464186 (now linked). The fix is in KDE Frameworks 5.104 which is expected to be released tomorrow (March 11.). I've not seen the release note yet but KDE Frameworks is a tightly coupled set of 83 source packages, and I'd prefer avoiding having to backport individual fixes as much as possible. If I could trick the Release Team into accepting 5.104 into bookworm as a whole, it would fix this issue and probably several others in a safe way. Frameworks 5.x is now in maintenance mode and upstream is supposed to include only bugfixes there. This particular fix is a oneliner so not an issue to backport in itself, but it would mean I'm going down the long hard path of patching on top of the 83 frameworks packages version 5.103, hoping not to (cross-)break more things. So patching is definitely not my preferred option. Let's see what the complete Frameworks 5.104 release note brings us and I'll file a proper unblock pre-approval bug to discuss this. Happy hacking, -- Aurélien