https://bugs.kde.org/show_bug.cgi?id=436350
--- Comment #25 from Vaclav Masin <[email protected]> --- (In reply to Konrad Materka from comment #23) > (In reply to Vaclav Masin from comment #20) > > I'm starting to feel a little bit bad for taking so much of your time on an > > issue that due the uniqueness of my setup is quite possibly experienced by > > no one else on the whole planet than me. > Quite possibly, but it already motivated me to fix similar problem :) It is > possible that your issue is totally valid and can happen for someone else. > It is also possible that is unique to your setup. We can try to investigate > a little bit more, fit it or leave it - at least there is a workaround. > Of course this works in other direction - if you are fed up with this just > comment we can leave it as it is. > I'd certainly prefer to get it fixed properly. As long as you're not fed up with me yet, I'm happy to cooperate. > > Anyway, the "ugly" patch indeed seems to prevent plasmashell from crashing. > > I'm going to attach the two outputs you ask for now. > > The output is quite suprising. It tries to use item with empty service id > (Data Source). It should be something like: ":1.9/StatusNotifierItem" (as in > listIcons.sh output). > > In this case there two processes: plasmashell (all panels, applets, desktop > etc) and kded5 (daemons with one important: StatusNotifierWatcher). > > I'll attach new patch with additional logs. Can you apply, restart > plasmashell (with QT_LOGGING_RULES) and try to reproduce? > > After that, can you restart kded5 and try to reproduce? > killall kded5; QT_LOGGING_RULES=*kde*=true kded5 & > > kded5 holds the state of Status Notifier Items (SNIs). Idea is that this > state is corrupted for whatever reason with some empty items. If that's the > case restart should fix it (temporarily). After restart in plasmashell logs > you should see entries like: > kde.dataengine.sni: Service "org.kde.StatusNotifierWatcher" status change, > old owner: "" new: ":1.144" > kde.dataengine.sni: Registering > ":1.125/org/ayatana/NotificationItem/discord1" So you want me to undo that "ugly" patch, apply the new one, restart with the logging enabled, try to trigger the crash and post the log? Although looking at the new patch the "ugly" code seems still there, so I shouldn't be able to reproduce the crash. So I guess you just want the extended log, with an attempt to start up e.g. ktorrent or so? -- You are receiving this mail because: You are watching all bug changes.
