Re: [plasma-workspace/Plasma/5.7] shell: don't try to load layout before kamd starts
> It can also be simply not installed. > That was the case for the Neon/Netrunner reports. Yes, Bhushan covered that part - not starting or crashing. I just specified the cases where kamd is 'expected' to crash. >> One of the things that we could try is for kactivities to return >> 'Unknown' as the state until it figures out whether there is kamd >> running or not. And then set it as Running or NotRunning. > > +100 Ok. This will make Plasma's life a bit easier I guess. I still expect corner-case problems, but there should be fewer of them. Cheers, Ivan ___ Plasma-devel mailing list Plasma-devel@kde.org https://mail.kde.org/mailman/listinfo/plasma-devel
Re: [plasma-workspace/Plasma/5.7] shell: don't try to load layout before kamd starts
On Sat, Jul 23, 2016 at 9:07 PM, Ivan Čukić wrote: > The only reasons for kamd to crash are inability to create WAL sqlite > database (it will return an assert with a detailed message), and to > crash on *exit* due to the QSql bug related to driver deinit. > > It can also be simply not installed. That was the case for the Neon/Netrunner reports. And whilst that is a downstream bug, plasma shell code does try to make it optional. if we want to make it mandatory we should make the code fully do that. > Other than that, I haven't had any crashes reported for quite a long time > now. > > Are there any backtraces or more info related to this issue? > > One of the things that we could try is for kactivities to return > 'Unknown' as the state until it figures out whether there is kamd > running or not. And then set it as Running or NotRunning. > +100 > But Plasma will still have to decide what to do in both cases. > > Cheers, > Ivan > > p.s. I've approved D2275 - the old script was indeed very wrong. > ___ > Plasma-devel mailing list > Plasma-devel@kde.org > https://mail.kde.org/mailman/listinfo/plasma-devel > ___ Plasma-devel mailing list Plasma-devel@kde.org https://mail.kde.org/mailman/listinfo/plasma-devel
Re: [plasma-workspace/Plasma/5.7] shell: don't try to load layout before kamd starts
On Sat, Jul 23, 2016 at 06:50:45PM +0200, Marco Martin wrote: > > I was surprised plasma-mobile was without kamd, and yes, the fact that > before it appeared to be "working", it was ctually wrong > I will adjust mobile packaging to install kactivitymanagerd as well, and perhaps plasma-workspace packaging in Neon should be adjusted to depend on kactivitymanagerd. Thanks -- Bhushan Shah http://bhush9.github.io IRC Nick : bshah on Freenode ___ Plasma-devel mailing list Plasma-devel@kde.org https://mail.kde.org/mailman/listinfo/plasma-devel
Re: [plasma-workspace/Plasma/5.7] shell: don't try to load layout before kamd starts
The only reasons for kamd to crash are inability to create WAL sqlite database (it will return an assert with a detailed message), and to crash on *exit* due to the QSql bug related to driver deinit. Other than that, I haven't had any crashes reported for quite a long time now. Are there any backtraces or more info related to this issue? One of the things that we could try is for kactivities to return 'Unknown' as the state until it figures out whether there is kamd running or not. And then set it as Running or NotRunning. But Plasma will still have to decide what to do in both cases. Cheers, Ivan p.s. I've approved D2275 - the old script was indeed very wrong. ___ Plasma-devel mailing list Plasma-devel@kde.org https://mail.kde.org/mailman/listinfo/plasma-devel
Re: [plasma-workspace/Plasma/5.7] shell: don't try to load layout before kamd starts
On Sat, Jul 23, 2016 at 6:50 PM, Marco Martin wrote: >> This commit seems to break plasmashell loading in various cases, I at >> least have one report from #kde-neon that plasmashell is not starting >> and I've locally also reproduced that if kactivitymanagerd is not >> starting or crashing, pplasmashell just sits there doing nothing.. > > also reported on > https://bugs.kde.org/show_bug.cgi?id=365989 so, in the case of that bug report, it's a crash (couldn't reproduce yet on a normal system) that happens when the very wrong layout.js is used (it's wrong upstream, so distros just based theirs upon it, making them wrong as well) i guess neon has a similar issue https://phabricator.kde.org/D2275 corrects layout.js upstream -- Marco Martin ___ Plasma-devel mailing list Plasma-devel@kde.org https://mail.kde.org/mailman/listinfo/plasma-devel
Re: [plasma-workspace/Plasma/5.7] shell: don't try to load layout before kamd starts
On Fri, Jul 22, 2016 at 5:12 PM, Bhushan Shah wrote: >> http://commits.kde.org/plasma-workspace/f7ef6ee87b8957bebc976b6fc9e0df279cea05f1 >> > > This commit seems to break plasmashell loading in various cases, I at > least have one report from #kde-neon that plasmashell is not starting > and I've locally also reproduced that if kactivitymanagerd is not > starting or crashing, pplasmashell just sits there doing nothing.. also reported on https://bugs.kde.org/show_bug.cgi?id=365989 the root of the problem then it seems that in some cases kactivitymanagerd doesn't start correct ly or crashes: that's the first thing we should make sure never, ever happens, because in that case, i have no idea how to guarantee how a plasmashell session could ever start correctly, the fact that this was probably happening since a long time but masked, is also quite worrying. > This commit also breaks plasmashell on mobile.. I was surprised plasma-mobile was without kamd, and yes, the fact that before it appeared to be "working", it was ctually wrong -- Marco Martin ___ Plasma-devel mailing list Plasma-devel@kde.org https://mail.kde.org/mailman/listinfo/plasma-devel
Re: [plasma-workspace/Plasma/5.7] shell: don't try to load layout before kamd starts
Hello Marco and co. On Wed, Jul 20, 2016 at 11:59:37AM +, Marco Martin wrote: > Git commit f7ef6ee87b8957bebc976b6fc9e0df279cea05f1 by Marco Martin. > Committed on 20/07/2016 at 11:57. > Pushed by mart into branch 'Plasma/5.7'. > > don't try to load layout before kamd starts > > if the status of kamd is anything but running, refuse to > execute load() > this fixes activity creation in default layout initialization > > M +4-2shell/shellcorona.cpp > > http://commits.kde.org/plasma-workspace/f7ef6ee87b8957bebc976b6fc9e0df279cea05f1 > This commit seems to break plasmashell loading in various cases, I at least have one report from #kde-neon that plasmashell is not starting and I've locally also reproduced that if kactivitymanagerd is not starting or crashing, pplasmashell just sits there doing nothing.. This commit also breaks plasmashell on mobile.. Thanks -- Bhushan Shah http://bhush9.github.io IRC Nick : bshah on Freenode ___ Plasma-devel mailing list Plasma-devel@kde.org https://mail.kde.org/mailman/listinfo/plasma-devel