> On Ago. 21, 2015, 8:08 a.m., David Faure wrote: > > I thought the idea was "there's no race, because as long as the recipient > > doesn't go to the event loop, it won't be processing any incoming messages > > anyway"? > > Albert Astals Cid wrote: > But with Thiago's patches dbus runs in a thread so that's not true > anymore. > > David Faure wrote: > Why don't Thiago's patches preserve behavior compatibility by > starting/enabling the dbus thread only when reaching the event loop? > > Otherwise it sounds like any case of "talking to an application that is > starting" will create races (am I making the call before or after the object > has been published?). Not just KUniqueApplication, but any other dbus call to > a starting app, including the dbus activation case. > > Albert Astals Cid wrote: > I don't know, he knows better, hope he has time to answer. > > And anyhow even if you disagree with his patches to Qt dbus, there's no > reason to discard this patch, saves us a dbus call on the startup case, makes > the app start marginally faster i guess :D
Right, the patch itself looks fine and I'd say "ship it". As for the behaviour-compatibility... it's because I had not thought of it until now. And if it weren't for the QDaemonThread, it wouldn't be doing that. I guess we can discuss what to do -- back to the ML for this. - Thiago ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/124856/#review84129 ----------------------------------------------------------- On Ago. 21, 2015, 4:06 p.m., Albert Astals Cid wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://git.reviewboard.kde.org/r/124856/ > ----------------------------------------------------------- > > (Updated Ago. 21, 2015, 4:06 p.m.) > > > Review request for KDE Frameworks and Thiago Macieira. > > > Repository: kdelibs4support > > > Description > ------- > > For the first invocation, call newInstance ourselves instead of relying on > the parent process to do it. > > With the new threading patches in qt dbus the current code races as the > parent may end up calling newInstance before the child has set up the object > that handles it. > > > Diffs > ----- > > src/kdeui/kuniqueapplication.cpp 713c6f4 > src/kdeui/kuniqueapplication_p.h de4b328 > > Diff: https://git.reviewboard.kde.org/r/124856/diff/ > > > Testing > ------- > > konsole starts fine with this + Thiago's Qt dbus threading patches > > > Thanks, > > Albert Astals Cid > >
_______________________________________________ Kde-frameworks-devel mailing list Kde-frameworks-devel@kde.org https://mail.kde.org/mailman/listinfo/kde-frameworks-devel
