antlarr added inline comments. INLINE COMMENTS
> graesslin wrote in composite.h:242 > I would just count down the m_maxFramesTestedForSafety till it reaches 0 ok > graesslin wrote in x11_platform.cpp:215 > You can use QMetaObject::invokeMethod with Qt::QueuedConnection. Oh, I thought QueuedConnection needed to get to the event loop to execute the slot method, but I just noticed it's the event loop **of the receiver object**, so it would be fine. I'll change that. > graesslin wrote in x11_platform.cpp:229-232 > I think you can merge the PostFrame with PostInit. So that the init test also > does the freeze testing. That would also solve the conceptual problem I > pointed out above. Do you prefer to merge PostInit with PostFrame or with PostLastGuardedFrame? The difference would be that if it's merged with PostFrame and a user sets KWIN_MAX_FRAMES_TESTED to 0 (so no freeze detection is done when painting frames) then the detection thread and timer wouldn't be deleted and if it's merged with PostLastGuardedFrame, then a new thread and timer will be created for the frame drawing. I would merge it with PostFrame, since it requires a user who probably knows what he's doing to interact and is more optimized for 99.9999% of cases, and still an idle thread and stopped timer shouldn't consume any resources in the rare case a user sets that, but I ask just in case. REVISION DETAIL https://phabricator.kde.org/D3132 EMAIL PREFERENCES https://phabricator.kde.org/settings/panel/emailpreferences/ To: antlarr, #kwin, #plasma, davidedmundson Cc: luebking, graesslin, kwin, plasma-devel, davidedmundson, lesliezhai, ali-mohamed, jensreuterberg, abetts, sebas
