https://bugs.kde.org/show_bug.cgi?id=378010
--- Comment #8 from David Edmundson <k...@davidedmundson.co.uk> --- That trace has a few interesting things in it: - we're still losing a tonne of time (30% or so) processing X events. We're getting over 670,000 of them in 30 seconds. They end up costing a lot because it then goes through all the nativeEventFitlers in KDE code that do a somewhat pointless string compare.. but the root question is what the hell are these events, and where's the real bug? That's come up before: https://bugs.kde.org/show_bug.cgi?id=373368 - we're losing 7% in some X window updating it's properties 313 times in 30 seconds. I've not seen that before. That window then propogates through the task manager changing, which builds up. Qt 5.9 will fix a lof of that propogation but again the root question which window is changing and what and why? Can you put some debug in KWidnwSystem::windowChanged and find out? Once we know we can break that out into a new bug. - we're losing another 15% in some JS evaluation. It's quite possibly related to the issue above; the evaluation needs to have some trigger. QML profiler could confirm that. -- You are receiving this mail because: You are watching all bug changes.
