There's something that leaves a blur on the screen. See the attached image, where the cursor is. The blur goes over any windows, but underneath the cursor. The blur goes away with a 'kwin --replace'. I'm not sure exactly what causes the blur. It only seems to happen after the screen saver/screen sleep has been active for a while (e.g. if I leave the computer on overnight). However, that's about the location where notifications appear, so maybe it's due to one of them that pops up while the screensaver/screen sleep is active. Reproducible: Sometimes Steps to Reproduce: 1. Leave computer for some time. 2. Come back to find blur.
Created attachment 74602 [details] Shows the blur on the upper left part of the screen -- above a window (obscuring it) and below the cursor.
please provide output of qdbus org.kde.kwin /KWin supportInformation when the problem is visible
and afterwards check whether it's also reproducible when disabling caching in the blur settings (looks like what i inintially ran into reg. bug #307112 but could not reproduce afterwards)
Thomas, could you provide more specific instructions? If I go to configure 'Blur' in the Desktop Effects system setting, I don't see anything about disabling caching, and I didn't see directions in skimming that bug report.
"Safe intermediate rendering results" - it's the only checkbox in the config dialog, below the slider.
Created attachment 74726 [details] Output from qdbus org.kde.kwin /KWin supportInformation
Ok, I've run in to the blur again, so I've attached the requested output. I've now disabled the blur caching, and I'll see if I run in to this again.
> kwin4_effect_fade > kwin4_effect_blur Actually sounds related to bug #306186 now (unless the fade was in action for sth. entirely different, like a popup) But also actually, the blur should take the opacity ("0") into account then.
The bug still happens with blur caching disabled. FWIW, I've also run in to that issue with the password dialog box; it was common for some time, but I haven't seen it in a little while now.
given that Bug #306186 has been fixed for 4.10: please provide an update whether it solved the issue once you tested out 4.10.
Will do. I may try out one of the release candidates before the final release.
Let's do it different: I assume it's fixed. In case you are still able to reproduce once you have had the chance to give 4.10 a try, please reopen. I think that's easier as you cannot properly say that it is fixed given that there is no concrete way to reproduce it.