https://bugs.kde.org/show_bug.cgi?id=510913
Duncan <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #1 from Duncan <[email protected]> --- OK, I've tried twice to reply and both times my reply got too complex and went off into the weeds. Trying for simpler... I have the problem here too, for a very long time... and I have a theory too, tho (I think) it differs a bit from yours. AFAIK this has been a *very* long term issue, since 2020/covid at least I believe as I used the covid down-time to switch to wayland and am quite sure I had it then (on X pre-switch). But if my memory is correct, the problem goes back much further than that, very possibly to the kde4 era and the origin of gwenview, so it's likely the problem was in the original code. =:^( Comparing systems: My core hardware is /ancient/, decade-plus old 6-core/6-thread AMD fx6100, only a bit newer AMD Radeon Polaris-11 (says dmesg, I've long forgotten the model number but IIRC its an rx4xxx series), 16 gig RAM, SATA3 storage tho now with SSDs attached... Software is Gentoo ~amd64 qt-6.10.2, but running live-git kde frameworks/plasma/gear, wayland as mentioned. With this old a bug that doesn't matter much except to confirm it's not fixed yet in live-git. As might be expected from my older/slower system I see the problem here with *much* smaller images than the 24 Mpx you mention, here down to sub-Mpx at times. Here it's not just the arrow keys affected, but at least the registered navigation shortcut keys are affected as well. (This matters because they have slightly different high-level code-paths, tho they converge at a lower level.) Additionally, it triggers well before key-auto-repeat would kick in. Without attempting the waaayyyy too detailed off-in-the-weeds explanation this time, I can note that htop shows gwenview with 13 threads, two of which are (input) event processing. The randomly-works/randomly-double-triggers spottiness of this smells (to this experienced but non-dev admin) like a thread-race bug, and of course qt's many-to-many signal/slot/event mechanisms make a missed or too-late-in-threaded-context event-accept a real possibility, in terms of possible double-triggers. -- You are receiving this mail because: You are watching all bug changes.
