https://bugs.kde.org/show_bug.cgi?id=517499
--- Comment #8 from Branislav Klocok <[email protected]> --- Thanks for the detailed week of data — this is a valuable independent corroboration. It confirms the bug is not specific to my setup: you're on AMD iGPU / Arch / Plasma 6.6.5, I'm on Intel UHD 620 / openSUSE Tumbleweed / Plasma 6.6.2, same symptom on both. In particular your 09:21 capture independently confirms comment 6: the panel's screen index gets re-resolved to a different output at runtime while docked, with no drm/connector event and no suspend in between. So the root cause is plasmashell pinning the panel to a screen *index* that can silently re-enumerate, rather than resolving it from KWin's current priority ordering. Your workaround (derive the priority-1 output from kscreen-doctor, then locate it in plasmashell by *position* rather than by screen index) makes sense given that positions survive the re-enumeration that flips the indices — useful pointer for anyone hitting this in the meantime. Hopefully the maintainers can use the two data points to pin down whether the proper fix is (a) re-introducing robust per-arrangement profiles or (b) having plasmashell always resolve the panel screen from KWin's live priority/ordering. -- You are receiving this mail because: You are watching all bug changes.
