https://bugs.kde.org/show_bug.cgi?id=517259
--- Comment #2 from Prajna Sariputra <[email protected]> --- I tried running a git bisect between KWin 6.5.5 and 6.6.2, which gave me the following output: dd43d93ccf6c4cd8a4f322fd822c98c3d2893c35 is the first bad commit commit dd43d93ccf6c4cd8a4f322fd822c98c3d2893c35 (HEAD) Author: Vlad Zahorodnii <[email protected]> Date: Mon Oct 13 17:02:48 2025 +0300 Make Workspace::windowActivated() signals more reasonable If workspace()->activateWindow() is called and there is another active window, the following signal sequence will be observed: - Workspace::windowActivated(nullptr) - Workspace::windowActivated(anotherWindow) The windowActivated() signal with nullptr window is unexpected. This change also untangles the dependency order between Window::setActive() and Workspace::setActiveWindow(). Fullscreen related code has been removed from Workspace::setActiveWindow() because it matches the logic in Window::setActive(). src/activation.cpp | 22 ++++++---------------- src/events.cpp | 8 +++++--- src/layershellv1window.cpp | 2 +- src/window.cpp | 1 - src/xdgshellwindow.cpp | 2 +- 5 files changed, 13 insertions(+), 22 deletions(-) Unfortunately reverting the above commit on KWin 6.6.2 doesn't work due to conflicts, but I can confirm that both the issue in the original report and the other variant of the issue I encountered in comment 1 disappears if I build KWin at the point right before the above commit, and that both issues appear if I rebuild KWin with that commit added. Both issues are also present in KWin git master at commit 8a5d6d97e729cbcdd2ac3f25b913a2f593712b01 dated Mon Mar 9 01:49:54 2026 UTC time. -- You are receiving this mail because: You are watching all bug changes.
