https://bugs.kde.org/show_bug.cgi?id=429400
Václav Ryšavý <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #10 from Václav Ryšavý <[email protected]> --- Can confirm this bug still reproduces with a current stack: kdeconnect-kde 26.04.3, Plasma 6.7.2 (Wayland session), xdg-desktop-portal-kde6 6.7.2, KDE Connect Android 1.35.9, openSUSE Tumbleweed. Active desktop keyboard layout is "us" (confirmed via kxkbrc / KWin layout D-Bus call), a secondary "cz" layout is also configured but inactive. Symptom: "@" is received as "2", "#" is received as "3" - matching the "symbol -> digit" pattern already described in comment #6 ([] -> 89, {} -> numbers), just on different keys. Traced the receive path with QT_LOGGING_RULES="*.debug=true" on kdeconnectd: the mousepad plugin authenticates against the RemoteDesktop portal and logs "Connected to ei" - so on this stack it goes through xdg-desktop-portal-kde -> libei, not the legacy /dev/uinput backend (confirmed no fd open on /dev/uinput during the test). No further per-keystroke keysym/keycode detail is exposed via Qt logging categories, so I couldn't pinpoint whether the mis-mapping happens in the Android app's key encoding or in kdeconnect-kde's char->keysym translation before it reaches libei. -- You are receiving this mail because: You are watching all bug changes.
