https://bugs.kde.org/show_bug.cgi?id=519627
--- Comment #2 from [email protected] --- Git commit 83aa6a6c0925c904dc0d1f41a71b716cc8c6fe8d by Tomaz Canabrava, on behalf of Timothy Redaelli. Committed on 10/05/2026 at 07:32. Pushed by tcanabrava into branch 'master'. Refactor Kitty keyboard CSI dispatch via processToken cases Move the four Kitty control sequences (CSI ?u, CSI >Nu, CSI <Nu, CSI =N;Mu) out of the early special-case in csi_dispatch() and into processToken() switch cases via the existing token_csi_pr/pg/pl/pq encoders, matching how every other CSI prefixed sequence is handled. Bare CSI u (Restore Cursor Position) needs no explicit handling: 'u' is in neither CPN nor CPS char classes, so it falls through to the last for-loop branch and dispatches as token_csi_ps('u', 0) -> restoreCursor(), unchanged from before the Kitty work. The for-loop branches for </=/> prefixes now forward params.value[0/1] (instead of 0,0) and break to avoid duplicate dispatch on multi-param sequences. Preexisting consumers (token_csi_pq('c'), token_csi_pg('c'), token_csi_pg('q')) ignore p/q so the change is transparent for them. Architectural pattern suggested in !1212. M +12 -22 src/Vt102Emulation.cpp https://invent.kde.org/utilities/konsole/-/commit/83aa6a6c0925c904dc0d1f41a71b716cc8c6fe8d -- You are receiving this mail because: You are watching all bug changes.
