labwc/include/input/key-state.h
tokyo4j 885db240f3 keyboard: simplify key state handling
We used to store both `pressed` keys and `bound` keys and derive
`pressed_sent` keys from them, which are notified to the client when
switching focus. But I think this was overengineered and we can remove
`pressed` keys and `bound` keys. Instead we now directly modify
`pressed_sent` keys just before calling `wlr_seat_keyboard_notify_key()`.

Technically we could even remove `pressed_sent` keys as they are
duplicated in wlr_seat->keyboard_state.keyboard->keycodes, but we keep
`pressed_sent` keys as it is easier to access.

The only place which required `bound` keys was `handle_modifiers()`.
This function checked `bound` keys to keep widnow switcher alive when
Alt-Tab is pressed and only the Alt modifier is released. The window
switcher is then finished when Tab key is released. I replaced the check
with `(seat->keyboard_group->keyboard.num_keycodes > 0)`. This slightly
changes the behavior; when Alt-Tab and any other key e.g. X are pressed
and Tab key and Alt modifier are released, the window switcher is now
kept alive. But I don't think this breaks any workflows for users.
2026-05-04 17:01:36 +09:00

16 lines
474 B
C

/* SPDX-License-Identifier: GPL-2.0-only */
#ifndef LABWC_KEY_STATE_H
#define LABWC_KEY_STATE_H
struct seat;
/*
* All keycodes in these functions are (Linux) libinput evdev scancodes which is
* what 'wlr_keyboard' uses (e.g. 'seat->keyboard_group->keyboard->keycodes').
* Note: These keycodes are different to XKB scancodes by a value of 8.
*/
void key_state_indicator_update(struct seat *seat);
void key_state_indicator_toggle(void);
#endif /* LABWC_KEY_STATE_H */