question: how to map from xcb key press/release event state field to xkb_mod_mask_t

[wez]

To deal with some timing issues in certain keyboard automation or high speed keyboard input scenarios when mixing an IME with xkb, my application is following the suggestion in https://github.com/ibus/ibus/issues/2600#issuecomment-1904396417 to compute and call update_mask with the state supplied in the incoming X event.

At the moment I am literally passing these u32 value straight from the X event and having those take precedence over the the parameters values last provided in an UpdateState X event.

This seems to be working well for the most part: automated key events (either via xdotool, or programmable mechanical keyboards) now appear to be handled correctly.

Is this actually valid? Are the key/button mask bits in the key press/release event state field actually directly compatible with the xkb_mod_mask_t in this context, or should I be taking additional steps to decode and map the values? Or just mask off some bits?

Some background/context:

• https://github.com/ibus/ibus/issues/2600#issuecomment-1904396417
• https://github.com/fcitx/fcitx5/issues/893

[wismill]

@wez check the interactive-x11 tool for a complete example of using xcb with xkbcommon: it passes xcb_xkb_state_notify_event_t masks to xkb_state_update_mask.

[bluetech]

Is this actually valid? Are the key/button mask bits in the key press/release event state field actually directly compatible with the xkb_mod_mask_t in this context, or should I be taking additional steps to decode and map the values? Or just mask off some bits?

Yes, this is documented and if we break it, it will be considered a breaking change.