Send sequence events to child if they don't match kitty.

[ppwwyyxx]

Fix #5840

Applications that use key sequences:

• tmux works, tested by the Ctrl+b>c case in Key sequence is not sent to children if a prefix is mapped by kitty #5840
• vim works, tested by inoremap <C-b>e abcde in vimrc
• modifier sequence is replayed correctly: Send sequence events to child if they don't match kitty. #5841 (comment)

[kovidgoyal]

If you expect to have this merged, it needs to work robustly, which means dealing with release events as well. The kitty keyboard protocol is getting more and more widespread adoption which means more and more terminal programs will become capable of responding to release events.

[ppwwyyxx]

The latest commit deals with release events: when ctrl+b>l is a kitty key binding and ctrl+b>c is pressed, the show_key kitten can receive press and release events of ctrl+b and c.

[kovidgoyal]

You have to record and play back modifier key events as well, currently
they are excluded in keys.c Also you cant use the w pointer after
calling process_sequence, since that could have invalidated it. Instead
use

w = window_for_window_id(active_window_id)

[ppwwyyxx]

Sure, now modifier sequence is also replayed (using the same test case):

And the window pointer is reset.

[ppwwyyxx]

Maybe it would be a good idea to avoid duplication by moving is_modifier_key to glfw.c and exposing it to Python? Let me know if you would like that.

[kovidgoyal]

On Fri, Dec 30, 2022 at 11:04:12PM -0800, Yuxin Wu wrote: Maybe it would be a good idea to avoid duplication by moving `is_modifier_key` to `glfw.c` and exposing it to Python? Let me know if you would like that.

yes, that would be good.

[ppwwyyxx]

OK, done. Not sure if I'm doing it in a good way though.