Passthrough unknown key codes

[jyn514]

Currently, except for a few select mouse buttons, keyd does not support remapping keys with an event code higher than 256. This is reasonable and I don't propose changing it. However, right now it discards all keys that it doesn't know how to remap. That's not ideal and it prevents using keyd with arbitrary USB devices.

This does two things:

• Passes through unknown event codes without modifying them.
• Changes the internal representation of keycodes from uint8_t to uint16_t. Otherwise, we simply can't store the keycodes, they wrap around.

I tested this with a Forty4 GC201 game controller and Celeste through Steam; before this change it did not register any inputs, and afterwards it works fine.

cc #842 (comment), hopefully this addresses some of your concerns there.

[Nekotekina]

I think you don't need to increase internal management arrays for simply passing them events through though.

[rvaiya]

I am aware of this limitation and chose to make keyd codes a single byte. A large number of evdev codes are not actually used in practice and only exist for obscure or obsolete devices. This blows up the memory requirements for little gain.

If there is a specific problem you have related to this, please create a dedicated issue. The other PR you linked appears to be unrelated.

[workingjubilee]

...huh. Are gamepads and art tablets "obscure", or "just something rvaiya doesn't use"? Everyone I know who does even small amounts of digital drawn art has a tablet and stylus combo of some kind...

[Nekotekina]

I couldn't resist it and implemented it in my fork, see latest commit.
https://github.com/Nekotekina/keyd/tree/master
@jyn514 @workingjubilee Please test if you're interested. Remapping support needs some more work (filling constants).

[Nekotekina]

By the way, memory usage:

It's literally nothing, I also used some bitfields to pack structs better.