search

zmkfirmware / zmk

ZMK Firmware Repository
https://zmk.dev/
MIT License
2.83k stars 2.86k forks source link

feat: Added a flag to key press behavior to allow it to be inverted #2554

Open Nick-Munnich opened 1 month ago

Nick-Munnich commented 1 month ago

Currently if two of the same binding are pressed and then released, the first release will be treated as both keys being released.

This adds a new optional flag to the key press behavior invert-if-active causing the behavior to invert its function if the binding is already pressed:

This flag resolves this inconsistent behavior, but is marked as optional for backwards compatibility and as it is less intuitive.

It is also a quite useful interaction, in my opinion:

EDIT: I have a feeling that this may interact poorly with combos. This is a note telling me to test that interaction in the future when I get around to it.

petejohanson commented 2 weeks ago

I am on the fence on this... I'm not convinced this is the correct solution for the issue at hand. Additionally, this seems potentially useful for some scenarios, but feels really counter intuitive for a basic behavior like key press. I'm not sure I love this, versus a different behavior that does this. Need to ponder this.

Nick-Munnich commented 1 week ago

I am on the fence on this... I'm not convinced this is the correct solution for the issue at hand. Additionally, this seems potentially useful for some scenarios, but feels really counter intuitive for a basic behavior like key press. I'm not sure I love this, versus a different behavior that does this. Need to ponder this.

Yeah, I was uncertain about it too. Originally I thought about making this a separate behavior in a module because I just wanted the functionality, but quickly realised that it's essentially just a slight variant of the standard &kp behavior. That put it more into the realm of "add a new flag" for me. Then while working on #2555 I stumbled across the weird interaction, and realised that this approach would also serve to bypass the interaction, in a way.

There are alternatives to resolving the interaction, which may be something to look into, but I would disagree with making the function a new behavior. I do agree that the description is very counterintuitive, but I find it very intuitive to use in practice, if that counts for anything.