What is the best way to put Layer Lock on a mod-tap?

[Dimagog]

Hi,
I want to use my left Ctrl as a LayerLock key when tapped.
How do I do it right given LLOCK cannot be used with MT() directly?
I’ve managed to get it to work in a convoluted way, but I’m wondering what would be a good way to do it.

Here is how I did it for now:

1. Made the key for Ctrl/LLOCK to be MT(MOD_LCTL, KC_F17)
2. Added unused key NOT_USED
3. Modified the code in process_record_user to call LayerLock like this:

    if (!process_layer_lock(keycode, record, record->tap.count > 0 ? MT(MOD_LCTL, KC_F17) : NOT_USED)) {
        return false;
    }

The :? conditional is to prevent layer lock activation on holding Ctrl.

Another option I've considered was to process MT(MOD_LCTL, KC_F17) myself and to recursively call process_record with a synthetic record for LLOCK, but I'm not sure such re-enterancy is supported/encouraged/tolerated by QMK. And there is no function in layer_lock.c to activate the lock.

[getreuer]

Thank you for the great question! You're right, it's natural to want to combine Layer Lock with some other function on hold.

That's a slick workaround you made with that conditional arg record->tap.count > 0 ? MT(MOD_LCTL, KC_F17) : NOT_USED. This makes sense. If you like, you could use the existing KC_NO keycode for the purpose of NOT_USED.

Another option I've considered was to process MT(MOD_LCTL, KC_F17) myself and to recursively call process_record with a synthetic record for LLOCK, but I'm not sure such re-enterancy is supported/encouraged/tolerated by QMK.

Yes, I've struggled a fair bit with this approach in Achordion and Repeat Key and have some comments about it. Recursively calling process_record() does work, and the Dynamic Macros and Repeat Key core features make use of this internally. When doing this, you need to ensure your logic is such that an infinite recursion is impossible, since getting caught in an infinite recursion would make the keyboard inoperable. The other caveat is when recursing, feature handlers that run before process_record_user() will process the event twice, which may confuse things. So yeah, it's dicey and worth avoiding when there's an alternative.

Generally, for combining non-basic keys with MT, there is a pattern suggested in "changing the tap function" in the mod-tap documentation. This appears to be QMK's recommended solution to the problem. For the case of Layer Lock, the handler should, on tap, call layer_lock_invert(get_highest_layer(layer_state)) to toggle the lock on the highest active layer. I've added a section describing this to the Layer Lock documentation.