Add SplitKB Aurora keymaps

[bobobo1618]

Sorry if I'm not going about this the right way. These are for SplitKB's variants of Corne and Sweep. The keyboard names should match ZMK. Sweep is already committed and a branch for Corne is in the works.

[manna-harbour]

Nice!

It looks like you haven't set up the author in git?

Also, please add the following lines to .github/workflows/test-all-promicro-shields.yml in the appropriate place:

        "splitkb_aurora_corne_left","splitkb_aurora_corne_right",
        "splitkb_aurora_sweep_left","splitkb_aurora_sweep_right",

[bobobo1618]

The lack of author is deliberate, I don't really like putting an email address or name in my public commits because it's resulted in spam before. Do you require an author or are you okay with it as-is?

Added the keyboards to the workflow.

However in testing I've run into a really weird problem: on Sweep at the very least, toggling external power or RGB only affects the left board. It's a key combo that requires pressing keys on both boards, so they're clearly communicating, but something about how Miryoku does it seems to break things.

When I flash a basic keymap for testing, everything works as it should again, then when I flash Miryoku back, it's broken again.

I've checked the debug and Miryoku builds and for both, the config and KConfig are identical, so the problem isn't there. I'm building locally with the same Zephyr and ZMK sources, so there's no difference there. The only apparent change is the DTS and I have no idea where to look.

Any idea where to look?

[manna-harbour]

The lack of author is deliberate, I don't really like putting an email address or name in my public commits because it's resulted in spam before. Do you require an author or are you okay with it as-is?

I don't suppose it would cause any problems, and if I include the PR id in the commit then it can still be attributed. However in that case I'd probably change the author to myself for aesthetics and to prevent any possible weirdness with GitHub.

But if the only reason is spam, you can just use the GitHub noreply address. That works well for this purpose.

[manna-harbour]

However in testing I've run into a really weird problem: on Sweep at the very least, toggling external power or RGB only affects the left board. It's a key combo that requires pressing keys on both boards, so they're clearly communicating, but something about how Miryoku does it seems to break things.

There's a ZMK issue where locality isn't preserved if the keycode is not used directly in the keymap: zmkfirmware/zmk#1494. In this case EP and RGB are not source specific, but they do use global locality so that the actions are invoked on both sides of a split, so it's likely that the global information is also being lost. To test, replace &u_out_tog with &out OUT_TOG in https://github.com/manna-harbour/miryoku_zmk/blob/master/miryoku/miryoku_babel/miryoku_layer_alternatives.h. If that's the case I'll have to remove the shift functions for those keys, and make them optional for use on non-spits as with https://github.com/manna-harbour/miryoku_zmk#double-tap-boot.

[bobobo1618]

Yep, zmkfirmware/zmk#1494 was the problem. bobobo1618@6ba964f fixed it for me, though the shift combos no longer work of course.

What would you prefer I do with the author? I can change it to you, to GitHub noreply, or leave it. Up to you.

[manna-harbour]

What would you prefer I do with the author? I can change it to you, to GitHub noreply, or leave it. Up to you.

I'd recommend that you use the GitHub noreply address in general. If you don't want to do that here then please change it to me Manna Harbour <51143715+manna-harbour@users.noreply.github.com>.

[bobobo1618]

Done, author updated

[manna-harbour]

Thanks, merged now.

zmkfirmware/zmk#1504 and zmkfirmware/zmk#1520 are both in ZMK main now.

Here's a sample build run: https://github.com/manna-harbour/miryoku_zmk/actions/runs/3492778615.

[manna-harbour]

Yep, zmkfirmware/zmk#1494 was the problem. bobobo1618@6ba964f fixed it for me, though the shift combos no longer work of course.

Here's a workaround. It's in a temporary branch for now. https://github.com/manna-harbour/miryoku_zmk/tree/globalshiftfunctions-tmp#global-shift-functions