Layer toggle behind layer hold

[lobre]

I am having a hard time trying to accomplish the following.

I have the [main] layer. On this layer, I do have a layer-hold on space which leads me to a [headside] layer. To note that on [main], I also have tab which does not have any mapping assigned, so it regularly inputs a tab character.

On [headside] (while keeping space maintained), I would then like to have the tab key toggling back and forth on tap (and not hold) between another layer [tailside] and this [headside].

And so if at any moment I decide to release space, I should go back to [main] whatever if [headside] or [tailside] was active.

As an example:

1. From [main] I hold space,
2. Landed on [headside], I type some keys that are mapped on this layer,
3. I single tap tab (while keeping space held) which activates [tailside],
4. I type some keys that are mapped on [tailside],
5. I single tap tab (while still keeping space held), which re-toggles [headside],
6. ... <loop from 3 to 5 as long as needed>
7. I decide to go back to [main] by releasing space.

Do you think such a behavior (mix of hold and tap) is achievable in keyd? Help would be greatly appreciated, and heads up for this awesome project!

[nsbgn]

The following sort-of achieves what you want:

[main]
space = overload(headside, space)

[headside]
tab = swap(tailside)
a = b

[tailside]
tab = swap(headside)
a = c

However, since space is a tricky key to overload, you probably want at least space = overloadt(headside, space, 200) or lettermod(headside, space, 100, 200) instead of plain overload. Unfortunately, swap doesn't seem to work with any of these. @rvaiya, that would be a bug, right?

[lobre]

Thank you, at first sight, I confirm that this behavior does what I expect.

But I notice the same bug as you. It is fine with overload, but not with overloadt 😐

[lobre]

I read the code a little bit yesterday, and I have the feeling that overloadt (and overloadt2) should have theoretically the same behavior as overload, apart from the timeout. Am I guessing right @rvaiya?

I could not yet identify where the bug is. I need to dig more but as I am not familiar with the codebase, it is hard for me to catch it for the moment.

[matcls]

changing this line
from
for (i = 0; i < CACHE_SIZE; i++)
to
for (i < 0; i <= CACHE_SIZE; i++)

makes swap work here.

[lobre]

@matcls , you mean this instead no?

 for (i = 0; i <= CACHE_SIZE; i++)

If so, do you understand why looping one step further is necessary? I am not sure to exactly grasp what is the purpose of this cache.

[rvaiya]

Sorry for the slow response, I haven't had much time to triage issues lately. This was indeed a bug and should be fixed in the latest commit.

[lobre]

No worries, open source does not need reactivity! Thank you a lot for fixing the bug.

I just tested with the commit you introduced, and it effectively resolves my issue 🥳.

I am closing this issue.