Implementing kind-of quasimode (introduced by Jef Raskin)

[santhoshr]

Requirements

• I've searched platform-specific issues, issues and discussions to see if this has been reported before.

Describe the bug

"not" operator doesn't work with list of keys, for e.g.

(defcfg
  process-unmapped-keys yes
  concurrent-tap-hold yes
)

(defsrc caps)
(deflayer default
    @swh
)

(defalias
  swh (switch
    ((not a b c )) S-a break
    () XX break
  )
)

curiously not works with input real
((not (input real a))) S-a break

Is there a way to check previous key input? for e.g. to check if I have pressed any of given list of keys and they can be pressed individually or in combination, for e.g. if a-z letters are configured and when switch returns I can either press escape or press return based on switch evaluation?
Key-history will not work because it will take single key and one should also give recency, if I'm typing any word then I cannot predict where 'a' comes also I need list of keys. Key-timing will not work because if I use switch for processing keys while holding some other key, if I hold alt key and use this logic key-timing will take alt key into account.

Relevant kanata config

No response

To Reproduce

Given as part of description

Expected behavior

"not" operator working

Kanata version

1.6.1

Debug logs

No response

Operating system

Windows 10

Additional context

No response

[jtroo]

Try (not (or a b c))

[jtroo]

Random thought, perhaps or should have the alternate name "any", likewise and could have the alternate name "all"

[santhoshr]

I realized "not" works on "key press", any way to make it work on key release?

[santhoshr]

Random thought, perhaps or should have the alternate name "any", likewise and could have the alternate name "all"

any / all & or / and won't make much difference "all" could even cause confusion

[jtroo]

I realized "not" works on "key press", any way to make it work on key release?

It's not clear to me exactly what you mean. My guess is you're looking for (on-release tap-vkey vk-switch) where vk-switch is defined to be the switch action you want.

[santhoshr]

With key-history I was able to achieve pseudo condition to implement kind-of quasimode (introduced by Jef Raskin).

Bit of explanation below,
Quasimode keys works like shift, one key is held while typing other keys. So I want some of my keys to have this behaviour, one such functionality is Search, on pressing capslock and typing I want to search, and on release it should trigger enter or escape based on whether I typed or not, since we are working with keyboard we can't check application input box if it is empty or not to trigger enter or escape. Running the below code and opening a browser page, holding capslock will allow for kind-of quasimode search, and on release, it would trigger escape or return.

(defcfg
	process-unmapped-keys yes
	concurrent-tap-hold yes)

(defsrc 
					caps)
(deflayer default
					@fin)

(defalias
	fin (multi (on-press tap-vkey cf) (on-release tap-vkey fc)))

(defvirtualkeys
	cf  C-f
	es  esc
	en  ret
	fc (switch
			 ((key-history f 1)) esc break
			 ((key-history ret 1)) esc break
			 () ret break)
	no (switch
			 ((key-history f 1)) esc break
			 ((not a b c d e f g h i j k l m n o p q r s t u v w x y z 0 1 2 3 4 5 6 7 8 9 lsft rsft)) esc break
			 ((key-history ret 1)) esc break
			 () ret break))

In the above,

• On alias @fin if you replace fc with no ( fin (multi (on-press tap-vkey cf) (on-release tap-vkey no))) ), it will work but you have to "hold" one of the letters to get return, to test open notepad, hold search key, start typing now release it will trigger esc not return because switch case fails in "not" since none of the key in the list is on "pressed" state, I have included lsft and rshft in the list so if you type after holding search key now before release if you hold shift then it will trigger ret properly,
• return is not required in incremental search in modern browsers but application like notepad you need to trigger manually
• is there better way to implement this in kanata

[santhoshr]

In the above comment switch case not fails but passes and triggers escape, since not itself is inverse logic, even though we have pressed search characters, not condition should not have passed, to get the condition fail, you have to hold one of the keys in list for e.g. lsft

[gerhard-h]

consider caps also sends you to an seach-layer where every key say e is replaced by (multi e nop1) now you can test for (key-history nop1 1) or in case of timing issues also test for (key-history nop1 2). This has the added benefit to be able to search for words ending on f and words with special characters