International Keycodes support for windows

[eokoshi]

I use a Japanese keyboard layout on Windows.
I would like to ask why international keycodes (INT1, INT2, etc...) are not supported on windows?
It was no issue in QMK... I would appreciate any guidance.
Mainly, I would like to be able to type / backslash and _ underscore...

[caksoylar]

Can you try using them and report? This was last tested 5 years ago in #597 and it seems that the user did not use CJK keyboards which these might apply to.

[eokoshi]

Thanks very much for your reply.
The issue is that there is no feedback whatsoever. When I press, its like I didnt press anything at all.
Do you have a specific keylogging program you recommend?

I have tried putting them in my keymap as keycodes.

&kp INT1

and also like so (saw this in a post)

#define JP_BSLH 0x87
#define JP_UNDER LS(0x87)

I can confirm that the issue also applies to other INTERNATIONAL keycodes, as I use INT5 in my keymap as well. (no effect)

In the referenced commit, it marked these codes as "unavailable" on Windows, but why is this?
Is this a windows issue, or a ZMK issue, or a me issue?

QMK is able to use KC_INT1, etc, without issue.
LLM said its due to wireless HID stuff... but then the rest of the response was hallucinated garbage so 🤷‍♀️

[Genteure]

I have tested on Windows with the Microsoft Japanese IME:

• かな = LANG1
• 英数 = LANG2
• 変換 = INT_HENKAN
• 無変換 = INT_MUHENKAN

I rarely use Japanese IME so not an expert. My understanding is this is only one out of many key codes mapped to each feature.
Other OSs and IMEs might use different key codes by default.

The Japanese community seems to have this figured out but I can't really find them documented anywhere. It would be nice to have them clearly listed on ZMK docs.

[eokoshi]

I searched japanese-language posts about this, but I could not find anything that solved the issue of not being able to use the backslash/underscore/ろ key (INTERNATIONAL_1, 0x87)

I tried setting MUHENKAN as INT5 to no avail, so are you saying that it only works if INT_MUHENKAN is used?

I see that LANG1 and LANG2 are documented as korean and LANG3-5 are for japanese, but is this incorrect?

[Genteure]

are you saying that it only works if INT_MUHENKAN is used

No INT_MUHENKAN, INT5 and INTERNATIONAL_5 are exactly the same. Do you have custom HID kconfig tweaks? https://zmk.dev/docs/config/system#hid
I have everything set to default.

I see that LANG1 and LANG2 are documented as korean and LANG3-5 are for japanese, but is this incorrect?

I would ignore the current docs on these since IME key codes are less tested, and it also depends on what the IME is configured to use. On my Ubuntu setup with Mozc, LANG1 and LANG2 don't do anything but INT_HENKAN and INT_MUHENKAN seem to work.

It would be really easy to confirm and improve the docs if someone with an off the shelf commercial Japanese keyboard run a HID capture on it to check the HID usages.

Searching on GitHub LANG path:*.keymap and INT path:*.keymap I see people have these in their keymap: LANG1, LANG_HANJA (LANG2), INT1, INT_YEN (INT3), LS(INT3), LS(INT1)

[Genteure]

There are also a few variants of #define JP_ going around, for example:

• https://github.com/johtani/sofle-v2-oled-zmk/blob/102df043ecbe6562e6a6d5e9a64fc8e23cfe95cb/config/sofle.keymap#L18-L56
• https://github.com/ataruno/zmk-MoooseMini/blob/74cf9523cc5a4d33cc7d7530d6638d7fdb108540/config/MoooseMini.keymap#L17-L53
• https://github.com/sonohenNoHito001/zmk-keyboard-toucan/blob/192995cf9a3e0aba7c1a4115c6e5180e7954ae4e/config/toucan.keymap#L6-L30

[eokoshi]

I really appreciate your response.

I checked those three links:

1. does not actually use backslash or underscore in their keymap (how??)
2. and 3. both define a mod-morph which seems to be incorrect??

        jis_symbole_unsco_backslash: jis_symbole_unsco_backslash {
            compatible = "zmk,behavior-mod-morph";
            label = "JIS_SYMBOLE_UNSCO_BACKSLASH";
            bindings = <&kp JP_UNDERSCORE>, <&kp JP_BACKQUOTE>;

            #binding-cells = <0>;
            mods = <(MOD_LSFT)>;
        };

backquote instead of backslash?

I also tried the LS(0x87) mapping and it did not work for me...

I just checked and config settings that could mess with HID and i saw that i have CONFIG_ZMK_HID_REPORT_TYPE_NKRO=y
looking at the docs in the link you sent, it actually does say that this may mess with INTL1-8 codes!!!

I will test to see if removing this setting fixes it.
I guess it never showed in my searches because I was searching for INT and INTERNATIONAL rather than INTL

[peterjc]

Can you check the ANSI vs ISO vs JIS setting of the various keyboards you are testing, both QMK and ZMK? That will make a difference on how some of the letters/symbols would normally be sent.

[eokoshi]

The problem seems to have been NKRO after all.
Disabling this in my settings allows for proper function.
I first tried setting CONFIG_ZMK_HID_KEYBOARD_NKRO_EXTENDED_REPORT=y as suggested in the docs, and this did finally enable me to type / and _, but / was still kind of buggy.
I disabled NKRO completely CONFIG_ZMK_HID_REPORT_TYPE_NKRO=n and now it is working great.

\\\\_____\\\\

[caksoylar]

Thanks for reporting. We need to fix the docs for https://zmk.dev/docs/config/system#hid to refer to INT keycodes correctly. It would also be nice to fix the Windows column in keycodes compatibility, though ideally I'd also like to add a warning footnote to those regarding NKRO.