Can't make it work with other language

[jsirex]

Am I missing something simple, but I can't switch language to another.

If I just change language from US to, for example, RU, I still see US letters.
If I try to choose something from keyboard -> keymap menu (like ru), I see Cyrillic_<XXX> instead.

Linux, Debian/testing

[AlynxZhou]

I am not sure whether you are talking about UI translation or symbols on different keymap.

We don't have RU translation for the interface, because I don't know that language, I need help.

And if you mean keys, they should be what they looks like on the keyboard, but for symbols only on specific keyboard, I don't know them, so they will just show XKBcommon key names. In this case you can tell me what symbol it is on your keyboard, and I'll add it.

[jsirex]

If I just change language it show English letters as on keyboard:

If I change keymap to ru, codes are displayed:

[AlynxZhou]

You need to tell me which character the key name is mapped, and then I can manually map them in code.

[jsirex]

I quite don't understand what are you asking about?

I have 2 languages installed in the system: US and RU. When I switch to RU I see Cyrillic_<XX> instead of just russian symbols. Mapping? IDK.. For example under buttons qwerty in russian layout it йцукен.

[AlynxZhou]

When I switch to RU I see Cyrillic_<XX> instead of just russian symbols.

I know nothing about russian symbols, nor xkbcommon, right? So if you want the app to show russian symbols, you need to tell me which russian symbol is from which key, so I could then replace things like Cyrillic_i into the symbol you need.

[AlynxZhou]

To be clear, I need a table like this:

code	char
Cyrillic_i	What you get in a text editor when you type that key
Cyrillic_<XX>	...

[jsirex]

Hi, the table will be really hard. It is in

grep Cyrillic /usr/include/X11/keysymdef.h

I'm not a pro here, so I start digging with LLM. Below I summarized what I've understood.

What is that “Cyrillic_sha / Cyrillic_de …” table?

On Linux (X11, and often Wayland compatibility layers), keyboard input works in layers:

1. Physical key → keycode
2. Keycode → keysym (symbolic name, e.g. Cyrillic_sha)
3. Keysym → Unicode character (ш)

showmethekey is showing you the keysym name, not the final Unicode character.

So:

• Cyrillic_sha → Unicode ш
• Cyrillic_de → д
• Cyrillic_u → у
• Cyrillic_shorti → й
• Cyrillic_tse → ц

What is happening (precisely)

code explains exactly why you see Cyrillic_sha instead of ш.
This line is the root cause:

xkb_keysym_get_name(xkb_key_sym, main_key, XKB_KEY_SYM_NAME_LENGTH);

xkb_keysym_get_name() returns the symbolic keysym name, not the Unicode character.

For example:

Keysym	Returned string	Unicode
XKB_KEY_Cyrillic_sha	"Cyrillic_sha"	ш
XKB_KEY_a	"a"	a

So for Latin layouts it looks correct by accident, but for Cyrillic layouts you get the keysym identifier, which is exactly what you’re seeing.

Nothing is “wrong” with your layout — the app is just stopping one layer too early.

Instead of (or in addition to) getting the keysym name, it should convert the keysym to Unicode.

With xkbcommon, the correct function is:

int xkb_keysym_to_utf8(xkb_keysym_t keysym, char *buffer, size_t size);

---

My conclusion, that probably any non-latin (non-english?) languages have these issues and it could be fixed at once for all.

[jsirex]

also from llm untrusted code:

replace this:

main_key = g_malloc(XKB_KEY_SYM_NAME_LENGTH);
xkb_keysym_get_name(xkb_key_sym, main_key,
                    XKB_KEY_SYM_NAME_LENGTH);

With something like:

main_key = g_malloc(8); // UTF‑8 max for one codepoint
int n = xkb_keysym_to_utf8(xkb_key_sym, main_key, 8);
if (n <= 0) {
    // fallback to symbolic name
    g_free(main_key);
    main_key = g_malloc(XKB_KEY_SYM_NAME_LENGTH);
    xkb_keysym_get_name(xkb_key_sym, main_key,
                        XKB_KEY_SYM_NAME_LENGTH);
}

[AlynxZhou]

That sounds correct and I'll do it in next few days (maybe, I am writing new feature of another project).