Re[2]: wrong keyboard layout

[Phil Krylov]

Hello Shachar,

Tuesday, August 12, 2003, 12:15:51 AM, you wrote:

SS> Dmitry Timoshkov wrote:
> > BTW, what wrong happens if your keyboard was misdetected? In the current
> > CVS all national characters should work fine regardless what keyboard
> > layout was detected. Is that only a warning message which makes you worry?

SS> Does this also apply to UTF-8? I, for one, still can't get any 
SS> characters recognized when the locale is UTF-8.

With the current CVS, I can't get any Cyrillic input with any setting
of LANG: ru_RU.KOI8-R, ru_RU.UTF-8, ru_RU.CP1251, and en_US.UTF-8.
Only ? characters are input.

wrong keyboard layout

[Dmitry Timoshkov]

"Phil Krylov"  wrote:
> With the current CVS, I can't get any Cyrillic input with any setting
> of LANG: ru_RU.KOI8-R, ru_RU.UTF-8, ru_RU.CP1251, and en_US.UTF-8.
> Only ? characters are input.

Even if you set LC_ALL=ru_RU.KOI8-R? A good test is to run 'xev' and
see whether it "works" for you.

[Phil Krylov]

Hello Dmitry,

Tuesday, August 12, 2003, 3:23:14 PM, you wrote:

DT> "Phil Krylov"  wrote:
> > With the current CVS, I can't get any Cyrillic input with any setting
> > of LANG: ru_RU.KOI8-R, ru_RU.UTF-8, ru_RU.CP1251, and en_US.UTF-8.
> > Only ? characters are input.

DT> Even if you set LC_ALL=ru_RU.KOI8-R? A good test is to run 'xev' and
DT> see whether it "works" for you.

Ah, it must be my fault - I used LANG instead of LC_ALL. Using
LC_ALL=ru_RU.UTF-8 and your latest patch (ignoring XK_ModeSwitch), my
keyboard started working perfectly.

Thanks!

Re[3]: wrong keyboard layout

[Phil Krylov]

Hello,

Tuesday, August 12, 2003, 7:14:16 PM, I wrote:

PK> Ah, it must be my fault - I used LANG instead of LC_ALL. Using
PK> LC_ALL=ru_RU.UTF-8 and your latest patch (ignoring XK_ModeSwitch), my
PK> keyboard started working perfectly.

sorry, i made a mistake. ru_RU.KOI8-R works, while ru_RU.UTF-8 (which
is my default setting) and en_US.UTF-8 don't.

PK> Thanks!

Re[4]: wrong keyboard layout

[Phil Krylov]

Hello,

Tuesday, August 12, 2003, 7:46:14 PM, I wrote:

PK>> Ah, it must be my fault - I used LANG instead of LC_ALL. Using
PK>> LC_ALL=ru_RU.UTF-8 and your latest patch (ignoring XK_ModeSwitch), my
PK>> keyboard started working perfectly.

PK> sorry, i made a mistake. ru_RU.KOI8-R works, while ru_RU.UTF-8 (which
PK> is my default setting) and en_US.UTF-8 don't.

But ru_RU.UTF-8 works with the following patch to keyboard.c (but it
won't work on XFree86 3.x):

--- keyboard.c  2003-08-12 19:23:29.000000000 +0400
+++ keyboard.c-mine     2003-08-12 19:39:21.000000000 +0400
@@ -1840,7 +1840,7 @@ INT X11DRV_ToUnicode(UINT virtKey, UINT 
     else TRACE("Found keycode %d (0x%2X)\n",e.keycode,e.keycode);

if (xic)
-        ret = XmbLookupString(xic, &e, lpChar, sizeof(lpChar), &keysym, NULL);
+        ret = Xutf8LookupString(xic, &e, lpChar, sizeof(lpChar), &keysym, NULL);
     else
         ret = XLookupString(&e, lpChar, sizeof(lpChar), &keysym, NULL);
     wine_tsx11_unlock();
@@ -1911,8 +1911,15 @@ INT X11DRV_ToUnicode(UINT virtKey, UINT 
        /* perform translation to unicode */
        if(ret)
        {
-           TRACE_(key)("Translating char 0x%02x to unicode\n", *(BYTE *)lpChar);
-           ret = MultiByteToWideChar(CP_UNIXCP, 0, lpChar, ret, bufW, bufW_size);
+           if (xic)
+           {
+               TRACE_(key)("Translating string %.*s from UTF-8 to UCS-2\n",
+                           ret, lpChar);
+               ret = MultiByteToWideChar(CP_UTF8, 0, lpChar, ret, bufW, bufW_size);
+           } else {
+               TRACE_(key)("Translating char 0x%02x to unicode\n", *(BYTE *)lpChar);
+               ret = MultiByteToWideChar(CP_UNIXCP, 0, lpChar, ret, bufW, bufW_size);
+           }
        }
     }

PK>> Thanks!

wrong keyboard layout

[Dmitry Timoshkov]

"Phil Krylov"  wrote:
> But ru_RU.UTF-8 works with the following patch to keyboard.c (but it
> won't work on XFree86 3.x):
>
> --- keyboard.c  2003-08-12 19:23:29.000000000 +0400
> +++ keyboard.c-mine     2003-08-12 19:39:21.000000000 +0400
> @@ -1840,7 +1840,7 @@ INT X11DRV_ToUnicode(UINT virtKey, UINT 
>      else TRACE("Found keycode %d (0x%2X)\n",e.keycode,e.keycode);
>  if (xic)
> -
>    ret = XmbLookupString(xic, &e, lpChar, sizeof(lpChar), &keysym, NULL);
> -
>    ret = Xutf8LookupString(xic, &e, lpChar, sizeof(lpChar), &keysym, NULL);
> else
>        ret = XLookupString(&e, lpChar, sizeof(lpChar), &keysym, NULL);

It must be a bug in the XmbLookupString implementation of XFree86. It
seems that it doesn't respect UTF-8 locale, while works fine in others.

Could you try to investigate what really returns XmbLookupString for
UTF-8 locale?

[Phil Krylov]

Hello Dmitry,

Wednesday, August 13, 2003, 6:28:21 AM, you wrote:

DT> Could you try to investigate what really returns XmbLookupString for
DT> UTF-8 locale?

It returns 0.

wrong keyboard layout

[Dmitry Timoshkov]

"Phil Krylov"  wrote:
> DT> Could you try to investigate what really returns XmbLookupString for
> DT> UTF-8 locale?
>
> It returns 0.

What version of XFree86 are you using? Could you send me a log using
current cvs: +x11drv,+event,+key,+keyboard?