Re: wine cvs notes and proposed keyboard detection fix.

[Dmitry Timoshkov]

"Shane Shields" <locutusenterprises(a)softhome.net> wrote:
> > There are two problems with current x11drv keyboard code:
> > 1. Very likely that Turkish keyboard layout there is incorrect.
>
> AFAICT The Turkish keyboard layout is correct.

X11 team tends to remove more than 2 characters per keysym. See
/etc/X11/xkb/symbols/tr. Turkish keyboard layout in Wine has more
than 2 for some of them. Just do 'setxkbmap tr' and run Wine with
+key,keyboard,+x11drv and look for mismatches.
> The original code doesnt get as 
> far as looking at the turkish definition. From what I could determine it 
> stopped searching when it got a wrong keysymb and if that combination 
> satisfied the selection that keyboard was chosen.

That shouldn't happen. Keyboard detection code always does a full round
of comparisons with every keyboard table in x11drv.
> It didnt matter if it was 
> the right keyboard or not. This was not completely verified through testing 
> as I only have one keyboard :)
> > 2. As stated before the keyboard code should be fixed to take into account
> > the real keyboard layout while building the keyc2vkey map, not the
> > hardcoded one.
>
> That would be a more elegant and portable fix. How do the other programs do it 
> (read window managers)?

Other programs have no constrains implied by implementing Win32 APIs on the top
of X11.

-- 
Dmitry.

wine cvs notes and proposed keyboard detection fix.

[Fabian Cenedese]

> That shouldn't happen. Keyboard detection code always does a full round
> of comparisons with every keyboard table in x11drv.

Ahmm... I don't have keyboard problems, just out of curiosity: Would it speed
up wine starting if the keyboard was configured fixed and wine wouldn't have
to test all possible layouts?

bye  Fabi

wine cvs notes and proposed keyboard detection fix.

[Dmitry Timoshkov]

"Fabian Cenedese" <Cenedese(a)indel.ch> wrote:
> Ahmm... I don't have keyboard problems, just out of curiosity: Would it speed
> up wine starting if the keyboard was configured fixed and wine wouldn't have
> to test all possible layouts?

You could try it by leaving only single layout in the global table
of keyboard layouts. But I doubt that it will change too much.

-- 
Dmitry.

wine cvs notes and proposed keyboard detection fix.

[Shane Shields]

On Tuesday 20 January 2004 5:23 am, Dmitry Timoshkov wrote:
> X11 team tends to remove more than 2 characters per keysym. See
> /etc/X11/xkb/symbols/tr. Turkish keyboard layout in Wine has more
> than 2 for some of them. Just do 'setxkbmap tr' and run Wine with
> +key,keyboard,+x11drv and look for mismatches.

I have done that and piped the output to a logfile. Attached is an abbreviated 
version with just the icelandic and turkish detection included. It seems to 
be mismatching on detecting the third keysym where the key is defined 
in /etc/X11/xkb/symbols/tr this could be a clue :)
> That shouldn't happen. Keyboard detection code always does a full round
> of comparisons with every keyboard table in x11drv.

I took another look at the code and with the trace I have to say that I was 
wrong :-(

The trace only shows 0x0000 for the third(?) keysymb so I dont know what was 
returned. I believe (gingerly) that is due to the trace having a precision of 
4.
          TRACE_(key)("mismatch for keysym 0x%04lX, keycode %d, 

in function X11DRV_KEYBOARD_DetectLayout where it should be 
 
         TRACE_(key)("mismatch for keysym 0x%08lX, keycode %d, 

or my version of X11 is returning a larger value than it should. I suspect 
this because of the large amount of calls to the function 
KEYBOARD_MapDeadKeysym. Maybe the X11 guys have changed a short to a long 
somewhere in my version of X11 (cvs 2003-11-24) or my gcc (3.2.2) has defined 
them differently. If that is the case then the line 

                ckey[i] = keysym & 0xFF;

in function X11DRV_KEYBOARD_DetectLayout would change the keysym and cause a 
mismatch. Another thought popped up is maybe the X11 guys extended the 
keysymb codes to allow for 6 instead of four combinations. I will have to 
check the changelogs to be sure.
> Other programs have no constrains implied by implementing Win32 APIs on the
> top of X11.

LOL
-- 
Shane Shields

Registered LFS Compiler: 7582
To drink the WINE of success you must first seek the sayings of source

Anyone sending unwanted advertising e-mail to this address will be
charged $25 for network traffic and computing time. By extracting my
address from this message or its header, you agree to these terms.