Re: Win32 problem with accels and layouts.

[Tor Lillqvist <tml iki fi>]

(Sorry for UTF-8/quoted-printable brokenness in the quote, I am using
an oldish XEmacs as my MUA)

> GTK+ does fuzzy-group matching for accelerators/mnemonics;=20 if you
> have a =D1=84/a key, as is common for Russian keyboards, if you=20
> have only a '=D1=84' or 'a' as a mnemonic, then that Alt-=D1=84/a
> will ac= tivate the mnemonic in either group.

> This is implemented in gtk/gtkkeyhash.c.

> If it isn't working for Windows, this probably means that work needs
> to be put into the backend function=20
> gdk_keymap_translate_keyboard_state() and/or
> gdk_keymap_get_entries_for_keyval().

But even if the keyboard has both Cyrillic and Latin engravings on the
keys, Windows doesn't tell that to GTK. GTK only knows about the
current keyboard layout.

OK, so it could check what other keyboard layouts are available
(switchable) on the machine. In fact, there is still code left in
gdkevents-win32.c (inside #if 0) that looks for a keyboard layout that
corresponds to a language that uses Latin characters. It's a leftover
from code that wasn't needed any longer. Hmmm.

Should GDK on Windows tell GTK about all the keyboard layouts
available on the machine, and not just the current one? Would the
different keyboard layouts correspond to "groups" on X11?

--tml