"Alexandre Julliard" julliard@winehq.org wrote:
And what is the proper fix in your opinion? I think that even support for keyboard layouts won't help for that particular case.
The proper fix is to determine the codepage from the current locale information, not from the keyboard layout. This is basically what the UTF8 hack does, except it should be done for all codepages.
And fallback to LOCALE_IDEFAULTUNIXCODEPAGE if the encoding information is missing in the locale environment?
Probably it's better to implement this as a part of init_default_lcid() in kernel32 then? But then the question arises how to pass that information to x11drv, or just completely duplicate init_default_lcid() in x11drv?