On Thu, 4 Jan 2001, Tamas Decsi wrote:
Hi,
I think locale is not configured properly - $LANG should be set to
hu_HU.ISO-8859-2 (of course if your system supports hungarian locale -
/usr/share/locale/{hu,hu_HU} should exist) - please try it and report
results. Anyway, it's a incorrect to use GdkEventKey->keyval - the right thing
is to use GdkEventKey->string.
If setting LANG properly solves the problem, bug#559 should be closed.
>
> Hello abi-dev,
>
> I couldn't stand and traced down the reason why abiword (unix) neglects
> some iso-8859-2 chars (eg. the 'nastiest' hungarian chars:
> o/O/u/U with double acute: charcodes 00f5, 00fb, 00d5, 00db).
>
> It turned out that in
> ev_UnixKeyboard::keyPressEvent()
> the passed
> GdkEventKey *e
> structure provides an empty string in
> e->string
> (at least) for the mentioned chars, while the
> e->charcode
> still gives the character code.
> Since that string is used as an input to ut_mbtowc (and then iconv), the
> whole thing results in an empty string, instead of those chars.
>
> My questions are:
>
> 1. this is not exactly bug #559, because the crash it notes did not happen.
> What should I do with this bug in bugzilla? (sorry, I've never used it
> before)
>
> 2. it turned out that the problem may be in gdk, and not in abi.
> I don't know much about gdk, so I need some hints:
>
> - Does gdk support iso-8859-2 encoding?
> - If yes, where it is told, that we use iso-8859-2 ?
> - If no, it seems to be abi's bug to use the GdkEventKey->string, and
> should be avoided somehow. Am I right?
>
> Any hints are welcome.
>
> Cheers,
> Tamas
>
Best regards,
-Vlad
