>>>>> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
Lars> I just did some tests (using libidn and the nice stringprep Lars> utility functions therein). Just by changeing Lars> Paragraph::value_type to uint32_t and adding Lars> stringprep_ucs4_to_utf8 on output, and some Lars> strigprep_locale_to_utf8 -> stringprep_utf8_to_ucs4 on input, I Lars> almost have it working. Very good. How does this compare to the CJK LyX patch? Did you look at it? Lars> Finishing this step and removing all code now rendered cruft, Lars> might also give us a better position to move forward with Lars> combining characters, ligatures, bidi etc. (cruft: intl stuff... Lars> translateAndInsert... stuff like that) Yes. Lars> I fear that XForms might need an upgrade to use either XwcLookup Lars> or XmbLookup to give us what we require in the keyhandler (we Lars> might we able to do it the event handler as well, unless xforms Lars> already ate some of our multibyte chars). Or some IM thingie Lars> (More work for you while in retirement Angus). We already have this patch from CJK LyX. But it probably needs some care (suse includes it, and it seems that it causes problems). Lars> So in conclusion: As a first step towards proper unicode support Lars> a change from the localespecific 8-bit representation we have Lars> used until now move to a basic uint32_t UCS-4 simple codepoint Lars> solution instead. We may then temporarily loose hebrew,arabic Lars> but as unicode support progresses those should come back nicely. While this maybe euro-centric, note that it probably solves 80% of the locale-related problems that people have reported... JMarc
