On Sun, Apr 04, 2004 at 08:17:13AM -0400, Nathanael Nerode wrote: > You could go with the language familes. These are mostly in ISO 639a, > though that doesn't have bs (Bosnian), or nb and nn (Norwegian was still > no). They're also accurate as far as I know linguistically. :-) > ASIAN: ja, ko, zh_CN, zh_TW > SEMITIC: ar, he > SLAVIC: bg, bs, cs, pl, ru, sk, sl, uk Slavic languages are split between Latin2 and Cyrillic for the character sets, so this may not an optimal grouping. > ROMANCE: ca, es, fr, gl, it, pt, pt_BR, ro Romanian also uses Latin2, so is probably better grouped with the Latin Slavic languages? > GERMANIC: da, de, en, nl, nb, nn, sv > FINNO-UGRIC: fi, hu And IIRC Finnish and Hungarian are both Latin1, so it probably doesn't make sense to split just the two of them off into their own group. > other families: cy, el, id, lt, sq, tr Hmm, we have Albanian translations? Not sure what charset that uses; the others, though, each require code points not shared by any of the others, yes. > Asian root (4 languages + lots and lots of non-Latin characters) > Semitic root (2 languages + BIDI + 2 non-Latin scripts) This suggests that we may want *two* slang udebs, one with bidi support and one without? Is this worth the space savings of not having libfribidi on the floppies? -- Steve Langasek postmodern programmer
Attachment:
signature.asc
Description: Digital signature