On Thu, 2012-01-26 at 14:33 +0100, Andras Timar wrote: > I still don't understand what's happening, i.e. in good case it reads > UI_SANS list for "hi" once, then several time for "en", and it looks > good, and in bad case it reads UI_SANS list for hi_IN several times > and still, we get corrupted characters.
I'm happy enough that there isn't any nasty hidden glyph/font replacement code bugs or locale matching regressions here that are papered over by renaming the font fallback lists and that renaming is the best fix so pushed the three commits of... a899ab36646ae0b8d078c510725d014332294099 91fa56cb98f21207e4bd559d0a821fc0b6e12526 756045dcffcdf2da3bc4d0c7bf0a55e5fc75106e to 3-5 and 3-4, need an additional ack for 3-5-0. The UI_SANS is "hi" because that's the name of the language pack basically, so it looks for default "hi" fonts for CTL text, doesn't find anything in the bad case seeing as we only have hi-IN ones listed so presumably falls back to "en" and use those fonts which don't have the appropiate glyphs. The other "en" are probably for default Western text and so forth. Later on when we get default fonts for document language we've converted the language id for hindi back to hi-IN and then we find hi-IN Anyway, I'm happy enough that there isn't any nasty hidden glyph/font replacement code bugs here that are papered over by renaming the font fallback lists and that that's the best fix so pushed to 3-4 and 3-5, need an additional ack for 3-5-0 I. We should in master generally shrink the rest of the tags down to the minimum correct lang-tag , e.g. hu-HU to hu, but zh-CN remains as zh-CN. And I'm confident that the zh-CN/zh-TW/pa-IN font entries currently should work for the windows UI fallback lists seeing as those langpacks are language_Territory. C. _______________________________________________ LibreOffice mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/libreoffice
