On Wed, Sep 19, 2007 at 06:25:55AM +0400, Andrey Chernov wrote: > I'll think about... It seems my first attempt was right, i.e. we need real UTF-8.src instead of UCS-4 mimic of it. All other locales keep their true wchar_t encodings, only UTF-8.src not following the rules. I'll send regenerated UTF-8.src a bit later. -- http://ache.pp.ru/