On Sat, Jun 09, 2001 at 07:23:32PM +0200, Denis Barbier wrote: > > > > > > It doesn't know how to handle characters like © (which should be > > > > > > sorted > > > > > > between S and T), though :o) Could that be fixed by tweaking the > > > > > > regexp? > > > > > > > > > > No, i believed it was handled by langcmp. Maybe you should fix this > > > > > routine? > > > > > > > > Hmm. I'm not quite sure how to change the function properly. > > > > > > > > Also, this looks like it could be a slippery slope: if I add an > > > > exception > > > > for those five Croatian letters, other people will do it too, and the > > > > file > > > > might get bloated with hardly important things... Maybe we could get a > > > > Unicode-aware cmp or something? :) Shouldn't there be a way to use the > > > > locale's sorting method ($LC_COLLATE)? > > > > > > Indeed, adding > > > > > > <perl> > > > use locale; > > > use POSIX; > > > setlocale(LC_COLLATE, "hr_HR"); > > > </perl> > > > > > > in template/debian/languages.wml fixes this problem. > > > Maybe you could add a variable in <lang>/.wmlrc when locales must be > > > changed? > > > > How do you mean? A variable that would hold the $LC_COLLATE string? > > Exactly, something like (in template/debian/languages.wml) > > <when "$(CUR_LOCALE)"> > <perl> > use locale; > use POSIX; > setlocale(LC_ALL, "$(CUR_LOCALE)"); > </perl> > </when> > > and then translators may add CUR_LOCALE definition in <lang>/.wmlrc when > necessary.
Sounds good, I'll add it. -- Digital Electronic Being Intended for Assassination and Nullification