[Monotone-devel] Re: [windows] locale problems

[Lapo Luchini]

Thomas Keller wrote:
> > It's weird because for example, "(révision de base)" is output as
> > "(rÚvision de base)" where:
> > - é is ISO8859-1 0xE9 or U+00E9
> > - Ú is ISO8859-1 0xDA or U+00DA
> > I can't recall in which version this problem appeared.
> Which package do you use? The native binary or the cygwin package? Maybe 
> one of the packager(s) knows whats wrong here...

That's very VERY weird, because AFAIR 'é' is not encoded 0xDA in any 
charset...

> > And this makes me wonder: when i import non-ascii filenames, how are
> > they stored ? if i checkout on a linux box with an UTF-8 locale, will
> > the filenames be transcoded ?
>
> monotone uses an internal normalized utf8 format for paths, however the 
> conversions from/to the various external character sets is what most 
> likely make problems (and throws out the logic error above), just 
> because its not really there (tm) yet. So you're safe if everything uses 
> utf8 around you - otherwise you're pretty much on your own for now.

BTW: cygwin doesn't support Unicode and by default doesn't define a 
charset in the env, I have to use CHARSET=CP1252 or something like that 
to "mtn add" files with accents.

    Lapo