Unicode support for non-unicode locales

[Jonathan S. Shapiro]

I don't know if this will be helpful, but we concluded in OpenCM that
Unicode would force us to do file type encoding. The problem goes much
further than file names.

If a unicode-encoded file is checked out on a shift-JIS system, how does
the user read it? Note that the encoding really has nothing to do with
the file extension.

Getting file names properly encoded is less difficult. Full support for
Unicode is hard.