dolecek@ics.muni.cz said:
> I still think that doing the conversion in kernel is more efficient,
> at least memory-wise.

Well possible. The implication - codeset determined at mount time -
is serious however. How can I do an "ls" (or use an X based
directory viewer) over multiple directories, each mounted with another
encoding? What happens if a CD contains multiple language versions
of something? How is the per-mount encoding communicated to the
userland utilities at all? I see lots of complications.

best regards
Matthias