> Just for record, I'm not in favour of a transparent, locale dependant
> file name transcoding.  IMO having file names change with user locale
> would be an administrative madness.

But what is the "file name"?  The octet sequence passed to the API?
The octet sequence that crosses the syscall boundary?  The octet
sequence on disk?  The string of characters (or, sometimes, octets)
that bear the intent behind one or more of the above?

I agree that having the API <-> on-disk mapping change with locale
is..suboptimal.

> My primary aim is to have a reasonable way to handle file systems
> which use Unicode as it's internal file name character set. UTF8 is
> clearly one of the best choices here, since it's the only standard
> UNIX-compatible way to handle full Unicode range.

Well, the only one that uses 8-bit chars, maybe - I'd prefer to switch
to 16-bit chars.  (Actually, wouldn't UTF-7 also qualify?  Yes, I think
UTF-8 is better for this purpose.)

> I stay fully content with FFS/ext2fs/lfs to be file name encoding
> agnostic.

I think you and I have nothing we disagree on here, then.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B