On Mon, Nov 22, 2004 at 21:09:56 +0100, Love wrote:

> der Mouse <mouse@Rodents.Montreal.QC.CA> writes:
> 
> > I for one am not prepared to throw that installed base away.
> 
> You are argueing for continued disorder, for us that already live in
> multi encoding in the same filesystem its a constant pain. Choosing
> one encoding that is prefered and supported in the system is a step
> toward order and sane-ness.
>
> Give that the world seems to have choicen utf8, lets go there and provide
> tools to unbreak filename (and not making decoding failure fatal).

ffs is encoding agnostic so you can use any encoding for file names.
If you standardize, as a adminsitration policy for your system, on
utf-8 - you get what you want - single system-wide encoding.  You
don't need any kernel tweaks to achieve that, you can do this today.
And someone who is happy with his current latin-1 (latin-2, koi8-r,
etc) file names on his system can happily continue to use them.  No
breakage for "legacy" users.  No obstacles for utf-8 adoption.

Ther are file systems that specify that file names are in a certain
character set, e.g. Joliet extension.  For these file systems I
propose we provide a transcoding option for mount_foo.  On my koi8-r
system I want file names transcoded into koi8-r to match the rest of
my file naming convention.  On your utf-8 system you specify
transcoding to utf-8 as the destination charset - to match your naming
convention (note, that while Joliet uses unicode, it uses ucs-2, not
utf-8).

SY, Uwe
-- 
uwe@ptc.spbu.ru                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen