On Tue, Nov 16, 2004 at 13:00:08 -0800, Jason Thorpe wrote:

> > And the easiest thing to do for ffs is to interpret the stored
> > data as utf8 always, which fits very well with Valerieys
> > suggestion.
> 
> I agree completely that UTF-8 is the ideal conversion target for 
> kernel<->userland interaction.  I *think* this is what Mac OS X does, 
> as well.

But our userland is still far from being utf-8 friendly,
unfortunately.  So for the time being I'd prefer a working "legacy"
8-bit solution.

Or, rather, a solution that would allow for any conversion table to be
supplied and let the kernel be agnostic about the actual charset -
making users responisble for ensuring that the charsets match.  Thus
users can continue to use their legacy 8-bit encodings that are
supported by exisitng userland.  And developers will be able to work
on utf-8 support. :)

wscons internals is an example of "unicodification" that went half-way
and was left cold in the wind.  I don't want filesystem stuff to end
up in a similar situation.

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