>  Why do you need Unicode support in the kernel?  The tables are a  
>lot of crap to stow in the kernel.
>  The Mac OS X filesystem API's are all still char*.  What we do is  
>we treat char* strings as UTF8.  Read Unicode from disk as UTF8 and  
>present that to userland.  Higher-level toolbox calls deal with the  
>translation buisiness in user space.

	I prefer this way (no kernel modification for charset conversion,
	do it in userland).  Unicode does not solve problem for everybody,
	as a matter of fact most Japanese and Chinese text are exchanged
	in non-Unicode encoding.
	So please do not hardcode something to Unicode.

itojun