I think this is an issue we should not try to solve for 2.0, but
put off to an internationization document.
It does seem like the MIME/HTTP charset and maybe language
information are going to be proxies for some other things,
unless we make the format negotiation more complex.
A natural interpretation to make is that you won't use
numerical character references outside the MIME charset.
But if that's the case, why use numeric references at all?
And what if someone wants to serve up Japanese (or multilingual)
documents encoded in US-ASCII with numeric references?
We may have to say that some situations require full Unicode
font support to display. But let's not solve that today.
-- Albert Lunde Albert-Lunde@nwu.edu