Nathaniel,
Sorry for the confusion. In a number of earlier notes, I've said that
your document must include US-ASCII, since that is the current system
behavior. (It certainly makes no sense for the new spec to support
an environment that is, in any way, less functional than the old.)
As to the concerns of the non-English-usage writers, I believe that my
suggestion leaves them certainly no worse off than they currently are,
but more likely much better off.
Let's note that while they have some tricks they play with the current
system, to get their extended character set information sent around,
none of that is standardized and I'm not even sure that any/all of it
is documented.
Your document provides a place to declare character set. I believe that
there is nothing to prevent the folks out there from citing whatever
character set they want, in each message. (X-myset, X-ISO-10646, or
whatever.) This strikes me as being significantly better than the
current world, without requiring that your document get tied to
the details.
In other words, your document should/must not preclude experimentation
with alternative character sets. However, neither should it depend
upon any specific one (except the only one that has extended, formal
Internet experience, namely US-ASCII.) When that experiment produces
a clear winner, it will be a simple matter to document the fact and
specify its handling within the world of RFC XXXX. But it must be only
AFTER the experimentation produces a clear winner.
Dave