I've since recanted parts of the note of mine which Rick
presumably dredged out of the archives (perhaps he's found some
of the later messages by now :-) ). In particular, Keith is
right: regardless of the protocols used at higher levels within a
UA, it is important to retain the notion of an octet-stream at
the transport layer, which is what Content-Transfer-Encoding has
primarily to do with.
Since that earlier note doesn't really have anything to do any
more with anything other than UA internals, it doesn't really
matter whether Rick likes it (thanks, though!) or whether
Ohta-san thinks it's "meaningless," and there's no reason to
reopen discussion on those points here now. I do have some more
wild & crazy ideas about charsets, which I may air on the new
ietf-charsets mailing list if I ever have the time and intestinal
fortitude to do so; I'd also be willing to discuss them privately
with anyone who is interested.
Steve Summit
scs(_at_)adam(_dot_)mit(_dot_)edu