Ah, but I am 99% sure this discussion is about extending addresses
(not the full name, but the real honest to goodness address) to include
non-USASCII characters. For instance the recent message from Tony Hansen
about this unnamed e-mail carrier who has a real customer requirement
to implement that exact same behaviour.
Further .. there is this tradition that login name == mail name. There
is lots of users having this `knowledge'. As more people experiment
with non-USASCII login names, there will be more attempts to put these
things into (at least) the local-part of addresses. Perhaps also
the domain-part?
Further .. It is said that Windows NT uses Unicode `everywhere'. Presumably
this is also true for their login-names. Again, using the same convention
this implies attempts to use Unicode in mail addresses.
Your suggestion only covers the `phrase' (conventionally used to hold the
users full name) part of the address.. *not* the address (local-part
nor domain-part) itself. RFC-1342 already covers the territory you
propose to cover (as I recall, one of the proposals leading to
RFC-1342 made use of secondary headers ...).
I have presented three extremely likely cases where non-USASCII addresses
(not phrases) will be attempted/implemented in, say, 1-2 years time.
David