Taxonomy of proposals so far

Given the lull in the discussion, it might be a good time to look at 
what is in front of us. A few orthogonal and related proposals have 
come forth. Some of what appears here has been suggested in off-list 
discussions.

Here is a first cut at a taxonomy of proposals for internationalizing 
the mailbox names and their transport. Note that the names appear in 
the From, To, Cc, and Received headers (and maybe others).

1. ASCII-encode the names.

2. UTF8-encode the names, protecting the result with an ESTMTP
    extension.

    2.1 If the SMTP server doesn't know the extension, bounce
        the message.

    2.2 If the SMTP server doesn't know the extension, send the
        message anyway, but fall back to non-UTF8 names.

        2.2.1 Fall back ASCII-encoding the names.

        2.2.2 Fall back using a different mailbox name that is
              all-ASCII.

              2.2.2.1 Get the second name from headers that are
                      sent with withe message.

              2.2.2.2 Get the second name by querying a server that
                      is supposed to know about that user.

3. Do nothing.

Before we go into the merits and problems of each of these, or how 
they might interact, it would be good to see if this is a 
reasonably-complete chart. Changes and additions are welcome.

--Paul Hoffman, Director
--Internet Mail Consortium