w> 1. The UASG cannot recommend any transformation of a mailbox at any time for any reason.
2. The UASG accepts that a replacement of a non-ascii mailbox name with an ascii mailbox name can be used in very limited circumstances - basically where the sending system 'owns' the sending mail address and an ascii address is pre-defined in a alias field and the first hop is clearly not EAI Ready.
Yes, those sound right. For #2 another plausible case is that the sender is doing the downgrade and has the aliases in the local address book.
3. Mailbox administrators can use any characters or set of characters not explicitly disallowed by the RFCs in their mailbox configurations. * Does the UASG recommend any approach to avoid confusion? I would rather point to someone else's well considered work in this area. Does such exist?
That's what the IETF PRECIS work was about. The username and nickname profiles are both plausible for local parts. I put references into the Google draft.
4. We may include an Appendix in the EAI Documentation with some history and examples of transformations that were discarded and why.
I suppose, but I worry that would just encourage people to argue that they shouldn't have been discarded. Is there any way to explain that there is a spec for EAI, if you do what it says you will interoperate, and if you do something else, even something else which you imagine to be better, you won't? R's, JOhn