Dave,
I think you've caught a problem.
I want to stress that I'm not an IDN expert. Patrik Fältström is, and
I'm copying him on this for further comment.
He points out that RFC-4952 properly states that-
"The domain part of email addresses is already internationalized
[RFC3490], while the local part is not."
But RFC-4952 is itself only an informational document, and there are
experimental documents that might seem to conflict. Specifically,
RFC-5335 does *not* reference 3490, and talks about an SMTP extension to
carry UTF8 headers. Worse- RFC 5505 doesn't reference 3490 either and
references RFC 2047 encoding. The key issue is the order of encoding (a
word in which we get into some trouble), from I can tell; because an
alternate header is otherwise provided, and that breaks DKIM.
For DKIM to work, I don't see how we can avoid saying that 3490 encoding
must occur prior to EAI (RFC 5335/5505) encoding. Otherwise, you have
to handle the case where both the signing and verifying DKIM
implementation must handle Downgrade- headers in a perhaps fruitless
effort to reconstruct the original headers so that the original domain
can be derived, in order to then encode THAT into 3490.
I suspect what is needed at the very least is a clarification; probably
a proposed standard would be nice.
FWIW, I don't have single MUA on my systems (and I have quite a number)
that supports ANY of this yet; including TB, Apple Mail, pine, and
associated tools.
Eliot
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html