"Stuart D. Gathman" suggested:
On Sat, 4 Jun 2005, Florian Weimer wrote:
I'm implementing the syntax part of draft-schlitt-spf-classic-01[1]
and noted that no charset is defined for the explanation string.
There's also no kind of language handshake. These days, this is
unacceptable for a message which should be presented to end users.
The exp message is both provided by and used by the sending domain.
The sending domain can put anything it wants in there since they
are the only ones who will see it. The receiver simply relays verbatim
back to the sender what the senders DNS provided.
So even if every sending domain had its own unique charset encoding,
everything
would work smoothly. There is no need for an encoding standard to
ensure interoperation.
A suggested standard would be useful to avoid reinventing the wheel,
but following it is not needed for interoperability.
What you say about the syntax + semantics being private to the sending domain is
only valid for messages rejected at SMTP-time.
For those 'bounced' later, the receiving MTA composes the reply mail, and it has
not idea what encoding has been used, or what language to use for any other
words it inserts - hence (sadly) the default to English.
Chris Haynes