"Mark" == Mark Lentczner <markl(_at_)glyphic(_dot_)com> writes:
Mark> If you are referring to the three headers, Delivered-To,
Mark> X-Envelope-To and Envelope-To, in my last meeting with the
Mark> authors of marid-core, it was agreed to remove them from the
Mark> algorithm. With reference to marid-core-01, you can
Mark> consider step 3 in section 4 to be deleted.
Thanks for the info.
Mark> We all agreed with your logic of using non-standard headers
Mark> which could be shown to be implemented differently in
Mark> different MTAs. I showed the way Postfix uses Delivered-To
Mark> as an existence proof of this problem.
One of my problems here is that I'm not intimately familiar with any
MTA that adds these headers. The only MTA I would claim to be
familiar with in any detail is sendmail, which uses none of these
headers.
In any case, that's by the by. I'm happy that they're gone from the
spec.
Incidentally, one further very minor question I'd like to raise. Is
it necessary to refer only to non-empty headers (ie to explicitly
ignore all empty headers)?
I left the word "non-empty" in my PRA proposal because it was in the
original. But none of the headers in question are allowed to be
empty, and it's an extra complication for implementations that may not
be necesary unless there are real concerns that empty header fields
occur significantly in the wild in otherwise well-formed messages.
-roy