I am mostly in agreement with the above. But the standard for MDNs
also specifies that the notification must go to the return-path address
*if it is automatically generated* (which vacation notices are).
This may have been your intent but the documents do not say this. 1894
is quite clear about sending the MDN to the envelope from but there is
no other restriction. I would further argue that except for a few
ambiguous references to a user agent, 1894 and its siblings refer to MTA
activities.
If we were to standardize a format for vacation notices, seems like
we'd make them an extension of MDNs, and we'd want their behavior to
be consistent with MDNs.
Consistent does not mean the same. I stand firm on the architectural
principle that a vacation notice is a user agent feature
(notwithstanding certain implementation preferences). The MTA is
responsible for routing and delivery to a message store. It is not
responsible for the behavior of a user or specific user preferences with
respect to that delivery (again, notwithstanding certain implementation
preferences).
You and I may have to agree to disagree on this point. We'll have to
wait until we create a "user agent MDN standard" to see which is in the
minority.
Jim