RE: HTTP Multipart Response, is this reasonable?

["Gregory W. Vesper" <gvesper@xxxxxxxxxxxxxxxxxxx>]

> Why would the message size affect use of async vs sync (in the http
> post reply) MDNs?

For a 10+MB file, by the time you decrypt, validate signatures, and generate
the MDN, you may have timed out your HTTP connection.

> (I assume the MDN wouldn't include a copy of the original-- just the
> MIC)

The MDN would be the the same as AS1 MDNs.

> It seems to me that separate transmission of an MDN would be mainly
> needed if a forwarding gateway was involved, and final delivery can't be
> confirmed until the forwarding is completed.

That's another potential scenario yes, but I think the response timeout
scenario will be more common.

> Note the way I read the spec, the choice of reply method is determined
> by the sender not the recipient. Did I miss something, or would there
> be a problem using HTTP if the sender requested an MDN using the
> 'default' method, but the http is a forwarding agent and can't issue
> immediate replies?

The MDN reply method is determined by the sender - I suppose this could be
made a suggestion rather than a mandate with a fallback to a default
behaviour.  Inasmuch as current EDIINT MDNs are async over SMTP, this might
be a nice default.

Cheers,
-Greg.