Graham Klyne (GK@dial.pipex.com)
Tue, 25 Jan 2000 10:37:35 +0000
I believe this debate has wandered somewhat from the original proposal that
sparked it, and I'll try to use Keith's comment to bring it back...
At 12:56 AM 1/25/00 -0500, Keith Moore wrote:
>What I don't think you can do - or at least, I don't see how to do -
>is define reasonable 'delivered' vs. 'not delivered' criteria for
>arbitrary multipart MIME messages.
This is exactly the point I was trying to address...
We have argued (I think we agree) that a partial NDN mechanism should be a
natural extension of DSN that can reasonably be interpreted by existing
DSN-aware senders. This requires that the receiving MTA must make a
judgement about overall success/failure for the purposes of DSN generation,
even if it will supplement that information with partial delivery
information (which, to partially address Ned's point, may help an NDN-aware
sender to make an appropriate report to the user).
I was (implicitly) agreeing with Eric that if the original message is
accompanied by critical part information (whatever that may be), then it is
reasonable to declare success if all the critical parts are delivered.
But in the absence of such information, and given choice between declaring
success or failure, then I was suggesting that failure to deliver _any_
part should result in a report of failure to deliver the message as a
whole. The alternative of declaring success seems unreasonable (but I am
open to hear other proposals).
#g
------------
Graham Klyne
(GK@ACM.ORG)
This archive was generated by hypermail 2.0b3 on Tue Jan 25 2000 - 13:16:08 IST