Why are we making a decision in SIEVE WG that MDNs are bad?
Separate the two parts of my note. The first part said what I think
consensus was in the meeting, and I think that's saying that we want to
provide alternatives and make sure implementers and script authors
understand what they're getting into when they refuse/reject a message.
The second part -- the part you quoted -- was my opinion. So it's more
overstated than the consensus was. It doesn't necessarily reflect
What I think we should do is simply recognize that a protocol level
reject is a useful feature missing from SIEVE, and allowing a user or
implementation to control whether they do that or issue an MDN is a
I think that's what we're doing... except that I *DO* think we should
advise about the drawbacks of sending MDNs. Maybe it's not our job,
but we bought it by getting stuck in this spot.