Hi everyone! I just wanted to speak up and say that in the case of my MTA (MDaemon) the question of whether to link in a web-client lib is a non-issue because (for me) that had to be done years ago. Perhaps that lib is not called by the SMTP delivery code but its in the EXE already at least. I have no serious issue with this and one way to handle it (as I think was mentioned before) is by deferring back to queue any delivery attempt which finds a previously unknown STS policy record until such time (not long) as a separate thread safely dedicated to the task can validate the policy and update the cache file. I will have to take this approach due to my MTA's internal design.

I also wanted to mention that I serve the world of the very small on-premise business community and I've received numerous emails stating the expectation and desire to have something like this implemented for them. This is a consequence of the STS articles etc recently put to press but also shows that its not only the big players who might push for it. Its true though that the vast majority are oblivious to these matters (which is fine) and they want us vendors to "just handle it" (which we should). So I hope that this effort moves forward and above all does not get overly complicated.

A couple of tiny minor spec things for starters: Section 3 says that MX patterns are required yet the example at the end of the doc doesn't have them. Also, "_.example.com" has the _ char as the wild-card. Maybe there's a reason not to use * for this? See, minor stuff. It would be good to have some reference STS records/hosts setup somewhere that implementers can test against.

Arvel





Attachment: signature.asc
Description: PGP signature

_______________________________________________
Uta mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/uta

Reply via email to