Keith Moore writes:
To nobody in particular:
IMO, it is a Bad Idea to try to embed one address in another, and an
especially Bad Idea to extend SMTP to try to facilitate that. If you want to
know whose mail bounced, sure, add a small token to the return address when
sending the mail. Associate that small token with the original recipient
address in a database, so when you get the NDN you can look up that token
and see whose mail bounced. But don't try to make it invertible. You're
only making the service less reliable by trying to do so, and also tempting
people to try to decode information which should be opaque to them.
No argument from me. Agreeing to all of the above, but having said that:
The VERP-style encoding of bounce addresses has been in use for at least 20
years now. I wasn't the one who came up with it. I don't really know when it
originated. Qmail's ezmlm used it, according to Wikipedia this dates it back
to 1997 at the latest.
If this ever caused any interoperability problems somewhere, I haven't heard
of it. I had to deal with more issues related to EAI (in IMAP space mostly)
than with VERP (namely, no issues at all).
I'm just pointing this out. I'm not married to my current implementation,
but it worked for me.
Description: PGP signature
ietf-smtp mailing list