On Thu, Dec 23, 2010 at 02:39:23PM +0100, lst_ho...@kwsoft.de wrote:
>
> To summarize:
>
> DSN as of RFC 3461 is only recommended as internal status indicator for
> message relayed out of the own scope. End-to-end status is neither
> supported nor technically possible at the moment.
This is a stronger statement than I would make. End-to-end DSN is
sometimes possible, but given the hostile environment in which we find
ourselves today, it is often either not available, or not advisable,
or both.
Some sites will support DSN all the way to the mailbox, I can't guess
what fraction these are. I've chosen to not be among them.
> So with this DSN is useless for us, as we don't have a complex or
> unreliable internal network with many hops. If the first server accepted
> mail from the client and no bounce is coming back, the mail has left our
> scope for sure (thanks reliability of Postfix) and therefore notify the
> sender of the obvious is superfluous.
DSNs are potentially useful for automated mail sending apps so they
can track message status in their database (either bounce or successful
delivery to the responsible org, though not necessarily mailbox).
How was end-to-end DSN potentially (more) useful for you? In most cases
once mail is accepted by an MX host, the next few delivery steps are
reliable. If the mail does not then bounce, it is either delivered or
quarantined. Sites that quarantine mail will not explicitly want to
expose this behaviour to senders. Some may unwittingly expose the
message destination to senders, but you should not rely on this...
--
Viktor.
