On 01/Feb/10 17:20, Chris Lewis wrote:
Steve Atkins wrote:
On Feb 1, 2010, at 6:59 AM, John Levine wrote:
Except that that isn't, I presume, the way these buttons currently
work - except when somebody is subscribed to a feedback loop. It
seems over-complicated and inefficient (even with BURL) to send an
ARF to your own system admin, and rather more simple to just set a
flag or annotation on the IMAP server.
However, it's not trivial to timely interpret changes in the IMAP folders.
You're right, for the minority of us who run IMAP. For everyone else
who uses POP, mailing an ARF report back to the POP server may be the
best we can do. The authserv-id from RFC 5451 isn't ideal to use as
the mailing address since the RFC says quite clearly that it has to
look like a FQDN but it doesn't actually have to be an FQDN.
If we go down this route, we could probably add a flag to 5451 to
say it's OK to send ARF reports.
I'd guess that hosts who don't support getting abuse reports can
choose an authserv-id which is not a mail domain. At any rate, they
will have to instruct their users on how to enable processing of that
header field.
... or put it in the the message, where there's one standard format
to deal with, rather than tying it to just one of the various ways
people retrieve messages. Most everyone already stashes an assortment
of metadata in the headers anyway.
Uh? The authserv-id from RFC 5451 is right there already...
Which gives you the opportunity to do more complicated things, like if
the IMAP/POP servers aren't the place to send the notification. In our
case, they're not.
If using ARF, more complicated behavior might be noted in specific
fields. In several cases --e.g. unsubscribe, FBL-- some kind of
forwarding is required anyway. Changes will be easier to manage if the
software is centralized at a few abuse@ handlers, than if complicated
behavior is carried out by each client.
A X- header that has machine-readable instructions on what the reader
should do if you hit the TiS button.
A savvy client should seek user's consent anyway.
I'd envisage emailing ARF or plain forwards or site-chosen identifiers
to a specified place. Plus something non-SMTP.
Sticking to ARF seems smoother, from the server handler POV.
_______________________________________________
Asrg mailing list
Asrg(_at_)irtf(_dot_)org
http://www.irtf.org/mailman/listinfo/asrg