On 31 Jul 1998, Jason L Tibbitts III wrote:
: >>>>> "RSW" == Randall S Winchester <rsw@Glue.umd.edu> writes:
:
: RSW> Well, as more sites turn off relaying it will get somewhat
: RSW> better. Once a big spammer gets your majordmo list, it will be more
: RSW> then one that goes though and bounced.
:
: True, but I think that trying to parse out bounces is not necessarily the
: best answer. For one, it's a heck of a lot of work and we're never
: guaranteed to be able to figure out everything (although this shouldn't
: stop us from parsing common formats). Some way for the owner to remove a
: particular message from considering in bounce calculations would be
: interesting.
:
: You always know know the message number that bounce, because it's always in
: the envelope if you have an MTA that supports it. (Look at the envelopes
: from any Mj2-run lists.) Sendmail and qmail do; someone tells me that MMDF
: does and I can't see why Exim and Zmailer wouldn't though I don't know what
: their separators are. (Sendmail uses '+', qmail uses '-', I've heard MMDF
: uses '='.) So the bounce processor can use this to its advantage, and the
: owner could theoretically tell it not to worry about bounces from a
: particular message.
My "Precedence:" hack does have some merit as well. If it can be done in a
non-hack fashion, it could make it into a future sendmail release. It
logically makes sence to handle rejects from a mailing list differently, as
there are "man in the middle" issues where they will have no knowledge of
others policies. As long as mailing lists use the "Precedence:" field, they
can be handled differently. I do not evaluate what if anything is to the
right of that field as there is no standard, but maillists are usually the
only things that set this field.
:
: Another thing to do is to make your input filters as strong as those
: downstream of you. (But of course you know this.) I have some play code
: to implement RBL lookups on every IP address mentioned in the Received:
: lines of a message and bounce transgressing messages to the owner. (It's
: pretty easy, really, and this is less Draconian than refusing to accept the
: messages at all.) If people stop relaying then we can actually trust this
: information.
Stopping the relay's is an important first step in the spam wars.
: An interesting side effect of bounce processing: I can forge bounces to the
: processor and remove someone I don't like from a list.
Yes, along with the "subscribe enemy" to all lists in the world, I keep
expecting to see the "unsubscribe enemy" ones...
Randall
References:
|
|
