Re: [exim] Sender callbacks are a deliberate "bug"

Hello,

On Tuesday 21 November 2006 21:18, Ian FREISLICH wrote:
> Mark Symonds wrote:
> > Hi.
> >
> > Apparently, sender callouts are a "bug".
>
> IIRC callouts were originally introduced to enable people to verify
> users on their own servers. An example would be for an exim front
> to an exchange server. They have subsequently been abused by kooks
> in search of the FUSSUP as an aid to their spam filtering.
>

I know it's a point of contention, but my take on it is this: If you are
emailing me from an unreachable email address and/or IP, then the only way
for me to let you know that bounces are going to fail is to refuse at smtp
time. For me it isn't about FUSSP, but rather making sure we can get
informative error messages back to the sender. Otherwise with are left with
users scratching their heads, "I emailed that to you last Friday... didn't
you get it?"

> It's relatively easy to circumvent: when an incoming envelope sender
> is <>, limit the number of recipients to 1 and only reject an unknown
> local_part in your smtp_pre_data acl. It doesn't stop the callout,
> but it means that you'll always answer in the affirmative for
> callouts. Much in the same way that you would always (if you have
> the time) respond to CR blowback so that the blowback sender gets
> the spam.
>

Yes - null sender was the problem. :-) The other end had their greylisting
configured to defer null senders as well, causing intermittency while their
caches purged/refreshed, etc. They updated their configuration to allow for
them and we're good.

Thanks much, and best,

Mark