On Wed, 2007-10-10 at 15:28 +0100, Alun wrote: > This is all done now and works according to plan. My question, which
> started this off, was whether it was possible, from the embedded perl
> interpreter, to identify the file descriptor of the socket that's
> connected to the remote host. Tom confirmed that I wasn't missing
> anything in the spec, and I implemented what was needed for my purposes!
>
> The blacklist now has 819 entries.
All good. But...
Given that you've explained three times now what you're doing, and I
still can't see where you're generating your blacklist, I can see a flaw
here. If a user sends a message to, for example, a person the BBC with a
sender address of auj@???, and the person at the BBC
responds, will that mean the BBC's outbound MX farm will end up
blacklisted?
I only ask as you seem to have missed a class of possible "spurious"
connections, being those in response to misconfigured messages sent by
your users. I suppose if you can map them onto a valid @aber.ac.uk
address then the problem won't happen.
