Sherwood asked in a private mail whether it would be better to build a
general API for message scanning...
I'm actually thinking along those lines, I did some proof of concept
stuff by request of Brightmail earlier in the year to integrate their
spamwall library into exim. I want to take that code, clean it up and
generalise it so that at each of the spam handling API callout points
(which are basically one on start of message receipt, one call per
address (sender/recipient), a couple of calls for building up headers
and a couple for building up the body). Each of these points would
handle a callout to the spamwall api, to an embedded perl function (you
would define the appropriate perl functions for each point from the
exim config or maybe they would be setup from the perl startup script).
If I could find a clean means of doing it I would also allow a general
call out - maybe we should look at stealing the Milter filtering stuff
off sendmail as well.
Additionally we have previously discussed on list the idea of a program
call out from within an exim filter which will return status to the
filter.
If you want to do content modification there is nothing in hand other
than the transport filters framework which exist.
Nigel.
--
[ - Opinions expressed are personal and may not be shared by VData - ]
[ Nigel Metheringham Nigel.Metheringham@??? ]
[ Phone: +44 1423 850000 Fax +44 1423 858866 ]
