On Wed, Apr 17, 2002 at 11:34:18PM +0100, Sean Rima wrote:
> Okay I see where you are coming from. And I can see there is a strong
> possibilty of time outs etc, although I wonder if the source for spamc
> was included into local_scan would this speed things up, I don't know,
> only throwing the idea out.
It could do. The difficulty comes with virus analysis when you have such
files as
http://colon.colondot.net/~mbm/42.zip (which you may already have
come across).
Ideally you want to do *some* spam processing at DATA time, but save the
more complicated bits, and virus processing for asynchronous use. Both
sender and receiver SMTPs should be trying to avoid duplicate mail at
that point, the sender by having a long timeout, and the receiver by
trying to be as quick as it can about processing the message, but
guaranteeing that it's written to disk.
MBM
--
Matthew Byng-Maddick <mbm@???> http://colondot.net/
