[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: gethostby* users

David Woodhouse wrote:
On Sat, 2007-10-13 at 20:50 -0500, Les Mikesell wrote:
These slides give the whole story but the big things are that you can do anything you want by writing a small amount of perl code, the scanning processes don't run under the same uid as the mailer, and it is very efficient.

Looks like it's just trying to provide sendmail with the kind of
functionality which is built in to Exim.

I doubt that Exim can match perl's functionality in the general case... How much does it know about MIME?

I find Exim's ACL language much
easier to use than perl, because it's actually designed for processing

I've never enjoyed having to learn a new language for every task. I suppose reusable code is a matter of taste. In fact, all the normal operations are supplied as perl subroutines so the part you customize consists of a few lines to call the subroutines in the right order and respond according to the returned values. For example, if you are running MimeDefang on a rely machine that doesn't have the user base, there is a routine supplied to check via smtp that the user exists on the delivery server before responding to the sender's recipient check. Also, per the name, MimeDefang splits out the mail parts into files once and optionally unzips zipped attachments (using a re-usable CPAN module, of course) so that the subsequent multiple passes of spam and virus scans don't have to repeat this work.

And of course you can call out to separate scanning processes
(running with their own uid) such as spamd, clamd etc.

Spamassassin, being perl itself, is more efficiently run inside the MimeDefang process. For clamd, it passes the location of the unpacked files over the socket.

  Les Mikesell
   lesmikesell gmail com

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]