-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Ewan Edwards writes: > Hi list, > > I'm curious about the mechanism for processing rules, specifically, is there > a distinct advantage to compiling rule subroutines into the > Mail::SpamAssassin::PerMsgStatus namespace compared to, say, using anonymous > subs? Hi Ewan -- the main reason that is done is so that we can use perl's Devel::DProf module to profile and identify slow rules. It works well, and there's not a whole lot of effort beyond using anon subs. > I'm trying to get per-user rules running within the spamd environment as it > is not feasible to invoke spamassassin per message delivery given current > mail volumes on our servers. Also, per-user rules are becoming a necessity > since the default rule sets (SA 2.6.3) are no longer very effective against > current spam. > > None of the user's rules will require the eval:method(args) facility, > however, multiple users may use the same name. Any problems with "allow_user_rules"? that should work.... > Unless there is an advantage to using named subroutines, I will be modifying > PerMsgStatus.pm to process rules directly. Should the eval: rule definition > facility become required, I would use anonymous subs, so that multiple > users' rules do not conflict with each other. > > If there are no real complaints about this sort of modification, would a > patch for the change be welcome? yep, unless allow_user_rules works OK ;) - --j. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) Comment: Exmh CVS iD8DBQFA71P/QTcbUG5Y7woRAktRAJ9gvn7Pd/UCvDUayKgNQh7oIEKZCQCgp1Qj m2jAleZDN72F297H2wFAHhw= =FzjF -----END PGP SIGNATURE-----
