On Tue, 30 Sep 2003, Steven Manross wrote:
> Would anyone find it beneficial to have this added to SpamAssassin, or
> is it just me? :)
>
> $status->check_internal_first($mail);
>
> Where check would be performed with "local_only" tests first... and
> then if it is tagged using locals only, you don't bother with RBL and/or
> external tests, because it's already been tagged, and you don't need to
> add the cost of the external tests if it's already spam..
>
> Just a thought.
>
> I've been running into performance issues while processing large volumes
> of mail through SA if I don't do this type of "check_internal_first"
> methodology.
>
> I have a prototype that I am playing with.
>
> Steven
Check the archives, variations on this idea come up time and again.
case #1:
It looks spammy, so you flag it and quit. BUT if you had checked against
BondedSender (a net test) you'd find that it was OK. (Some eBay messages
can look spammy).
case #2:
It looks mildy spammy (just makes it over your threshold) so you flag
it and quit. BUT if you had hit it with the net checks its score would
have gone way up and thus been added to your Bayes auto-learn.
So you miss out on an opportunity to better train your Bayes.
The SA scoring set is based upon all the tests getting a chance
to say their piece (that's why there are different sets of values
for different environments; net, no-net, no-bayes, etc).
--
Dave Funk University of Iowa
<dbfunk (at) engineering.uiowa.edu> College of Engineering
319/335-5751 FAX: 319/384-0549 1256 Seamans Center
Sys_admin/Postmaster/cell_admin Iowa City, IA 52242-1527
#include <std_disclaimer.h>
Better is not better, 'standard' is better. B{
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
