Sandy,
The utility in this is that the tests failed would be centralized in
one application, Delude, and available for combo tests and easier
diagnosis within that framework.
Of course SpamAssassin has a multitude of different tests and scores
within those tests, and it isn't practical to do everything in bitmask
fashion, but I could see some utility in separating out a bayesian test
as BAYESIAN-LOW, BAYESIAN-MED and BAYESIAN-HIGH for instance, or
lumping together multiple Nigerian/Advance Fee Fraud filters into one
separate test trackable by Declude.
I understand that this probably isn't within the scope of your current
plans, but it does offer some interesting potential for the future. I
personally find the combo filtering technique within Declude to be
very, very valuable. For instance finding out that a message hit both
a DUL list and SPAMCOP is enough to delete a message, but these tests
are not reliable enough to even hold on individually (IMO of course).
Matt
Sanford Whiteman wrote:
The latest Declude's support bitmasked result codes. You might
consider creating a user maintained map from tests failed in SA to
bitmapped result codes returned to Declude. I'm not sure what the
limit might be in Declude for the number of unique tests, but I have
managed to use 16 so far without issue. It probably is limited to
either 16 or 32 if it is limited. It might very well not be.
Interesting possibility, and I can see the usefulness. . . would be a
certain bandwidth saver vs. having different SPAMD instances (and
different SPAMC32 runs) for different rulesets, but what you save in
bandwidth might be offset be a nearly ridiculous amount of complexity
on the client side. One SPAMD might have tens of rulesets and
thousands of rules, so trying to bitmask to identify the rules would
be pretty crazy. Bitmasking at the ruleset level would be doable, but
you wouldn't get the actual offending RegEx line from that, just the
file that contained it.
Rather than trying to match result codes up to a matrix of all
possible results, I think such a full-featured external test as
SpamAssassin is really a better push for a "dumb" report.txt kind of
feature, which could simply gulp up a TESTSFAILED-style string
returned from SPAMC32 and either put it in the logs without
inspection, put it into a Declude variable for further use, or
whatever.
In the meantime, SPAMC32's upcoming local logging by queuefile name
should help immensely with tracking results by rule. Remember that SA
has its own weighting features, so tweaking rule-level weights and
inter-rule/inter-ruleset dependencies lets you do weight management
like Declude's before returning an aggregate SPAMC32 weight. . .
albeit at a management cost vs. doing as much as possible within
Declude.
I'd see bitmasking in the far future, though I can't see putting aside
time for it right now.
Thanks,
Sandy
------------------------------------
Sanford Whiteman, Chief Technologist
Broadleaf Systems, a division of
Cypress Integrated Systems, Inc.
e-mail: [EMAIL PROTECTED]
SpamAssassin plugs into Declude!
http://www.imprimia.com/products/software/freeutils/SPAMC32/download/release/
Defuse Dictionary Attacks: Turn Exchange or IMail mailboxes into IMail Aliases!
http://www.imprimia.com/products/software/freeutils/exchange2aliases/download/release/
http://www.imprimia.com/products/software/freeutils/ldap2aliases/download/release/
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail". The archives can be found
at http://www.mail-archive.com.
|
