On 3/26/10 7:39 AM, Michael Scheidell wrote:
OnI can't think of a way for the GA to know that the rule contains the
same info as a DNSBL test. There are rule overlap stats, but I don't
think that would be enough with only a small number of ham occurrences.
https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6390
mostly because all the rest of the KHOP_SC* rules already take into
account the fact that it also triggers the spamcop rule anyway
score KHOP_SC_CIDR16 2.200 1.291 2.200 1.291
score KHOP_SC_CIDR24 2.649 1.198 2.649 1.198
score KHOP_SC_TOP200 4.000 3.944 4.000 3.944
score KHOP_SC_TOP_CIDR16 2.797 0.728 2.797 0.728
score KHOP_SC_TOP_CIDR24 0.001 0.000 0.001 0.000
score KHOP_SC_TOP_CIDR8 2.200 0.001 2.200 0.001
this would be one of those 'manual tweaks', and you can't be in
khop_sc_top200 unless you are in spamcop.
(eg: khop_sc_200 is the top 200 spamming ip's in spam cop), so , by
that definition, and with that logic, you don't need a meta.
if its in khop, it would have hit spamcop. if you run local tests,
khop will run but spamcop rbl won't, thus the two different scores,
one with local tests only (3.999) and the other for with network tests
(2.65)
I think the correct thing to do would be to modify the KHOP_SC_TOP200 to
be a meta that doesn't fire if the corresponding spamcop DNSBL rule
fires. Perhaps you could open a bug at
http://issues.apache.org/SpamAssassin/ about it.
Daryl
______________________________________________________________________
This email has been scanned and certified safe by SpammerTrap(r).
For Information please see http://www.secnap.com/products/spammertrap/
______________________________________________________________________
