http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5503
Summary: FR: autolearn_force will force Bayes learning according
to tflag setting
Product: Spamassassin
Version: unspecified
Platform: All
OS/Version: other
Status: NEW
Severity: critical
Priority: P1
Component: spamassassin
AssignedTo: [email protected]
ReportedBy: [EMAIL PROTECTED]
>From the SA doc I haven't been able to figure out how to get bayes to
autolearn spam if it hits a rule.
tflags SYMBOLIC_TEST_NAME [ {net|nice|learn|userconf|noautolearn|multiple} ]
For example,
- I have network tests enabled
german txt stock spams hit a score of 25 due to my custom rules but as
there were no network hits, the msg was never learnt, and Bayes remained
bayes_00.
I had to enable a crappy RBL with a score of 0.001 and then autolearn
started catching them.
It would be great if we could set a tflag autolearn_force
imho, to use autolearn_force with dead safe header rules, metas, etc.
would greatly help keep feeding bayes in a powerful way (mainly to avoid
the bayes_00 effect.)
so if the rules has a negative score, Bayes will learn the msg as HAM ...
a positive score would do the obvious spam learn.
(pls pardon any non technical, unorthodox language)
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
