On Tue, 2013-10-22 at 13:49 -0700, Quanah Gibson-Mount wrote:
> --On Tuesday, October 22, 2013 12:24 PM -0700 John Hardin
> <[email protected]> wrote:
>
> > On Tue, 22 Oct 2013, Quanah Gibson-Mount wrote:
> >> X-Spam-Status: No, score=0.348 tagged_above=-10 required=3
> >> tests=[BAYES_50=0.8, DKIM_SIGNED=0.1, DKIM_VALID=-0.1,
> >> DKIM_VALID_AU=-0.1, HTML_IMAGE_RATIO_02=0.437, HTML_MESSAGE=0.001,
> >> RP_MATCHES_RCVD=-0.8, T_HEADER_FROM_DIFFERENT_DOMAINS=0.01]
> >> autolearn=ham
> >
> > What are your thresholds set to? You might want to lower your ham
> > learning threshold and zero the RP_MATCHES_RCVD score.
> 10_default_prefs.cf:ifplugin Mail::SpamAssassin::Plugin::AutoLearnThreshold
> 10_default_prefs.cf:bayes_auto_learn_threshold_nonspam 0.1
> 10_default_prefs.cf:bayes_auto_learn_threshold_spam 12.0
>
> However, as I read the docs, the score is supposed to be lower for it to be
> autolearned. Last I checked, 0.348 > 0.1, so why was this autolearned as
> HAM if the cutoff is 0.1?
The Description section of the AutoLearnThreshold doc explains it:
Certain tests are ignored when determining whether a message should be
trained upon. Most notably that includes the BAYES_xx rules.
Moreover, auto-learning occurs using scores from either scoreset 0 or 1,
depending on what scoreset is used during message check. It is likely
that the message check and auto-learn scores will be different.
In other words: Non-Bayes ruleset scores may differ from the scores
listed above. The score for BAYES_50 definitely needs to be subtracted.
Which results in a negative score...
The usefulness of RP_MATCHES_RCVD is currently under discussion. I
suggest to zero out that rule, or assign it a negative zero.
--
char *t="\10pse\0r\0dtu\0.@ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1:
(c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}
