On 17/02/06, W B Hacker <wbh@???> wrote:
> Chris Bolton wrote:
>
> > Peter Bowyer wrote:
> >
> >> On 17/02/06, Nigel Metheringham
> >> <nigel.metheringham@???> wrote:
> >>
> >>
> >>> On Fri, 2006-02-17 at 09:29 +0000, Chris Bolton wrote:
> >>>
> >>>
> >>>> X-Spam-Status: No, score=0.4 required=7.5
> >>>> tests=AWL,BAYES_00,FORGED_RCVD_HELO,---^^^---
> >>>>
> >>>> HTML_90_100,HTML_MESSAGE,HTML_SHORT_COMMENT,HTML_TITLE_EMPTY,
> >>>> MISSING_SUBJECT autolearn=no version=3.1.0
> >>>>
> So long as you have 'autolearn=no' you might even want to drop
> the BAYES completely. Or at least it's presence in the
> announcement string.
autolearn=no means that this message wasn't either hammy enough or
spammy enough to be worth learning from (thresholds adjustable in the
SA config). autolearn=disabled would mean it's turned off.
Peter
--
Peter Bowyer
Email: peter@???
Tel: +44 1296 768003
VoIP: sip:peter@???
VoIP: *5048707000@???
FWD: **275*5048707000
VoipTalk: **473*5048707000
