[Mimedefang] spamassassin vs mimedefang/spamassassin - different results

[Michael H Moran]

On Tue, May 06, 2003 at 03:08:17PM -0400, David F. Skoll wrote:
> On Tue, 6 May 2003, Michael H Moran wrote:
> 
> > Spamassasin reports the following tests:
> 
> >         BAYES_60,FROM_ENDS_IN_NUMS,MAILTO_TO_SPAM_ADDR,
> >         MSG_ID_ADDED_BY_MTA_3,NO_REAL_NAME,UPPERCASE_75_100,
> >         USER_AGENT_IMP,US_DOLLARS
> 
> MIMEDefang by default turns off the BAYES rules.  In a system-wide
> installation, it's hard to train the Bayesian analyzer.

Interesting.. that would explain a great deal.   

Is that always the case, or can it be overridden in local.cf or 
sa-mimedefang.cf?

Given that this is being run on my laptop, my id and root share the
same BAYES rules, owner root, group defang - 775 (symlinks), I wouldn't 
think it would be a problem.

On a side note, I just upgraded to 2.33 an the first piece of spam I 
recieved hit the following tests:

BAYES_60,MAILTO_TO_SPAM_ADDR,MIME_BOUND_MANY_HEX,
MSG_ID_ADDED_BY_MTA_3,NIGERIAN+_BODY,US_DOLLARS

Thanks again.. great tool.

Mike
mhm at austin.ibm.com