[Mimedefang] Including an email address in an action_bouncereply message.

[Martin Ferguson]

> 
> I was probably going to start bouncing mail which 
> SpamAssassin marks at 20
> or above.  I was then going to slowly lower that number down. 
>  But, I don't
> imagine I would ever go below 10 to bounce mail.  Anything 
> else that makes
> it through would be marked as SPAM.
> 

For info, I've been running a filter for three weeks now, which tags
mail as spam which have a score of 5 or over, only 2 messages out of
several thousand that were marked as spam have been marked incorrectly
and both of these were html mails from mailing lists (I suppose anyone
who runs a mailing list and sends their announcement mails in html
deserves them to be bounced ;-))

That said, perhaps bouncing at 5 is a bit harsh, does anyone have a
filter which will tag as spam at 5 and bounce at 10?

Sorry my perl is not very good, I found one at
http://thompsonic.com/util/antispam/mimedefang/script1.html but I can't
seem to get it to work.

Thanks for any help
Martin.




> -----Original Message-----
> From: mimedefang-admin at lists.roaringpenguin.com 
> [mailto:mimedefang-admin at lists.roaringpenguin.com] On Behalf 
> Of Jeff Grossman
> Sent: 10 March 2003 18:07
> To: mimedefang at lists.roaringpenguin.com
> Subject: Re: [Mimedefang] Including an email address in an 
> action_bouncereply message.
> 
> 
> on 3/10/03 9:07 AM, Nels Lindquist at nlindq at maei.ca wrote:
> 
> > On 10 Mar 2003 at 14:51, Martin Ferguson wrote:
> > 
> >> I'm currently using action_bounce to bounce mails which get a
> >> spamassassin score of higher than 5, I'd like to include 
> an alternate
> >> email address in the reply message in case I bounce 
> anything which isn't
> >> spam, something along the lines of....
> > 
> > Bouncing at 5 is going to result in a significant number of legit
> > messages being bounced.  Note that the SpamAssassin developers very
> > strongly recommend against bouncing *at all,* (though some of us
> > still do it :-) ).  Our MD/SA setup bounces at 10 and includes spam
> > score headers for everything else, which still rejects the vast
> > majority of the spam, while allowing through the marginal false-
> > positives (which tend to cluster very close to 5).
> Jeff
> -- 
> Jeff Grossman (jeff at stikman.com)
> 
> _______________________________________________
> MIMEDefang mailing list
> MIMEDefang at lists.roaringpenguin.com
> http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
>