question on multiple wordlists

David Relson relson at osagesoftware.com
Fri Oct 11 20:40:22 CEST 2002


At 01:31 PM 10/11/02, Eric Seppanen wrote:
>On Fri, Oct 11, 2002 at 01:20:18PM -0400, David Relson wrote:
> > You're right, -n and -s aren't needed with --wordlist.  Each of the 
> options
> > implies operations on a single wordlist.
> >
> > However consider the following:
> >
> > bogofilter -d mydirectory -n <goodmsg
> > bogofilter -d mydirectory -s <spammsg
> >
> > Given the possibility of typing -n when -s was meant, -N and -S will 
> remain
> > useful with -d.  Given that the user may do:
> >
> > bogofilter --wordlist myspamlist <spammsg
> > bogofilter --wordlist mygoodlist <goodmsg
> >
> > There's still a need for -N and -S which take two wordlists.  This 
> leads to
> > the value of --spamlist and --goodlist.
>
>I still disagree.  Remember, -N and -S exist for people that are
>automatically updating the lists.  There's no reason to automatically
>update user-defined lists, that's what the stock goodlist/spamlist are
>for.

Not so.  -N and -S are for correcting mistakes.  Whether it was a mistake 
in automatic classification or a mistake in manually running bogofilter, 
mistakes can and will happen.


>Hand-maintained db-files will be updated infrequently, if at all.  The
>only obvious purpose for such a file would be to download big spam DBs
>from the net, and there's no reason to ever update those.

It seems reasonable to treat a standard wordlist as read-only.  We have the 
ignore-list and (possibly white and black lists) to counter disliked 
elements of a standard list.  "countering" might be more efficiently 
implemented by correcting the list, rather than having another list to 
supercede the first one.

FWIW, as I write this, I'm counting word lists - system good, bad, ignore, 
white, and black; user ....  The count could go far beyond our basic two.

>Plus, I think there's easy ways to screw it up and accidentally mess up
>your goodlist or badlist too:
>
>bogofilter -N --spamlist myspamlist <msg
>
>If you insist, I think that an option to subtract a message from a given
>list could be implemented, without risking the trouble:
>
>bogofilter --subtract --spamlist myspamlist <msg
>
>This is the only possible way to correct a situation where you're only
>building a spam list (for net distribution), with no matching goodlist.
>
>---------------------------------------------------------------------
>FAQ: http://bogofilter.sourceforge.net/bogofilter-faq.html
>To unsubscribe, e-mail: bogofilter-unsubscribe at aotto.com
>For summay digest subscription: bogofilter-digest-subscribe at aotto.com
>For more commands, e-mail: bogofilter-help at aotto.com


For summay digest subscription: bogofilter-digest-subscribe at aotto.com



More information about the Bogofilter mailing list