registering spammails

[David Relson]

At 01:16 PM 2/4/03, Oleksandr Moskalenko wrote:

>Yes, I run "bogofilter -S" on every spam message I get from within mutt.
>Should I change it to "bogofilter -s"? I remember reading a note on the
>web about some bugs in the bogofilter that made using "-N" and "-S"
>undesirable and where there was a warning to use "-n" and "-s" instead.
>Has that part been worked out by now? Is it safe to use "-S" and "-N".
>
>Alex.
>
>P.S. I just piped a message through "-S" and now .MSG_COUNT shows
>
>                        spam   good
>.MSG_COUNT                1   1334
>
>So it looks like it was good to start from scratch whatever was the
>cause. I think I might have inadvertently run A LOT of messages through
>"bogofilter -N" some time ago when I was sorting through one of my
>mailing list mailboxes, tagging tens of messages at once, and saving
>them elsewhere as the 's' keymap was bound to the command above. It
>couldve hit the spamlist.db hard or something of that kind, couldn't it?
>
>Alex.

Alex,

An incorrect keymap could cause your wordlists to be not what you 
want.  Bogofilter doesn't really care as its job is to use the wordlists 
_you_ have given to it and evaluate messages with those wordlists.  Of 
course feeding messages to the incorrect wordlist will affect the 
results.  An incorrect keymap could definitely cause problems :-(

As to the registration options, -s, -n, -S, -N, and -u, they are all 
working properly.

-s and -n are for simply adding messages to the spamlist and the 
goodlist.  They are the options to use for new messages that bogofilter 
hasn't seen before.  -S and -N are for messages that were registered 
incorrectly and that you want to correct.  Effectively -S and -N take 
messages out of one wordlist and add them to the other.  What they really 
do is decrement and increment counts.  The process may leave behind a token 
with a count of 0, which is O.K.

-u is used as part of classification and will feed the message to the 
spamlist or the goodlist, depending on whether the message is classified as 
spam or as ham.  You don't have to use -u and some people say it shouldn't 
be used, while others use it happily.

David