bogofilter (-o) & bogoutil (-c) bugs?
David Relson
relson at osagesoftware.com
Mon Feb 17 15:37:49 CET 2003
Bill,
Thanks for your message. Detailed responses follow...
David
At 11:35 PM 2/16/03, W M Brelsford wrote:
>I'm new to this list, although I've been using bogofilter
>successfully for several months. A few problems I've
>encountered:
>
>1. bogofilter -o <value> seems to have no effect.
'-o' sets the spam_cutoff value. Messages with scores above spam_cutoff
are classified as spam. In the test below, I've classified a message that
has a score of 0.509410, which is less than the default spam_cutoff of
0.54. The message was classified as "No" (not spam). I then reran
bogofilter with "-o 0.500" to lower the spam_cutoff value and the messagee
is classifed as "Yes" (spam). Here's the test:
[relson at osage 0103]$ bogofilter -C -v < msg.test
X-Bogosity: No, tests=bogofilter, spamicity=0.509410, version=0.10.3
[relson at osage 0103]$ bogofilter -C -v -o 0.500 < msg.test
X-Bogosity: Yes, tests=bogofilter, spamicity=0.509410, version=0.10.3
>2. The bogofilter man page lists two meanings for "-c". Using it
> to specify a config file works; using it to set a minimum
> count has no effect (and yields no error message).
The man page is wrong. It looks like some of bogoutil's options were put
into bogofilter's man page. I'll correct that.
>3. bogoutil -c1 should remove tokens with 0 counts, but also
> removes counts of 1. -c0 removes nothing.
The code is keeping tokens with counts above the specified value. I'll
correct the documentation.
>I still have these problems with 0.10.3 (using the i586 rpm).
The documentation will be corrected later today and the fixes will be in
the next release.
More information about the Bogofilter
mailing list