Upgrade woes - Debian

[David Relson]

On Fri, 10 Dec 2004 19:58:27 -0600
Karl Schmidt wrote:

> David Relson wrote:
> > On Fri, 10 Dec 2004 19:20:58 -0600
> > Karl Schmidt wrote:
> > 
> > 
> >>Ok - I got things working -
> >>Deleted old data base, retrained
> >>
> >>Had to change the file ownerships to Debian-exim.Debian-exim
> >>
> >>Then the fun.
> >>
> >>The default X-bogosity line contains the word "spamicity" which 
> >>mozilla's filters don't see as any different than just "spam"
> >>
> >>I changed "spamicity" to "s-pamicity" in /etc/bogofilter.cf and have

> >>things working again.
> >>
> >>Might want to change the default.
> > 
> > 
> > Hi Karl,
> > 
> > So, mozilla isn't cooperating, eh?  Does it allow case sensitive
> > checking, i.e."Spam" != "spamicity", or inclusion of punctuation,
i.e
> > "spam," ??? 
> 
> Both Mozilla and the newer Mozilla-thunderbird mail filters are case 
> insensitive. I'm not sure exactly what grep function they use, but it 
> really doesn't matter because there are many clients that will all
have 
> something different. In any case, there will be a lot fewer people 
> running into this if the default line changes.
> 
> BTW there is a grave bug listed on Debian which needs to be tended to 
> for all  - those log files - at 10meg each could bring down a lot of 
> systems. Might need an auto deletion method to keep things from
getting 
> too big. I archive my spam (for training AND if there ever is a class 
> action law suit I will have the evidence to collect <grin>) along with

> my ham so I can simply retrain in the case of a failure. I break my
spam 
> and archives down by year an only train on the last year or two of
email.
> 
>  >
>  > Good work in using bogofilter.cf!  You could also change the
>  > Spam/Ham/Unsure tags to alpha/beta/gamma or any other threesome.
> 
> Once I hand sorted my normal inbox mail from the 200 probable spams I 
> got today I went and edited the mbox file and replaced spamicity with 
> s-pamicity.

Changing the X-Bogosity: line "spamicity=..." to "score=..." mightn't be
a terrible thing to do.

> See:
>
http://bugs.debian.org/cgi-bin/pkgreport.cgi?which=pkg&data=bogofilter&archive=no

If you don't care to preserve the logfiles, the "massive memory leak"
can be addressed by running "db_archive -dh /your/bogofilter/dir"
periodically.  As I think further, on debian it's probably "db42_archive
..."

> BTW2
> I have thought about using bogofilter to sort my mail for many other 
> jobs - so that training on my manual sorts will cause it to get 
> automatic. I could plug bogofilter into sylpheed to see what I
> forwarded to where and based on the word statistics could start
> guessing what actions I will probably take based on content. (Sure
> wish sylpheed worked better with IMAP).
> 
> 
> One last question - I see examples of training based on mbox files
> that don't use the -M  Does it matter?

No.

> 
> Anyway - thanks for the software.

My pleasure...