[Spambayes] Ver 1.0.4 (downgrade from 1.1a1) sb_server.pystarterror

[Tony Meyer]

> So I launched the commandline 
> learning (as I have my spam saved into files):
> 
> -----------------------
> [vys0 at agnich spambayes-1.0.4]$ sb_mboxtrain.py -g ~/tmp/ham1 
> -g ~/tmp/ham2 -g 
> ~/tmp/ham3 -g ~/tmp/ham4 -g ~/tmp/ham5 -s ~/tmp/spam
> Training ham (/home/vys0/tmp/ham1):
>   Reading as Unix mbox
>   Trained 0 out of 19 messages
[...]
> Strange - I worked on those files before, and they were OK. 
> Spambayes was able to learn on them. What could be the reason?

I don't use sb_mboxtrain myself, but I believe it rewrites the mboxes that
it reads with "X-Spambayes-Trained" headers, so that it knows which messages
it has trained.  Try using the "-f" switch to force training:

    sb_mboxtrain.py -f -g ~/tmp/ham1 -g ~/tmp/ham2 -g ~/tmp/ham3 -g
~/tmp/ham4 -g ~/tmp/ham5 -s ~/tmp/spam

BTW, bulk training may not give the best results.
<http://entrian.com/sbwiki/TrainingIdeas> has a lot more details.

> And, as for CVS version - will it overwrite "corrupted" 
> files? As I use 1.0.4 
> ATM - will it work fine with this version ? :>

I'm not sure exactly what you mean - could you elaborate?  1.0.4 should work
fine.  If you later upgrade to 1.1 or CVS, it should use your existing
configuration/training without any work on your part.

=Tony.Meyer

-- 
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes.
http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this.