[Spambayes] detailed differences between branches
Neale Pickett
neale@woozle.org
Sun Nov 24 09:48:01 2002
In case anyone was wondering, the protagonist of the movie won.
So here's what I think will be different--what will change when I
merge. I'm guessing on a few things here because I'm really sleepy and
I don't know what all these modules do. When in doubt, I'm assuming
what's in HEAD is newer than what's in the playground branch.
Tim, care to comment on Corpus.py, FileCorpus.py, and CostCounter.py?
Options.py
* No more show_best_discriminators
* persistent_storage_file becomes pop3proxy_persistent_storage_file and
hammiefilter_persistent_storage_file
* ditto for persistent_use_database
Tester.py
* Changed some calls to new classifier.Classifier arguments
classifier.py
* PICKLE_VERSION is now 4
* WordInfo stores only spamcount, hamcount
* New MetaInfo class stores nspam, nham
* class name change: Bayes -> Classifier
* Classifier class computes spamprob on demand and caches result. This
means you don't need to call update_probabilities() anymore. But
it's still there if you really really want to call it.
* As a result, the third argument to Classifier.learn() has been
removed.
* New Classifier.probability() method takes a record (WordInfo object)
as an argument and returns the probability for that record. This is
what does the caching.
dbdict.py
* DBDict class moved out of hammie.py and into here
* We know it's ugly right now.
hammie.py
* Just contains Hammie class now, if you try to run this as __main__,
it will run main() from hammiebulk.py for compatibility.
* Also has a handy open() function to open a database or pickle and
return a Hammie object.
hammiefilter.py:
* funtions moved into HammieFilter class
* default db is "~/.hammiedb". I don't know what's reasonable for the
Windows camp, so I'm leaving it up to you folks to figure that out.
pop3proxy.py:
* Uses Classifier class instead of Hammie class
hammiebulk.py:
* The canonical executable Hammie front-end. Now does dishes!
* You must pass in -D or -d, there is no default.
Man. I've got my work cut out for me.
'Night
Neale
More information about the Spambayes
mailing list