[spambayes-dev] Speedup for full retrain when using DB dict

[Tim Peters]

[Tim[
> Wouldn't it be simpler to do the full retrain using a
> PickledClassifier instance, then populate a DBDictClassifier
> from the result?  That would also skip the extra layers of
> code (and time) to maintain the changed_words dict during
> the retrain.

[Skip Montanaro]
> Perhaps.  Are you suggesting I detect the zero-length shelve object
> before instantiating a classifier or instantiating a
> PickledClassifier from within the DBDictClassifier code?

They should both be simple enough to try and time, and then let your sense
of beauty guide you <wink>.  What I'd like to avoid is stuff trying to make
a class's implementatinon act like a different class's implementation, and
especially when that latter class already exists and does a fine job as-is.