Token Timestamp Updates

[RW]

I notice that Bogofilter doesn't update token timestamps
on classification.   Is this just for Berkley DB? Is there any way of
changing it?


The problem is that without updates, or train-on-everything, it isn't
sensible to purge tokens by age. If you try to do it you could end-up
with odd results. For example if you train from corpora and then train
on unsure/error you may find that almost all of the wordlist expires on
the same day.