db maintenance "delete oldest least used tokens, but maintain count of x"

[Boris 'pi' Piwinger]

David Relson <relson at osagesoftware.com> wrote:

>> > I'm not sure what to do about .MSG_COUNT either.  My educated guess
>> > is"don't worry.  do nothing."
>> 
>> Won't work in the long run. One half of your tokens have expired,
>> .MSG_COUNT is way too large.
>
>Is it?  Or does it reflect shorter messages.  When "xyz" expires, it's
>comparable to saying that the message was 1 token shorter than
>bogofilter originally thought it to be.

Imagine one of the deleted tokens returns. It will be seen
as far less likely than it should be.

>As a guess, tokens will expire more or less evenly from ham and spam.

I have really no idea what will happen here. But I would not
make that assumption.

pi