On 07/09/06, Tim <ignored_mailbox@xxxxxxxxxxxx> wrote:
The only difficult part is working how you keep that data (what
identifies the spam). It can rack up quite quickly, so you might want
to consider this a short-lived thing (dispose of the information
rapidly).
We run our own pyzord and feed mail from spamtraps to it. By default the records stay in pyzord's db for 3 months. This suits us and checking against it is much quicker than the whole SA Bayes thing.
Keith Brazington
Quetz Limited