Matt> I'm very ready to believe that having the cache persist makes a
Matt> lot of difference given how Skip uses it.
Correct. I run my train-to-exhaustion script which retrains on my entire
training database. Not having the ip addresses cached to disk means that
runs slow every time. I suspect the same would be true of any full-retrain
mechanism with x-lookup-ip enabled. If you're just training or scoring on
individual messages or have a long-running process like the pop3 proxy that
keeps the cache alive in memory the incremental cost is probably not bad.
Skip
_______________________________________________
[email protected]
http://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.net/faq.html
