View Single Post
  #2 (permalink)  
Old 04-20-2006, 01:49 PM
KevinH KevinH is offline
Zimbra Employee
 
Posts: 4,792
Default

You'll see the test produce the exact same result except for these two:

AWL=0.356, BAYES_00=-2.599

vs

AWL=-0.440, BAYES_05=-1.11

The first is the auto-white list - http://wiki.apache.org/spamassassin/AutoWhitelist

You can read from the wiki that it will vary over time and try to push the score in the direction of it's most recent scores. So that guy is just trying adjust for the difference in the BAYES_X scores. The Bayes tests will vary by the headers as Bayes tokenizes everything. You can read more here:

http://wiki.apache.org/spamassassin/BayesFaq


The bottom line is it's no big deal since both messages are clearly good mail and they result is such. What your seeing is just a little fluctuation in SA and this is expected and why there are so many tests ran on each message before something is declared as spam or not.
__________________
Bugzilla - Wiki - Downloads - Offline Client
Reply With Quote