Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Matt Kettler
At 07:13 PM 8/14/2003 -0700, Ricardo Kleemann wrote: Thanks Dan, I have taught it with --spam with a significant number of messages, although I haven't taught it with --ham yet. Is that a problem? If the bayes database doesn't have 200 ham messages in it's learning database, the use of the bayes

Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Ricardo Kleemann
Hi, In this case, the user nobody does have a valid home, and it also has the .spamassassin directory with the bayes files in it. So again, I'm confused as to why bayes rules are not being accounted for in the SA report headers. - Original Message Follows - > At 07:15 PM 8/14/2003 -0700

Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Matt Kettler
At 07:15 PM 8/14/2003 -0700, Ricardo Kleemann wrote: Hi, Is there a problem running spamc as "nobody"? I'm actually using "nobody" as the user running it, and I have run sa-learn as "nobody" as well. I figured it doesn't matter which user it is, as long as that particular user has the bayes datab

Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Ricardo Kleemann
Thanks Dan, I have taught it with --spam with a significant number of messages, although I haven't taught it with --ham yet. Is that a problem? Ricardo - Original Message Follows - > Hi, > > I'm pretty sure that Bayes analysis doesn't kick in until > the program has had a chance to le

Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Ricardo Kleemann
Hi, Is there a problem running spamc as "nobody"? I'm actually using "nobody" as the user running it, and I have run sa-learn as "nobody" as well. I figured it doesn't matter which user it is, as long as that particular user has the bayes database and tokens setup. Ricardo - Original Messa

Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Martin Radford
At Thu Aug 14 17:40:19 2003, Ricardo Kleemann wrote: > > The output from spamc didn't even show a score for bayes. > Shouldn't it at least show a percentile score for bayes? > Here's what it reported: > > X-Spam-Status: No, hits=1.7 required=5.0 > > tests=MAILTO_TO_SPAM_ADDR,MSGID_HAS_NO_

Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Matt Kettler
At 01:23 PM 8/14/2003 -0700, Ricardo Kleemann wrote: This is a message that has been run through sa-learn with --spam, and yet I get nothing reported back from spamc. I did restart spamd after altering local.cf. I'm running spamc as the same user that ran sa-learn. I'd try running spamd with debugg

Re: [SAtalk] Help understanding spamc and bayes

2003-08-14 Thread Ricardo Kleemann
Hi, Thanks for the suggestion. I added the settings to local.cf as suggested, then ran the message through spamc again, however it reports the exact same thing as before. :-( This is a message that has been run through sa-learn with --spam, and yet I get nothing reported back from spamc. I did r