The Tuesday 2004-01-27 at 10:35 +0200, isofroni@cc.uoi.gr wrote:
Alright, it finally worked.
Good! :-)
Another issue is about spamassassin "decisions" for the same mail. Two similar mails can score different points (one paased and the other failed)
Do you know why is that happening?
Look carefully at the report - this can be hidden as headers, or in clear text. For one that passed, you will see something like this: |X-Spam-Checker-Version: SpamAssassin 2.61 (1.212.2.1-2003-12-09-exp) on | nimrodel.valinor |X-Spam-Level: **** |X-Spam-Status: No, hits=4.8 required=5.0 tests=BAYES_90,MIME_BOUND_MANY_HEX | autolearn=no version=2.61 This one got 4.8 points, less than 5, so it passed; You can see the tests that indicate the level of "spamminess". For one marked as spam, there is a report in clear (or no report, depends on your personal configuration), giving details of each test. First, the headers: |X-Spam-Flag: YES |X-Spam-Checker-Version: SpamAssassin 2.61 (1.212.2.1-2003-12-09-exp) on | nimrodel.valinor |X-Spam-Level: ********** |X-Spam-Status: Yes, hits=10.1 required=5.0 tests=BAYES_99,MIME_BOUND_MANY_HEX, | NIGERIAN_BODY1 autolearn=no version=2.61 And then, the report: |Content analysis details: (10.1 points, 5.0 required) | | pts rule name description |---- ---------------------- -------------------------------------------------- | 2.7 MIME_BOUND_MANY_HEX Spam tool pattern in MIME boundary | 5.4 BAYES_99 BODY: Bayesian spam probability is 99 to 100% | [score: 1.0000] | 2.0 NIGERIAN_BODY1 Message body looks like a Nigerian spam message 1+ Finally, you can do more things. You can "teach" the bayesian filter about what you consider spam and what not - you will find details about this on spamassassin documentation, because it is long to explain. It has been commented as well on the list many times, simply browse the list archive looking for spam or bayes in the subject line. Look also at the unofficial SuSE FAQ. -- Cheers, Carlos Robinson