https://bz.apache.org/SpamAssassin/show_bug.cgi?id=8510
Bill Cole <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #2 from Bill Cole <[email protected]> --- If you're referring to mail-tester.com, to the best of our knowledge they have historically used stale rules and unknowable configurations. The results from their checks are not useful for us in validation of rules. The fact that they are giving PDS_OTHER_BAD_TLD a score indicates that *at least* they are using obsolete rules. Our validation of rules is done based on SA users submitting results of "mass check" runs of their classified mail, which are used to determine the scoring and enablement of rules. The results are at https://ruleqa.spamassassin.org. There is a current test rule breaking out .xyz from the broader set of TLDs covered by the rules you cited: T_SCC_TLD_XYZ. It currently shows that 84% of mail using .xyz addresses were spam in recent mail. That's an improvement from past runs but it is still bad. The numeric domain name is significant as well and PDS_NAKED_TO_NUMERO is a meta-rule includes that as a sub-test but also requires a "naked" To address, i.e. one email address without enclosing angle-brackets and no display name. That is likely to be idiosyncratic to the mail you sent to mail-tester.com for testing and unlikely to be seen in normal email. As Sidney notes, the FROM_SUSPICIOUS_NTLD* overlap is intentional. I will just add that while the scores for those are determined by the rescoring algorithm, they both are strong enough that we've had to cap their scores at 0.5 and 2.0 because otherwise their false positives would be overwhelming. SA *by design* has rules that mostly match spam but sometimes match ham. that's why so few rules are scored over 2 with a standard threshold of 5. It's also true that we have many "nice" rules which score negatively (less likely to be spam) and match substantially more ham than spam. -- You are receiving this mail because: You are the assignee for the bug.
