https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7735
--- Comment #61 from Henrik Krohns <apa...@hege.li> --- (In reply to Loren Wilton from comment #60) > My mind keeps coming back to the idea of marking unrun rules that are > dependencies as zero score, and running the dependent metas. > > Or simplify that to marking all unruns as zero score and as having run. I > mean, if the rule didn't run, then it didn't hit, and a rule that doesn't > hit has zero score. Not sure I understand why you bring up "scores" here. Either a rule hits or doesn't hit. Or neither if it's unrun. > I think unrun is a good internal concept for ordering dependent rule > evaluations, but I'm much less convinced that it is a good external concept. > I tend to think that it is a bit of mechanics that should be hidden. It has been a logical enough concept from the beginning. If we don't know how an unrun rule would affect a meta, then we either can't run the meta or must accept that the result will not be "reliable". The latter probably not being a deal breaker, given how the rules are usually used. But as the discussion is slow and bugs keep piling on, I don't know how soon we can resolve this at this snail pace. Please vote on the open bugs atleast to get them committed. Then we can talk about reverting the meta logic. -- You are receiving this mail because: You are the assignee for the bug.
