http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5545
------- Additional Comments From [EMAIL PROTECTED] 2007-07-05 03:12 ------- argh. guys, if you oppose the idea, why didn't you all say so when we were discussing it a month ago? sheesh ;) The problem is, then, that we have two kinds of rules in sandboxes: 1. rules that should be put in updates, published etc. if they perform well enough in the nightly mass-checks 2. rules that are just in there for testing in the mass-checks, which we don't want to see published at all. What often happens is that we get a ruleset for testing from an external contributor. this may contain hundreds of rules in class 2. we don't want to have to laboriously rename them all to use a T_ prefix; that gets hairy very fast, esp if the contributor does a couple of round trips with changes. I think I like Daryl's idea best; allow the policy to be set on a file-by-file basis. it'd be easy to fix build/mkrules to support that. in the meantime I'll revert the changes I've made to trunk. btw: 'Whoa! Hold on here, there's nothing restricting this to sandbox rules. Rules that only appear in the legacy rules directory (like all the BAYES_* rules) are also affected by this.' that shouldn't be the case -- they may not wind up listed in the "rules/active.list" file, but that has no effect, since that file just controls promotion of rules from the rulesrc sandboxes -- and the legacy rules don't need to be promoted, they're always in the "rules" dir anyway. ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.
