> >>> Nor does it make sense to use a tool, even if supplied > with SpamAssassin, > >>> that is broken for performing updates. > > > >> what's the "broken" part? > > > > Well, this may not qualify as broken, but I would say it's an > > undesirable behavior that, upon successful download of the new > > set of rules, it immediately deletes your old set of rules. > > What happens if the new set is broken? There's no easy way > > to revert to the last known good state. > > > > I would prefer a system where it downloads every update to a new > > directory, then just changes a symlink to point to the newest > > one, leaving the old one in place in case you want to revert. > > Of course, this would require a system for expiring old updates > > (since you don't want to have 100 copies of the rules sitting > > around), but that shouldn't be too hard. > > One would presume an intelligent system for doing such updates would > play directory rename tricks or simply copy off active rules to an > archive directory so that if a "spamassasssin --lint" fails on the > newly downloaded files recovery could be effected easily. One would > further presume that update would do this. Is this cyberunit faulty > for making this presumption?
Sa-update lints prior to updating, and updates only if the rules lint successfully. Bret
