http://bugzilla.spamassassin.org/show_bug.cgi?id=3082
------- Additional Comments From [EMAIL PROTECTED] 2004-04-13 10:00 ------- Subject: Re: auto-whitelist database needs expiry On Tue, Apr 13, 2004 at 09:37:00AM -0700, [EMAIL PROTECTED] wrote: > did i need to add --expire-awl to spamassassin.raw instead of sa-learn.raw? > i > just figured sa-learn was a good place for it because AWL is really > a "learning" tool in itself, and since bayes and awl will both be sql driven, > having functions for them in the same binary make sense.. eh? Well....just because Bayes and AWL have SQL backends doesn't mean that they are the same and you can have one file driven and another SQL driven, or driven by some totally different storage backend. sa-learn has nothing to do with AWL so the command should not go there. I'm not 100% sure it should go in spamassassin either but can't off the top of my head think of a better place. It does however need to be extended to work with all backends, which is basically the purpose of this bug. Unfortunately, your solution is too SQL specific, and I believe too MySQL specific. We need to back out one level and set the last update time in AutoWhitelist.pm then just store that value. That would allow us to support this in all backends and multiple SQL implementations. > i have fully tested the expiry code on mysql, but not postgresql... i'm not > sure how postgresql handles a TIMESTAMP entry... I know in mysql, any > modifications done to that row will cause the TIMESTAMP to be set to NOW(). > > I was reading somewhere that postgresql needs > t TIMESTAMP default now, > in its table definition, but I'm not 100% positive. I'm no PostgreSQL expert, but I don't believe similar functionality is available out of the box. You might be able to do a trigger to handle the update. However, the point is moot since we need to handle this at a higher level than the storage layer. Michael ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.
