http://bugzilla.spamassassin.org/show_bug.cgi?id=3331
------- Additional Comments From [EMAIL PROTECTED] 2004-08-11 23:49 ------- Subject: Re: [review] Bayes option to keep original token as db data (not key). On Wed, Aug 11, 2004 at 11:21:22PM -0700, [EMAIL PROTECTED] wrote: > I think leaving it out, removes functionality that a lot of people *do* > expect. > we'd be pretty much the only bayes impl apart from dspam and CRM-114 (the > n-gram > systems) that don't support dumping, in that case. Well, that's not really true. There are tons of Bayes-style filters out there (how many MUAs have built-in Bayes-style filters?) which don't do dumping. If you compare us to other OSS projects, you could be right. However, I do agree with Quinlan in that normal users won't care. The more technial folks, or the folks who are more involved with things like running sa-learn may care. For example, I was upset when the hashing first went in, but then I realized I have better things to do with my time than going through the tokens in my Bayes DB. The rest of the users on my box, however, haven't said anything, leading me to think they don't ever look at the dump output. > However I could compromise by setting it off by default, and noting clearly in > the doco how to turn it on... The issue, though, is the performance penalty even with the setting off. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
