On Sat, May 28, 2005 at 03:50:19PM -0400, Duncan Findlay wrote: > > On the other hand 3.0.4 seems to be *done*. How many bugs fixed now? > > Why not release? > > Because 3.1.0 is so close and we'll very soon be telling everyone to > upgrade to 3.1.0. Why make them upgrade to 3.0.4 first?
You don't HAVE to upgrade to 3.0.4 to get to 3.1... I was talking with JM the other day, and my thought process is that since SA is becoming more of an engine with the rules being "separate" (they're not so dependent on code anymore, there are plugins, etc,) and since we now have sa-update which can be used to update rules and such, there's less of a reason for people to upgrade to the latest version. Sure, 3.1 is better than 3.0, but 3.0 works, and if the people don't want/need the new features, there's no reason to make them upgrade. Other than that, there are numerous distros and such that will continue only providing 3.0 for a while, so I'd like to see us get as stable and bug-free a version out as possible, within reason. > According to Dan, for reasons I'm not yet aware, we only need one > mass-check for 3.1.0 instead of 3 which we needed for the 3.0.x > releases. In 3.0 we had to do set0/set1, then set2, then set3. In 3.1, it's set3 with "random sampling" for bayes learning (instead of autolearn based on score). That way, we can easily get sets 0-2 by filtering out certain rule hits from the set3 output. At least, I believe that was the idea. :) -- Randomly Generated Tagline: Well, if you believe THAT, I have a nice bridge for sale...
pgpFqKuvwcV5z.pgp
Description: PGP signature
