Nate Williams <na...@sri.MT.net> wrote:

So, it sounds like goal should be to reduce the differences between the "current" tree and the "stable" tree. One question I did have - when a real release happened, why wouldn't you make -release and -stable the same tree at that instant? Maybe I'm oversimplifying, but at some instant, shouldn't -current, -stable and -release be the same thing? Or maybe the -stable tree is the "main trunk" where -release is snapshot'd off of at release time and -current is a major development branch that gets merged into the -stable tree as it is stablized? Ehhh...now that I think on it...that sounds too simplistic.

Or keep a -current tree (or collection of them acting as "token generators" as in Jordan's message) that lags the real -current tree by a week or two and snapshot that into a -stable if enough "it was good" "tokens" are received over the one week period?

Why does my mind wander back to college and the Heisenberg principle? :-))