On Wed, Feb 23, 2005 at 09:57:42AM -0600, David Engel wrote: > On Wed, Feb 23, 2005 at 09:55:49PM +1100, Eyal Lebedinsky wrote: > > This is not a good approach. A release should be a more heavily tested > > and stabilized package than the usual snapshots. > > In my experience, almost nobody will test anything until a release is > made. >
That's not true. Once a package reaches a certain user base size -- and I think mythtv has reached that size though I don't know for sure -- it develops a core of people who will indeed download and test a release candidate (but who won't compile from CVS.) Among the most popular packages, like Mozilla, or Java, immense numbers of people download the release candidates -- what in the old days used to be called a beta but that term is being re-appopriated. Almost more than they need. If you have a package with a small user base (and also without enough difference in features and fixes to incentivize the beta testers) then this won't work, but there is a level above which it works well, and I think Myth has made it. The term beta has changed in part because lots of packages, myth included, release to the public software with a verion number that starts with "0." That usage tries to mean, "If this were a commercial product, we would not ship this yet." In the old parlance, the CVS is alpha -- meant for internal testers, the release candidate is beta, and the release is, well, the release. However, a version 0 product is not at the "release" state of the commercial vocabulary so it gets confusing.
_______________________________________________ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users