On Wed, 2005-08-03 at 20:49 +0100, Stewart Williams wrote: > I didn't mean you refer to them as stable, I mean't the Fedora project. > They usually issue 2 or 3 test releases then a release that they > consider stable, Well, stable enough not to arrive DOA anyway. I'm not an FC developer, but I do lurk around the test and devel mailing lists, and this is my explanation: The released versions of FC are not stable in the same sense that, say, a RHEL or FreeBSD release is stable. Notice the release schedule is _time_ based and is not driven by how stable the code is at any given point. The test1-3 are attempts to squash the show stopper bugs before the release date. You wont see Fedora saying "hey, we'll ship FCx when it's ready, whenever that may be (ala FreeBSD)" but we will see "we'll ship FCx on March 14 (or thereabouts, because we want to include OO.org2 beta :)". Releases are just snapshots of development a couple of weeks before the release date. Before the release date so they can be spun into isos. Ever wonder why immediately after install there are lots of updates? developers just keep working. One can continually use the very latest code by using the development tree [affectionately known as 'rawhide'] as your yum repo. Right after a release, rawhide and the isos are pretty close to the same. The gap widens as development continues and the release ages. Bear in mind, I'm not advocating using rawhide, as it's the very edge and is known to 'eat babies '. > but people keep saying they are not stable enough and > are too bleeding-edge. So why do people use it on servers then? 'stable enough ' for production systems is, clearly, in the eye of the beholder. But I don't think anyone worth listening to would recommend using FC in a production system without the caveat that it's a test bed for new technologies and is liable to have some rough, bleeding edges. -- Craig Thomas
