On Mon, Oct 03, 2005 at 07:45:03PM +0200, Hans Verkuil wrote: > On Monday 03 October 2005 18:26, Axel Thimm wrote: > > What I would suggest is a versioning scheme that will make it > > clear what a release is supposed to be. Perhaps 0.even ist stable > > and 0.odd is not? So 0.3.9 should perhaps become 0.4.0 with 0.4.1 > > etc small bugfixes, and 0.5.0 the new branch with new features? > > I never liked the odd/even schemes. My idea is to have two active > series: one stable (0.3 branch), one for development (0.4 > trunk). When 0.4 is stable that becomes a stable branch, 0.3 is > discontinued and 0.5 is opened for development. As long as the > website states that clearly I see no problems.
It would be nic to have some deterministic mapping that will make it easy to check whether say 0.4.18 as found in Mandriva 2006 was a cut from the stable or development branch. I wouldn't know w/o browsing history, whether the change from development to stable status was on 0.4.17 or 0.4.18 etc. Even/odd isn't the only scheme that can offer that, there are others like having stable going 1.0.0, 1.0.1, ... and developement towards "2" going 1.90.0, 1.90.1, .... Or reserve the subminor != 0 for development releases, and have stable releases always look like 0.4, 0.5, ... similar to what Intel does. There are endless versioning schemes to distinguish stable from development releases. But you're the piano man, so you play the music :) > Also, this stable/development distinction is only temporarily. The goal is > after all inclusion in the kernel, at which point there is no development > series anymore. That would assume we reach the state of perfection :) I believe that the kernel will accept some proven version of ivtv, but not allow active developement of new features to happen there. You can see this in almost any project that made it into the kernel independent of its size, or origin: The development continues in the project's domain and when the developers feel like they have reached a stable codebase again they submit the changes (or part of it) upstream to the kernel. So there will always be some development of ivtv going on, and if it's not just new tuners, it will not go directly into the kernel. -- Axel.Thimm at ATrpms.net
pgpwplWEE9d1N.pgp
Description: PGP signature
_______________________________________________ ivtv-devel mailing list [email protected] http://ivtvdriver.org/mailman/listinfo/ivtv-devel
