There haven't been any new opinions on this issue in a while and I'm leaning toward Mike's suggestion of only using Mx to publish to the snapshot repository. Unless I hear a strong opinion otherwise, I'll remove the milestone identifier from version, allowing users to more easily consume the daily snapshot builds. Thanks, Craig, for raising this issue. +1 for creating JIRA milestone releases w/o branching. I too think it'll help with issue tracking.
-Jeremy On Fri, Dec 5, 2008 at 9:11 AM, Michael Dick <[EMAIL PROTECTED]>wrote: > Hi Jeremy, > > On Fri, Dec 5, 2008 at 8:50 AM, Jeremy Bauer <[EMAIL PROTECTED]> wrote: > > > Mike, > > I was thinking the same thing. Providing milestone releases as an > OpenJPA > > published, mostly stable, tested release with clear definition of the new > > function it includes. Based on Craig's observation, I don't know that > > means > > we have to include the -Mx- identifier in the builds though. I think it > > would help us (dev) and users identify which milestone version of code is > > in > > development and use, but it does have obvious maintenance implications. > If > > we do not publish official maven releases maybe Mx it isn't required - > > could > > we add the Mx at the point when we do the milestone build? > > > > There's no reason why we can't add Mx at a later time. That would be a > benefit to users who always want the latest snapshot, regardless of the > milestone. > > Users who want a specific milestone can specify the latest one from the > snapshot repos (or central if we do a formal release). > > > > > > Regarding a maven publish, is it possible to publish a release to the > maven > > repository, say an M1 release and not branch off or create a > corresponding > > M1 release in JIRA? It may be useful to publish milestone releases in > > maven > > (for those who use maven), but not add the overhead of creating an svn > > branch which we do not intend (and wouldn't make sense) to maintain. > > > > JIRA versions and Maven releases are totally separate entities (pun not > intended). Similarly a release in JIRA is not necessarily tied to a branch > in SVN. > > I'd be inclined to publish Mx builds to the SNAPSHOT repository (less > stringent checking - quicker to publish). Create a JIRA release for it (so > we can track when issues crop up and when they're fixed). Create a tag in > SVN for easy access, but do not create a branch. > > I'm open to other ideas though, > > -mike > > > > > > -Jeremy > > > > On Thu, Dec 4, 2008 at 9:50 PM, Michael Dick <[EMAIL PROTECTED]> wrote: > > > > > Hi Craig, > > > > > > I think it depends on how official we want to make the milestone > > releases. > > > I > > > was thinking of the milestones being "smaller" convenience releases, > not > > > "official" releases that get published to maven central. When we cut a > > > milestone release we'd leave it in the SNAPSHOT repository so that > folks > > > can > > > test a stable(ish) version. At the same time it wouldn't be an official > > > release with all the overhead and additional work that implies > > (maintenance > > > branch?). > > > > > > What do other people think? > > > > > > -mike > > > > > > > > > On Thu, Dec 4, 2008 at 2:44 PM, Craig L Russell <[EMAIL PROTECTED] > > > >wrote: > > > > > > > Hi Jeremy, > > > > > > > > I don't understand why the version needs to be changed from simply > > > > 2.0.0-SNAPSHOT to 2.0.0-Mx-SNAPSHOT. Seems like we would be making > > > trouble > > > > for folks who want to use the release. Wouldn't we want to make the > > > change > > > > at the time we want to release, e.g. from 2.0.0-SNAPSHOT to 2.0.0-M2? > > > > > > > > Regards, > > > > > > > > Craig > > > > > > > > > > > > On Dec 4, 2008, at 8:11 AM, Jeremy Bauer wrote: > > > > > > > > Three in favor, no oppose. Motion passes. :-) > > > >> The build artifacts for each milestone will be named > > > >> openjpa*-2.0.0-Mx-SNAPSHOT.jar, where x is the milestone number. > I'll > > > >> commit the pom updates for M1 shortly. > > > >> > > > >> -Jeremy > > > >> > > > >> On Wed, Dec 3, 2008 at 9:40 AM, Albert Lee <[EMAIL PROTECTED]> > > wrote: > > > >> > > > >> +1. > > > >>> > > > >>> Albert Lee. > > > >>> > > > >>> On Wed, Dec 3, 2008 at 9:03 AM, Jeremy Bauer <[EMAIL PROTECTED]> > > > >>> wrote: > > > >>> > > > >>> OpenJPA dev's, > > > >>>> Now that we have a few iteration periods defined and are getting > > > content > > > >>>> into iteration 1, I think we should consider planning OpenJPA 2.0 > > > >>>> > > > >>> milestone > > > >>> > > > >>>> releases. Based on a 3 (sometimes 4) week iteration schedule and > > the > > > >>>> > > > >>> fact > > > >>> > > > >>>> that it takes ~ a week to create and publish a release, how does a > > > >>>> milestone > > > >>>> release every other (ie. after each even numbered) iteration > sound? > > > If > > > >>>> > > > >>> we > > > >>> > > > >>>> discover that a milestone release every other iteration is not > > > optimal, > > > >>>> > > > >>> we > > > >>> > > > >>>> can adjust as appropriate. > > > >>>> > > > >>>> -Jeremy > > > >>>> > > > >>>> > > > >>> > > > >>> > > > >>> -- > > > >>> Albert Lee. > > > >>> > > > >>> > > > > Craig L Russell > > > > Architect, Sun Java Enterprise System http://db.apache.org/jdo > > > > 408 276-5638 mailto:[EMAIL PROTECTED] > > > > P.S. A good JDO? O, Gasp! > > > > > > > > > > > > > >
