Hi, I don't think Maven allows you to create both hadoop1 and hadoop2 artifacts from the same execution. And we obviously can't just rename the built artifacts because the jars contain Maven metadata and the POMs are signed. I've never needed to do this before but my guess is that we have to configure the maven-jar-plugin to set the classifier to hadoop1 or hadoop2 depending on the profile.
We'll have to check whether we can run "mvn release:perform" twice (once for each profile) or if that requires some extra trickery. Regards, Matthias [1] http://maven.apache.org/plugins/maven-jar-plugin/jar-mojo.html On Wednesday, 2013-07-24, Josh Wills wrote: > We haven't really been in the habit of planning the releases-- my sense has > been that we do a release when it feels like there's enough good stuff in > the snapshot to merit doing one, and I certainly feel like that's the case > now. > > My one question for the devs on the next release would be if (and how) we > should go about creating both hadoop1 and hadoop2-based binary > releases/Maven artifacts. If anyone has strong feelings on it either way, > I'd like to hear them, along with a recommendations of the best way to > express it in the version for the release (e.g., 0.7.0-hadoop1 and > 0.7.0-hadoop2? Or maybe use profiles in some way that I don't really > understand?) > > J > > > On Wed, Jul 24, 2013 at 1:37 PM, Micah Whitacre <[email protected]> wrote: > > > Is there a documented release plan for the upcoming 0.7 version of Crunch? > > Or an expected date? > > > > There are a few fixes that have been resolved (CRUNCH-217) specifically > > that I know would be useful to have released. > > > > Micah > > > > > > -- > Director of Data Science > Cloudera <http://www.cloudera.com> > Twitter: @josh_wills <http://twitter.com/josh_wills>
