Maven's support for "classifiers" in dependencies seems to be targeted at this kind of case: http://maven.apache.org/pom.html#Dependencies
I'm not sure how exactly that works with publishing artifacts though. It may just amount to appending the "classifier" as a suffix anyway. But may be worth looking at in more detail. On Wed, May 16, 2012 at 11:35 AM, Jonathan Hsieh <j...@cloudera.com> wrote: > Andy, > > Ah, ok that sounds reasonable. Some this would be similar to how the > security build used to have a "-security" suffix but for hadoop2 we'd have > something like a "-hadoop2" suffix instead. > > Jon. > > On Wed, May 16, 2012 at 11:27 AM, Andrew Purtell <apurt...@apache.org>wrote: > >> On Wed, May 16, 2012 at 11:24 AM, Jonathan Hsieh <j...@cloudera.com> wrote: >> > I've gotten pinged by folks working on Apache Flume, a project that >> depends >> > directly upon hbase and hadoop hdfs jars about how to get the proper >> hbase >> > jars that work against hadoop 1.0 and hadoop 0.23/2.0. >> > Unfortunately, the transition from hadoop 1.0.0 to hadoop 0.23.x/2.0 >> > requires hbase to be recompiled to run against the different hadoop >> > version. ("compile compatible" but not "binary compatible"). >> > >> > Currently, we build and publish hbase jars compiled against hadoop 1.0.x. >> > >> > What is the right way to publish poms/jars for those who want use an >> hbase >> > jars compiled against hadoop 0.23/2.0? Is there a right way? >> >> This requires we add a version suffix for the Hadoop version used during >> build? >> >> Best regards, >> >> - Andy >> >> Problems worthy of attack prove their worth by hitting back. - Piet >> Hein (via Tom White) >> > > > > -- > // Jonathan Hsieh (shay) > // Software Engineer, Cloudera > // j...@cloudera.com