All: I've created a fork of mrunit from Hadoop's contrib directory at https://github.com/esammer/mrunit
Why: * MRUnit could benefit from a separate development and release cycle than that of Hadoop itself. This would allow MRUnit to evolve independently of Hadoop. * Moving projects out of contrib simplify Hadoop's build and the artifacts that need to be maintained. * I've heard rumblings that contrib is generally frowned upon by some of the committers. I agree. What is already done: * Build system ported to Maven. * No more build warnings re: generics or deprecation. I think it makes sense to remove mrunit from contrib (after a deprecation period) but I'm curious to hear others' opinions. If the community at large decides it makes sense to retain mrunit in contrib I'm happy to provide contributions back. I'd also like to offer commit status on the fork to any current Hadoop committers who are interested. The project is being hosted on Github to ease collaboration. Pull requests are greatly appreciated. Regards. -- Eric Sammer twitter: esammer data: www.cloudera.com
