At least make sure it does work with newer version of Hadoop. I feel that's the only objectif.
MR being stable, MRUnit should be the same. I don't foresee major changes in the project as a consequence. But newer hadoop versions might introduce dependencies changes that could break an old MRUnit. I feel that's the only thing to watch for. Bertrand Dechoux On Tue, Apr 21, 2015 at 1:46 AM, Jarek Jarcec Cecho <jar...@apache.org> wrote: > Hi Brock, > thank you for opening the topic! I share the same feeling that people are > mostly moving away from mapreduce and going to SQL or Spark. I do however > feel that still plenty of people are using mapreduce and even though that > we don’t see much new requests often, I would see a benefit of us providing > support for those for foreseeable future as well. Hence my vote would be to > continue running the project as we did so far and revisit the question at > some point in the future (~6 months, year or something like that). > > Jarcec > > > On Apr 16, 2015, at 2:42 PM, Brock Noland <br...@apache.org> wrote: > > > > Hi, > > > > The MapReduce API has been stabilized and rarely changes these days. > > Additionally I feel the trend is away from writing MapReduce and > > towards using higher level languages such as SQL and Apache Spark. > > > > Do you accept this premise? If so, what should we do next? > > > > If not, why and what should we do next? > > > > Cheers, > > Brock > >
