There are a bunch of so called tools in hadoop-mapreduce-project/src/tools - DistCp, HadoopArchives, Rumen etc. And contrib projects are in src/contrib in all of common, hdfs and mapred source trees. Not sure how the distinction was ever made.
The last time we had a discussion about moving contrib projects out of the core, we didn't reach any consensus - * http://s.apache.org/HadoopContribDiscussion*. Do we want to revive that discucssion now? Or we want to keep the status-quo, imitate the source structure of the present day tools and contrib, but move them to appropriate maven modules and then have that discussion separately? I personally prefer the later, given the length and the eventual failure of the previous discussion. HADOOP-7590 is a related issue where the src location of contribs like gridmix, streaming etc is being talked about. I suppose that issue and this thread ought to converge. Thanks, +Vinod
