On Fri 28 Sep 2012 09:39:13 AM EDT, Harsh J wrote:
Modularity!
Exactly! Write a mapper that operates as a filter on something about your keys, then use it in whatever jobs you want. Your job needs to operate on data subset A? chain it with the filter mapper that picks out A. Your next one needs to operate on subset B? chain it with the filter that picks out B!
