And make the default output a little better (use CSV if multiple values are present, otherwise just print value-per-line). Delete the temporary files off of hadoop after we've successfully completed output.
This is good. csv output by default makes my stuff simpler. I was gonna worry about json encoding, but my map/reduce output actually works better as a csv.
And make the default output a little better (use CSV if multiple values are present, otherwise just print value-per-line). Delete the temporary files off of hadoop after we've successfully completed output.