dione
dione copied to clipboard
Use Hadoop's file splits instead of start_pos,end_pos
we can also use Spark's PartitionedFile.