The program creates a RDD and a DataFrame containing an array of integers.  It then performs a 
simple map operation to add 1 to every element as a normal spark program would do, and shows 
the result.  It then does the same thing, adding 1 to every element, but through JNI, passing 
the RDD partitions as an array through JNI to some C++ code, which performs the addition, and 
then returns a new array for the output of the map step.  This is then printed, and should match 
the output of the previous map operation.  The same task is then repeated for a DataFrame in 
place of the RDD (not yet completed).

There are scripts to build the program locally, run in local mode, and run in distributed mode. 
You should modify the first line or two to point to your own instalation of Scala and Spark.