[SPARK-12150] [SQL] [Minor] Add range API without specifying the slice number

[gatorsmile]

For usability, add another sqlContext.range() method. Users can specify start, end, and step without specifying the slice number. The slice number is based on the sparkContext's defaultParallelism. It just makes consistent with the RDD range API.

[SparkQA]

Test build #47206 has finished for PR 10149 at commit 72860c4.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):\n * public abstract static class PrefixComputer\n

[gatorsmile]

@marmbrus @cloud-fan This PR changes the external API. Not sure if this will be merged or we should revisit it after the release of 1.6? Thank you!

[marmbrus]

This seems reasonable, but if we are going to do work on sqlContext.range it would be nice if we could take care of a bigger issue. Right now this code is much slower than it needs to be.

I think that is is because we are boxing and creating a new Row for each element. Instead, we should consider creating an actual logical/physical operator for range. the implementation of doExecute() can mirror the code in sparkContext.range, but would allocate one UnsafeRow per partition and just call setLong for each element (reusing the same buffer).

Does that make sense?

Benchmark code here: https://github.com/databricks/spark-sql-perf/blob/master/src/main/scala/com/databricks/spark/sql/perf/DatasetPerformance.scala

[gatorsmile]

That sounds an interesting work! Let me think about it.

Do you want me close this PR?

[gatorsmile]

I will make a try this week. Thanks for your insights! : )

[marmbrus]

Cool, let me know if you run into any problems.

[gatorsmile]

Thanks! Will do my best. : )