Details
-
Sub-task
-
Status: Closed
-
Major
-
Resolution: Later
-
1.6.0
-
None
-
None
Description
Basically follows the logic as that for Python UDF evaluation (org/apache/spark/sql/execution/python.scala). Will extract and reuse common logic between Python UDF and R UDF evaluation.
Serialization/deserialization is different from Python UDF. R UDF will use R SerDe to directly serialize a batch of InternalRows into bytes (that is, in SerializationFormats.ROW format)