Details
-
Improvement
-
Status: Open
-
P3
-
Resolution: Unresolved
-
None
-
None
-
None
Description
ParquetIO today uses Avro's GenericRecord to materialize reads from Parquet records, this implies an extra conversion from GenericRecord to Beam's Row. We can 'push' this conversion into the source read process by implementing a RecordMaterializer that builds Rows at read time xtending Parquet's ReadSupport.