Github user cloud-fan commented on the issue:
https://github.com/apache/spark/pull/21118
parquet scan doesn't need unsafe row because it outputs `ColumnarBatch`.
Note that, `UnsafeRow` is the data format Spark uses to exchange data between
operators, but whole-stage-codegen can merge several operators into one. So in
your case, a parquet scan followed with some simple operators like filter,
project, is still one operator, so you won't see `UnsafeRow`.
I believe if you add aggregate, you will see `UnsafeRow` in the final
aggregate's generated code, which comes from the `ShuffleExchangeExec` operator.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
