Github user cloud-fan commented on the issue:
https://github.com/apache/spark/pull/22197
I agree moving the `clipSchema` to the beginning is cleaner, but I feel it
goes too far away from our original target: making parquet filter pushdown case
insensitive when spark is case insensitive.
I'd expect a very small change: at the beginning, go through the parquet
schema and find duplicated field names. We can do the refactoring later, to
make this PR small.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
