Github user cloud-fan commented on the issue:
https://github.com/apache/spark/pull/21319
Yea this is still a workaround before having stats in the physical plan. My
major concern is, moving stats to physical plan is a huge change and I may not
have enough free time to drive it. It will be great if @wzhfy can drive it, but
I'm not confident whether we can finish it before Spark 2.4.
I think we should not be blocked by this, and I hope we can have a good
data source v2 operator pushdown story in Spark 2.4. This is my major
motivation for these refactoring/cleanups.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
