Github user rdblue commented on the issue:
https://github.com/apache/spark/pull/19568
@DonnyZone, I don't know of any cases that use codgen after the fix for
`CodegenFallback`, but I think this is still a good idea.
If Spark is going to generate code, it should generate correct code. That
means either we remove the codegen implementation from `CodegenFallback`, or we
fix the row passed in by sort-merge join. I'd rather fix sort-merge join
because we may want to change the behavior to codegen as much of the condition
as possible later on.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
