Github user cloud-fan commented on the issue: https://github.com/apache/spark/pull/22029 So, the goal here is to make the behavior consistent between multi-column IN-subquery and multi-column normal IN for Spark. That said, I feel it's reasonable to change the behavior of `(a, b) in (struct_col1, struct_col2, ...)` to return null if a field is null, but it seems pretty weird to also apply this behavior to `input_struct_col in (struct_col1, struct_col2, ...)`. It's OK to treat the `(...)` syntax specially, but it's strange to treat struct type different.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org