Github user maryannxue commented on the issue:
https://github.com/apache/spark/pull/21699
> Actually I am mostly worry of the pivotColumn. Specifying multiple
columns via struct is not intuitive I believe.
It depends on whether we'd like to add extra interfaces for multiple
columns. I don't have a preference between reusing this interface for multiple
pivot columns or adding new ones. And we can always decide later.
But back to this interface, I'd assume this is for more advanced users, and
the pivot column, even just being a single column, can have complex types, so
the "literal object" values might be insufficient. Plus, this is a new
interface we haven't pushed out yet, but once we have, we are more likely to
end up adding a new one than changing it if we want to make it more
sophisticated later on.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
