Github user maryannxue commented on the issue:
https://github.com/apache/spark/pull/21699
@MaxGekk Yes, it was caused by my previous PR. The change in my PR was a
walk-around for an existing problem in either Aggregate or PivotFirst (I
suspect it's Aggregate) with struct-type columns. The change itself worked as
designed because Pivot SQL support wouldn't allow any function (like
"lowercase") in the pivot column. However it broke your PR coz it aimed to
allow any expression.
That said, we have two options here:
1) Give up the PivotFirst approach and fall back to "else" branch for
struct-type pivot columns, i.e., multiple column in pivot FOR clause.
2) Fix the bug for Aggregate or PivotFirst.
I will do a little investigation into option 2) tomorrow and get back to
you :)
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
