Github user cloud-fan commented on the issue: https://github.com/apache/spark/pull/21611 There are 2 problems: 1. `Option` is mistakenly mapped to a struct type. It's not only for aggregate, but join too ``` scala> val a = Seq(Some(1)).toDS a: org.apache.spark.sql.Dataset[Some[Int]] = [value: int] scala> val b = Seq(Some(1.2)).toDS b: org.apache.spark.sql.Dataset[Some[Double]] = [value: double] scala> a.joinWith(b, lit(true)) res2: org.apache.spark.sql.Dataset[(Some[Int], Some[Double])] = [_1: struct<value: int>, _2: struct<value: double>] ``` 2. `Aggregator` can't use option of product. I think these 2 problems can be solved separately, what do you think?
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org