Github user srowen commented on the issue: https://github.com/apache/spark/pull/22383 Oh, hm: ``` Serialization stack: - object not serializable (class: java.util.Optional, value: Optional[x]) - field (class: scala.Tuple2, name: _2, type: class java.lang.Object) - object (class scala.Tuple2, (1,Optional[x])) - field (class: scala.Tuple2, name: _2, type: class java.lang.Object) - object (class scala.Tuple2, (1,(1,Optional[x]))) - element of array (index: 0) - array (class [Lscala.Tuple2;, size 5) ``` So `java.util.Optional` isn't `Serializable`. Well, that may scuttle this whole idea. I think we're going to find a number of instances where Spark or user apps need to `collect()` `Optional` objects. Unless someone has a bright idea I think we can't do this. Same reason I was unable to change Spark to use `java.util.function` interfaces -- lambdas aren't otherwise `Serializable` in Java!
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org