Github user viirya commented on a diff in the pull request:
https://github.com/apache/spark/pull/15780#discussion_r90191825
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/ScalaReflection.scala
---
@@ -590,7 +591,11 @@ object ScalaReflection extends ScalaReflection {
"cannot be used as field name\n" +
walkedTypePath.mkString("\n"))
}
- val fieldValue = Invoke(inputObject, fieldName,
dataTypeFor(fieldType))
+ // primitive take only non-null or struct takes non-null object
guarded by isNull
--- End diff --
The concept of `None` can't well fit with Tuple, except for Tuple1.
For a Tuple 2, for example, we encoder it as a row of 2 columns. If it is
None, should we encode it as `[null]` or `[null, null]`? Conceptually, `[null]`
looks the correct answer. However, practically it becomes a row with only one
column and there is a conflict in data format.
Currently, if given a None for a Tuple2 data, we will encode it as `[null,
null]`, as you seen in the test @ueshin mentioned.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
