[
https://issues.apache.org/jira/browse/DRILL-1767?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14220412#comment-14220412
]
Steven Phillips commented on DRILL-1767:
----------------------------------------
I think this is expected, based on how Drill nested data support is designed.
Drill does not have the concept of a nullable Map, but scalars can be null.
Of course, this is something we could potentially revisit if it there were a
good reason to add it.
> kvgen ignores null scalars but not empty maps in the output
> -----------------------------------------------------------
>
> Key: DRILL-1767
> URL: https://issues.apache.org/jira/browse/DRILL-1767
> Project: Apache Drill
> Issue Type: Bug
> Components: Functions - Drill, Storage - JSON
> Reporter: Rahul Challapalli
>
> git.commit.id.abbrev=108d29f
> Dataset :
> {code}
> {
> "id":1,
> "keys":{"m1":{"v7":{"k1":"k2"}},"m2":{"v1":"abc","v2":{"w1":0.24}}}
> }
> {
> "id":1,
> "keys":{"m1":{"v7":{"k1":"k3"}},"m2":{"v1":"abc","v2":{"w1":0.2}}}
> }
> {code}
> Query :
> {code}
> select kvgen(keys) from `json_kvgenflatten/temp.json`;
> +------------+
> | EXPR$0 |
> +------------+
> |
> [{"key":"m1","value":{"v7":{"k1":"k2"},"v2":{}}},{"key":"m2","value":{"v7":{},"v1":"abc","v2":{"w1":0.24}}}]
> |
> |
> [{"key":"m1","value":{"v7":{"k1":"k3"},"v2":{}}},{"key":"m2","value":{"v7":{},"v1":"abc","v2":{"w1":0.2}}}]
> |
> +------------+
> 2 rows selected (0.508 seconds)
> {code}
> In the above output , drill seems to be ignoring the empty scalars (v1) but
> not empty maps (v2)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
