[
https://issues.apache.org/jira/browse/HIVE-391?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12696330#action_12696330
]
Raghotham Murthy commented on HIVE-391:
---------------------------------------
can you add a test where there is no map-side aggregate and there is a count
distinct? this case will be useful when the cardinality of the distinct column
is very high and the data is not partitioned on the distinct column.
> udafcount merge does not handle nulls
> -------------------------------------
>
> Key: HIVE-391
> URL: https://issues.apache.org/jira/browse/HIVE-391
> Project: Hadoop Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 0.3.0
> Reporter: Namit Jain
> Assignee: Namit Jain
> Priority: Blocker
> Attachments: hive.391.1.patch
>
>
> udafcount merge does not handle nulls
> If the mapper does not emit any row on null input, i.e both count and count
> distinct are present, and the aggregation function is count,
> it will get a null pointer
> select count(1), count(distinct x.value) from src x where x.key = 9999;
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
