[
https://issues.apache.org/jira/browse/HIVE-17247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16121180#comment-16121180
]
Rui Li commented on HIVE-17247:
-------------------------------
I think {{spark_dynamic_partition_pruning_mapjoin_only}} needs to be updated
with the change here.
> HoS DPP: UDFs on the partition column side does not evaluate correctly
> ----------------------------------------------------------------------
>
> Key: HIVE-17247
> URL: https://issues.apache.org/jira/browse/HIVE-17247
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
> Reporter: Sahil Takiar
> Assignee: Sahil Takiar
> Fix For: 3.0.0
>
> Attachments: HIVE-17247.1.patch
>
>
> Same problem as HIVE-12473 and HIVE-12667.
> The query below (uses tables from {{spark_dynamic_partition_pruning.q}})
> returns incorrect results:
> {code}
> select count(*) from srcpart join srcpart_date on (day(srcpart.ds) =
> day(srcpart_date.ds)) where srcpart_date.`date` = '2008-04-08';
> {code}
> It returns a value of 0 when DPP is on, when it is disabled it returns 1000
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
