[
https://issues.apache.org/jira/browse/HIVE-8970?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14230169#comment-14230169
]
Chao commented on HIVE-8970:
----------------------------
I just regenerated all qfile outputs, and noticed some result differences for
join38.q, join_literals.q, join_nullsafe.q and subquery_in.q.
Look like they are not related to map-join work, since if I reset my branch to
the HIVE-8946 commit then the results are correct.
Maybe it's caused by the recent merge or RSC commits.
> Enable map join optimization only when hive.auto.convert.join is true [Spark
> Branch]
> ------------------------------------------------------------------------------------
>
> Key: HIVE-8970
> URL: https://issues.apache.org/jira/browse/HIVE-8970
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
> Affects Versions: spark-branch
> Reporter: Chao
> Assignee: Chao
> Fix For: spark-branch
>
> Attachments: HIVE-8970.1-spark.patch, HIVE-8970.2-spark.patch
>
>
> Right now, in Spark branch we enable MJ without looking at this
> configuration. The related code in {{SparkMapJoinOptimizer}} is commented
> out. We should only enable MJ when the flag is true.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
