You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@hive.apache.org by "Chao Sun (JIRA)" <ji...@apache.org> on 2017/04/12 21:02:41 UTC

[jira] [Created] (HIVE-16428) Refactor & fix the logic in HoS mapjoin optimization

Chao Sun created HIVE-16428:
-------------------------------

             Summary: Refactor & fix the logic in HoS mapjoin optimization
                 Key: HIVE-16428
                 URL: https://issues.apache.org/jira/browse/HIVE-16428
             Project: Hive
          Issue Type: Improvement
            Reporter: Chao Sun
            Assignee: Chao Sun


[The logic for mapjoin optimization in HoS|https://github.com/apache/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/optimizer/spark/SparkMapJoinOptimizer.java#L276] seems unnecessarily complex and (perhaps) buggy. For instance, If {{bigInputStat}} is null AND the position is not in {{bigTableCandidateSet}}, then the size for the current position will not be counted into the {{totalSize}}. This seems wrong.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)