[
https://issues.apache.org/jira/browse/HIVE-3550?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13472176#comment-13472176
]
Hudson commented on HIVE-3550:
------------------------------
Integrated in Hive-trunk-h0.21 #1730 (See
[https://builds.apache.org/job/Hive-trunk-h0.21/1730/])
HIVE-3550 Early skipping for limit operator at reduce stage
(Navis via namit) (Revision 1395796)
Result = FAILURE
namit : http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1395796
Files :
* /hive/trunk/ql/src/java/org/apache/hadoop/hive/ql/exec/ExecReducer.java
> Early skipping for limit operator at reduce stage
> -------------------------------------------------
>
> Key: HIVE-3550
> URL: https://issues.apache.org/jira/browse/HIVE-3550
> Project: Hive
> Issue Type: Improvement
> Components: Query Processor
> Reporter: Navis
> Assignee: Navis
> Priority: Trivial
> Fix For: 0.10.0
>
> Attachments: HIVE-3550.D5943.1.patch
>
>
> Queries with limit clause whose final result is from reduce stage could not
> be early exited like map stage. But it seemed possible to just skip remaining
> lines. It can be accomplished by adding just two lines. Is there any reason
> we should not to do that?
> {noformat}
> select key from src order by key limit 10;
> 1. current
> ExtractOperator: 3 forwarded 500 rows
> LimitOperator: 4 forwarded 10 rows
> 2. early skipping
> ExtractOperator: 3 forwarded 12 rows
> LimitOperator: 4 forwarded 10 rows
> {noformat}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
