Mailing-List: contact dev-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hive.apache.org
Date: Wed, 4 Jul 2012 15:24:35 +0000 (UTC)
From: "Ruben de Vries (JIRA)" <jira@apache.org>
To: hive-dev@hadoop.apache.org
Message-ID: <58869574.3571.1341415475065.JavaMail.jiratomcat@issues-vm>
In-Reply-To: 
 <77497440.13034.1335876828925.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] [Commented] (HIVE-2992) JOIN + LATERAL VIEW + MAPJOIN fails
 to return result (seems to stop halfway through and no longer do the final
 reduce part)
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HIVE-2992?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13406584#comment-13406584 ] 

Ruben de Vries commented on HIVE-2992:
--------------------------------------

Here are the queries with their explain: https://gist.github.com/3047871
                
> JOIN + LATERAL VIEW + MAPJOIN fails to return result (seems to stop halfway through and no longer do the final reduce part)
> ---------------------------------------------------------------------------------------------------------------------------
>
>                 Key: HIVE-2992
>                 URL: https://issues.apache.org/jira/browse/HIVE-2992
>             Project: Hive
>          Issue Type: Bug
>    Affects Versions: 0.8.1
>         Environment: ubuntu - hadoop 0.20.2-cdh3u0 - hive 0.8.1
>            Reporter: Ruben de Vries
>
> Okay first off; I know JOIN + LATERAL VIEW together isn't working so I moved my JOIN into a subquery and that makes the query work properly
> However when I added a MAPJOIN hint for the JOIN in the subquery it will also stop doing the reducer for the main query!
> This only happens when there's a LATERAL VIEW in there though, if I remove the LATERAL VIEW then the main query still get's a reducer to do grouping
> Here's a gist: https://gist.github.com/2499436 Containing the queries and a PHP script which you can run to execute the test case I'm using, which does;
>   - setup a database called hive_mapjoin
>   - setup tables
>   - load some test data
>   - do the selects
> You'll need the https://github.com/rcongiu/Hive-JSON-Serde/downloads json-serde-1.1-jar-with-dependencies.jar with it though and change the path ;)
> I guess looking at the queries you guys can probally figure out a better testcase, but maybe it's helpful :-)
> Not sure if this is a bug or me doing something that just isn't supposed to be working, but I can't seem to find any pointers that this wouldn't be supported...
> Here's another gist with the plan.xml: https://gist.github.com/2499658

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira