[
https://issues.apache.org/jira/browse/HIVE-2466?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13134440#comment-13134440
]
Namit Jain commented on HIVE-2466:
----------------------------------
A few high level comments:
Instead of making the dump prefix optional - why dont you always have it in
hashtablesinkdesc and
mapjoindesc.
This way, you can get rid of all the checks : if dumpdescriptor is not null.
The logic will be simpler - the names of map files will be :
mapfile1
mapfile2
..
etc
Also, it might be nicer to add the static function in PlanUtils.java instead of
QBJoinTree.java.
> mapjoin_subquery dump small table (mapjoin table) to the same file
> -------------------------------------------------------------------
>
> Key: HIVE-2466
> URL: https://issues.apache.org/jira/browse/HIVE-2466
> Project: Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 0.7.1
> Reporter: binlijin
> Assignee: binlijin
> Priority: Blocker
> Attachments: hive-2466.1.patch, hive-2466.2.patch
>
>
> in mapjoin_subquery.q there is a query：
> SELECT /*+ MAPJOIN(z) */ subq.key1, z.value
> FROM
> (SELECT /*+ MAPJOIN(x) */ x.key as key1, x.value as value1, y.key as key2,
> y.value as value2
> FROM src1 x JOIN src y ON (x.key = y.key)) subq
> JOIN srcpart z ON (subq.key1 = z.key and z.ds='2008-04-08' and z.hr=11);
> when dump x and z to a local file,there all dump to the same file, so we lost
> the data of x
--
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
