[
https://issues.apache.org/jira/browse/PIG-110?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12571860#action_12571860
]
Johannes Zillmann commented on PIG-110:
---------------------------------------
I attached a patch with a junit test which reproduces this bug.
The test class execute pig in local mode and in mapreduce mode.
Local pass, mapreduce fails.
> Jobs using the combiner and flatten of group keys produce wrong results
> -----------------------------------------------------------------------
>
> Key: PIG-110
> URL: https://issues.apache.org/jira/browse/PIG-110
> Project: Pig
> Issue Type: Bug
> Components: impl
> Affects Versions: 0.0.0
> Reporter: Alan Gates
> Assignee: Antonio Magnaghi
> Attachments: PIG-110_test.patch
>
>
> A job such as the following:
> a = load 'mydata';
> b = group a by ($0, $1);
> c = foreach b generate flatten(group), COUNT($1)
> currently produces wrong results. Instead of returning the count in the 3rd
> column, it returns a repeat of the value in the second column. If the
> combiner is forced off, this does not occur, so I assume this is in the
> combiner logic somewhere.
> This bug was introduced in revision 617338 (the checkin for PIG-32).
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
