[
https://issues.apache.org/jira/browse/PIG-4406?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14302295#comment-14302295
]
Brian Johnson commented on PIG-4406:
------------------------------------
If you do something like this
A = LOAD 'hbase...'
B = GROUP A BY f1 USING 'collected'
C = FOREACH A GENERATE f1, f2
D = JOIN A by f1, B by group
E= STORE ...
it produces an implicit SPLIT which then breaks the collected group because
it's now loading the records from a file instead of from hbase for B and C
> Splits prevent optimized joins/groups
> -------------------------------------
>
> Key: PIG-4406
> URL: https://issues.apache.org/jira/browse/PIG-4406
> Project: Pig
> Issue Type: Bug
> Reporter: Brian Johnson
>
> If you have either an explicit or plan generated split you can't do any
> optimized joins or groups because it sees the split as the loader instead of
> the actual loader
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
