[
https://issues.apache.org/jira/browse/PHOENIX-671?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15178680#comment-15178680
]
James Taylor commented on PHOENIX-671:
--------------------------------------
Hello again, [~RCheungIT]. Yes, this is a good one and of smaller scope than
PHOENIX-154.
bq. 1. should I work follow the roadmap that you describe in the comments or
start the work on calcite branch
This one has very little requirements in the way of parsing or compilation -
it'd just be a matter of allowing {{UNION}} instead of just {{UNION ALL}} and
remember that so that when we put together the query plan (UnionPlan), we use
the correct iterators. I think it could be done in either branch.
bq. 2. it is a pure client-side job. Is that correct?
We push as much work into the cluster as we can, but I believe the remaining
work to support UNION (we already support UNION ALL), would be client side as
mentioned above. You'd essentially be using the above mentioned iterators to
dedup rows that are being retrieved from the parallel scans that are executing
on the server side.
Anything to add, [~maryannxue]?
> Support UNION
> -------------
>
> Key: PHOENIX-671
> URL: https://issues.apache.org/jira/browse/PHOENIX-671
> Project: Phoenix
> Issue Type: Task
> Reporter: James Taylor
> Labels: gsoc2016
>
> This is particularly useful for Pentaho support.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
