[
https://issues.apache.org/jira/browse/FLINK-2105?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14639681#comment-14639681
]
ASF GitHub Bot commented on FLINK-2105:
---------------------------------------
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/907#issuecomment-124270766
Very nice work! :+1: I had just a few minor style regards.
It would be nice if you could check the overhead of having all outer join
cases handled by the same class vs. having a dedicated class for LEFT, RIGHT,
and FULL OUTER.
I would also like to run a major regression benchmark to double-check that
the refactoring of the current sort-merge join didn't cause a performance
regression. I can take care of that once you addressed the comments.
If all lights are green after that, I would be happy to merge this PR.
> Implement Sort-Merge Outer Join algorithm
> -----------------------------------------
>
> Key: FLINK-2105
> URL: https://issues.apache.org/jira/browse/FLINK-2105
> Project: Flink
> Issue Type: Sub-task
> Components: Local Runtime
> Reporter: Fabian Hueske
> Assignee: Ricky Pogalz
> Priority: Minor
> Fix For: pre-apache
>
>
> Flink does not natively support outer joins at the moment.
> This issue proposes to implement a sort-merge outer join algorithm that can
> cover left, right, and full outer joins.
> The implementation can be based on the regular sort-merge join iterator
> ({{ReusingMergeMatchIterator}} and {{NonReusingMergeMatchIterator}}, see also
> {{MatchDriver}} class)
> The Reusing and NonReusing variants differ in whether object instances are
> reused or new objects are created. I would start with the NonReusing variant
> which is safer from a user's point of view and should also be easier to
> implement.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
