You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "Fabian Hueske (JIRA)" <ji...@apache.org> on 2017/08/28 08:56:00 UTC

[jira] [Commented] (FLINK-7509) Refactorings to AggregateCodeGenerator

    [ https://issues.apache.org/jira/browse/FLINK-7509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16143526#comment-16143526 ] 

Fabian Hueske commented on FLINK-7509:
--------------------------------------

Yes, with more special cases being added it might make sense to split the code for the different aggregation types:

- batch group aggregation
- stream non-windowed group aggregation
- stream group windowed aggregation
- stream over windowed aggregation

But maybe another separation is more meaningful.

Do you have concrete plans for the split [~jark]?


> Refactorings to AggregateCodeGenerator
> --------------------------------------
>
>                 Key: FLINK-7509
>                 URL: https://issues.apache.org/jira/browse/FLINK-7509
>             Project: Flink
>          Issue Type: Improvement
>          Components: Table API & SQL
>            Reporter: Jark Wu
>
> I think the `AggregateCodeGenerator#generateAggregations` is too long with 500+ LOC currently and hard to extend. I would like to refactor it if you have no objection.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)