[jira] [Commented] (FLINK-984) Distinct operator for POJO DataSets without KeySelector and on UnsortedGrouping

["Markus Holzemer (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-984?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14047490#comment-14047490 ] 

Markus Holzemer commented on FLINK-984:
---------------------------------------

I agree with the first point.
With the second point I am not sure. For me distinct is not a real group operation like for example a reduce. The current syntax is not distinct(GROUPING) but distinct(KEY) and I am not sure if it would confuse people if they would need to specify a grouping to get distinct values that consider only some fields.

Perhaps we can get some more opinions on that?

> Distinct operator for POJO DataSets without KeySelector and on UnsortedGrouping
> -------------------------------------------------------------------------------
>
>                 Key: FLINK-984
>                 URL: https://issues.apache.org/jira/browse/FLINK-984
>             Project: Flink
>          Issue Type: Improvement
>            Reporter: Ufuk Celebi
>            Priority: Minor
>
> Two issues regarding the distinct operator:
> - The distinct operator does not support {{distinct()}} with POJO datasets. I think it should be possible to implement a reflection based KeySelector which selects all public fields as key?
> - All other grouped operators work by {{groupBy(GROUPING).operator(...)}} whereas the distinct operator does {{distinct(GROUPING)}}. I think it would be more consistent to allow {{groupBy(GROUPING).distinct()}}.



--
This message was sent by Atlassian JIRA
(v6.2#6252)