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 2018/11/15 11:44:00 UTC

[jira] [Commented] (FLINK-10889) Semantic inconsistency between DataSet#print and DataStream#print

    [ https://issues.apache.org/jira/browse/FLINK-10889?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16687904#comment-16687904 ] 

Fabian Hueske commented on FLINK-10889:
---------------------------------------

I think there were good reasons to not follow the same semantics as in the DataSet API.

In order to support this, the TMs would either need to directly connect to the client and stream the results or the JM would need to act as a proxy. Either way, this would not work well in most cluster environements.


> Semantic inconsistency between DataSet#print and DataStream#print
> -----------------------------------------------------------------
>
>                 Key: FLINK-10889
>                 URL: https://issues.apache.org/jira/browse/FLINK-10889
>             Project: Flink
>          Issue Type: Improvement
>          Components: DataSet API, DataStream API
>            Reporter: Jeff Zhang
>            Assignee: vinoyang
>            Priority: Major
>
> DataSet#print will print the result on client side, while DataStream#print will print the result on TM. This inconsistency will confuse users. IMHO, we should make the behavior consistency between DataSet and DataStream, I prefer to print the result on client side.  Regarding DataStream#print, we can use DataStreamUtils#collect to print it on client side.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)