You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Andy Bryant (JIRA)" <ji...@apache.org> on 2018/11/08 04:23:00 UTC
[jira] [Comment Edited] (KAFKA-6641) Consider auto repartitioning
for Stream.transform() API
[ https://issues.apache.org/jira/browse/KAFKA-6641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16679272#comment-16679272 ]
Andy Bryant edited comment on KAFKA-6641 at 11/8/18 4:22 AM:
-------------------------------------------------------------
This is done now by the looks. In {{KStreamImpl.transform}} it generated a new {{KStreamImpl}} with {{repartitionRequired}} flag set to {{true}}. Similarly for {{transformValues}} it just propagates the current flag value as expected.
was (Author: kiwiandy):
This is done now by the looks. In KStreamImpl.transform it generated a new KStreamImpl with repartitionRequired flat set to true. Similarly for transformValues it just propagates the current flag value as expect.
> Consider auto repartitioning for Stream.transform() API
> -------------------------------------------------------
>
> Key: KAFKA-6641
> URL: https://issues.apache.org/jira/browse/KAFKA-6641
> Project: Kafka
> Issue Type: Improvement
> Components: streams
> Reporter: Guozhang Wang
> Priority: Major
> Labels: api
>
> Today with map / mapValues of Streams DSL, we will set a flag / not set a flag for the underlying topology builder; but for transform / transformValues, we do not make such marking choices. Maybe the topology builder can still make such flagging for transform() to indicate that since the key maybe changed, we should issue a repartition for the downstream stateful operators when necessary.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)