You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Guozhang Wang (JIRA)" <ji...@apache.org> on 2017/11/01 20:20:00 UTC
[jira] [Commented] (KAFKA-3907) Better support for user-specific
committing in the Streams DSL
[ https://issues.apache.org/jira/browse/KAFKA-3907?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16234681#comment-16234681 ]
Guozhang Wang commented on KAFKA-3907:
--------------------------------------
We considered adding this support as part of KIP-159 but found out that adding {{commit}} in the {{RecordContext}} is not a good implementation. A better approach maybe providing some syntax-sugar of `process(ProcessorSupplier)` to let users easily specify some customized commit operator on the topology. This way we can better educate the user that `commit` is not on a specific record or operator, but on the whole topology / task.
> Better support for user-specific committing in the Streams DSL
> --------------------------------------------------------------
>
> Key: KAFKA-3907
> URL: https://issues.apache.org/jira/browse/KAFKA-3907
> Project: Kafka
> Issue Type: Bug
> Components: streams
> Reporter: Guozhang Wang
> Priority: Major
> Labels: api
>
> Currently for user-specifically committing the current processing state, users can make use of the {{ProcessorContext}} object, which is exposed in the {{Processor}} API. Other than that, the application will also automatically committing the processing state based on the configured interval.
> Hence in the Streams DSL, if a user wants to explicitly call {{commit}}, she needs to use a {{process(ProcessorSupplier)}} API to get a customized processor instance in order to access the {{ProcessorContext}}. We should think of a better way to support user-specific committing interfaces inside the high-level Streams DSL.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)