[
https://issues.apache.org/jira/browse/FLINK-14501?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16957645#comment-16957645
]
Yang Wang commented on FLINK-14501:
-----------------------------------
It make sense that the `CustomCommandLine` just parse the cli options to flink
configuration. And then use the configuration to create the corresponding
`ClusterDescriptor`.
> The correct {{ClusterClientFactory}} can be discovered based on Service
>Discovery.
I'm not clear about the Service Discovery. Do we put the
createClusterDescriptor in the `{{ClusterClientFactory}}`?
> Move ClusteDescriptor/ClusterSpecification creation from the
> CustomCommandLine to a ClusterClientFactory
> --------------------------------------------------------------------------------------------------------
>
> Key: FLINK-14501
> URL: https://issues.apache.org/jira/browse/FLINK-14501
> Project: Flink
> Issue Type: Sub-task
> Components: Client / Job Submission, Command Line Client
> Affects Versions: 1.9.0
> Reporter: Kostas Kloudas
> Assignee: Kostas Kloudas
> Priority: Major
>
> This issue targets to make the {{ClusterDescriptor}} and the
> {{ClusterSpecification}} independent from the {{CommandLine}} . Currently
> this is not the case, as the {{CustomCommandLine}} is responsible for
> instantiating them.
>
> The correct {{ClusterClientFactory}} can be discovered based on Service
> Discovery.
>
> This decoupling will enable the further decoupling of the cluster execution
> from the command line-related code, which apart from unnecessary, it is also
> a bad separation of concerns. Now the {{CustomCommandLine}} is only
> responsible for parsing the CLI options to configuration options.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
