[
https://issues.apache.org/jira/browse/FLINK-23498?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Timo Walther updated FLINK-23498:
---------------------------------
Description:
The configuration story in Flink is not very consistent at the moment. Ideally,
we should have a layered approach where pipeline executor, DataStream API, and
Table API store their configuration and add it to a global configuration on
request. However, this is a big change that we would like to avoid at this
point.
Instead, we partially follow this approach by adding:
- {{Executor.getConfiguration}} to access the config of lower layers
- {{Planner.getConfiguration}} to access a global configuration during planning
This is required to access properties stored in
{{StreamExecutionEnvironment.configuration}}.
was:
The configuration story in Flink is not very consistent at the moment. Ideally,
we should have a layered approach where pipeline executor, DataStream API, and
Table API store their configuration and add it to a global configuration on
request. However, this is a big change that we would like to avoid at this
point.
Instead, we partially follow this approach by adding:
- {{Executor.configure}} to propagate changes to the
{{StreamExecutionEnvironment}} layer
- {{Executor.getConfiguration}} to access the config of lower layers
- {{Planner.getConfiguration}} to access a global configuration during planning
This is required to access properties stored in
{{StreamExecutionEnvironment.configuration}}.
> Use a layered configuration in Executor and Planner
> ---------------------------------------------------
>
> Key: FLINK-23498
> URL: https://issues.apache.org/jira/browse/FLINK-23498
> Project: Flink
> Issue Type: Sub-task
> Components: API / DataStream, Table SQL / API
> Reporter: Timo Walther
> Assignee: Timo Walther
> Priority: Major
> Labels: pull-request-available
>
> The configuration story in Flink is not very consistent at the moment.
> Ideally, we should have a layered approach where pipeline executor,
> DataStream API, and Table API store their configuration and add it to a
> global configuration on request. However, this is a big change that we would
> like to avoid at this point.
> Instead, we partially follow this approach by adding:
> - {{Executor.getConfiguration}} to access the config of lower layers
> - {{Planner.getConfiguration}} to access a global configuration during
> planning
> This is required to access properties stored in
> {{StreamExecutionEnvironment.configuration}}.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
