You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Satyam Shekhar (Jira)" <ji...@apache.org> on 2020/08/04 16:45:00 UTC
[jira] [Commented] (FLINK-18095) Custom Execution Environment for
TableEnvironment
[ https://issues.apache.org/jira/browse/FLINK-18095?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17170950#comment-17170950 ]
Satyam Shekhar commented on FLINK-18095:
----------------------------------------
For testing purposes, I create a mini-cluster that can outlive a query. I would like to use a TableEnviroment connected to the mini-cluster in tests. The ability to inject a StreamingEnvironment provides the right interface to allow users to easily connect to real Flink cluster in production or in-process mini-cluster in tests.
> Custom Execution Environment for TableEnvironment
> -------------------------------------------------
>
> Key: FLINK-18095
> URL: https://issues.apache.org/jira/browse/FLINK-18095
> Project: Flink
> Issue Type: New Feature
> Components: Table SQL / API
> Affects Versions: 1.10.0
> Reporter: Satyam Shekhar
> Priority: Minor
>
> The recommended pattern for creating a TableEnvironment that uses Blink planner in batch mode is the following -
>
> var settings = EnvironmentSettings.newInstance()
> .useBlinkPlanner()
> .inBatchMode()
> .build();
> var tEnv = TableEnvironment.create(settings);
>
> The above configuration, however, does not allow injecting an ExecutionEnvironment in TableEnvironment. Instead, TableEnvironment is hardcoded to always get the ExecutionEnvironment via StreamExecutionEnvironment.getExecutionEnvironment(). This limits users to connect to a remote environment while working with TableEnvironment.
>
> The workaround suggested by [~godfreyhe] and [~jark] is to directly create a StreamTableEnvironmentImpl instance through StreamTableEnvironmentImpl constructor instead of create method.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)