[
https://issues.apache.org/jira/browse/SPARK-56572?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Andreas Neumann updated SPARK-56572:
------------------------------------
Affects Version/s: 4.1.1
4.1.0
(was: 4.2.0)
(was: 4.1.2)
> Inject the Spark session into python pipeline code
> --------------------------------------------------
>
> Key: SPARK-56572
> URL: https://issues.apache.org/jira/browse/SPARK-56572
> Project: Spark
> Issue Type: Improvement
> Components: Declarative Pipelines
> Affects Versions: 4.1.0, 4.1.1
> Reporter: Andreas Neumann
> Priority: Major
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> In Declarative Pipelines, all Python scripts run as separate modules but have
> to share the same Spark session. This session is created by the framework,
> and every script needs to bring it into its own scope by declaring
> {code:java}
> spark = SparkSession.active() {code}
> This bears the risk that users might create their own Spark session instead,
> and that could break dependencies between different Python scripts of the
> same pipeline. It would be better to inject this session directly into the
> module, so user code does not need to worry about obtaining it.
> Also, change `spark-pipelines init` to omit that line from the generated
> sample code.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
