[
https://issues.apache.org/jira/browse/FLINK-3852?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15279399#comment-15279399
]
Tzu-Li (Gordon) Tai commented on FLINK-3852:
--------------------------------------------
Hi Mark,
I think replacing the StreamingJob skeleton as the main class doesn't match
what this issue is aiming for.
I see 2 approaches here:
1. BatchJob skeleton and StreamingJob skeleton within same quickstart module (
i.e. Batch & Streaming Java / Batch & Streaming Scala)
In the javadoc within each skeleton, you can guide the user to use `./bin/flink
run -c <either BatchJob or StreamingJob> <quickstart jar>` as instructed in
https://ci.apache.org/projects/flink/flink-docs-release-1.0/apis/cli.html.
Instead on defining a mainClass in the shade plugin, comment the transformers
for the mainClass setting out with instructions that the user can add this back
if they decide to simply use `./bin/flink run <quickstart jar>` with either one
of the skeletons as main class entry point.
2. Have 4 separate quickstarts, i.e. Batch Java / Batch Scala / Streaming Java
/ Streaming Scala.
I like the first approach better, since it also has a natural way of letting
the user know that there are different submission options for a packaged Flink
jar job.
> Use a StreamExecutionEnvironment in the quickstart job skeleton
> ---------------------------------------------------------------
>
> Key: FLINK-3852
> URL: https://issues.apache.org/jira/browse/FLINK-3852
> Project: Flink
> Issue Type: Task
> Components: Quickstarts
> Reporter: Robert Metzger
> Labels: starter
>
> The Job skeleton created by the maven archetype "quickstart" is still setting
> up an ExecutionEnvironment, not a StreamExecutionEnvironment.
> These days, most users are using Flink for streaming, so we should reflect
> that in the quickstart as well.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
