[
https://issues.apache.org/jira/browse/FLINK-3788?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15247924#comment-15247924
]
Stefano Baghino commented on FLINK-3788:
----------------------------------------
Yes, I experimented this very problem when trying Flink for the first time. I
believe it depends on how the {{App}} trait is implemented in Scala
([DelayedInit
Scaladoc|http://www.scala-lang.org/api/current/#scala.DelayedInit]), perhaps
this won't be an issue with Scala 2.12+.
> Local variable values are not distributed to job runners
> --------------------------------------------------------
>
> Key: FLINK-3788
> URL: https://issues.apache.org/jira/browse/FLINK-3788
> Project: Flink
> Issue Type: Bug
> Components: DataSet API
> Affects Versions: 1.0.0, 1.0.1
> Environment: Scala 2.11.8
> Sun JDK 1.8.0_65 or OpenJDK 1.8.0_77
> Fedora 25, 4.6.0-0.rc2.git3.1.fc25.x86_64
> Reporter: Andreas C. Osowski
> Attachments: FLINK-3788.tgz
>
>
> Variable values of non-elementary types aren't caught and distributed to job
> runners, causing them to remain 'null' and causing NPEs upon access when
> running on a cluster. Running locally through `flink-clients` works fine.
> Changing parallelism or disabling the closure cleaner don't seem to have any
> effect.
> Minimal example, also see the attached archive.
> {code:java}
> case class IntWrapper(a1: Int)
> val wrapped = IntWrapper(42)
> env.readTextFile("myTextFile.txt").map(line => wrapped.toString).collect
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
