[
https://issues.apache.org/jira/browse/FLINK-5865?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15926567#comment-15926567
]
ASF GitHub Bot commented on FLINK-5865:
---------------------------------------
Github user StephanEwen commented on the issue:
https://github.com/apache/flink/pull/3380
I think you are right about the argument that we do not expect users to
"handle" the exceptions in any way, so they may as well be unchecked exceptions.
Should we then throw `FlinkRuntimeException` or define a specific
`StateException extends FlinkRuntimeException` that we throw there?
> Throw original exception in states
> ----------------------------------
>
> Key: FLINK-5865
> URL: https://issues.apache.org/jira/browse/FLINK-5865
> Project: Flink
> Issue Type: Improvement
> Components: State Backends, Checkpointing
> Affects Versions: 1.3.0
> Reporter: Xiaogang Shi
> Assignee: Xiaogang Shi
>
> Now all exception thrown in RocksDB states are converted to
> {{RuntimeException}}. It's unnecessary and will print useless stacks in the
> log.
> I think it's better to throw the original exception, without any wrapping.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
