[
https://issues.apache.org/jira/browse/FLINK-5883?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15928447#comment-15928447
]
ASF GitHub Bot commented on FLINK-5883:
---------------------------------------
Github user StephanEwen commented on the issue:
https://github.com/apache/flink/pull/3392
@greghogan You are right, this is a bit of an undefined zone here.
In the distributed runtime, all the `Iterables` can be used only once
(because of once consumable streams behind them). This affects the "collection
execution" where in theory no such restriction exists. The PR makes the
behavior consistent across the two execution models.
Strictly speaking, it does change user-facing behavior for cases that rely
on the fact that the "collection execution" allows multiple traversals. We
never defined whether such "environment-specific behavior" is considered part
of the stability contract or not.
What do you think?
> Re-adding the Exception-thrown code for ListKeyGroupedIterator when the
> iterator is requested the second time
> -------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-5883
> URL: https://issues.apache.org/jira/browse/FLINK-5883
> Project: Flink
> Issue Type: Improvement
> Components: DataSet API
> Reporter: lincoln.lee
> Assignee: lincoln.lee
>
> Originally, ListKeyGroupedIterator ensured that a TraversableOnceException
> was thrown when the iterator is requested the second time within FLINK-1023,
> it was lost from FLINK-1110 unexpectedly, so add it back.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
