[
https://issues.apache.org/jira/browse/FLINK-3634?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15205986#comment-15205986
]
ASF GitHub Bot commented on FLINK-3634:
---------------------------------------
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1817#issuecomment-199694039
Changes look good.
Shall we add that `zipWithIndex` requires two passes and is hence not
pipelined and that `zipWithUniqueId` should be preferred if possible?
> Fix documentation for DataSetUtils.zipWithUniqueId()
> ----------------------------------------------------
>
> Key: FLINK-3634
> URL: https://issues.apache.org/jira/browse/FLINK-3634
> Project: Flink
> Issue Type: Improvement
> Components: Documentation
> Affects Versions: 1.1.0
> Reporter: Greg Hogan
> Assignee: Greg Hogan
> Priority: Minor
> Fix For: 1.1.0, 1.0.1
>
>
> Under FLINK-2590 the assignment and testing of unique IDs was improved but
> the documentation looks to still reference the old implementation.
> With {{parallelism=1}} there is no difference between zipWithUniqueID and
> zipWithIndex. With greater parallelism the results of zipWithUniqueID are
> dependent on the partitioning.
> The documentation should demonstrate a possible result that is different from
> the incremental sequence of zipWithIndex while noting that results are
> dependent on the parallelism and partitioning.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
