[
https://issues.apache.org/jira/browse/HIVE-19116?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16429007#comment-16429007
]
Matt McCline commented on HIVE-19116:
-------------------------------------
Patch #2 committed to master. [~djaiswal] thank you for your code review.
> Vectorization: Vector Map data type doesn't keep the order of the key/values
> pairs as read
> ------------------------------------------------------------------------------------------
>
> Key: HIVE-19116
> URL: https://issues.apache.org/jira/browse/HIVE-19116
> Project: Hive
> Issue Type: Bug
> Components: Hive
> Affects Versions: 3.0.0
> Reporter: Matt McCline
> Assignee: Matt McCline
> Priority: Critical
> Fix For: 3.0.0
>
> Attachments: HIVE-19116.01.patch, HIVE-19116.02.patch
>
>
> The VectorExtractRow class does not preserve the order of the key/value pairs
> when going from MapColumnVector to a Map object. This causes Q file
> differences in tests with the MAP data type making it seem like we are
> getting Wrong Results (well, actually we are).
> When LazyMap class (for example) adds key/value pairs to its "map" it uses a
> LinkedHashSet to preserve insert order.
> FYI: [~teddy.choi]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
