[
https://issues.apache.org/jira/browse/HIVE-12574?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15037992#comment-15037992
]
Aihua Xu commented on HIVE-12574:
---------------------------------
Patch#2: include the fix for count() and collect_set(). We need to check if
rowToProcess is negative before processing the row.
> windowing function returns incorrect result when the window size is larger
> than the partition size
> --------------------------------------------------------------------------------------------------
>
> Key: HIVE-12574
> URL: https://issues.apache.org/jira/browse/HIVE-12574
> Project: Hive
> Issue Type: Sub-task
> Components: PTF-Windowing
> Affects Versions: 2.0.0
> Reporter: Aihua Xu
> Assignee: Aihua Xu
> Attachments: HIVE-12574.2.patch, HIVE-12574.patch
>
>
> In PTF windowing, when the partition is small and the window size is larger
> than the partition size, we are seeing incorrect result. It happens for max,
> min, first_value, last_value and sum functions.
> {noformat}
> CREATE TABLE sdy1(
> ord int,
> type string);
> {noformat}
> The data is:
> {noformat}
> 2 a
> 3 a
> 1 a
> {noformat}
> The result is as follows for the query {{select ord, min(ord) over (partition
> by type order by ord rows between 1 preceding and 7 following)}}
> {noformat}
> 1 1
> 2 1
> 3 1
> {noformat}
> The expected result is:
> {noformat}
> 1 1
> 2 1
> 3 2
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
