[
https://issues.apache.org/jira/browse/HIVE-10650?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14542566#comment-14542566
]
Ashutosh Chauhan edited comment on HIVE-10650 at 5/13/15 7:52 PM:
------------------------------------------------------------------
Hey [~aihuaxu] I see that you have deleted
{code}
select f, sum(f) over (partition by ts order by f rows between 2 preceding and
1 preceding) from over10k limit 100;
{code}
But you have added same query with additional projection of ts in another q
file. But, results are quite different. This led me to think if we had bug in
current Hive and have wrong results checked in. I then tested this query and
others you have in patch on postgres and results matched with what you have.
So, it seems like we have a correctness bug in current trunk, which this patch
is fixing. Is that correct?
I am still going through code changes.
was (Author: ashutoshc):
Hey [~aihuaxu] I see that you have deleted
{code : sql}
select f, sum(f) over (partition by ts order by f rows between 2 preceding and
1 preceding) from over10k limit 100;
{code}
But you have added same query with additional projection of ts in another q
file. But, results are quite different. This led me to think if we had bug in
current Hive and have wrong results checked in. I then tested this query and
others you have in patch on postgres and results matched with what you have.
So, it seems like we have a correctness bug in current trunk, which this patch
is fixing. Is that correct?
I am still going through code changes.
> Improve sum() function over windowing to support additional range formats
> -------------------------------------------------------------------------
>
> Key: HIVE-10650
> URL: https://issues.apache.org/jira/browse/HIVE-10650
> Project: Hive
> Issue Type: Sub-task
> Components: PTF-Windowing
> Reporter: Aihua Xu
> Assignee: Aihua Xu
> Attachments: HIVE-10650.patch
>
>
> Support the following windowing function {{x preceding and y preceding}} and
> {{x following and y following}}.
> e.g.
> {noformat}
> select sum(value) over (partition by key order by value rows between 2
> preceding and 1 preceding) from tbl1;
> select sum(value) over (partition by key order by value rows between
> unbounded preceding and 1 preceding) from tbl1;
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
