You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "godfrey he (JIRA)" <ji...@apache.org> on 2019/04/12 09:41:00 UTC
[jira] [Updated] (FLINK-12170) Add support for generating optimized
logical plan for Over aggregate
[ https://issues.apache.org/jira/browse/FLINK-12170?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
godfrey he updated FLINK-12170:
-------------------------------
Description:
This issue aims to generate optimized plan for Over aggregate queries on both Batch and Stream, e.g.
{code:sql}
SELECT a, b, c, RANK() OVER (PARTITION BY b ORDER BY c) FROM MyTable
{code}
currently, Stream requires all over aggregate functions must be computed on the same window, e.g.
{code:sql}
SELECT c,
COUNT(a) OVER (PARTITION BY c ORDER BY proctime RANGE UNBOUNDED PRECEDING),
SUM(a) OVER (PARTITION BY b ORDER BY proctime RANGE UNBOUNDED PRECEDING)
from MyTable
{code}
the above sql is not supported because the partition keys is different.
Batch does not have such limitation.
was:
This issue aims to generate optimized plan for Over aggregate queries, e.g.
{code:sql}
SELECT a, b, c, RANK() OVER (PARTITION BY b ORDER BY c) FROM MyTable
{code}
currently, Stream requires all over aggregate functions must be computed on the same window, e.g.
{code:sql}
SELECT c,
COUNT(a) OVER (PARTITION BY c ORDER BY proctime RANGE UNBOUNDED PRECEDING),
SUM(a) OVER (PARTITION BY b ORDER BY proctime RANGE UNBOUNDED PRECEDING)
from MyTable
{code}
the above sql is not supported because the partition keys is different.
Batch does not have such limitation.
> Add support for generating optimized logical plan for Over aggregate
> --------------------------------------------------------------------
>
> Key: FLINK-12170
> URL: https://issues.apache.org/jira/browse/FLINK-12170
> Project: Flink
> Issue Type: New Feature
> Components: Table SQL / Planner
> Reporter: godfrey he
> Assignee: godfrey he
> Priority: Major
>
> This issue aims to generate optimized plan for Over aggregate queries on both Batch and Stream, e.g.
> {code:sql}
> SELECT a, b, c, RANK() OVER (PARTITION BY b ORDER BY c) FROM MyTable
> {code}
> currently, Stream requires all over aggregate functions must be computed on the same window, e.g.
> {code:sql}
> SELECT c,
> COUNT(a) OVER (PARTITION BY c ORDER BY proctime RANGE UNBOUNDED PRECEDING),
> SUM(a) OVER (PARTITION BY b ORDER BY proctime RANGE UNBOUNDED PRECEDING)
> from MyTable
> {code}
> the above sql is not supported because the partition keys is different.
> Batch does not have such limitation.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)