[
https://issues.apache.org/jira/browse/FLINK-24070?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17407286#comment-17407286
]
zl commented on FLINK-24070:
----------------------------
Hi [~lzljs3620320]，f2 is a time field. and the code above is just to show two
different over windows in a query statement. the difference between the two
over window is that the first over window uses _*2 PRECEDING*_ rows and the
second over window uses *_5 PRECEDING_* rows.
> Support different over window aggregates in streaming queries
> -------------------------------------------------------------
>
> Key: FLINK-24070
> URL: https://issues.apache.org/jira/browse/FLINK-24070
> Project: Flink
> Issue Type: Improvement
> Components: Table SQL / Planner
> Reporter: zl
> Priority: Major
>
> Currently, Flink does not support using different over window aggregates in a
> query statement, such as:
>
> {code:java}
> // code placeholder
> select
> f1, f2, f3,
> sum(f3) over (PARTITION BY f1 ORDER BY f2 ROWS BETWEEN 2 PRECEDING AND
> CURRENT ROW) as sum_1,
> sum(f3) over (PARTITION BY f1 ORDER BY f2 ROWS BETWEEN 5 PRECEDING AND
> CURRENT ROW) as sum_2
> from table_a;{code}
>
> However, this feature is commonly used in feature engineering for generating
> some new features, do we have plan to support this?
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
