[
https://issues.apache.org/jira/browse/CALCITE-1646?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16036725#comment-16036725
]
Zhiqiang He commented on CALCITE-1646:
--------------------------------------
https://github.com/apache/calcite/pull/468
please review it, thanks.
> Partition by and order by syntax support for MATCH_RECOGNIZE
> ------------------------------------------------------------
>
> Key: CALCITE-1646
> URL: https://issues.apache.org/jira/browse/CALCITE-1646
> Project: Calcite
> Issue Type: Sub-task
> Components: core
> Affects Versions: 1.11.0
> Reporter: Zhiqiang He
> Assignee: Zhiqiang He
> Labels: features
>
> h1. PARTITION BY: Logically Dividing the Rows into Groups
> Typically, you want to divide your input data into logical groups for
> analysis. In the examples with stocks, the pattern matching is divided so
> that it applies to just one stock at a time. To do this, use the PARTITION BY
> clause. PARTITION BY specifies that the rows of the input table are to be
> partitioned by one or more columns. Matches are found within partitions and
> do not cross partition boundaries.
> If there is no PARTITION BY, then all rows of the row pattern input table
> constitute a single row pattern partition.
> h1. ORDER BY: Logically Ordering the Rows in a Partition
> The ORDER BY clause is used to specify the order of rows within a row pattern
> partition. If the order of two rows in a row pattern partition is not
> determined by ORDER BY, then the result of the MATCH_RECOGNIZE clause is
> non-deterministic: it may not give consistent results each time the query is
> run.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
