Mark Andreev created SPARK-58980:
------------------------------------
Summary: Adds support to `groups` type in window clauses
Key: SPARK-58980
URL: https://issues.apache.org/jira/browse/SPARK-58980
Project: Spark
Issue Type: Improvement
Components: SQL
Affects Versions: 5.0.0
Reporter: Mark Andreev
Spark SQL supports ROWS and RANGE window frames, but not SQL-standard GROUPS,
which counts peer groups defined by the complete window ORDER BY key.
*Proposal*
Add GROUPS as a third window frame type with the standard boundaries:
- GROUPS BETWEEN <lower-bound> AND <upper-bound>
CURRENT ROW represents the current peer group, while n PRECEDING/FOLLOWING
moves by peer groups. GROUPS requires ORDER BY and supports multiple ordering
expressions.
Implementation should reuse Spark's existing peer semantics and window
infrastructure. No new logical/physical operator, sort, or shuffle is required.
Mail list discussion:
[https://lists.apache.org/thread/0pcwflbjnykxjybbmy277lb045wognsp]
Design document:
[https://docs.google.com/document/d/1BOVWBHz6vQi39uQ_8SJfMWg55oJrUqNMQAgtHXmTZhA/edit?tab=t.0#heading=h.11jrf3z6k51f]
Related ticket: https://issues.apache.org/jira/browse/SPARK-28648
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
