TsReaper opened a new pull request #13897:
URL: https://github.com/apache/flink/pull/13897
## What is the purpose of the change
Consider the following SQL and the execution plan.
```sql
WITH
T1 AS (SELECT COUNT(*) AS cnt FROM x GROUP BY a),
T2 AS (SELECT COUNT(*) AS cnt FROM y GROUP BY d),
T3 AS (SELECT b AS cnt FROM x INNER JOIN y ON x.b = y.e)
SELECT cnt FROM
(SELECT cnt FROM T1)
UNION ALL
(SELECT cnt FROM T2)
UNION ALL
(SELECT cnt FROM T3)
```
```
MultipleInputNode(readOrder=[1,0,0,0], members=[\nUnion(all=[true],
union=[cnt])\n:- Union(all=[true], union=[cnt])\n: :- Calc(select=[CAST(cnt)
AS cnt])\n: : +- HashAggregate(isMerge=[true], groupBy=[a], select=[a,
Final_COUNT(count1$0) AS cnt])\n: : +- [#3]
Exchange(distribution=[hash[a]])\n: +- Calc(select=[CAST(cnt) AS cnt])\n:
+- HashAggregate(isMerge=[true], groupBy=[d], select=[d, Final_COUNT(count1$0)
AS cnt])\n: +- [#4] Exchange(distribution=[hash[d]])\n+- Calc(select=[b
AS cnt])\n +- HashJoin(joinType=[InnerJoin], where=[=(b, e)], select=[b, e],
build=[right])\n :- [#1] Exchange(distribution=[hash[b]])\n +- [#2]
Exchange(distribution=[hash[e]])\n])
:- Exchange(distribution=[hash[b]])
: +- Calc(select=[b])
: +- LegacyTableSourceScan(table=[[default_catalog, default_database, x,
source: [TestTableSource(a, b, c, nx)]]], fields=[a, b, c, nx])
:- Exchange(distribution=[hash[e]])
: +- Calc(select=[e])
: +- LegacyTableSourceScan(table=[[default_catalog, default_database, y,
source: [TestTableSource(d, e, f, ny)]]], fields=[d, e, f, ny])
:- Exchange(distribution=[hash[a]])
: +- LocalHashAggregate(groupBy=[a], select=[a, Partial_COUNT(*) AS
count1$0])
: +- Calc(select=[a])
: +- LegacyTableSourceScan(table=[[default_catalog, default_database,
x, source: [TestTableSource(a, b, c, nx)]]], fields=[a, b, c, nx])
+- Exchange(distribution=[hash[d]])
+- LocalHashAggregate(groupBy=[d], select=[d, Partial_COUNT(*) AS
count1$0])
+- Calc(select=[d])
+- LegacyTableSourceScan(table=[[default_catalog, default_database,
y, source: [TestTableSource(d, e, f, ny)]]], fields=[d, e, f, ny])
```
The two unions here in multiple input here is actually redundant, as the
amount of data shuffled will not change even if they're moved out of the
multiple input node. We should remove such redundant union from multiple input
nodes.
## Brief change log
- Remove redundant union from multiple input node
## Verifying this change
This change added tests and can be verified by running the added tests.
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): no
- The public API, i.e., is any changed class annotated with
`@Public(Evolving)`: no
- The serializers: no
- The runtime per-record code paths (performance sensitive): no
- Anything that affects deployment or recovery: JobManager (and its
components), Checkpointing, Kubernetes/Yarn/Mesos, ZooKeeper: no
- The S3 file system connector: no
## Documentation
- Does this pull request introduce a new feature? no
- If yes, how is the feature documented? not applicable
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
