[
https://issues.apache.org/jira/browse/CALCITE-4876?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17439678#comment-17439678
]
ZheHu commented on CALCITE-4876:
--------------------------------
Hi, [~julianhyde], you are right about this.
When I change the precedence of *INTERSECT_ALL* from {color:#FF0000}18 to
14{color}(same with UNION and EXCEPT), I get the right SQL expression. Just as
you mentioned above, I also tried the following two testcases:
(1) Converting the below RelNode to SQL with HiveSqlDialect(LIMIT syntax), SQL
expression is correct.
{code:java}
RelNode rel =
builder.scan("DEPT")
.project(builder.field("DEPTNO"))
.scan("EMP")
.project(builder.field("DEPTNO"))
.minus(true)
.limit(0, 2)
.build();
new RelToSqlConverter(SqlDialect.DatabaseProduct.HIVE.getDialect())
.visitRoot(rel)
.asStatement()
.toSqlString(SqlDialect.DatabaseProduct.HIVE.getDialect())
{code}
(2) Running the following SQL in JdbcTest.java, I also get correct results.
{code:java}
CalciteAssert.hr()
.query("select \"deptno\" from \"hr\".\"depts\"\n"
+ "intersect all\n"
+ "select \"deptno\" from \"hr\".\"depts\"\n"
+ "limit 2")
.returns("deptno=40\ndeptno=10\n");
{code}
I might need a bit more time to look further into this precedence issue.
> Converting RelNode to SQL with CalciteSqlDialect gets wrong result while
> EnumerableIntersect is followed by EnumerableLimit
> ---------------------------------------------------------------------------------------------------------------------------
>
> Key: CALCITE-4876
> URL: https://issues.apache.org/jira/browse/CALCITE-4876
> Project: Calcite
> Issue Type: Bug
> Components: core
> Affects Versions: 1.28.0
> Reporter: ZheHu
> Priority: Major
>
> When I convert the following RelNode to SQL(with CalciteSqlDialect) in
> ToLogicalConverterTest.java:
> {code:java}
> @Test void testIntersect() {
> final RelBuilder builder = builder();
> RelNode rel =
> builder.scan("DEPT")
> .project(builder.field("DEPTNO"))
> .scan("EMP")
> .project(builder.field("DEPTNO"))
> .intersect(true)
> .limit(0, 2)
> .build();
> System.out.println(new
> RelToSqlConverter(SqlDialect.DatabaseProduct.CALCITE.getDialect()).visitRoot(rel).asStatement());
> {code}
> I get:
> {code:java}
> SELECT *
> FROM SELECT `DEPTNO`
> FROM `scott`.`DEPT`
> INTERSECT ALL
> SELECT `DEPTNO`
> FROM `scott`.`EMP`
> FETCH NEXT 2 ROWS ONLY
> {code}
> But the expected SQL should be:
> {code:java}
> SELECT *
> FROM (SELECT `DEPTNO`
> FROM `scott`.`DEPT`
> INTERSECT ALL
> SELECT `DEPTNO`
> FROM `scott`.`EMP`)
> FETCH NEXT 2 ROWS ONLY
> {code}
> Other SetOperators like UNION、MINUS can be converted correctly.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
