You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Maksim Zhuravkov (Jira)" <ji...@apache.org> on 2023/03/15 05:14:00 UTC

[jira] [Updated] (IGNITE-19031) Sql. Joins with USING condition do not use type coercion rules.

     [ https://issues.apache.org/jira/browse/IGNITE-19031?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Maksim Zhuravkov updated IGNITE-19031:
--------------------------------------
    Summary: Sql. Joins with USING condition do not use type coercion rules.  (was: Sql. Joins with USING condition do not use type coercion.)

> Sql. Joins with USING condition do not use type coercion rules.
> ---------------------------------------------------------------
>
>                 Key: IGNITE-19031
>                 URL: https://issues.apache.org/jira/browse/IGNITE-19031
>             Project: Ignite
>          Issue Type: Bug
>          Components: sql
>    Affects Versions: 3.0.0-beta2
>            Reporter: Maksim Zhuravkov
>            Priority: Minor
>              Labels: calcite3-required, ignite-3
>
> JOINs with USING condition do not use type coercion and that causes some queries to fail  with ClassCastException, even though equivalent JOINs complete successfully.
> {code:java}
> sql("CREATE TABLE T11 (c1 int primary key, c2 INTEGER)");
> sql("CREATE TABLE T12 (c1 BIGINT primary key, c2 BIGINT)");
> Transaction tx = CLUSTER_NODES.get(0).transactions().begin();
> sql(tx, "INSERT INTO T11 VALUES(1, 2)");
> sql(tx, "INSERT INTO T11 VALUES(2, 3)");
> sql(tx, "INSERT INTO T12 VALUES(1, 2)");
> sql(tx, "INSERT INTO T12 VALUES(2, 4)");
> tx.commit();
> sql("SELECT * FROM t11 JOIN t12 USING (c1)");
> {code}
> Error:
> {code:java}
> Caused by: java.lang.ClassCastException: class java.lang.Integer cannot be cast to class java.lang.Long (java.lang.Integer and java.lang.Long are in module java.base of loader 'bootstrap')
> 	at org.apache.ignite.internal.util.ColocationUtils.append(ColocationUtils.java:71)
> 	at org.apache.ignite.internal.sql.engine.util.HashFunctionFactoryImpl$TypesAwareHashFunction.hashOf(HashFunctionFactoryImpl.java:116)
> 	at org.apache.ignite.internal.sql.engine.trait.Partitioned.targets(Partitioned.java:47)
> 	at org.apache.ignite.internal.sql.engine.exec.rel.Outbox.flush(Outbox.java:242)
> 	at org.apache.ignite.internal.sql.engine.exec.rel.Outbox.push(Outbox.java:151)
> 	at org.apache.ignite.internal.sql.engine.exec.rel.SortNode.flush(SortNode.java:193)
> 	at org.apache.ignite.internal.sql.engine.exec.rel.SortNode.end(SortNode.java:154)
> {code}
> An equivalent query passes with no issues:
> {code:java}
> sql("SELECT * FROM t11 JOIN t12 ON t11.c1 = t12.c1");
> {code}
> *Solution*
> Because JOINs that use USING join condition are equivalent to JOINs that use ON condition, It would be better to replace USING condition with equivalent ON condition prior to optimisation (even at the parsing stage) to leverage the code that handles type coercion.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)