Evgeny Stanilovsky created IGNITE-25504:
-------------------------------------------
Summary: Sql. Confusing explain output for join condition
Key: IGNITE-25504
URL: https://issues.apache.org/jira/browse/IGNITE-25504
Project: Ignite
Issue Type: Task
Components: sql ai3
Reporter: Evgeny Stanilovsky
Confusing condition found:
{noformat}
MergeJoin
condition: =(VAL1, VAL10) <--- VAL10 what does it mean ?
joinType: inner
est: (rows=1)
IndexScan
table: PUBLIC.T1
index: IDX1
type: SORTED
fields: [ID, VAL1]
collation: [VAL1 ASC]
est: (rows=1)
IndexScan
table: PUBLIC.T2
index: IDX2
type: SORTED
fields: [ID, VAL1]
collation: [VAL1 ASC, ID ASC]
est: (rows=1)
{noformat}
to reproduce:
with @WithSystemProperty(key = COLOCATION_FEATURE_FLAG, value = "true")
{noformat}
sql("CREATE TABLE T1 (ID INT, VAL1 VARCHAR, VAL2 VARCHAR, PRIMARY
KEY(VAL1, VAL2)) COLOCATE BY(VAL1);");
sql("CREATE TABLE T2 (ID INT, VAL1 VARCHAR, VAL2 VARCHAR, PRIMARY
KEY(VAL1, VAL2)) COLOCATE BY(VAL1);");
sql("CREATE INDEX IDX1 ON T1(VAL1, VAL2);");
sql("CREATE INDEX IDX2 ON T2(VAL1, ID);");
List<List<Object>> res = sql("explain plan for SELECT T1.ID as T1ID,
T2.ID as T2ID FROM T1 JOIN T2 USING (VAL1);");
System.err.println("plan: " + res.get(0));
{noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
