db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dag H. Wanvik (JIRA)" <j...@apache.org>
Subject [jira] Updated: (DERBY-4712) Complex nested joins problems
Date Tue, 06 Jul 2010 21:47:52 GMT

     [ https://issues.apache.org/jira/browse/DERBY-4712?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Dag H. Wanvik updated DERBY-4712:
---------------------------------

    Attachment: assert-bind-opt-trees.txt
                assert-bind-opt-trees.log

Attaching derby.log which dump of the query tree after bind and optimization of a simplified
query which also gives the ASSERT error, and an annotated drawing of the query tree after
bind and optimize. The drawing shows that the column reference which gives the ASSERT during
code generation points to a result column of a result column list which has been replaced
during the optimization phases. This is the immediate cause of the error.

The column reference is part of a join condition (T2.X=1) on the outermost left join. This
has been pushed into the right side of the outermost left join so after optimization, the
predicate is part of the PRN above the inner join which is the right side of the outermost
left join.

> Complex nested joins problems
> -----------------------------
>
>                 Key: DERBY-4712
>                 URL: https://issues.apache.org/jira/browse/DERBY-4712
>             Project: Derby
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 10.6.1.0
>            Reporter: Thomas Mueller
>            Priority: Minor
>         Attachments: assert-bind-opt-trees.log, assert-bind-opt-trees.txt
>
>
> I ran a randomized test for nested joins against PostgreSQL, Derby, and H2, and found
some problems with Derby. The queries below throw NullPointerExceptions; the last query an
assertion. The test case is here: http://code.google.com/p/h2database/source/browse/trunk/h2/src/test/org/h2/test/db/TestNestedJoins.java
. There are probably shorter queries that are problematic, but I can't test it because Derby
closes the connection after the assertion.
> create table t0(x int);
> create table t1(x int);
> create table t2(x int);
> create table t3(x int);
> create table t4(x int);
> insert into t4 values(0);
> insert into t4 values(1);
> insert into t4 values(2);
> insert into t4 values(3);
> create table t5(x int);
> insert into t5 values(0);
> insert into t5 values(1);
> insert into t5 values(2);
> insert into t5 values(3);
> insert into t5 values(4);
> create table t6(x int);
> insert into t6 values(0);
> insert into t6 values(1);
> insert into t6 values(2);
> insert into t6 values(3);
> insert into t6 values(4);
> insert into t6 values(5);
> create table t7(x int);
> insert into t7 values(0);
> insert into t7 values(1);
> insert into t7 values(2);
> insert into t7 values(3);
> insert into t7 values(4);
> insert into t7 values(5);
> insert into t7 values(6);
> create table t8(x int);
> insert into t8 values(0);
> insert into t8 values(1);
> insert into t8 values(2);
> insert into t8 values(3);
> insert into t8 values(4);
> insert into t8 values(5);
> insert into t8 values(6);
> insert into t8 values(7);
> create table t9(x int);
> insert into t9 values(0);
> insert into t9 values(1);
> insert into t9 values(2);
> insert into t9 values(3);
> insert into t9 values(4);
> insert into t9 values(5);
> insert into t9 values(6);
> insert into t9 values(7);
> insert into t9 values(8);
> insert into t0 values(1);
> insert into t1 values(2);
> insert into t0 values(3);
> insert into t1 values(3);
> insert into t2 values(4);
> insert into t0 values(5);
> insert into t2 values(5);
> insert into t1 values(6);
> insert into t2 values(6);
> insert into t0 values(7);
> insert into t1 values(7);
> insert into t2 values(7);
> insert into t3 values(8);
> insert into t0 values(9);
> insert into t3 values(9);
> insert into t1 values(10);
> insert into t3 values(10);
> insert into t0 values(11);
> insert into t1 values(11);
> insert into t3 values(11);
> insert into t2 values(12);
> insert into t3 values(12);
> insert into t0 values(13);
> insert into t2 values(13);
> insert into t3 values(13);
> insert into t1 values(14);
> insert into t2 values(14);
> insert into t3 values(14);
> insert into t0 values(15);
> insert into t1 values(15);
> insert into t2 values(15);
> insert into t3 values(15);
> select t0.x , t1.x , t2.x , t3.x , t4.x , t5.x , t6.x , t7.x , t8.x from (((t0 inner
join ((t1 right outer join (t2 inner join t3 on t2.x = t3.x ) on t1.x = t2.x ) left outer
join (t4 inner join t5 on t4.x = t5.x ) on t1.x = t4.x ) on t0.x = t2.x ) left outer join
(t6 inner join t7 on t6.x = t7.x ) on t1.x = t6.x ) inner join t8 on t5.x = t8.x );
> select t0.x , t1.x , t2.x , t3.x , t4.x , t5.x , t6.x , t7.x from ((t0 right outer join
t1 on t0.x = t1.x ) inner join (((t2 inner join (t3 left outer join t4 on t3.x = t4.x ) on
t2.x = t3.x ) right outer join t5 on t2.x = t5.x ) left outer join (t6 inner join t7 on t6.x
= t7.x ) on t4.x = t6.x ) on t0.x = t5.x );
> select t0.x , t1.x , t2.x , t3.x , t4.x , t5.x , t6.x , t7.x from ((((t0 left outer join
t1 on t0.x = t1.x ) right outer join t2 on t0.x = t2.x ) right outer join t3 on t0.x = t3.x
) inner join ((t4 inner join t5 on t4.x = t5.x ) right outer join (t6 right outer join t7
on t6.x = t7.x ) on t4.x = t6.x ) on t1.x = t4.x );
> select t0.x , t1.x , t2.x , t3.x , t4.x , t5.x from (((t0 inner join t1 on t0.x = t1.x
) right outer join (t2 right outer join t3 on t2.x = t3.x ) on t0.x = t2.x ) inner join (t4
left outer join t5 on t4.x = t5.x ) on t1.x = t4.x );
> select t0.x , t1.x , t2.x , t3.x , t4.x , t5.x , t6.x from ((t0 right outer join (t1
right outer join (t2 left outer join (t3 left outer join t4 on t3.x = t4.x ) on t2.x = t3.x
) on t1.x = t3.x ) on t0.x = t1.x ) left outer join (t5 inner join t6 on t5.x = t6.x ) on
t2.x = t5.x );

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message