[teiid-issues] [JBoss JIRA] (TEIID-4175) Certain nested dependent join structure will cause fewer results than expected
Steven Hawkins (JIRA)
issues at jboss.org
Wed Apr 27 16:53:00 EDT 2016
[ https://issues.jboss.org/browse/TEIID-4175?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Steven Hawkins updated TEIID-4175:
----------------------------------
Description:
With a query such as:
{code}
SELECT pm1.g1.e1, pm1.g2.e2 from /*+ makeind */ pm1.g1 inner join /*+ preserve */ (/*+ makeind */ pm1.g2 inner join pm1.g3 on pm1.g2.e2 = pm1.g3.e2) on pm1.g1.e1 = pm1.g2.e1
{code}
There will be a join structure like:
{code}
JoinNode(1) [Dependent]
AccessNode(2)
JoinNode(3) [Dependent] [MERGE JOIN (ALREADY_SORTED/ALREADY_SORTED)]
DependentAccessNode(4)
DependentAccessNode(5)
{code}
Such that there is a dependent join as a left child of a merge join that is marked as having that child already sorted. When there is more than 1 query needed for DependentAccessNode(4), the DependentAccessNode will mistakenly mark the right child as needing to be sorted.
was:
With a query such as:
SELECT pm1.g1.e1, pm1.g2.e2 from /*+ makeind */ pm1.g1 inner join /*+ preserve */ (/*+ makeind */ pm1.g2 inner join pm1.g3 on pm1.g2.e2 = pm1.g3.e2) on pm1.g1.e1 = pm1.g2.e1
There will be a join structure like:
JoinNode(1) [Dependent]
AccessNode(2)
JoinNode(3) [Dependent] [MERGE JOIN (ALREADY_SORTED/ALREADY_SORTED)]
DependentAccessNode(4)
DependentAccessNode(5)
Such that there is a dependent join as a left child of a merge join that is marked as having that child already sorted. When there is more than 1 query needed for DependentAccessNode(4), the DependentAccessNode will mistakenly mark the right child as needing to be sorted.
> Certain nested dependent join structure will cause fewer results than expected
> ------------------------------------------------------------------------------
>
> Key: TEIID-4175
> URL: https://issues.jboss.org/browse/TEIID-4175
> Project: Teiid
> Issue Type: Bug
> Components: Query Engine
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Priority: Blocker
> Fix For: 9.0, 8.12.5, 8.13.5
>
>
> With a query such as:
> {code}
> SELECT pm1.g1.e1, pm1.g2.e2 from /*+ makeind */ pm1.g1 inner join /*+ preserve */ (/*+ makeind */ pm1.g2 inner join pm1.g3 on pm1.g2.e2 = pm1.g3.e2) on pm1.g1.e1 = pm1.g2.e1
> {code}
> There will be a join structure like:
> {code}
> JoinNode(1) [Dependent]
> AccessNode(2)
> JoinNode(3) [Dependent] [MERGE JOIN (ALREADY_SORTED/ALREADY_SORTED)]
> DependentAccessNode(4)
> DependentAccessNode(5)
> {code}
> Such that there is a dependent join as a left child of a merge join that is marked as having that child already sorted. When there is more than 1 query needed for DependentAccessNode(4), the DependentAccessNode will mistakenly mark the right child as needing to be sorted.
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)
More information about the teiid-issues
mailing list