[
https://issues.apache.org/jira/browse/OPENJPA-2764?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16731852#comment-16731852
]
ASF subversion and git services commented on OPENJPA-2764:
----------------------------------------------------------
Commit 260042ba5184f1f415f2f0975e3449f13611b84a in openjpa's branch
refs/heads/master from Mark Struberg
[ https://gitbox.apache.org/repos/asf?p=openjpa.git;h=260042b ]
OPENJPA-2764 fix tests with random behaviour
The behaviour was random due to the undefined order in which
real databases return data.
> Map path expression tests behave random
> ---------------------------------------
>
> Key: OPENJPA-2764
> URL: https://issues.apache.org/jira/browse/OPENJPA-2764
> Project: OpenJPA
> Issue Type: Bug
> Components: build / infrastructure
> Affects Versions: 3.0.0
> Reporter: Mark Struberg
> Assignee: Mark Struberg
> Priority: Major
> Fix For: 3.0.1
>
>
> Most of the tests under org.apache.openjpa.persistence.jdbc.maps behave
> random when they test against a real database.
> Those tests capture the JQPL path navigation for Maps (Covered in the spec in
> 4.4.4.1).
> {code}
> public class @Entity Division {
> private Map<Division, VicePresident> orga;
> }
> {code}
> Such structures can be navitated via KEY(), VALUE(), and ENTRY().
> Our tests did create 2 Divisions with 2 orga entries. And using
> {{query.getResultList().get(0)}} to verify the results. And this was exactly
> the problem. using {{get(0)}} leads to random behaviour with real databases.
> On the default Derby database it didn't make any difference as the result
> from the index query was always in the order in which the data got written to
> disk. But this is not guaranteed for performance tuned databases like
> PostgreSQL, MariaDB and MySQL. In those cases we got random errors.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
