[
https://issues.apache.org/jira/browse/OPENJPA-2940?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18080594#comment-18080594
]
ASF subversion and git services commented on OPENJPA-2940:
----------------------------------------------------------
Commit a0674cd1147e56600ee3880527c32d8e76359ccb in openjpa's branch
refs/heads/OPENJPA-2940 from Richard Zowalla
[ https://gitbox.apache.org/repos/asf?p=openjpa.git;h=a0674cd11 ]
[OPENJPA-2940] Fix TestPessimisticLocks fragility on PG17
The non-ORDER-BY query "select e from Employee e where e.id < 10" with
OFFSET 1 may return either id=1 or id=2 depending on PG row order. The
test asserted both 'first.1' or 'first.2' for the result but then
hardcoded em2.find(Employee.class, 2, ...) which only blocks when em1
locked id=2. Lock whichever employee survived the OFFSET and pass that
id to em2.find so the test no longer depends on flush-order side effects
(seen on PG17 with supportsLockingWithOuterJoin=false where
@SequenceGenerators metadata loading shifted HashMap iteration).
> Implement Jakarta Persistence 3.2
> ---------------------------------
>
> Key: OPENJPA-2940
> URL: https://issues.apache.org/jira/browse/OPENJPA-2940
> Project: OpenJPA
> Issue Type: Improvement
> Reporter: Paulo Cristovão de Araújo Silva Filho
> Priority: Major
> Attachments: image-2026-03-28-14-55-26-463.png
>
>
> Implement Jakarta Persistence API 3.2 as prescribed by
> [specification|https://jakarta.ee/specifications/persistence/3.2/] and
> required by Jakarta EE 11.
> This change requires JDK17 or higher compatibility and needs considerable
> checks.
> I've already pushed a branch (WIP) without the proper API bump, focusing on
> updating dependencies while keeping the project passing its current tests at
> least in default profile.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
