[hibernate-issues] [Hibernate-JIRA] Updated: (HHH-6620) @TableGenerator-governed sequence table insertion problems

[Laird Nelson (JIRA)]

     [ http://opensource.atlassian.com/projects/hibernate/browse/HHH-6620?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Laird Nelson updated HHH-6620:
------------------------------

    Attachment: hhh-6620.zip

Attaching self-contained test case that shows the problem. I've turned show_sql on and dialed up the H2 diagnostics so that you can see exactly what the H2 driver sees.  Note that the ID being inserted is 0, not 5000, i.e. that the initialValue attribute is being ignored as described.

> @TableGenerator-governed sequence table insertion problems
> ----------------------------------------------------------
>
>                 Key: HHH-6620
>                 URL: http://opensource.atlassian.com/projects/hibernate/browse/HHH-6620
>             Project: Hibernate Core
>          Issue Type: Bug
>          Components: annotations, core
>    Affects Versions: 3.5.0-Final
>         Environment: May affect other versions besides 3.5.0-Final
>            Reporter: Laird Nelson
>              Labels: jpa2
>         Attachments: hhh-6620.zip
>
>
> The JPA specification is admittedly mum on this, but Hibernate serves up some surprising behavior.
> Suppose you have a sequence table that is created according to the requirements of a particular {{@TableGenerator}} annotation.  So two columns, storing a sequence name and a sequence value.  Suppose this sequence table is created by some DDL script somewhere, and Hibernate's automatic schema creation is deliberately switched off.
> Hibernate will helpfully insert new rows into this table as required, but when it does so it does not honor the {{initialValue}} attribute of the {{@TableGenerator}} annotation in question.
> I will supply a test case once I get the bug number.
> The workaround is obviously to manually insert a row into this table with the appropriate initial value, but then of course the {{initialValue}} attribute is largely useless.
> OpenJPA handles this case flawlessly.
> EclipseLink doesn't even attempt to perform the insertion.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira