[
https://issues.apache.org/jira/browse/OPENJPA-528?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12580516#action_12580516
]
Jeremy Bauer commented on OPENJPA-528:
--------------------------------------
The original fix provided by Michael has been enhanced to eliminate the
problems encountered in the TCK.
In short, when schema mapping occurs columns are inserted into a table object,
including foreign key columns. Then, when foreign keys are mapped into the
same table the column name validator finds the column already exists and
generates a new unique column with a numeric value appended. This column is
mapped to the foreign key. Later, when the entity is persisted, the
extra/incorrectly generated column name is not found in the database table.
The result is a SQLException.
To eliminate this problem, a new method was added to the DBDictionary to
enable/disable the column name uniqueness check. The method to populate the
foreign key column was changed to disable uniqueness checking - since the FK
column is already in the table. Thus, an extra/incorrect column does not get
generated.
A patch is provided as OPENJPA-528.patch. Please review by tomorrow morning
(3/20/08).
> FK columns which contain reserved workds are renamed incorrectly.
> -----------------------------------------------------------------
>
> Key: OPENJPA-528
> URL: https://issues.apache.org/jira/browse/OPENJPA-528
> Project: OpenJPA
> Issue Type: Bug
> Reporter: Michael Dick
> Assignee: Michael Dick
>
> The check for reserved words in FK column names happens before we append the
> _ID suffixes. As a result if a reserved word is found we end up appending a
> number to the reserved word and then appending _ID.
> Whats worse is that the check for uniqueness also occurs before we append the
> _ID, which could lead to other problems down the road.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
