[hibernate-issues] [Hibernate-JIRA] Closed: (HHH-2073) many-to-one join fetch with optional parent causing stray query to fire

[Steve Ebersole (JIRA)]

     [ http://opensource.atlassian.com/projects/hibernate/browse/HHH-2073?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Steve Ebersole closed HHH-2073.
-------------------------------

    Resolution: Rejected

<quote>This is not reflected by a foreign key constraint in the database (hence it allows child to have no parent)</quote> 
This is true simply because of the bad design choices of your db.  Had the db been designed properly and used nulls to represent a non-existent relation, then an FK could very well have been defined to cover/protect this association.

The only "work around" I am aware of is to design the database properly.  As much as you or the original designer wants to think otherwise, a non-null value is not a null value.

The only way to not have the second query issued (aside from fixing the db design) is to add as new feature to explicitly deal with that scenario.  Not saying I am necessarily against that idea, just saying that adding code to work around bad db design decisions is not high on our priority list.  If you want to take a stab at implementing such an enhancement and the work is good, I would certainly consider accepting such a patch.

> many-to-one join fetch with optional parent causing stray query to fire
> -----------------------------------------------------------------------
>
>                 Key: HHH-2073
>                 URL: http://opensource.atlassian.com/projects/hibernate/browse/HHH-2073
>             Project: Hibernate3
>          Issue Type: Bug
>            Reporter: Adam Hardy
>            Priority: Minor
>         Attachments: bugtest.zip
>
>
> I have a call centre application which uses a mature database and is therefore pushing Hibernate to accomplish the mappings I need. The issue here is extra queries that Hibernate is firing where I set up the mapping to do a join in one hit. 
> I am fetching child objects which may or may not have a parent. The child's foreign key points to a non-primary key on the parent. This is not reflected by a foreign key constraint in the database (hence it allows child to have no parent). 
> If the child object has the foreign key 0 instead of null, which is the protocol in this database, then the parent which doesn't exist is not (& cannot) be instantiated, no problem, but then after processing the resultset, Hibernate fires off another query to fetch the parent that it could not find data for in the first query. 
> This is obviously unnecessary in this context and degrades the performance on large searches on this child. 
> I have classified it as minor since I suspect there may be work-arounds but I have not researched that yet. 
> I didn't know whether I could put the mapping XML in the query, so I have attached a simplistic testcase example that demonstrates the problem. 
> The following forum messages are relevant and the latter contains the mappings, which is easier than extracting them out of the testcase zip:
> http://forum.hibernate.org/viewtopic.php?t=960522
> http://forum.hibernate.org/viewtopic.php?t=964211

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira