Composite id with manyToOne references (from a legacy db)

ge0ffrey · **Joined:** Thu Oct 13, 2005 4:19 am **Posts:** 98

I've tried and errored all the propositions in the hibernate and hibernate-annotations manual, but I can't get this to work.
(Might I humble point out that http://www.hibernate.org/hib_docs/annot ... identifier could use more examples on mapping composite keys?)

OrderLine has a composite key of OrderLine and Product.

Code:

@Entity
???
public class OrderLine {
    ...

    ???
    @ManyToOne()
    @NotNull
    @JoinColumn(name = "order", nullable = false)
    public Order getOrder() {...}

    ???
    @ManyToOne()
    @NotNull
    @JoinColumn(name = "product", nullable = false)
    public Product getProduct() {...}

    ...
}

What do I place at the ???
I 've also made a OrderLineId class to use @IdClass, but that didn't work either.

All the examples I 've found don't use associations as part of their composite keys. I hope that isn't illegal?

emmanuel · **Posted:** Mon Nov 13, 2006 6:54 pm

I'm afraid I never tested your case. It should technically work.
Open a JIRA issue with a running test case. But this is probably not my priority so a complete patch might be your way out :-)

ge0ffrey · **Joined:** Thu Oct 13, 2005 4:19 am **Posts:** 98

What replacements of ??? should technically work?

I've tried a great number of combinations (some logical, some ridiculous) of @Id, @IdClass, @Embeddable, @EmbeddableId with @ManyToOne, @JoinColumn.

ge0ffrey · **Joined:** Thu Oct 13, 2005 4:19 am **Posts:** 98

It took me a good while to get it all setup, but here's a testcase patch :)

http://opensource.atlassian.com/project ... se/ANN-492

The error is the same as my app gets in legal configs using IdClass: "repeated column"

emmanuel · **Posted:** Tue Nov 14, 2006 11:35 am

When I mean technically, I mean it should work or this is a bug

chuisman9955 · **Joined:** Thu Dec 02, 2004 2:13 pm **Posts:** 10

I'm experiencing the same problem, is there any resolution to this?

emmanuel · **Posted:** Sun Dec 17, 2006 6:35 pm

Go and watch (the function) the JIRA issue
http://opensource.atlassian.com/project ... se/ANN-492

sguillory · **Joined:** Fri Aug 06, 2004 5:16 pm **Posts:** 7

I had the same problem, and finally realized it was a bug in Hibernate's implementation of JPA. So after trying every single combination over a couple of days like Geoffrey did, I finally just switched to Toplink's implementation, and got past the problem in 15 minutes.

It's a shame this is not a higher priority bug since pretty much any legacy database is going to utilize composite natural keys with associations on parts of the key. The inability to model this makes Hibernate's implementation basically unusable for legacy databases.

emmanuel · **Posted:** Mon Jan 08, 2007 1:25 pm

sguillory wrote:

The inability to model this makes Hibernate's implementation basically unusable for legacy databases.

Hibernate is the most flexible JPA implementation regarding legacy and exotic mappings. Do not let a single bug hidding the global feature set.

@IdClass is barely used contrary to @EmbeddedId when dealing with composite it.

sguillory · **Joined:** Fri Aug 06, 2004 5:16 pm **Posts:** 7

Hibernate's implementation of JPA may be the most flexible for legacy and exotic mappings, however, currently it is incapable of modeling composite keys with associations on parts of the key. It doesn't matter if you use the @IdClass annotation style or the @EmbeddedId annotation style.

So I currently can't use hibernate's implementation. However, I am watching this issue and will try it again when it is fixed. I am just a bit dissappointed to hear that this bug is not a high priority since it renders Hibernate Annotations useless for legacy databases.

Stan

emmanuel · **Posted:** Wed Jan 10, 2007 10:04 am

sguillory wrote:

however, currently it is incapable of modeling composite keys with associations on parts of the key. It doesn't matter if you use the @IdClass annotation style or the @EmbeddedId annotation style.

You are wrong. There is a pretty comprehensive support released already http://fisheye.labs.jboss.com/browse/Hibernate/branches/Branch_3_2/HibernateExt/metadata/src/test/org/hibernate/test/annotations/idmanytoone

A patch was probided, and marked to be fixed for 3.2.2 (ie the next version), I can't see how it is not a high priority.

sguillory · **Joined:** Fri Aug 06, 2004 5:16 pm **Posts:** 7

I guess there is some confusion on my part. There appear to be two bugs opened on this issue: 492, and 499. 492 is still marked open while 499 is marked closed.

As for my assumption that the bug was not a high priority, I refer to your quote from November 13, 2006:

"But this is probably not my priority so a complete patch might be your way out :-)"

I think even you will admit that your english isn't the best, so I took this to mean ". . . not *high* priority . . .", since the sentence didn't make any sense as written.

At any rate, I'm glad to hear the issue has been resolved. I probably will stick with TopLink for now sense it seems to be a more robust implementation at this point, but I will definitely review the 3.2.2 release when it becomes available.

Thanks,
Stan

sguillory · **Joined:** Fri Aug 06, 2004 5:16 pm **Posts:** 7

I guess there is some confusion on my part. There appear to be two bugs opened on this issue: 492, and 499. 492 is still marked open while 499 is marked closed.

As for my assumption that the bug was not a high priority, I refer to your quote from November 13, 2006:

"But this is probably not my priority so a complete patch might be your way out :-)"

I think even you will admit that your english isn't the best, so I took this to mean ". . . not *high* priority . . .", since the sentence didn't make any sense as written.

At any rate, I'm glad to hear the issue has been resolved. I probably will stick with TopLink for now sense it seems to be a more robust implementation at this point, but I will definitely review the 3.2.2 release when it becomes available.

Thanks,
Stan