Hibernate version: 3.1.3

Hi,

we have two tables in our database, modeling habitats and districts. There is a many-to-one relationship from habitats to districts. We have a fixed set of districts, that is the district table is read-only as far as our system is concerned.

Now the problem is that the "foreign key" in the habitats table isn't a true foreign key - the relationship is of the form

foreign-key-value-in-habitats = primary-key-value-in-districts - 8000

That is, when a habitat row has a district number 215, the primary key of the actual district in the district table is 8215.

In the object model, we have a District class and a Habitat class, and I like a Habitat object to have a reference to a District object.

I tried to solve this by using a formula in the mapping (stripped down to the relevant parts):

(short lesson in German: Biotop == Habitat, Kreis == district ;) )



Again, remember that the Kreis table is read only - we never need to write into it. We *need* to write the KREIS_NR column in the Biotop, though.

I think *logically* it should be possible to do this, but it seems that Hibernate doesn't support formulas in properties that are referenced from many-to-one associations - at runtime, I get ClassCastException posted below.

Am I missing something? Is this just not (yet?) supported by Hibernate? Is it a bug?

Any hint would be highly appreciated.

java.lang.ClassCastException: org.hibernate.mapping.Formula
at org.hibernate.mapping.ManyToOne.createPropertyRefConstraints(ManyToOne.java:62)
at org.hibernate.cfg.HbmBinder$ManyToOneSecondPass.doSecondPass(HbmBinder.java:2673)
at org.hibernate.cfg.Configuration.secondPassCompile(Configuration.java:1012)
at org.hibernate.cfg.Configuration.buildSessionFactory(Configuration.java:1168)[/code]

I wonder the business use case behind this. I would use a simple HQL query for it.

Consider this example.
It works fine for me in MySQL and Hibernate 3.1

Class B analogous to your habitats and Class C analogous to your districts.




See the Class files.

Class C

private java.lang.Long c; //with setter and getter


Class B

private java.lang.Long b; //with setter and getter

private C refC; //with setter and getter
[/code]

It's really just an inconsistent database design. The district table could hold districts for the whole country. The habitat table only holds habitats for one federal state, for which the district number always begins with an eight, so they omitted that digit in the habitats table.



The example below isn't using a HQL query, is it?



It never occured to me that I could mix the column and formula attributes. I tried mixing column and formula *tags*, but that seemed to imply the semantics of a composite id.

What are the exact semantics of mixing column and formula attributes? And where can I learn more about it? Neither the online documentation, nor the books I use seem to say anything about these cases.

Nope.



I use eclipse to write my XML with the DTD in place. So I use the aut-complete feature to list me the options that are available. But that doesn' t make you learn exactly. There are lots of online examples which have typical business cases and how they are tackled using HBM.

Moreover using these tags and attributes becomes easier the more you work with them. You would probably know the exact tags and attributes to use as and when you are familiarised with them.
Thinking retrospectively, I was competely dumb on using these. Later as and when I started using them, you learn how they are designed and sometimes you can even use certain tags to cheat the HBM to map to an underlying E-R in database which is not in sync with the O-R in java world (That again drives the inconsistent E-R design we have, but sometimes changing something which is existing might impact business and involve lots of time to change the existing model and schema).

Fine end of story. Take away, three things,

1. Try learning lots of examples.
2. Use tags and attributes and play around with it.
3. Try Using an XML document builder.

Nope.



I use eclipse to write my XML with the DTD in place. So I use the aut-complete feature to list me the options that are available. But that doesn' t make you learn exactly. There are lots of online examples which have typical business cases and how they are tackled using HBM.

Moreover using these tags and attributes becomes easier the more you work with them. You would probably know the exact tags and attributes to use as and when you are familiarised with them.
Thinking retrospectively, I was competely dumb on using these. Later as and when I started using them, you learn how they are designed and sometimes you can even use certain tags to cheat the HBM to map to an underlying E-R in database which is not in sync with the O-R in java world (That again drives the inconsistent E-R design we have, but sometimes changing something which is existing might impact business and involve lots of time to change the existing model and schema).

Fine end of story. Take away, three things,

1. Try learning lots of examples.
2. Use tags and attributes and play around with it.
3. Try Using an XML document builder.