Re: [ojb-users] Problems with M-N mappings (newbie)

[Jakob B. <jbr...@ho...>]

hi martin,

2)
ojb 0.9 supports bidirectional m:n mappings only.
for unidirectional m:n you have to use the latest from repository.
there's also a testcase for unidirectional m:n between person and project.

hth jakob


----- Original Message -----
From: "Martin Fuzzey" <fu...@be...>
To: <obj...@li...>
Sent: Wednesday, June 12, 2002 6:20 PM
Subject: [ojb-users] Problems with M-N mappings (newbie)


> Hi all,
>
> I've just started using objectbridge (the 0.9 version) and am having a few
> problems with M-N mappings :
>
> 1) Unable to perform non-decomposed N:M mapping
> ======================================
>
> I have two classes StringFDP and AndFDP.
> StringFDP just contains a string field, wheras AndFDP contains a
Collection
> (of StringFDPs) (field name m_fdps)
> I also added a Collection field (named _containers) to StringFDP (although
> I don't really want this - see below)
>
> Following the test cases I did this :
>
>     <class-descriptor
>       class="prestudy.persist.bo.StringFdp"
>       table="STRING_FDP"
>     >
>        <field-descriptor id="1"
>           name="_uid"
>           column="objid"
>           jdbc-type="INTEGER"
>           primarykey="true"
>           autoincrement="true"
>        />
>        <field-descriptor id="2"
>           name="m_val"
>           column="val"
>           jdbc-type="VARCHAR"
>        />
>        <collection-descriptor
>           name="_containers"
>           element-class-ref="prestudy.persist.bo.AndFdp"
>           indirection-table="AND_FDP_ASSOC"
>        >
>           <fk-pointing-to-this-class column="FDP_ID"/>
>           <fk-pointing-to-element-class column="AND_ID"/>
>        </collection-descriptor>
>     </class-descriptor>
>
>
>     <class-descriptor
>       class="prestudy.persist.bo.AndFdp"
>       table="AND_FDP"
>     >
>        <field-descriptor id="1"
>           name="_uid"
>           column="objid"
>           jdbc-type="INTEGER"
>           primarykey="true"
>           autoincrement="true"
>        />
>        <collection-descriptor
>           name="m_fdps"
>           element-class-ref="prestudy.persist.bo.StringFdp"
>           indirection-table="AND_FDP_ASSOC"
>        >
>           <fk-pointing-to-this-class column="AND_ID"/>
>           <fk-pointing-to-element-class column="FDP_ID"/>
>        </collection-descriptor>
>
>     </class-descriptor>
>
> This gave me a NPE during a lock operation (I had to modify
TransactionImpl
> to print the cause of the exception rather than just the
> LockNotGrantedException) :
>
> java.lang.NullPointerException
>          at
> ojb.broker.metadata.ObjectReferenceDescriptor.getForeignKeyFieldDescr
> iptors(ObjectReferenceDescriptor.java:250)
>          at ojb.odmg.TransactionImpl.lockCollections(Unknown Source)
>          at ojb.odmg.TransactionImpl.register(Unknown Source)
>          at ojb.odmg.TransactionImpl.lock(Unknown Source)
>          at prestudy.persist.ojb.OjbFdpStorer.save(OjbFdpStorer.java:72)
>          at prestudy.persist.ojb.OjbFdpStorer.test(OjbFdpStorer.java:119)
>          at prestudy.persist.ojb.OjbFdpStorer.main(OjbFdpStorer.java:94)
> org.odmg.LockNotGrantedException
>          at ojb.odmg.TransactionImpl.lock(Unknown Source)
>          at prestudy.persist.ojb.OjbFdpStorer.save(OjbFdpStorer.java:72)
>          at prestudy.persist.ojb.OjbFdpStorer.test(OjbFdpStorer.java:119)
>          at prestudy.persist.ojb.OjbFdpStorer.main(OjbFdpStorer.java:94)
>
> It seems OJB wants a foreign key definition. But the unit test case the
> non-decomposed mapping PEOPLE, PROJECTS using indirection_table works
fine.
> The only difference I can see is that in the unit test there is _also_ a
> decomposed mapping between the same two classes (via ROLE).
> In tutorial 3 a seemingly dummy reverse_fk_descriptor is used (999) but
> when I add this I get an ArrayIndexOutOfBoundsException.
>
>
> 2) How do I do a unidirectional N:M mapping?
> =================================
> In fact my object model only needs navigability from AndFdp to StringFdp
> and not the inverse. Why do I have to add an unneeded Collection to my
> StringFdp class.
>
> 3) What I'm really trying to do.
> ======================
> In fact I'm really trying to map a composite pattern structure of objects
> where each object is either a simple object (such as StringFdp) or a
> composite (such as AndFdp). The idea being to have several composites
> representing logical operations (Or, Nand etc).
>
> Is it possible to map this type of model?
>
>
> Help much appreciated,
>
> Martin Fuzzey
>
>
> _______________________________________________________________
>
> Sponsored by:
> ThinkGeek at http://www.ThinkGeek.com/
> _______________________________________________
> Objectbridge-users mailing list
> Obj...@li...
> https://lists.sourceforge.net/lists/listinfo/objectbridge-users
>