Hibernate Community • View topic - one-to-one -missing update in one case- misunderstanding?

View unanswered posts | View active topics

Board index » Hibernate & Java Persistence » Hibernate Users

All times are UTC - 5 hours [ DST ]

one-to-one -missing update in one case- misunderstanding?

Page 1 of 1

[ 7 posts ]

Previous topic | Next topic

Author

Message

curio

Post subject: one-to-one -missing update in one case- misunderstanding?

Posted: Wed Nov 03, 2004 10:40 am

Expert

Joined: Tue Oct 05, 2004 9:45 am
Posts: 263

Hibernate version: 2.1.6

Hi,

i know, that i should show my mapping-documents etc. but i first want to ask if i have a mistake in thinking ... could be ;)

I've got a one-to-one-relationship (cascading=all;outer-join=true) between ClassA and ClassB (using primary-key - composite-ids on both side and version-columns).
Everything works fine ... insert, update and delete works perfectly.

Only one thing i wasn't able to get running ...

1) Load ClassA from Database (incl. ClassB)
2) Set reference to ClassB to 'null' (classA.setClassB(null))
3) saveOrUpdate(classA)

When i set classA.setClassB(null) the entry for ClassB in Database isn't deleted ... it doesn't matter if it's the same session or a new one ...

First of all it would be a great help to know if this is possible with hibernate? Or should i use a "many-to-one"-relationship in those cases?

thx!
curio

Top

ernst_pluess

Post subject:

Posted: Wed Nov 03, 2004 10:58 am

Expert

Joined: Thu Jan 29, 2004 2:31 am
Posts: 362
Location: Switzerland, Bern

try

Code:

cascade=all-delete-orphan

if you have a bidirectionl association you have to add

Code:

classB.setClassA(null)

HTH
Ernst

Top

curio

Post subject:

Posted: Wed Nov 03, 2004 11:33 am

Expert

Joined: Tue Oct 05, 2004 9:45 am
Posts: 263

hi ernst,

thanx for the hints ...

Code:

cascade=all-delete-orphan

is not allowed by dtd, i've tried this.
I'm using: http://hibernate.sourceforge.net/hibern ... ng-2.0.dtd

in both cases (bidirectional) association or not, there's no update to ClassA and no delete to ClassB.

The strange thing everything works fine ... only deleting the entry for ClassB from DB doesn't work ...

But i think that it must be possible ... otherwise "outer-join=true" wouldn't make sense ... well, perhaps i'll have to build a small example and add it here ... perhaps i really have some small things wrong (but what if everything else works? ) :)

thx!
curio

Top

ernst_pluess

Post subject:

Posted: Wed Nov 03, 2004 3:41 pm

Expert

Joined: Thu Jan 29, 2004 2:31 am
Posts: 362
Location: Switzerland, Bern

Hi curio

I'm sorry for posting rubish. I should have checked the DTD first myself :-(.

From the DTD I guess that a parent/child relation with all-delete-orphan is supported only for Collections.
Can one of the experts confirm this?

As a work around you can
- use a list/bag/set mapping
- or check in the onFlushDirty method of an Interceptor whether there was a B and delete it.

HTH
Ernst

Top

curio

Post subject:

Posted: Wed Nov 03, 2004 6:02 pm

Expert

Joined: Tue Oct 05, 2004 9:45 am
Posts: 263

Hi ernst,

no problem! i've thought that "all-delete-orphan" might work, too, because i've read a comment from gavin somewhere in this forum that this might be possible ...

hmm ... using an interceptor is a possibility, too ... yes ... never thought about this ... but sure ... i must try this ...

using a list was one of the possibilities i've thought about, too ... but at the moment i don't like it, because it's then possible to add more than only one instance of classB .... ok. i could hide the list and only support setting one element ...

i'll try both possibilites tomorrow when i'm back at work ...

thx for your help!
curio

P.S.: But if it's not possible to set a one-to-one-association to "null", i misinterpretated this kind of association the whole time ... i think i will debug a little bit tomorrow :)

Top

ernst_pluess

Post subject:

Posted: Thu Nov 04, 2004 3:56 am

Expert

Joined: Thu Jan 29, 2004 2:31 am
Posts: 362
Location: Switzerland, Bern

Quote:

P.S.: But if it's not possible to set a one-to-one-association to "null", i misinterpretated this kind of association the whole time ...

I'd expect that the foreign key will be set to NULL.

Hapy debugging
Ernst

Top

curio

Post subject: possible bug?

Posted: Fri Nov 05, 2004 9:34 am

Expert

Joined: Tue Oct 05, 2004 9:45 am
Posts: 263

i think i've got it right ... and found a bug ... but i can be wrong ...

Mapping files:

Code:

<hibernate-mapping>
   <class
    name="ClassA"
    table="testClassA">
    
    <id name="idClassA"
        column="id"
        type="integer"
        unsaved-value="0">
      <generator class="increment"/>
    </id>
    
    <property 
       name="dummy"
       column="dummy"
       type="string"/>
    
    <one-to-one
      name="classBRef"
      class="ClassB"
      outer-join="true"
      cascade="all"/>
  </class>
</hibernate-mapping>

<hibernate-mapping>
   <class
    name="ClassB"
    table="testClassB">
    
    <id name="idClassB"
        column="id"
        type="integer"
        unsaved-value="0">
      <generator class="foreign">
         <param name="property">classARef</param>
      </generator>
    </id>
    
    <property 
       name="dummy"
       column="dummy"
       type="string"/>
    
    <one-to-one
      name="classARef"
      class="ClassA"
      outer-join="true"
      constrained="true"/>
  </class>
</hibernate-mapping>

Two Classes with a one-to-one-Relationship ...
The code i use (quick and dirty):

Code:

    Session s = super.createSession();
    Transaction t = s.beginTransaction();
    ClassA classA = (ClassA)s.load(ClassA.class, new Integer(1));
    boolean aHasB = classA.getClassBRef() != null;
    boolean bHasA = classA.getClassBRef().getClassARef() != null;
    System.err.println("--------[classA has classB?]--->"+aHasB);
    System.err.println("--------[classB has classA?]--->"+bHasA);

    // remove classB-Ref
    classA.getClassBRef().setClassARef(null);
    classA.setClassBRef(null);
    
    s.saveOrUpdate(classA);
    t.commit();
    
    // reload classA from first-level-cache!
    classA = (ClassA)s.load(ClassA.class, new Integer(1));

    aHasB = classA.getClassBRef() != null;
    if (aHasB)
      bHasA = classA.getClassBRef().getClassARef() != null;
    else
      bHasA = false;
    System.err.println("--------[classA has classB?]--->"+aHasB);
    System.err.println("--------[classB has classA?]--->"+bHasA);
        
    s.close();

And the log:

Code:

[main] DEBUG net.sf.hibernate.impl.SessionImpl  - initializing non-lazy collections
--------[classA has classB?]--->true
--------[classB has classA?]--->true
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - saveOrUpdate() persistent instance
[main] DEBUG net.sf.hibernate.transaction.JDBCTransaction  - commit
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - flushing session
[main] DEBUG net.sf.hibernate.engine.Cascades  - processing cascades for: ClassA
[main] DEBUG net.sf.hibernate.engine.Cascades  - done processing cascades for: ClassA
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - Flushing entities and processing referenced collections
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - Processing unreferenced collections
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - Scheduling collection removes/(re)creates/updates
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - Flushed: 0 insertions, 0 updates, 0 deletions to 2 objects
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - Flushed: 0 (re)creations, 0 updates, 0 removals to 0 collections
[main] DEBUG net.sf.hibernate.impl.Printer  - listing entities:
[main] DEBUG net.sf.hibernate.impl.Printer  - ClassB{classARef=null, dummy=testDummyB, idClassB=1}
[main] DEBUG net.sf.hibernate.impl.Printer  - ClassA{idClassA=1, dummy=testDummyA, classBRef=null}
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - executing flush
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - post flush
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - transaction completion
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - loading [ClassA#1]
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - attempting to resolve [ClassA#1]
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - resolved object in session cache [ClassA#1]
--------[classA has classB?]--->false
--------[classB has classA?]--->false
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - closing session
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - disconnecting session
[main] DEBUG net.sf.hibernate.connection.DriverManagerConnectionProvider  - returning connection to pool, pool size: 1
[main] DEBUG net.sf.hibernate.impl.SessionImpl  - transaction completion

Nothing will be deleted ...

BUT ... like you can se in the log, the reference is deleted at object side, and even in the first-level-cache of the session ...

This can be a bug ... or? What are the hibernate-experts saying?
I'm using the current cvs_snapshot ...

thx!
curio

Top

Page 1 of 1

[ 7 posts ]

Board index » Hibernate & Java Persistence » Hibernate Users

All times are UTC - 5 hours [ DST ]

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum