[
https://issues.apache.org/jira/browse/OPENJPA-286?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Pinaki Poddar resolved OPENJPA-286.
-----------------------------------
Resolution: Fixed
This style of modeling persistent states as a name-value pair does not notify
OpenJPA which field has been dirtied and hence the observed behavior.
Inform OpenJPA when a field has been dirtied and things should work.
The added lines in TestMargeHashEntity (around Line 79) are marked with + and
additional comments
em = emf.createEntityManager();
em.getTransaction().begin();
+ parent = em.merge(parent);
// merge the detached entity to the current context
+ OpenJPAPersistence.cast(em).dirty(parent, "children"); // mark
it dirty because OpenJPA did not understand that the name-value pair has been
updated
// now proceed as usual by merging it again
so that it can detect the new child added in detached state
parent = em.merge(parent);
em.getTransaction().commit();
em.close();
> Can't merge detached One-To-Many child instance. The entities were implement
> by an internal Hash container.
> -----------------------------------------------------------------------------------------------------------
>
> Key: OPENJPA-286
> URL: https://issues.apache.org/jira/browse/OPENJPA-286
> Project: OpenJPA
> Issue Type: Bug
> Components: jdbc
> Affects Versions: 0.9.7, 1.0.0, 1.1.0
> Environment: java version "1.5.0_10"
> Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_10-b03)
> Java HotSpot(TM) Client VM (build 1.5.0_10-b03, mixed mode, sharing)
> Reporter: Gene Wu
> Assignee: Pinaki Poddar
> Attachments: hashpojo.zip
>
>
> Maybe you know, one of the entity implementation is composite of a hash set,
> which stores key-value pairs.
> It's a common design of Entity Bean in EJB2.x ages. Does any one use it in
> OpenJPA? I am using it right now, and encounter an issue in entity cascade
> manipulate. Here is the issue description.
> Base.java : the super class of entities.
> protected final Object getAttributeValue(String attributeName) {
> return _values.get(attributeName);
> }
> protected final void setAttributeValue(String attributeName, Object
> value) {
> _values.put(attributeName, value);
> }
> // To hold the {attributeName, value} pair of the value object.
> private HashMap<String, Object> _values = new HashMap<String,
> Object>();
> C.java extends Base.java : Entity C has a set of Entity D.
> public void setDs(Set<D> ds) {
> setCollection("Ds", ds);
> }
> @OneToMany(mappedBy = "c", fetch = FetchType.LAZY, cascade =
> CascadeType.ALL)
> public Set<D> getDs() {
> return (Set<D>)getAttributeValue("Ds");
> }
> D.java extends Base.java as well.
> The relationship between C and D is one-to-many. I use following code to
> reproduce the issue.
> C c = em.find(C.class, 1);
> logger.debug(c.getDs().size());
> D d = new D();
> d.setC(c);
> c.getDs().add(d);
> em.merge(c);
> After commit the transaction, you will find the new D instance was not
> inserted into the database(there is no insert sql log as well). After you
> add, em.persist(c). That new instance will be added!
> Meanwhile, I wrote a couple of classes, which do not use a HashSet. I got the
> expected result after merge is done.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
