Hi,

I am thinking of technique similar to this http://opensource.atlassian.com/projects/hibernate/browse/HHH-552 but applied to collections. I would like to define a association in hbm but I don't need it at the POJO level. The only purpose of this association (collection) is to define cascade="delete" which will delete all associated objects when its parent is deleted.

There is no update="false" attribute available on collection. I can try to define inverse="true" on the collection together with noop accessor but i am not sure if this is valid solution or rather kind of hack. If one could give me some hint I would be grateful.

This is the view of the situation:

I believe u have dummy collection for cascade delete operation.
Instead of using dummy collection maintain the list of objects getting removed before saving main POJO object. And iterate through removed object in DAO save method and remove associated entity manually.

something like

public void saveOrUpdate(<Object> r) throws CuicHibernateException
{
Transaction transaction = getTransaction();

try
{
getSessionFactory().getCurrentSession().saveOrUpdate(r);

//Delete Fields associated to deleted fields
List<String> removedFields = r.getRemovedFields();

if(removedFields.size() > 0) {
StringBuffer buff = new StringBuffer(32);
for(String str : removedFields)
buff.append(str + " ");

buff.deleteCharAt(buff.length() - 1); //delete last space

/* To delete Field I need to walk association to Filter, and remove it from there,
* otherwise the Field remains in object graph */
String className = Field.class.getName();
Query q = getSessionFactory().getCurrentSession().createQuery("from " + className + " obj where obj.fieldId in (?)");
q.setString(0, buff.toString());

for(Object f1 : q.list() )
{
f1.getParent().removeField(f);
}