Setting orphanRemoval to true while migrating chil

2019-01-19 19:24发布

问题:

Important Notice : If you are reading this post, then consider looking into this post too for in-depth discussions.


It is a quite usual practice/situation/requirement where children of a parent may be migrated to another parent. What happens, if orphanRemoval is set to true on the inverse side of such relationships?

Consider as an example, any simple one-to-many relationship as follows.

Inverse side (Department) :

@OneToMany(mappedBy = "department", fetch = FetchType.LAZY, cascade = CascadeType.ALL, orphanRemoval = true)
private List<Employee> employeeList = new ArrayList<Employee>(0);

Owning side (Employee) :

@JoinColumn(name = "department_id", referencedColumnName = "department_id")
@ManyToOne(fetch = FetchType.LAZY, cascade = {CascadeType.PERSIST, CascadeType.MERGE, CascadeType.REFRESH, CascadeType.DETACH})
private Department department;

While merging an operation/action like the following (where department is a detached entity supplied by a client),

Employee employee = entityManager.find(Employee.class, 1L);
Department newDepartment = entityManager.contains(department) ? department : entityManager.merge(department);

if (!newDepartment.equals(employee.getDepartment())) {
    employee.getDepartment().getEmployeeList().remove(employee);
    // Since orphanRemoval is set to true, 
    // this should cause a row from the database table to be removed inadvertently
    // by issuing an addition DELETE DML statement.
}

employee.setDepartment(newDepartment);
employee.setEmployeeName("xyz");        

List<Employee> employeeList = newDepartment.getEmployeeList();

if (!employeeList.contains(employee)) {
    employeeList.add(employee);
}

entityManager.merge(employee);

Of course, adding and removing an employee may better be done/handled using defensive link (relationship) management methods in the associated entities.

A department instance is supplied by a client. It is a detached entity. It can be the same or a different department depending upon an administrative action performed by the client in question. As a result, if the department instance supplied by a client is different from the one held by the current Employee, then it should be removed first from the list of employees (employeeList) held by the current old department on the inverse side associated with that Employee before adding it to the list of employees held by the new department supplied.

As a guess, the Employee row should be removed inadvertently from the database while removing the Employee instance from the list of employees currently being referred to by the employee's department - old department (before this operation has been triggered) i.e. while migrating a child from its parent to another parent, the child needs to be removed from its native parent before it is adopted by another parent and that child row is supposed to be removed inadvertently from the database (orphanRemoval = true).

The employee row in the database table, however, remains intact with the updated column values. No DML statements except an UPDATE statement are generated.

Can I consider, migrating children from their parent to another parent in this way, does not inadvertently remove those children from the database table as they should not be?

Currently using EclipseLink 2.6.0 having JPA 2.1.


EDIT:

If an Employee entity is only deleted (thus, not added to the list after it was deleted - not migrated to another parent but just deleted) from the list on the inverse side, then its corresponding row is deleted from the database too as usual (orphanRemoval = true) but the row is simply updated, when an Employee entity (child) is added to the list of another parent after it was deleted from the list of its native parent (migration of the entity).

The provider appears to be smart enough to detect, the migration of children from their parent to another parent, as an update.

The behaviour can be seen identical on both Hibernate (4.3.6 final) and EclipseLink (2.6.0) but it cannot be relied upon, if it is a provider specific behaviour (not portable). I cannot find anything about this behaviour in the JPA spec.

回答1:

This is documented in the JPA specification.

Section 3.2.4 (excerpt):

The semantics of the flush operation, applied to an entity X are as follows:

  • If X is a managed entity, it is synchronized to the database.
    • For all entities Y referenced by a relationship from X, if the relationship to Y has been annotated with the cascade element value cascade=PERSIST or cascade=ALL, the persist operation is applied to Y

Section 3.2.2 (excerpt):

The semantics of the persist operation, applied to an entity X are as follows:

  • If X is a removed entity, it becomes managed.

orphanRemoval JPA javadoc:

(Optional) Whether to apply the remove operation to entities that have been removed from the relationship and to cascade the remove operation to those entities.

orphanRemoval Hibernate docs:

If an entity is removed from a @OneToMany collection or an associated entity is dereferenced from a @OneToOne association, this associated entity can be marked for deletion if orphanRemoval is set to true.

So, you remove the employee E from the department D1 and add her to the department D2.

Hibernate then synchronizes the department D1 with the database, sees that E is not in the list of employees and marks E for deletion. Then it synchronizes D2 with the database and cascades PERSIST operation to the list of employees (section 3.2.4). Since E is now in this list, the cascade applies on it and Hibernate un-schedules the delete operation (section 3.2.2).

You may want to look at this question as well.

"What happens, if orphanRemoval is set to true on the inverse side of such relationships?"

You have already set it on the inverse side (the inverse side is the one which declares mappedBy). If you mean what if it were set on the other side (@ManyToOne in this case), then it would not make sense and that's why there is no such an attribute in @ManyToOne and @ManyToMany.