Hibernate ORM
  1. Hibernate ORM
  2. HHH-5855

The method "merge" from the EntityManager causes a duplicated "insert" of a child entity.

          Details

          • Type: Bug Bug
          • Status: Open
          • Priority: Critical Critical
          • Resolution: Unresolved
          • Affects Version/s: 3.3.1, 3.6.0
          • Fix Version/s: None
          • Component/s: annotations, core, entity-manager
          • Labels:
            None
          • Environment:
            Hibernate-Core 3.6.0, Hibernate EntityManager 3.6.0, Derby, Spring 3.0.5 (Configuration), JUnit 4.5
          • Bug Testcase Reminder (view):

            Bug reports should generally be accompanied by a test case!

          • Last commented by a user?:
            true

            Description

            The method "merge" from the EntityManager causes a duplicated "insert" of a child entity.
            Related to HHH-3332

            Observed Behavior:

            When you have an Entity A with a OneToMany(fetch=LAZY, cascade=ALL or (MERGE and PERSIST)) relationship to another Entity B and attempt to add an new B instance to A hibernate will generate a duplicate insert command upon commit.

            I have witnessed this problem occurring when this procedure is followed:

            1. retrieve an instance of A from persistence (A a = em.find(A.class, id))
            2. add a new B to A (a.getBs.add(new B()))
            3. merge instance 'a' back into persistence (em.merge(a))

            Upon transaction commit hibernate will generate an additional "insert" statement and insert the new instance of b into the data source twice.

            Expected behavior:

            I would expect hibernate to only perform a single insertion when 'a' is merged back to persistence and the B's are created through the cascade.

            Notes (Discoveries):

            This problem can be worked around by:

            • Changing the OneToMany relationship from a List to a Set
            • Only using either Cascade MERGE or PERSIST not both of ALL
            • Add a new B to A.getBs by adding to the '0' index (a.getBs.add(0,new B()))
            1. Zip Archive
              CascadeMergeTest.zip
              210 kB
              Seth Martin
            2. Zip Archive
              CascadeMergeTest.zip
              6 kB
              Seth Martin

              Activity

              Ascending order - Click to sort in descending order
              8 older comments
              Hide
              Permalink
              Mihalcea Vlad added a comment -

              Check my HHH-3332-again.zip attachement to see it still replicates on 4.x version too.

              Or:

              https://github.com/vladmihalcea/vladmihalcea.wordpress.com/blob/master/hibernate-facts/src/test/java/com/vladmihalcea/HibernateBagDuplicateTest.java

              Starting at line 75:

              //assertEquals(2, parent.getChildren().size());
              if (parent.getChildren().size() == 4)

              { LOG.error("Duplicates rows generated!"); }
              Show
              Mihalcea Vlad added a comment - Check my HHH-3332 -again.zip attachement to see it still replicates on 4.x version too. Or: https://github.com/vladmihalcea/vladmihalcea.wordpress.com/blob/master/hibernate-facts/src/test/java/com/vladmihalcea/HibernateBagDuplicateTest.java Starting at line 75: //assertEquals(2, parent.getChildren().size()); if (parent.getChildren().size() == 4) { LOG.error("Duplicates rows generated!"); }
              Hide
              Permalink
              Seth Martin added a comment -

              @Mihalcea Thank you for uploading the test case for hibernate 4.

              Show
              Seth Martin added a comment - @Mihalcea Thank you for uploading the test case for hibernate 4.
              Hide
              Permalink
              Seth Martin added a comment -

              Also, if the bug is still present in version 4 my test cases should still fail. i will dbl check them when i have some extra time.

              Show
              Seth Martin added a comment - Also, if the bug is still present in version 4 my test cases should still fail. i will dbl check them when i have some extra time.
              Hide
              Permalink
              Mihalcea Vlad added a comment -

              For the past 5 years this was my workaround:

              http://vladmihalcea.com/2013/10/16/hibernate-facts-favoring-sets-vs-bags/

              Show
              Mihalcea Vlad added a comment - For the past 5 years this was my workaround: http://vladmihalcea.com/2013/10/16/hibernate-facts-favoring-sets-vs-bags/
              Hide
              Permalink
              Rowland Mynhardt added a comment -

              Confirmed on Version 4.3.4.Final using HHH-3332 Again code attached

              Show
              Rowland Mynhardt added a comment - Confirmed on Version 4.3.4.Final using HHH-3332 Again code attached

                People

                • Votes:
                  14 Vote for this issue
                  Watchers:
                  16 Start watching this issue

                  Dates

                  • Created:
                    Updated:

                    Development