Hi!
I'd like to clear out a doubt.
I have 2 tables with manyToMany relation.

Tables are:
CREATE TABLE operations
(
id_operation serial NOT NULL,
...)


CREATE TABLE skill_types
(
id_skill_type serial NOT NULL,
...)


CREATE TABLE r_operations_skills
(
id_r_operation_skill serial NOT NULL,
id_operation int4 NOT NULL,
id_skill_type int4 NOT NULL)



I mapped the relation as following.

Operation.java





SkillTypes.java




It works as I expect: if I remove a skill used by operation I have an exception. Ok.

My doubt is: why when I remove a skill associated to an operation hibernate removes all entries in the relation tables and then does a lot of inserts?

for example r_operations_skills(id_operation, id_skill_type) contains:
(6,1)
(6,2)
(6,3)

when I remove skill #3 from operation 6 it does:

delete from r_operations_skills where id_operation=6
insert into r_operations_skills (id_operation, id_skill_type) values (6, 1)
insert into r_operations_skills (id_operation, id_skill_type) values (6, 2)

Is it a wrong mapping I made or is this a correct behaviour ?

Thank you
Nicola

There is a useful discussion on the consequences of how you map collections when it comes to updating their contents (eg by removing a skill associated with an operation):

http://www.hibernate.org/hib_docs/v3/re ... ollections

Looking at your schema, your collection might be following 'bag' semantics, ie a given skill could be associated with a given operation more than once (whether or not you do so, as far as hibernate can tell from your mapping, it's possible). From the documentation:



So the behavior is correct if you really want bag semantics. To make it more efficient, try either adding an index column to your join table so hibernate can treat it as an actual list (as declared in your java), or explicitly declare the collection as a Set, so hibernate knows each element is unique. Either way, it won't have to delete and reinsert the whole collection when you change one element, like it's doing now.