I’m not sure if the title fits. I had trouble finding a good one.

This bug leads to multiple duplicates appear in the list of a OneToMany relation that was initialized manually in order to prevent N+1 query problem.

There are three entities, A, B and C.

A <11> B
B <1N> C

A has one B, B has multiple C. Loading is configured as lazy.

Given situation: I have 4 {{A}}which all point to the same B. This {{B}}has two C. So if I do {{a.getB().getCList()}}the size should be 2. This works out of the box with lazy loading.

However, if I add the following query, in order to load all {{C}}into the cache and a prevent N+1 query problem, then suddenly each {{B}}has eight entries of C.

List<A> resultNotNeeded = entityManager.createQuery("""
                select distinct a from A a
                join fetch a.b b
                left join fetch b.cList
                where a in (:aList)
                """, A.class)
                .setParameter("aList", aList)
                .setHint(org.hibernate.jpa.QueryHints.HINT_PASS_DISTINCT_THROUGH, "false")
                .getResultList();

It doesn’t make a difference if I remove the distinct and query hint. {{aList}}is just the list of the four a entities.

A test case is attached.