I am trying to figure out if there is a way to prevent Hibernate from adding in left outer joins of child tables when querying for entities with ManyToMany relationship with said children. So, I have a parent table A with a ManyToMany relationship with table B. I want to find all A'a so I...

hibernateSession.createCriteria(A.class).list();

Note that I am not including any restrictions involving B or any other relationship and that I have the ManyToMany annotation parameterized with fetch = FetchType.LAZY.

I am aware of the duplicate parent entity confusion that is covered in the FAQ, and understand why I would get duplicate A records for every B record if the Criteria would employ a left outer join on B because of a restriction or projection involving B. But, I do not understand why that left out join would be used in a straight select all. Also of interest, this is not new code and was working prior to us migrating from Hibernate 3.X to Hibernate 4.1. Can anyone point me to a resource (doc, forum post, blog post, etc.) that explains why Hibernate now behaves this way? Are there ways to change this behavior?

There's a way to suppress the automatic join by specifying the fetchmode:

criteria.setFetchMode(relationAttributeName ,org.hibernate.FetchMode.SELECT);