design decision: variation of session fassade

[nimo frey]

Actually, I have all my jpa-queries within a Stateless Session Bean.

Look at that:

// my persistence-bean (version 1)
@Entity
@NamedQuery (
    name = "getUsers",
    query = "SELECT u FROM Users u"
)
public Users {

...

}

// my session bean (version 1)
@Stateless
public void MySessionBean{

@PersistenceContext 
EntityManager em;

public List<Users> getUsers{

return em.createNamedQuery("getUsers").getResultList();

}
}

Now, I am wondering what will be the drawbacks,
if I use a static method within the PersistenceBean returning the related Query-Instance to my session bean:

// my persistence-bean (version 2)

import javax.persistence.Query;
...
@Entity
@NamedQuery (
    name = "getUsers",
    query = "SELECT u FROM Users c"
)
public Users {

...

// I use this static method to return a query instance of all users
// Note that I use EntityManager as a parameter
public static Query getUsers(EntityManager em){

 em.createNamedQuery("getUsers")
}
}

}

// my session bean  (version 2)
@Stateless
public void MySessionBean{

@PersistenceContext 
EntityManager em;

public List<Users> getUsers{

// I invoke the static method with my actual entity-manager-instance to get all users
return Users.getUsers(em).getResultList();

}}

My Questions:

1. What version is better, version 1 or version 2 ? Why?

2. Are there any pitfalls when using entityManager-Instance within a static method?

3. Are there any lacks of performance when using a static method to retrieve a query-instance instead of retrieving it via a typical (non-static) ejb-method (for example, because of pooling or transaction, ..)?

4. In generally, are there any pittfalls, drawbacks, performance losses when using or invoking static methods within a ejb-bean ?

In my opinion, version 2 is better because of maintenance: I have all my query-related things of a object within the object itself. But I am really in doubt about my opinion.