1 Reply Latest reply on Oct 3, 2005 12:12 PM by Emmanuel Bernard

    Questions about Best Practice for multiple PAR

    Per Norman Newbie

      I'm working on a new J2EE-application, and is evaluating EJB3 as persistence layer. I'm quite new to J2EE and have a question about what's best practice regarding my PAR.

      Now I've split my @Entity classes into two different PARs. One PAR contains entity A which depends on entity B in the second PAR (with a forreign key in the database). This works fine if I create both object A and B with new in the same Session in the same Stateless bean.

      If I instead uses a different Stateless bean to retreive object B, I can't persist entity A (which references entity B) without first calling manager.refresh(b).

      The following example will give me an Exception "detached entity passed to persist" on last line:

      B entityB = otherStatelessBean.findMyB();
      A entityA = new A(entityB);

      To solve this I can write:

      B entityB = otherStatelessBean.findMyB();
      A entityA = new A(entityB);

      The refresh() will create a second (unnecessary) SELECT to the database.

      I split my classes because I wanted a modularized application with one PAR and one EJB3 (with session beans) in every module. The above example made me unsure if this is a good idea.

      My questions:
      * What is best practice when modularizing EJB3-applications?
      * Should I split entities or keep them together in one big PAR?
      * Is there another solution or pattern to use to solve my problem above?

      Thanks in advance!