Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [eclipselink-users] Pre-existing managed entity instances in the persistence context: pre-existing on disk? Or...?
Merge searches persistence context for entity with the same id as the detached being merged to put merge result into it (say, it has been read before through the same entity manager and em.clear hasn't been called). 
If none found then merged result placed into a new entity.

On 7/25/2013 1:19 PM, Laird Nelson wrote:

I had a specification-related question and was curious how EclipseLink
chose to interpret it.

Suppose I have a table on disk to which I've mapped an entity class.
  Let's just say there's a row in the table with a primary key of 6.
  Simple straightforward INTEGER primary key mapped to a simple
straightforward int field.

Now suppose I do this:

    final SomeEntity x = new SomeEntity();
    x.setID(6); // set the @Id -- the persistent identity

    //

    // Assumption: x is now "detached"; it is not "new"

    // Please correct this assumption if it is false.

    //

    em.clear(); // empty persistence context

    // 1

    final SomeEntity xPrime = em.merge(x); // 2

    em.flush(); // 3 -- Duplicate primary key error? If not, why not?


At point (1) above, there are no instances of anything in the
persistence context.  It is entirely empty.

At point (2), the merge operation must EITHER locate a "pre-existing
managed entity instance X' of the same identity [6]" OR create "a new
managed copy X' of X".  Can the provider choose which strategy to use
here?  Won't this have an effect on whether an INSERT or an UPDATE
statement is ultimately issued at point (3)?  Or is that irrelevant?

At point (3), will I get a primary key duplication error?  Is it your
understanding that merge() is required by the specification to do an
UPSERT?  If so, where is that requirement mandated?  Does "pre-existing
managed entity instance" actually mean "pre-existing managed entity
instance loaded just-in-time into the persistence context behind the
scenes by the provider if there's a corresponding row in the database"?

The relevant section in the specification is 3.2.7.1 <http://3.2.7.1>:

    "The semantics of the merge operation applied to an entity X are as
    follows:
    "If X is a detached entity [in my example it is], the state of X is
    copied onto a *pre-existing managed entity instance* X' of the same
    identity [bear in mind in my example I just cleared the persistence
    context so to my knowledge there are no instances at all of anything
    in there] or a new managed copy X' of X is created [the 'or' part
    makes me think the provider could just create a new one in all cases]."


Best,
Laird

--
http://about.me/lairdnelson


_______________________________________________
eclipselink-users mailing list
eclipselink-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipselink-users

Back to the top