Re: [eclipselink-users] NullPointerException in sequence of persistent p

[Tom Ware <tom.ware@xxxxxxxxxx>]

Hi Laird,

  You're right, for persist we don't have the same limitations.  This is likely a different issue.

-Tom

On 06/09/2013 12:44 PM, Laird Nelson wrote:

On Fri, Sep 6, 2013 at 8:38 AM, Tom Ware <tom.ware@xxxxxxxxxx> wrote:

  The line from the spec is here:

"If X is an entity merged to X', with a reference to another entity Y, where cascade=MERGE
or cascade=ALL is not specified, then navigation of the same association from X' yields a
reference to a managed object Y' with the same persistent identity as Y."

Sure, on a merge().  What about on a persist()?  Or is it implicitly understood that persist() is just a special form of merge()?  Section 3.2.2 is quite different from section 3.2.7.1.

 

The spec asks us to create a Y' and we need a way to do it.  I suspect what other providers do comes down to a combination of how they have interpreted the spec, and what is tested in the TCK.

Having said that, I think there is an argument that there should be a way to indicate that an object should not be touched by a merge and perhaps the lack of CASCADING combined with a read-only setting would be a good way to determine that.  Please feel free to enter an enhancement request for this behavior.

Thanks; I will.

Best,

Laird

--
http://about.me/lairdnelson

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