Re: [eclipselink-users] lazy loading overrides attached object

[Michael Hönnig <michael@xxxxxxxxxx>]

Hi Christopher,

Thank you very much for this input, even though I don't like this 
tradeoff that Cascade.REFRESH even means if the refresh is implicit as 
in this case.

One more information: With EclipseLink 2.3.2 the B instance was not 
refreshed implicitely.  Just after we upgraded to 2.4.2 this problem did 
come up.

Best regards
... Michael

> Check that you have not done a refresh on B and that the mappings
> involved are not marked to cascade the refresh.  If B was refreshed 
> by
> any means, it will cascade over to the A when the relationship is
> triggered, which will in turn cascade the refresh over the
> relationship of Bs if it has also been set, wiping out changes.
>
> If this is occurring, do not use cascade refresh or cascade all
> settings on the affected mappings, especially lazy mappings.
>
> On 2013-06-13, at 5:28 PM, Michael Hoennig <michael@xxxxxxxxxx> 
> wrote:
>
> > Hi everybody,
> >
> > Ever anybody had such a situation:
> >
> > - A with @OneToMany EAGER to B
> > - B with @ManyToOne LAZY to A
> > (yes, other than usual, but with good reason in this case)
> >
> > And now this situation:
> > - some B is retrieved separately, modified and still attached
> > - now we navigate to its A, thus all B are loaded EAGERLY
> > - this loading includes the B which is still attached
> > - and the modification is overridden by the EAGER loading!
> >
> > Thanks for any hints!
> > ... Michael
> >
> > --
> > http://michael.hoennig.de
> >
> > _______________________________________________
> > eclipselink-users mailing list
> > eclipselink-users@xxxxxxxxxxx
> > https://dev.eclipse.org/mailman/listinfo/eclipselink-users
> _______________________________________________
> eclipselink-users mailing list
> eclipselink-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-users

--
http://michael.hoennig.de