Re: [eclipselink-users] Can't deploy when using LAZY, can deply when usi

["Michael O'Brien" <michael.obrien@xxxxxxxxxx>]

Michael,
   For dynamic weaving in Glassfish (there is now a known issue with 
war/ejb module precedence in application.xml for some types of 
applications) - Try changing the order of your modules in 
application.xml in your EAR as in some cases (JSF managed-bean with an 
@EJB injection for example is one case) - the jar containing an injected 
PersistenceContext (usually a SSB in an ejb.jar) must be loaded first 
(where it is weaved/instrumented) before any WAR or other ejb.jar module 
that injects the SSB that contains the persistence unit.
   See bug # 323148 for GlassFish and WebLogic in this case - if it 
helps your situation.
http://bugs.eclipse.org/323148

   The NullPointerException in this case is thrown by the JVM itself 
instead of a MethodNotFound Exception when trying to run a weaving 
function of the form _*_vh.
   We will add a better root cause warning about the EE module 
processing order that causes this issue.

   thank you
   /michael
Tom Ware wrote:
> Hi Tim,
>
>   Here is the bug that tracks our spring issues:
>
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=288383
>
>   The weaving issue is related to how early Spring loads the 
> application classes (in a Java EE application server, JPA gets a hook 
> before the application classes get loaded that allows weaving to be 
> set-up - not quite that simple on Spring)
>
>   We cannot think of a reasonable way for us to solve this problem 
> without changes in Spring itself.
>
> -Tom
>
> Tim Hollosy wrote:
> > Looks like I ran into this exact "bug":
> >
> > http://old.nabble.com/OneToOne-Mapping-works-with-FetchType.EAGER-but-not-with-FetchType.LAZY--%3E-Exception-when-deploying-%28SGES-2.1%29-td21711849.html#a22969114 
> >
> >
> > I'm using the J2EE Preview Server and weaving with - spring-agent.jar
> > from Spring 2.5.5, everything else works fine, but it fails when I do
> > the OneToOne with lazy.
> >
> > I've switched directions and implemented things a different way, I
> > could test this out on my real environment which is OC4J 10.1.3.3
> > though.
> >
> > ./tch
> >
> >
> >
> > On Mon, Nov 9, 2009 at 2:08 PM, Tom Ware <tom.ware@xxxxxxxxxx> wrote:
> > > Hi Tim,
> > >
> > >  Is there a filed-bug for this?  Can you point me at the post?
> > >
> > > -Tom
> > >
> > > Tim Hollosy wrote:
> > > > Thanks Tom,
> > > > I did a search through the list and it looks  like this is some other
> > > > bug relating to using a PK as an FK to another entity, it has similar
> > > > results though with the odd weaved methods in the exception trace.
> > > >
> > > > ./tch
> > > >
> > > >
> > > >
> > > > On Mon, Nov 9, 2009 at 1:54 PM, Tom Ware <tom.ware@xxxxxxxxxx> wrote:
> > > > > Hi Tim,
> > > > >
> > > > >  The bug here is specific to GlassFish DDL generation.  Without DDL
> > > > > generation enabled, if you are seeing an issue, it is likely 
> > > > > something
> > > > > else.
> > > > >
> > > > >  The fix added to EclipseLink mentioned below was simply to 
> > > > > acknowledge
> > > > > an
> > > > > old TopLink Essentials property that was being passed to EclipseLink
> > > > > prior
> > > > > to DDL generation in order to work around the issue.
> > > > >
> > > > > -Tom
> > > > >
> > > > > Tim Hollosy wrote:
> > > > > > FYI I believe this bug appears in Jetty (The J2EE Preview) server 
> > > > > > used
> > > > > > in Eclipse as well. The previous work-arounds don't really work for
> > > > > > me, as I don't have any ddl generation turned on, changing the
> > > > > > relationship to eager fixed it, It is annoying though.
> > > > > >
> > > > > > ./tch
> > > > > >
> > > > > >
> > > > > >
> > > > > > On Wed, Oct 21, 2009 at 12:33 AM, Michael Bar-sinai
> > > > > > <mich.barsinai@xxxxxxxxx> wrote:
> > > > > > > Update (for future searchers):
> > > > > > > The problem was found (kudos to Tom and Mitesh) to be an issue with
> > > > > > > Glassfish 2.1's PersistenceProcessor. The fix will go into GF 2.2.
> > > > > > > Another
> > > > > > > workaround will go into Eclipselink 2. Another workaround is 
> > > > > > > described
> > > > > > > in
> > > > > > > previous mails.
> > > > > > > Customers with paid support might be able to get a fix now (contact
> > > > > > > your
> > > > > > > support etc).
> > > > > > > --Michael
> > > > > > > _______________________________________________
> > > > > > > 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
> > > > > _______________________________________________
> > > > > 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
> > > _______________________________________________
> > > 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
> _______________________________________________
> eclipselink-users mailing list
> eclipselink-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-users