Re: [mdr-users] createExtent(String, RefObject, RefPackage[])

From the description of the method one would expect that as soon as
the array includes existing extents they will be shared.
http://www.netbeans.org/download/dev/javadoc/MdrAPIs/org/netbeans/api/mdr/MDRepository.html#createExtent(java.lang.String,%20javax.jmi.reflect.RefObject,%20javax.jmi.reflect.RefPackage[])
This is not the case.

Originally, I expected that I could even load a standalone extent and
then pass it to the clustering extent.  As you stated, MDR could
support this but does not do this:
http://mdr.netbeans.org/servlets/ReadMsg?list=users&msgNo=3259

What I did afterwards was using extents loaded with the clustering
metamodel only (instead of using instances of the standalone version
of the clustered metamodels).  However, I passed such an extent twice
as member of two different arrays.  MDR did not check whether the
array members were already clustered.  I would expect this to happen,
according to the Javadoc description.

Regards,
Pieter

On 9/5/05, Martin Matula <[email protected]> wrote:
> Sorry, I do not understand the problem. What is the significance of the
> array identity?
> Martin
> 
> Pieter Van Gorp wrote:
> 
> >Hi all,
> >I looked at the problem again with a fresh mind and found its cause.
> >
> >Apparantly, MDR does not check the actual contents of the array passed
> >to createExtent(String, RefObject, RefPackage[])... therefore, when
> >you want to cluster only one metapackage (say "C") into two different
> >extents (say "A" and "B") and pass this C two times with an inline
> >array initialization like:
> >A= createExtent("A", A_M2, RefPackage[]{C}));
> >B= createExtent("B", B_M2, RefPackage[]{C}));
> >
> >then MDR represents the content of C into B two times.  This leads to
> >reference problems when storing/loading B to/from XMI.
> >
> >I solved the problem by using the same array variable for creating the
> >extent for A and B but I'd suggest to let MDR check the content of the
> >array passed to createExtent instead of comparing its address...
> >
> >Regards,
> >Pieter.
> >
> >On 8/10/05, Martin Matula <[email protected]> wrote:
> >
> >
> >>Have you tried debugging it?
> >>
> >>Pieter Van Gorp wrote:
> >>
> >>
> >>
> >>>Hi again,
> >>>
> >>>On 8/9/05, Pieter Van Gorp <[email protected]> wrote:
> >>>
> >>>
> >>>
> >>>
> >>>>On 8/8/05, Martin Matula <[email protected]> wrote:
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>>To fix this, you need to load the tracing metamodel. Now find UML
> >>>>>package in this tracing metamodel and instantiate it. If you now find
> >>>>>tracing package, instantiate it and pass in the instance of UML package,
> >>>>>the instance of UML package will be reused.
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>Good to know,
> >>>>thanks for the insight, I'll try it right away!
> >>>>
> >>>>
> >>>>
> >>>>
> >>>After changing merely one line of code, the clustered extent was
> >>>indeed filled up with the content of the UML-XMI: I replaced
> >>><original>appMextent= (RefPackage)
> >>>repo.createExtent(appMurl.toExternalForm(), getMainMP(appMMurl, "UML",
> >>>xmiReader));</original>
> >>>
> >>>by:
> >>>
> >>><new>appMextent= (RefPackage)
> >>>repo.createExtent(appMurl.toExternalForm(), getMainMP(traceMMurl,
> >>>"UML", xmiReader));</new>
> >>>
> >>>Strange enough, I'm now confronted with the opposite problem: all UML
> >>>elements seem to be available twice!  Moreover, this leads to
> >>>"duplicate IDREF" exceptions when using XMI modules for importing UML
> >>>profiles.  I started debugging with a UML file without XMI modules (by
> >>>storing the UML profiles within my UML test project) and can now open
> >>>the XMI that MDR writes from the clustered UML extent to a file called
> >>>"standalone.xmi".  Luckily, MagicDraw ignores the duplicate model
> >>>elements... however, I believe I should remove the error in my code
> >>>that causes the duplication of model elements... Martin, am I still
> >>>using the API wrong?
> >>>
> >>>Please refer to yesterday's mail for reproducing my problem:
> >>>http://mdr.netbeans.org/servlets/ReadMsg?list=users&msgNo=3255
> >>>
> >>>Regards,
> >>>-- Pieter.
> >>>
> >>>
> >>>
> >>>
> >>
> >>
> 
>