| EMF examples break VE (Bugzilla 79078) [message #72407] |
Mon, 22 November 2004 12:05  |
Eclipse User |
|
|
|
Hi all,
Bug 79078 was filed against EMF last Friday. Apparently, our Java model
example breaks VE. I've taken a quick look, and the problem seems to be
the following registration in its plugin.xml:
<extension point = "org.eclipse.emf.ecore.extension_parser">
<parser
type="java"
class="org.eclipse.emf.java.util.JavaResourceFactoryImpl"/ >
</extension>
This is globally registering our custom Java resource factory against
the "java" file extension.
I'm seeing the following exception in /.metadata/.log:
!ENTRY org.eclipse.ve.java.core 4 0 Nov 22, 2004 10:57:34.516
!MESSAGE Exception thrown.
!STACK 0
java.lang.ClassCastException: org.eclipse.emf.java.util.JavaResourceImpl
at
org.eclipse.ve.internal.java.codegen.core.JavaSourceTranslat or.createJavaInstances(JavaSourceTranslator.java:705)
at
org.eclipse.ve.internal.java.codegen.core.JavaSourceTranslat or.buildCompositionModel(JavaSourceTranslator.java:774)
at
org.eclipse.ve.internal.java.codegen.core.JavaSourceTranslat or.decodeDocument(JavaSourceTranslator.java:881)
at
org.eclipse.ve.internal.java.codegen.core.JavaSourceTranslat or.loadModel(JavaSourceTranslator.java:656)
at
org.eclipse.ve.internal.java.codegen.editorpart.JavaVisualEd itorPart$Setup.run(JavaVisualEditorPart.java:1444)
at org.eclipse.core.internal.jobs.Worker.run(Worker.java:66)
Here, createJavaInstances() is obtaining the resource of a model object
(URI=platform:/resource/Test/MyFrame.java), and casting it to
XMIResource. It's not expecting a JavaResourceImpl.
So, it would seem that VE isn't happy when .java files are parsed using
custom non-XMI resource implementation. Are you just expecting it to be
handled by the default XMIResourceImpl, or registering a different
subclass of XMIResourceFactoryImpl()? I didn't find that any VE
registration against an extension, but I *did* see a registration of
JavaXMIFactoryImpl() against the "java" *scheme*. Some background on
VE's needs and expectations in this realm would be most helpful.
My first instinct is that it might not be best to depend upon the global
registry like this. The safer thing to do might be to properly
initialize the registry local to your resource set, so as to suit your
needs.
Obviously, the fact that we're breaking you makes us all look bad.
Hopefully, we can resolve this quite quickly.
Thanks,
Dave
|
|
|
| Re: EMF examples break VE (Bugzilla 79078) [message #73147 is a reply to message #72407] |
Mon, 29 November 2004 16:22  |
| Eclipse User |
|
|
|
Originally posted by: richkulp.us.NO_SPAM.ibm.com
We already resolved it in our 1.0.2 builds.
https://bugs.eclipse.org/bugs/show_bug.cgi?id=71473
What I did was to override the extension map in my own resource set. I
really didn't like that, but after talking with Ed that is the only
thing we could come up with.
It is too late now, but shouldn't the scheme be higher priority than the
extension? But unfortunately you probably can't change that now because
someone may rely on it.
--
Thanks,
Rich Kulp
|
|
|
| Re: EMF examples break VE (Bugzilla 79078) [message #603484 is a reply to message #72407] |
Mon, 29 November 2004 16:22 |
| Eclipse User |
|
|
|
Originally posted by: richkulp.us.NO_SPAM.ibm.com
We already resolved it in our 1.0.2 builds.
https://bugs.eclipse.org/bugs/show_bug.cgi?id=71473
What I did was to override the extension map in my own resource set. I
really didn't like that, but after talking with Ed that is the only
thing we could come up with.
It is too late now, but shouldn't the scheme be higher priority than the
extension? But unfortunately you probably can't change that now because
someone may rely on it.
--
Thanks,
Rich Kulp
|
|
|
Powered by
FUDForum. Page generated in 0.03092 seconds
Home