Community
Participate
Working Groups
See Bug 95193 for background. The handlerSubmission element in the "org.eclipse.ui.commands" extension point was added during the 3.0 development cycle. It was always marked as experimental, as was shipped as experimental with 3.0. Now, a complete replacement for it exists (i.e., the 'org.eclipse.ui.handlers" extension point) in 3.1. Is it time to remove the handlerSubmission element entirely, and make a note in the 3.1 migration guide?
If it shipped as API in a release, we need to support it. However if this is going to cause major pain, we can deprecate it and document it in the migration guide.
Jim actually was the one who wanted it removed.
Read the background bug.
Sorry, I missed the background bug, and my initial reaction was the same as yours. I'm OK with the decision there.
Created attachment 21645 [details] Patch to "org.eclipse.core.commands" This removes the final keyword from two methods so that the AbstractHandler in "org.eclipse.ui.workbench" can re-implement them (and thereby extend this AbstractHandler).
Created attachment 21646 [details] Patch to "org.eclipse.ui" Removes the handlerSubmission element from the commands extension point.
Created attachment 21647 [details] Patch to "org.eclipse.ui.workbench" Makes the AbstractHandler in "org.eclipse.ui.workbench" extend the AbstractHandler in "org.eclipse.core.commands". This is intended as a response to Bug 96070 comment #3.
I'm looking for approval from my component lead. Nick?
+1. We'll need an entry in the migration guide documenting this change from 3.0.
Created attachment 21665 [details] Patch to "org.eclipse.platform.doc.isv" A note to the migration guide.
Andrea: Can you apply the patch to "org.eclipse.platform.doc.isv"? I do not have commit rights to that plug-in.
I have applied the patch to "org.eclipse.platform.doc.isv". Thanks, AC
Patches committed to CVS.
It seems that the API methods core.commands.AbstractHandler.fireHandlerChanged () and hasListeners() are no longer final. This opens up the possibility that any AbstractHandler subclass could override these methods. Was this the intent? If so, then they should have subclass contracts that state what subclasses are allowed to do with them.
It was intentional. Added additional javadoc to the methods.
Verified by looking in the extension point documentation for org.eclipse.ui.commands. handlerSubmission is no longer there.