Is there a test case I could use to see the problem?

run
ant -Dtest.config=ruby test
in ide.kit

Here is a workaround.
NewProjectWizardOperator.invoke().cancel();

As the priority is coming from the blocked P1 against commit validation and, with this workaround it will be fixed, I decrease the priority.

Moving back to P1, the MemoryValidationTest which is part of commit validation fails on Mac.

I have narrowed it to a race condition in TemplateWizard. 

A comment in instantiateImpl method says:
        // here can return newObjects because instantiateNewObjects() was called
        // from WizardDescriptor before close dialog (on Finish)

Well, on mac it is not allways true. I did this change:
$ hg diff openide.loaders/src/org/openide/loaders/TemplateWizard.java
diff -r 950f7af22983 openide.loaders/src/org/openide/loaders/TemplateWizard.java
--- a/openide.loaders/src/org/openide/loaders/TemplateWizard.java       Fri Oct 31 15:41:34 2008 +0300
+++ b/openide.loaders/src/org/openide/loaders/TemplateWizard.java       Fri Oct 31 15:59:29 2008 +0300
@@ -395,6 +395,7 @@ public class TemplateWizard extends Wiza
     }

     Set<DataObject> instantiateNewObjects (ProgressHandle handle) throws IOException {
+        System.out.println("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! instantiateNewObjects called");
         progressHandle = handle;
         try {
             // #17341. The problem is handling ESC -> value is not
@@ -499,6 +500,7 @@ public class TemplateWizard extends Wiza

         // here can return newObjects because instantiateNewObjects() was called
         // from WizardDescriptor before close dialog (on Finish)
+        System.out.println("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! instantiateImpl exits");
         return newObjects;
     }

and the output confirms my findings.

To reproduce, simple run org.netbeans.jellytools.NewProjectWizardOperatorTest test from IDE.

Jirko, I guess this is your code. Will you investigate the problem?

changeset:   bb276c9cffeb
user:        jrechtacek@netbeans.org
date:        Sun Aug 06 15:58:35 2006 +0000
summary:     #58889: Enhance wizard framework to display progress indication on Finish

> Jirko, I guess this is your code. Will you investigate the problem?
> 
> changeset:   bb276c9cffeb
> user:        jrechtacek@netbeans.org
> date:        Sun Aug 06 15:58:35 2006 +0000
Jardo, you are saying these test failing since Aug 06 15:58:35 2006 +0000? I don't think so. Assigning back.
> summary:     #58889: Enhance wizard framework to display progress indication on Finish

Ivan, what means current version? NB6.1? NB6.5? NB7.0? Thanks

No, I am saying that this is work for owner of dialogs API. The TemplateWizard has always been maintained by that 
module owner.

Okay, I was confusing by your comment. Do you know from when there is this problem? I doubt I did some changes there for
ages.

I do not know for how long this is around, actually.

My be it was there for a long time and manifested itself just recently by tests on Mac OS only. The platform did not 
have the highest priority, honestly.

As you can see it the issue is entered in Aug. I remember it was narrowed down from commit validation at that moment. 
I am not sure for how long commit validation was failing because of that.

I do not think it has to be solved for NB 6.5. Ivan, do you agree ?

yes, it is not necessary to fix it in 6.5

There is the known workaround NewProjectWizardOperator.invoke().close(); which is performing for ages in
IDEValidation.testMainMenu(). The original problem is some race condition between WizardDescriptor and TemplateWizard
callback calling. But no one reported such problem in IDE, it's visual test specific, moreover Mac specific (on other
platforms are tests passing). I decreasing the priority to P2.

The workaround is mentioned in the issue comments already - we know about it.

P2 is fine.

It's the known bug, which doesn't affect IDE usability to end-user's.  I guess its fix will destabilize TemplateWizard
code and probably make some regressions. I don't plan to fix it sometime presently.

NetBeans.org Migration: changing resolution from LATER to WONTFIX