Community
Participate
Working Groups
If a test times out the result may be Exception in thread "Press Finish" java.lang.NullPointerException at org.eclipse.ocl.examples.test.xtext.FileNewWizardTest$1$1.run(FileNewWizardTest.java:127) from getShell().getDisplay().asyncExec(new Runnable() Presumably shell/display mau be null after too long. ?? wasit really too long or is there a problem in ocl-compatibility-2020-06
(In reply to Ed Willink from comment #0) >> getShell().getDisplay().asyncExec(new Runnable() > > Presumably shell/display mau be null after too long. getDisplay() can never be null. getShell() can never be null wrt the Javadoc but can actually be null post close(). Bug 572392 raised.
(In reply to Ed Willink from comment #0) > ?? wasit really too long or is there a problem in ocl-compatibility-2020-06 Eventually got a success taking 42 mins rather than the original 30 mins limit, see Bug 572158. All tests from oxygen to 2021-03 pass so there doesn't seem anything fundamentally flaky about FileNewWizardTest "Press Finish"; just bad behaviour if the build time terminates during the wait.
(In reply to Ed Willink from comment #1) > (> getShell() can never be null wrt the Javadoc but can actually be null post > close(). Bug 572392 raised. Javadoc is wrong wrt 20 years of legacy, so all calls to getShell() reviewed to accommodate null / isDisposed(). @Nullable EEA also added but doesn't seem very effective.
