PartOnTopManagerTest.test_PlaceholderOnTopStackSwitch fails in I-builds

[trancexpress]

See: https://download.eclipse.org/eclipse/downloads/drops4/I20260415-1800/testresults/html/org.eclipse.e4.ui.tests_ep440I-unit-macosx-aarch64-java21_macosx.cocoa.aarch64_21.html

	expected: <false> but was: <true>

org.opentest4j.AssertionFailedError: expected: <false> but was: <true>
at org.junit.jupiter.api.AssertionFailureBuilder.build(AssertionFailureBuilder.java:151)
at org.junit.jupiter.api.AssertionFailureBuilder.buildAndThrow(AssertionFailureBuilder.java:132)
at org.junit.jupiter.api.AssertFalse.failNotFalse(AssertFalse.java:63)
at org.junit.jupiter.api.AssertFalse.assertFalse(AssertFalse.java:36)
at org.junit.jupiter.api.AssertFalse.assertFalse(AssertFalse.java:31)
at org.junit.jupiter.api.Assertions.assertFalse(Assertions.java:231)
at org.eclipse.e4.ui.tests.workbench.PartOnTopManagerTest.test_PlaceholderOnTopStackSwitch(PartOnTopManagerTest.java:175)
at java.base/java.lang.reflect.Method.invoke(Method.java:580)
at java.base/java.util.ArrayList.forEach(ArrayList.java:1596)
at java.base/java.util.ArrayList.forEach(ArrayList.java:1596) 

Workspace log: https://download.eclipse.org/eclipse/downloads/drops4/I20260415-1800/testresults/ep440I-unit-macosx-aarch64-java21_macosx.cocoa.aarch64_21/org.eclipse.e4.ui.tests.UIAllTests.txt

org.eclipse.e4.ui.tests.UIAllTests.txt

There are no test start/end in the log so no idea what is important.

I think I see this fail frequently, or at least fails that look very similar.

[trancexpress]

Same fail also here:

• https://download.eclipse.org/eclipse/downloads/drops4/I20260414-1800/testresults/html/org.eclipse.e4.ui.tests_ep440I-unit-macosx-aarch64-java21_macosx.cocoa.aarch64_21.html
• https://download.eclipse.org/eclipse/downloads/drops4/I20260413-1800/testresults/html/org.eclipse.e4.ui.tests_ep440I-unit-macosx-aarch64-java21_macosx.cocoa.aarch64_21.html
• https://download.eclipse.org/eclipse/downloads/drops4/I20260412-1800/testresults/html/org.eclipse.e4.ui.tests_ep440I-unit-macosx-aarch64-java21_macosx.cocoa.aarch64_21.html
• https://download.eclipse.org/eclipse/downloads/drops4/I20260410-1800/testresults/html/org.eclipse.e4.ui.tests_ep440I-unit-macosx-aarch64-java21_macosx.cocoa.aarch64_21.html
• https://download.eclipse.org/eclipse/downloads/drops4/I20260407-1040/testresults/html/org.eclipse.e4.ui.tests_ep440I-unit-macosx-aarch64-java21_macosx.cocoa.aarch64_21.html

So seems like a stable fail in macosx-aarch64 builds.

The last recorded build without the fail is:

https://download.eclipse.org/eclipse/downloads/drops4/I20260330-0010/reports.html#tests

[trancexpress]

At least in platform UI, I'm not seeing anything that looks related. Maybe this, but it seems like a stretch: #1338

I also can't run the test locally even on Linux, there are tons of errors like this (and a ton of windows get opened):

!ENTRY org.eclipse.ui 4 0 2026-04-16 08:04:47.635
!MESSAGE Unhandled event loop exception
!STACK 0
org.eclipse.e4.core.di.InjectionException: Unable to process "WorkbenchWindow.model": no actual value was found for the argument "MTrimmedWindow".

[HeikoKlare]

Looks like the failure started to happen when switching to new macOS agents in the infrastructure (somewhen between 30th March and 7th April).

The high number of windows being opened is something that happens on Windows as well. And I found that it at least seems to happen for every test using the WorkbenchContextExtension. E.g., for MWindowTest that happens for me as well.
The error message regarding WorkbenchWindow.model occurs for me in every test method of PartOnTopManagerTest.

[HeikoKlare]

The high number of windows being opened is something that happens on Windows as well.

Just saw that those tests are executed without UI harness. Unfortunately, you only see that when looking into the CI test configuration (in build.properties / test.xml), so you need to change the application for test execution to "headless mode" when running the tests in the IDE. Then you do not experience those many unexpected windows.

[vogella]

I think the many unexpected windows are a bug in the clean-up code, I push a PR for this. This will most likely not help with the Mac.

[HeikoKlare]

As an additional finding: the failure only occurs when executing the tests inside the IDE / via Ant in I-Builds. Execution with Tycho reproducibly succeeds on my system (and also does so in CI builds).