RFR: 8346222: SwingNodePlatformExitCrashTest fails with JUnit 5.11.3
Kevin Rushforth
kcr at openjdk.org
Fri Dec 13 21:36:41 UTC 2024
On Fri, 13 Dec 2024 19:46:54 GMT, Kevin Rushforth <kcr at openjdk.org> wrote:
> This PR fixes a latent test bug in test bug in `SwingNodePlatformExitCrashTest` by setting a flag to skip shutdown rather than relying on a quirk of JUnit 5.8.1.
>
> I discovered this while testing PR #1662 which updates JUnit to 5.11.3.
>
>
> $ gradle sdk shims
> $ gradle -PTEST_ONLY=true -PFULL_TEST=true -PUSE_ROBOT=true :systemTests:test --tests SwingNodePlatformExitCrashTest
> ...
> SwingNodePlatformExitCrashTest > executionError FAILED
> org.opentest4j.AssertionFailedError: Exceeded timeout limit of 10000 msec
> at app//test.util.Util.runAndWait(Util.java:168)
> at app//test.util.Util.runAndWait(Util.java:139)
> at app//test.util.Util.shutdown(Util.java:316)
> at app//test.robot.javafx.embed.swing.SwingNodeBase.teardownOnce(SwingNodeBase.java:81)
> ...
>
>
> The abstract `SwingNodeBase` base class calls `Util::shutdown` from its static `tearDownOnce` method, annotated with `@AfterAll`. The `SwingNodePlatformExitCrashTest` shuts down the FX toolkit as part of the test so must not call shutdown a second time. The current test logic attempts to "override" the static method in the base class with an @AfterAll annotation on a static method of the same name in the subclass. This accidentally works in JUnit 5.8 by hiding the method in the parent class, but no longer does in JUnit 5.11.
>
> The right fix is for the subclass to set a flag such that the superclass will skip the call to `Util::shutdown`.
As an FYI, the behavior change that exposed this latent bug in our test was a deliberate change in JUnit 5.11.
https://github.com/junit-team/junit5/issues/3553
Indeed it was the case that before 5.11, a static method annotated with `@BeforeAll` or `@AfterAll` in a subclass would hide the same named method in the superclass, and starting in 5.11, it no longer does.
-------------
PR Comment: https://git.openjdk.org/jfx/pull/1664#issuecomment-2542417832
More information about the openjfx-dev
mailing list