RFR: JDK-8277396: [TESTBUG] In DefaultButtonModelCrashTest.java, frame is accessed from main thread

[Alexey Ivanov]

On Fri, 19 Nov 2021 05:22:48 GMT, Prasanta Sadhukhan <psadhukhan at openjdk.org> wrote:

> I am not sure on this as those 100+ tests which accesses frame variable outside EDT before it calls dispose is not failing in CI testing and also locally so to modify this particular test only does not make sense to me.
Most tests are short-lived, that's probably the reason why it doesn't cause any issues. Yet it doesn't mean that it's the way it should be.

My point is that JDK regression tests could be used as a sample code for writing Swing code. That is JDK code as well as its tests should be a role model.

I wouldn't commit into fixing all 100+ tests: it requires a considerable amount of time and effort. However, I consider fixing them one by one is a good thing to do. I spent around 15-20 minutes on this issue, including submitting it and raising code review. I admit both of us have wasted more time discussing whether it's worth integrating…

> If you still want to take it forward, regarding the test, I will request to add a delay of 1sec after l56 after we make the frame visible, same way we normally do for other swing reg tests.

I added 1 sec delay in the latest iteration of the review.

-------------

PR: https://git.openjdk.java.net/jdk/pull/6455