We currently have some dangling Swing tests. I have the suspicion that Popper finds "old" (not visible) windows, whereby then potentially just such a window before the current (visible) window with the same title is preferred.
Swing seems to manage the window list based on WeakReferences, so the test result depends on
What other tests ran in the same JVM (explains why the tests run locally or in the Daily)
Whether a GC has been run since the last test (random)
To solve one would have to org.popper.fw.jemmy.JemmyPageObjectHelper.AllWindowsSearchContextProvider.getContainer ()
not all windows, but only those where IsVisible () is true ...
We currently have some dangling Swing tests. I have the suspicion that Popper finds "old" (not visible) windows, whereby then potentially just such a window before the current (visible) window with the same title is preferred.
Swing seems to manage the window list based on WeakReferences, so the test result depends on
To solve one would have to org.popper.fw.jemmy.JemmyPageObjectHelper.AllWindowsSearchContextProvider.getContainer () not all windows, but only those where IsVisible () is true ...