This seems to happen when running the full Selenium suite and it tries to tear down the session. It finally falls apart when cleaning up after window_tests.py::test_should_minimize_the_current_window.
<rdar://problem/42822671>
Created attachment 351887 [details] Patch
Comment on attachment 351887 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=351887&action=review > Source/WebKit/ChangeLog:9 > + Forcibly exit fullscreen in resetState(), while the VideoFullscreenManagerProxy is still valid. Should we be catching Objective-C exceptions in more places?
Created attachment 351889 [details] Patch
(In reply to Simon Fraser (smfr) from comment #3) > Comment on attachment 351887 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=351887&action=review > > > Source/WebKit/ChangeLog:9 > > + Forcibly exit fullscreen in resetState(), while the VideoFullscreenManagerProxy is still valid. > > Should we be catching Objective-C exceptions in more places? This is a C++ problem; we're calling a null pointer (because the VideoFullscreenManager has been cleared). So in addition to this patch, we should probably add a bunch of (self._manager) null checks in functions which use it.
Comment on attachment 351887 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=351887&action=review > Tools/TestWebKitAPI/Tests/WebKitCocoa/CloseWebViewAfterEnterFullscreen.mm:79 > + Nit: After
Comment on attachment 351889 [details] Patch r=me
Comment on attachment 351889 [details] Patch Just realized that this test will likely not succeed on iOS, since the fullscreen delegate is different there. I'll put up a new patch which runs the test only on Mac.
Created attachment 351911 [details] Patch for landing
Comment on attachment 351911 [details] Patch for landing Clearing flags on attachment: 351911 Committed r236984: <https://trac.webkit.org/changeset/236984>
All reviewed patches have been landed. Closing bug.