Reading this test made me cry for about 10 minutes. This test is attempting to inject its own timeout detection mechanism to catch the cases where the postMessage API is not accessible in an iframe. However, timeout is a valid failure state for a test, and our test harness is quite capable of catching such timeout cases, so that we wouldn't need every single test to come up with its own timeout detection solution. This test should just stop trying to be smart, and start playing well with our test framework.
Created attachment 526554 [details] [diff] [review] Patch (v1)