So, if the correct behavior was known, why did you not write a test asserting the correct behavior? I understand that it can get complicated, just legitimately curious if the behavior could or could not be easily captured by the test suite.