> We (SWITCH) could reproduce a similar case during testing back then, but we > couldn't find an explanation for the occurrence of the problem. (In the comment in the > code mentioned by Scott, "observed" refers to this test case). Thanks Daniel. -- Scott