Added testcases for the double byte string read error. We need to careful about regressions, particularly in HEAD