I've put that fix in. I'm not quite sure what you are suggesting for a new test, but test cases that fail on known bug regressions (i.e. one that fails before your fix and works after it) are always welcome. Thanks, Roland