#16831 has occurred because of a missed regression. This PR adds a simple test to try to prevent this occuring again. Signed-off-by: Andrew Thornton <art27@cantab.net>