Noam Postavsky writes: > Thanks, could you add a regression test for this too, please? Since > we've already had two mistakes independently generated, it seems likely > we'll make more if this code is updated again. Of course. Is the attached up to scratch? Thanks, -- Basil