Kathey Marsden wrote:
I think setting the type should not change the nullability of the column.
I agree; we have a separate ALTER TABLE command for modifying nullability. However, I don't recall this specific question being discussed during the 10.3 ALTER TABLE work, so it's quite possible that 10.3 isn't behaving correctly, either. Certainly looks like you've found a soft spot worthy of more investigation. thanks, bryan
