I do not understand your problem. No one will stop you from explicitly dropping an index first and the column afterwards. This is always possible, even now. Having DROP COLUMN should be an easement compared to the current procedure. And having to programatically check for indexes, remember them and recreate them is no way easier than the current situation. And yes, indexes *are* a banality. They are in no way an existential part of any database, they only speed things up. Notabene, we are talking about indexes, not constraints!