That only applies if the column being dropped comprises the ONLY column in the index. That is to say that for the following schema: ``` create table x(a, b, c); create [unique] index x_b on x(b); ``` then if you do an `ALTER TABLE x DROP COLUMN b;` I would assert that mayhaps the index `x_b` can be dropped automagically, or mayhaps not (arguments can be made for either position). However for the following schema: ``` create table x(a, b, c); create [unique] index x_bc on x(b,c); ``` that `ALTER TABLE x DROP COLUMN b;` should fail. There is absolutely nothing which can be done to make the schema transition to a consistent state. You (the user) should have to drop the index x_bc manually before the DROP COLUMN can be processed successfully.