Well, I would hope that

create procedure ...

declare variable i2 TYPE OF COLUMN test2.id1;

declare variable i TYPE OF COLUMN test.id;

begin

  i2 = 12345678;

  i = 12345678;

...

would complain about the assignment to i and not i2.

Although I agree with you that it sounds rather useless to allow foreign keys to be of a type that is a superset, it also sounds rather harmless (if properly implemented). Does the SQL standard say anything about this?

Set