By "hypotheses on the schema" I mean we are assuming the schema is already in a very normalized form. I fully agree most database schemas are awful, this is a deliberate choice to avoid getting us into that kind of trouble.
The "intermediate layer" role is played by the IR I was talking about in my previous comment. The user would never be able to see anything not provided by the IR because it's not even representable.
By "hypotheses on the schema" I mean we are assuming the schema is already in a very normalized form. I fully agree most database schemas are awful, this is a deliberate choice to avoid getting us into that kind of trouble.
The "intermediate layer" role is played by the IR I was talking about in my previous comment. The user would never be able to see anything not provided by the IR because it's not even representable.
Our scope is much narrower, really.