The problem is you need a boat load of additional code to cover unset. Not just in SQL (syntax is different for NULL than it is for comparing any other type) but often also in your importing language too (eg some languages will cast NULL to a nil value that can actually raise exceptions or even crash your application if not handled correctly).
Capturing those edge cases is non-trivial compared checking the value of a char.
In an idea world your unit tests and CI pipelines would catch all of those but that depends on well written tests. Thus in my experience having fewer hidden traps from the outset is automatically a better design than one that perfectly fits an academic theory but is harder to support in practice.
Capturing those edge cases is non-trivial compared checking the value of a char.
In an idea world your unit tests and CI pipelines would catch all of those but that depends on well written tests. Thus in my experience having fewer hidden traps from the outset is automatically a better design than one that perfectly fits an academic theory but is harder to support in practice.