Re: Does Codd's view of a relational database differ from that of Date & Darwin? [M.Gittens]

Marshall Spight wrote:
>
> Ah! But pehaps when you say user-defined domains were
> too restricted, you meant that they left off union types:
>
> data MyParticularSituation = IHaveOne of int
> | ElseIDont;

Yes, that is what I meant.

> I think that covers the "special values" situation pretty
> well, but it seems overly manual for capturing the
> cardinality=0 case.

True. But I would argue that this is because it makes the true complexity of the situation explicit. You can try to sweep that under the rug but then it will (1) complicate the meta-model because you need a logic and domain operations that deal with null values and (2) it may come back to bite you on unexpected moments because the system doesn't force you always to take into account that a certain value may or may not be there.

Note that especially (1) is important because not only does it mean that it becomes harder for users / programmers to understand in detail how the system works, but it makes it harder to implement it efficiently and correctly.

• Jan Hidders