Re: NULLs: theoretical problems?

> > Which recursive structures?
>
> You've discounted the notion before, it's basically just a relation that
> has an attribute that is of the same type as that same relation. For
> some reason that I can't explain very well, it's long bugged me that the
> RM seems to need a fundamental operator for to get transitive closure,
> so if that could ever be avoided, it just seemed to me that structures
> must aid it. Maybe such a relation is just a paradox that I can't see.

To recurse well, a data model should be built on a single basic modeling element. Recursion in RM will always be limited due the fact that its has several basic modeling elements (relation, tuple, attribute, value). In contrast, a database like dbd has only one modeling element, the node. Received on Wed Aug 22 2007 - 04:45:53 CEST