Re: object algebra

"Neo" <neo55592_at_hotmail.com> wrote in message news:4b45d3ad.0402231640.2248ccac_at_posting.google.com...
> > > Under RDM, closure requires meeting rather strict criteria (chapter 6)
> > > in comparision to criteria for closure in TDM.
> >
> > What strict criteria? I just finished reading the 8th edition of his
book,
> > and have no idea what you mean.
>
> From pg 145 of Date's 6th Ed, "we require two input relations to be
> what we might loosely call 'the same shape'. A more precise term for
> 'same shape' concept is type-compatibility. We will say two relations
> are type-compatible if they have identical headings. Meaning
> precisely: 1) Each have same set of attribute names 2) Corresponding
> attributes are define on the same domain. Union, intersection, and
> difference all require their operands to be type compatible."
>
> In TDM, things do not have to meet these criteria for closure.

Several things:
1) I believe Date has refined his position considerably with the 8th edition, and especially in The Third Manifesto.

2) Your "headings" must be type-compatible, or your answer makes no sense. Note that if union relation1 (R1) and union R2 have an attributes typed as T1 in R1 and T2 in R2, where T2 is a subtype of T1, then the union might make sense as having an attribute of T1. However, there are some intricacies that others may be better able to point out than me.

The point is: if your result has an arbitrary shape, because your inputs have arbitrary shape, then your answer is shapeless and meaningless. You can execute a query and "join" A to B using the ASCII value of the 3rd character of A.x and the square root of B.y if you really want to - you just can't expect it to make any sense. But at least in that case the answer would be consistent with your predicates and data, unlike a "shapless" result set.

• Eric