Re: MV Keys

mAsterdam wrote:
> Jon Heggland wrote:
> >
> > mAsterdam says...
> >
> >>Jon Heggland wrote:
> >>
> >>>dcressey says...
> >>>
> >>>>An alternative example would be:
> >>>>Is (onions, mushrooms) equal to (mushrooms, onions)?
> >>>>I bleieve I've aske this question several times.
> >>>
> >>>I think the answer is obvious: It is whatever we define it to be. We
> >>>just have to make the decision at some point.
> >>
> >>This is like choosing the right or the left side of the road.
> >>You don't. You may choose the country where you are going to
> >>drive, but that's it.
> >
> >
> > I don't get this analogy at all. If I have a pizza database, and for
> > some reason want to model "toppings" as a single attribute, I will use a
> > list if I (or the pizza place I'm making it for) want the topping order
> > to matter, and a set if I don't. Where does the national government
> > enter into it?
> >
> > Or are you quibbling about the specific notation (parenthesis, commas)
> > that was used in the example?
>
> No, not at all, sorry If I gave that impression.
>
> I really would like to have it this way:
>
> "I will use a list if I want the order
> to matter, and a set if I don't."

Don't forget bags (multisets). This cannot be shrugged off. If you have a bag (unordered, but can repeat values), then how do you want to implement it? It really works quite well for the implementation of properties that have list, bag, or set values to be implemented as lists, although I, too, would like to be able to specify whether the ordering matters or not.

> (pizza specifics skipped)
> , but in reality I don't have it like that.
> In most environments either sets (RDBMS) or lists
> (XML, Pick) are heavily favoured.

Pick has logical sets at the top level (as files, which can be modeled as functions/relations/sets) and lists as a type of attribute (but not sets).

> Now this is c.d.t, not c.d.p, so one could say that in
> theory it's just a choice. OTOH I think that these
> environments have theoretical and cultural roots
> (RM, Ontologies, Documents) yet to be investigated
> in a unified way.

Agreed. Cheers! --dawn Received on Tue Mar 07 2006 - 22:24:12 CET