Re: Objects and Relations

On 15 fév, 20:21, "JOG" <j..._at_cs.nott.ac.uk> wrote:
> On Feb 15, 7:05 pm, "Cimode" <cim..._at_hotmail.com> wrote:
>
>
>
> > On 15 fév, 18:05, "JOG" <j..._at_cs.nott.ac.uk> wrote:> Sigh. Let me take you back to what you initially wrote Cimode:
>
> > > Cimode wrote:
> > > > >>What I meant is that LegoBlock
> > > > >>and Location concepts should be separated if LegoBlock is to be
> > > > >>considered a relation. A relation *must* have a stable primary key.
> > > > >>Location is not a stable primary key therefore it does not identify
> > > > >>LegoBlock.
>
> > > For a legoBlock, location is perfectly acceptable as an identifier. It
> > > is unstable, and so potentially a bad design choice, but still
> > > perfectly valid identifier at any instant. My complaint at what you
> > > wrote has nothing to do with hollow spheres (which require a location
> > > +radius identifier), and I will not get dragged off by a red herring.
>
> > You have initially assumed that XYZ as a location. Now, you have
> > changed it to XYZR. My puposed was not to drag u into anything but to
> > point out an initial mistake you made in selecting XYZ only.
>
> No still not good enough I'm afraid. LegoPieces are not hollow
> spheres, and do not have a radius. In their case picking XYZ as an
> identifier is valid.
>
> > Given
> > the reaction of hostility it has triggered, I begin to regret such
> > effort.
>
> > > I assume that you will show the same sincerity in acknowledging that
> > > bad design choice != contravening RM.
>
> > Agreed.
If you ask me, XYZ can merely identify a point in a very specific referential. A normalization would probably dictate a far more complex design and keys (shape, color).

But you have not answered my question now that I have answered yours. Do you undestand my intentions now? Received on Thu Feb 15 2007 - 20:55:22 CET