Re: Constraints and Functional Dependencies

On 25 fév, 14:07, mAsterdam <mAster..._at_vrijdag.org> wrote:
> Cimode wrote:
> > mAsterdam wrote:
> > [Snipped]
> >> But what exactly is the reference referencing?
> >> A set of S-tuples, not just one.
> >> That makes it a sort of hand_wave instead of a reference, right?
>
> >> Is there something I am missing?
> > Maybe the preeminence of RA domain associative characterization over
> > structuralist perspective based on attributes.
>
> I need some help to decipher this.
It is not a complicated as it sounds to be. I believe that, to be able to apply formal set theory math definitions to computing, Codd had to find a way to associate domains in an *explicit* manner so that computing operations would be applied to mechanized systems. As naive set theory lacked conceptual tools that allowed such association in an explicit manner, Codd's prime genius was to create a structuralism to apply set naive associations as inter-domain operations. In such process, structuralism simply became a way to embody inter-domain operations through actors: relations. That way, he characterized relation structural aspects (tuples, attributes, degree). Only once these tools were available, he was able to handle more explicitely naive set operations which led to creation of new operators that were reflecting his structuralism. Problem is : he had to make a choice because he was starting from very little. The chosen structuralism made it easier to apply his model but it remainced somehow arbitrary. For instance, the structurally attribute based characterization of relation was favored over the relation variable being a tuple or tuple set. Attribute based characterization and stucturalism solved many problems but left many unsolved algebrically speaking: the entire debate triggered is a proof of such limitation of Codd's structuralism. My main focus these years have been to identify some areas of RM where Codd's structuralim of set theory has shown limitations. In such areas, such as *prime domains* or *key domain* (domain of values representing candidate keys) require coming back to fundamental domain definitions and set theory formalism.

In other words, it seems clear that RM goes way beyond its currently adopted structuralism.

> >> <handwave>
> >> (i a)
> >> A relation R with attribute a (written as R(a)) having
> >> a as a handwave into S(b)
> >> is expressed as follows:
>
> >> forall R(a): exists S(b): a = b
>
> >> Note that b need not be a ck to S, hence 'into', not 'to'.
> >> </handwave>
> > Again, attribute-based characterization of domains and relations leads
> > to a conceptual dead end and a circular argumentation about key (As
> > you can see).
>
> I really don't know if I am reading this right.
>
> I don't see circularity in Marshall's FD/ck expression.
> Do you?
Tryinfg to define fd throught attribute based structuralism leads to need for *keyness* definition (Marshall own words) but *keyness* can hardly be defined because of attribute based structuralism...Do you see what I am getting at? Received on Sun Feb 25 2007 - 16:13:02 CET