Re: Why are [Database] Mathematicians Crippled ?

Jan

> On Wednesday, 4 February 2015 20:50:15 UTC+11, Jan Hidders wrote:
> Op dinsdag 3 februari 2015 14:46:14 UTC+1 schreef Derek Asirvadem:
> > Jan
> >
> > > > We are progressing very slowly.
> > > >
> > > > Now I realise, you probably did not understand my first post about this video.
> > > >
> > 1. So you have the Codd definition, right ? Predates the donkey by decades; well-understood and accepted in the physical universe, etc. The 99%. Again, the unchanging law is, since you are not aware of it, since 1971, Codd's definition of Third Normal Form.
> > ____ "Every non-key attribute is Functionally Dependent on the Key, the whole Key, and nothing but the Key"
>
> Roughly, yes. This formulation is good as mnemonic device, but should not be confused with the real definition since it has several technical problems, But for the current discussion I think it's close enough.

The difference is definitions is severely hindering progress in any discussion with you. I repeat, this does not happen in other industries; it is unacceptable for theoreticians to have definitions that are different to the implementers, that pre-date the new "definitions" that the theoreticians have; it is fraud to call some isolated fragment of X, "X".

There is only one definition, for 3NF (Codd's 1971, quoted above).

I do not accept the theoreticians "definition" for 3NF. I have detailed that in the Normalisation thread. But you are free to mess with that non-definition. It places you in a position of capability far less that the implementers, as proved, and as will continue to be proved.

So the onus is on the theoreticians to use honest labels for their fragments. To label a fraction of X "X", is not only fraud, it means you are not working in this space, you are working in outer space, and as evidenced in many posts, you can't communicate with the people who you allege to be serving. We sit here and argue definitions and whores dressed up as virgins, and never get to dealing with the problem on the table.

Therefore (since Codd 1971 predated your fragment), and it is widely accepted and used, it is crazy to label your new kid "the real definition". you crippled kid is a fragment of the real thing.

Whether it has technical problems or not is another matter. If and when the definition of 3NF *does not work*, then we might expose the problems you allege. Until then, it is an unproved claim. And thus far, it has not been exposed in this thread. SO please stop wandering off into tangents.

> > 2. You have the nutty professor, who gives:
> > "<trigger fear response>
> > <trigger self-importance>
> > <trigger future security>
> > Given a relation R{x-instance, y-instance}
> > if they (whatever "they" means) agree on the x-attributes, then they /must/ agree on the y-attributes
> > if that is the case then we say x functionally determines y"
> >
> > Do I have that correct ?
>
> He doesn't restrict restrict the relation to the X attributes and Y attributes, but yes, his definition is that it holds for the rows in the relation that if they agree on the X attributes (the attributes in the set X) then they agree on the Y attributes.
>
> > Good.
>
> Yes. :-)
>
> > 3. So, using all your powers as a human, do you perceive any difference between [1] and [2] ?
>
> Er .. well, yes. One is the definition of the concept of 3NF and the other of the concept of functional dependency. They *should* be different. So I'm not sure what you are trying to say here. I assumed you were busy showing that the same concepts are differently defined by Codd and Boetticher.

Oh God.

[1] is the definition of 3NF. (NB. not "the concept of")

[1] is the definition of Functional Dependency.

Yes, it gives both. I didn't know that you didn't know that. So much for theoreticians in this space.

[2] is the fragment of the definition for FD, that the theoreticians use. It is not the definition for FDs, which we have had since 1971. So you, not Codd, not I, have to use a different label. Otherwise you commit fraud and cripple the communication process.

Since we seem to have accepted "Normalisation" vs "Normalisation Theory" in that thread, which is moving slightly less slowly than this thread, since yours came decades afterwards, since yours is a fragment, I suggest labelling it "the Theoretical Functional Dependency Fragment" TFDF. We (those of who know, love, and use The Definition, the 99%, do not accept that a fragment is a definition, so you can't use the word "definition" in the name.

(Note also the veracity of [1]. Your fractured, fragmented non-definitions cause a primitive Record Filing system to be passed off as "Relational" (5NF means it "satisfies" your TFDF). But in fact, it fails 3NF (separate to failing the RM requirements), so I rejected the disgusting data model that you passed. So whatever it is that you do with your various fragments, is farr less than than the capability of an implementer who has not been infected with them.)

In the hope of a bit of progress, let me restate (not change) question 3:
> > 3. So, using all your powers as a human, do you perceive any difference between the FD Definition contained in the 3NF Definition, and TFDF ?

My point being, since I opened this sub-thread, that under the label "Functional Dependency" for the unit, the nutty professor is teaching some disgusting fragment of FDs.

(In other units, he teaches how to dance with fragments, how to kiss and cuddle them, but this video was chosen by my colleagues because it is where he "defines" the fragment. It was along the lines of "look at what crap they are teaching the young at university level these days, no wonder our intake grads are incompetent", that the link was sent to me.)

And he is not even giving the TFDF correctly, which I both identify [4][6], and correct [7]. I am working streets ahead of you. I want to deal with the issue that TFDF is substantially different to FDs, and that is one of the reasons theoreticians in this space are crippled.

Cheers
Derek Received on Sat Feb 07 2015 - 01:34:24 CET