Re: On Normalisation & the State of Normalisation

On Tuesday, 3 February 2015 19:13:01 UTC+11, Erwin wrote:
> "The developer declares that the proposed extension satisfies 5NF. Is that correct ?"
>
> (a) as is usually the case when unskilled people present a drawing of boxes and ask whether this satisfies xNF, the question isn't answerable because the dependencies haven't been stated.
>
> (b) Assuming certain external predicates for the relvars that would at the logical level correspond to the boxes in your conceptual model, and also assuming a logical structure for the database that is kinda one-to-one with your drawing of boxes and lines, and also assuming the dependencies are as real life has them, in Belgium, the answer is no.

???

1. There is a legend at the bottom of the page. Please read. If the dependencies are still not clear, please post again, and I will spell out all the little squiggles and notches.

b.1. As stated, and as you seem to understand, it is the developer's model, not mine. If I erect a model like that Heaven will strike me dead !!!

b..2. It is not a conceptual model.

b.3. It is a logical model, minus the datatypes, which are irrelevant to the purpose of this thread.

b.4. No idea why you think the boxes are "kinda" one-for-one. The point of a model, at the logical level is one-for-one, at the logical level. When moving to the physical, which, again, is not relevant to this thread "Normalisation" of data, please be advised that the only change to the one-for-one is that Associative Tables are added to resolve each n:m relationship. They are none such in this model, so it is one-for-one at the physical level as well.

b.5. "Relvars" scare, and confuse, developers. I had previously assured him that they are nothing to be concerned about, because it is a trick that theoreticians use to avoid definition in the physical universe, so he did not use it in our conversations. Neither he nor I have seen "relvars" mentioned in the texts defining the normal forms or Normalisation. Neither he nor I give a rats about them. If your point is relevant to Normalisation, please explain what you mean in those terms.

b.6. The predicates are in the dependencies, _read_ as a model is _read_. Refer my response (a). If you would like assistance reading the model, let me know. It is IDEF1X, the standard for modelling Relational Databases, that we have been using in the implementation universe, since 1985. A pleasant coincidence: the best product that provides it, is called ERwin (Entity Relation modelling for win-doze), your namesake! But of course, anyone with a drawing program can construct a good diagram, they will be understood as long as they use the standard notation. Only in the implementation universe, of course.

b.7. The existential predicates are usually not stated nowadays, I am happy to provide them for you at your request. I was hoping to focus on the Normalisation issues.

b.7. He apologises for the relation names instead of proper Verb Phrases, but I thought they were pretty obvious, and allowed him to get away with it. I shall demand a correction in his next iteration.

I would think, once you have been able to _read_ the model; to identify the dependencies conveyed therein; and thus notice the predicates are explicit, your categorical response might change. Therefore I will not address that at this stage.

Cheers
Derek Received on Tue Feb 03 2015 - 11:05:01 CET