Re: On Normalisation & the State of Normalisation

From: Derek Asirvadem <derek.asirvadem_at_gmail.com>
Date: Mon, 23 Feb 2015 02:33:39 -0800 (PST)
Message-ID: <f8855e5d-85a5-41d8-b963-3d7ced9d5bfb_at_googlegroups.com>


James

> On Tuesday, 17 February 2015 22:53:24 UTC+11, Derek Asirvadem wrote:
> > On Monday, 16 February 2015 12:08:01 UTC+11, James K. Lowden wrote:
>
> > The Securities table has {ID, Type, Name, Active} ...
> > Type is a discriminator, one of {'equity', 'bond', 'swap', etc.} ...
> > We called this a supertype for ERwin purposes ...
> > Equities, Bonds, Swaps, etc. We called this a subtype in ERwin ...

> > You could eliminate the Type
> > column, and require each subtype have a FK relationship to Securities.
> > But that would not require a Securities row to have a related subtype,
> > and would not prevent one Securities row from being referred to by more
> > than one subtype table (e.g., "being" both Equity and Bond).

What ?

<big snip>

Over the weekend, it became clear to me that your post re Securities was really confused, and my responses (directed at your points) did not address the real problem.

The real problem is two-fold. First, and here is the issue re the different "definitions" that theoreticians use, raising its filthy head, and biting us both on the bum. When you used the terms "supertype" and "subtype", I thought you meant the established industry terms supertype and subtype. Upon reflection, you do not know what those terms mean, precisely. (You may well have a skewed and fragmented understanding of them, from your teachers and their putrescent books.)

That understanding of Subtypes, whatever it is, it is wrong, wrong, wrong. Twisted, crippled, brain-damaged, like the ones who teach it.

More important, the result of that marvellous "teaching" is, that once again as evidenced here, it cripples a perfectly competent implementer such as you, and prevents you from a. understanding the data, Relationally, and b. implementing it Relationally.
Prevents you from implementing a simple Subtype cluster (which we have had since 1970) with full integrity and control in a Relational database (which we have had since 1984). The result is you have a non-relational Record Filing System, with no Referential Integrity. That is their success, to cripple humans in their normal thought processes.

So please throw all that broken, fragmented, information re Subtypes out, and let's start again.

Not knowing how to implement Subtypes, is very much the third issue.



Security Cluster

Before we discuss the Security cluster, we need to get clear on what Subtypes are, ala the real universe (Codd, the RM, IDEF1X, the implementers).



Subtypes

Before we discuss Subtypes, we need to get clear on what relationships are are, ala the real universe (Codd, the RM, IDEF1X, the implementers). This is especially important because the maggot-ridden ones who transform humans into schizophrenics have destroyed the definition of this rather basic building block.

I have cut-pasted different sections from course notes for you (Relationship write-up, then Subtype write-up), sorry if it doesn't flow.

http://www.softwaregems.com.au/Documents/Article/Database/Relational%20Model/Subtype.pdf

There are links that will take you to example code segments.

When you are done with that, please ask any and all questions, re those definitive elements, before diving into ...



Security Cluster

The Security cluster you have is a classic Exclusive Subtype cluster

- for human understanding purposes
- for Relational Data Modelling purposes
--- Logically
--- Physically
- for "ERwin purposes"

Currently it has no Referential Integrity, and all the Files are independent.

Here is what it should be:
http://www.softwaregems.com.au/Documents/Article/Normalisation/Security%20DM.pdf

Fully Relational, full integrity.

No "superkeys", no "distwibuted keys", no double indices that do nothing.



Cause

I hate to harp on it, but this one must not go without specific reference.

> the maggot-ridden ones who transform humans into schizophrenics have destroyed the definition of this rather basic building block.

I am not kidding. Hugh Darwen, TweedleDumb, the grand master of schizophrenics himself, teaches various falsities and absurdities at the asylum for indoctrination at Warwick (which used to be a place if higher education). In his CS253 class, in the How To Handle Missing Information Without Using Nulls document (available online, and at the TTM site), he actually:

  • teaches that exclusive subtypes are ordinary fare, without ever using the technical term, and with stupefying ignorance
  • without teaching how Exclusive Subtypes are modelled, or implemented in the RM, or in SQL
  • teaches a non-relational, Record Filing System method for the cluster
  • includes some (not all) constraints
  • teaches, in addition to the first set of constraints, an implementation of those constraints backwards
  • teaches circular references
  • teaches the record-breaking "distributed keys" (p13), which is he says is "loose", no doubt like his bowels, and
  • does not provide the "constraint" that he says he is trying to implement, not to mention the additional indices, etc, etc
  • teaches an insane method (schizophrenic, covering up for his schizophrenia by doubling up on the "constraints") to implement the rather simple Exclusive Subtypes
  • then cries about the various problems of his abortion
  • falsely claims the problem is SQL (Straw Man)

Now, in your Security implementation briefing, you did not mention that level of insanity, but the you did provide me with some of it. And we know where you got it from. While TweedleDumb's document constitutes a veritable cesspool of knowledge, the Exclusive Subtype anti-implementation being a central part, I am not addressing the whole of it here, or even the central part, but I do want to point out one thing. As a matter of course, this cancer-causing agent teaches that a relationship is established by implementing TWO Foreign Keys, one each on both "sides". Which of course, causes a circular reference quite unnecessarily; contradicts normal human logic; as well as SQL; as well as breaching the Relational Model.

Each and every time. Teaches humans to be as retarded as he is. The state of "higher education" today.

He is setting the stage such that everyone who follows him has a circular reference in every relationship, and thus has a demand for "deferred constraint checking". Making sure everyone has cancer, and needs his "remedy".

> > Wicked

Yes. Evil. Cursed by God. Crying out to heaven for vengeance.

I don't know what your religion is, but we are required to do justice where there is iniquity. And we are not required to be nice to maggots that are feeding on us.

Cheers
Derek Received on Mon Feb 23 2015 - 11:33:39 CET

Original text of this message