You have many questions. It cannot be answered right away, if I am to give it the treatment that you expect. I will do so over time. I may open new threads, to keep each subject separate.
Post by y***@gmail.comit's a great honor to receive a reply from you.
Thank you. I am not particularly important in this venture. First the truth is. For that, and for the discernment of what is not, we have Aristotle.
Second, Dr E F Codd ... Sure, over the decades, due to purist practice and the elimination of the filth that others promote as the "relational model", I have progressed his work, by the Grace of God.
I stand on the shoulders of giants.
Post by y***@gmail.comPost by Derek Ignatius AsirvademThat is a big ask, and it will take years. It is best done with a mentor, and a real world project. Please feel free to open a thread with an example.
Sure! I think only by practice and exercise it is possible to improve, not by wasting time on cumbersome, redundant "academic" and "theoretical" only book/material with zero application in real life.
To be clear.
- I am a theoretician (but not an academic), an engineer by training
- Practice without a foundation of solid theory is for people who live in mud huts or in sewers
- Theory is thus essential for any practical endeavour to succeed ... but in the last 100 years, and particularly after we lost the war, it has been perverted into something quite insane
- I dismiss theory that does not have a practical purpose, as fantasy. Good for masturbation, but one should keep it to oneself. Writing papers about it, is porn, attractive to similarly perverted minds ... but as you know imposed on the young, indoctrination under the guise of "education".
- There is a particularly evil set of "theory" in this science, database design, in that it is intended to suppress real theory and retard progress of the science. Thus the Relational Model is suppressed, and the pre-relational Record ID based Filing Systems are promoted and marketed as "relational". Thus that set of theories is in fact anti-theory.
So the correct sequence in this era of perverted science is:
- Study the theory
--- determining and eliminating anti-theory
--- ignoring theory that has no practical purpose
- then implement the theory faithfully
--- noting all issues, and the requirements to overcome them
- Modulate the theory
Post by y***@gmail.comPost by Derek Ignatius AsirvademI am in the process of writing one, but it has to be expanded to overcome the mountain of pig poop that is established as the "science".
Best news! Whenever you are done, please update us. A practical book (in the style of your StackOverflow replies - comprehensive, well explained, step by step etc) is VERY needed.
Yes. Far more explanations than on SO. Note that my Answers on SO are limited to the question, and to what can be explained in that medium. Eg. it does not work to tell every person that Record IDs are wrong, wrong, wrong, I cannot go into the whys and wherefores, I just give them the data model that they need. Whereas in the book the explanations are complete. It is actually a course.
Post by y***@gmail.comPlease also include all the various phases of modeling, maybe accompanied by a real world example.
Yes. I use three full-blown Case Studies, throughout the course.
- a simple data model [43 tables] for all the basics
- a moderate DM [62 tables] for all the different types of constraints; etc
- and a complex DM [83 tables], including scheduling and resource matching
These are from the courses that I have taught over the decades, and thus quite mature. The second and third Case Studies are models that solve problems that the pig-poop eating freaks cannot solve, and precisely. ( I have challenged the freaks to solve such problems in this forum, but they do not. )
Post by y***@gmail.com1. Project initiation
2. Entity definition
3. Relationship definition
4. Key definitions
5. Attribute definition
I'm pretty sure you indeed follow there phases since IDEF1X is the standard for modelling Relational Databases
Yes and no. Of course I followed IDEF1X for decades, but the exercise requires much more than just a standard for the *notation*. It is not a Method, and a standard, as such, cannot be expected to provide the Method. Eg. the pig poop Gulag screams and wails that Codd did not specify this or Codd failed to specify prohibitions for that. All quite idiotic, and proof in and of itself that the freaks suckle sows at night, that they are total failures as "theoreticians". Because any genuine academic knows that a paper only defines the new material. Codd cannot be expected to define database design, only to define the RM as applied to database design.
I give the Method. It is those 5 points in the context of 30 years maturation, which becomes an hierarchy, not just 5 points, more like 20 points in a proper arrangement.
I also enforce the RM, which IDEF1X cannot do. I have these published (paying customers) as Extensions, but given that the OO/ORM freaks have diluted IDEF1X, really, it is fit for a new Standard.
Post by y***@gmail.com(and not UML like I was taught at University and 95% of StackOverflow believes)
1. UML is broken all on its own, without regard to how or where or upon what it is used. One single symbol for every **thing** (object or not), with 42 notations, which are never enough. So each modeller adds a notation of their own, in order to document what he has designed, the result being one symbol for everything with 400 notations, that no one understands. It is a study in absurdity.
2. It fails to differentiate between Control and Data, each of which have separate sciences for implementation.
//
The nerves are physically separate from the muscles that they control. Only a lunatic (as "taught" by the "theoreticians") builds a transport for material, and then places the control for that transport, *IN* the transport ... it must, needs be, separate.
//
So the imbeciles treat Data (tables) as Objects. And they remain clueless about the richness of (a) how to design Data, (b) how Data is defined, (c) in the context of the Relational Model.
3. The main reason UML is totally useless, for both Data and Objects, for any sort of Analysis of a system (before the fact of design), for design of the solution, and for documentation of a system after implementation, is this. It has no facility for decomposition, which is essential to understanding a system, both before; during; and after design.
Thus whatever one creates using UML (the OO or OO/ORM mindset) is a complex flat design, with no structure. A non-architecture. The application is one great big single stack, with no structure, a monolith, the very opposite of Architecture.
http://geek-and-poke.com/geekandpoke/2013/7/13/foodprints
Use:
a. IDEF1X for data modelling (plus my Extensions for greater precision and compliance with the RM)
b. SSADM for Process Modelling (full decomposition) for moderately complex systems. Easy to understand for both non-technical and technical people
c. IDEF0 for Process Modelling (full decomposition) for complex systems. Non-technical people need a bit of education.
d. UML for Object definition only (the idiotic "inheritance" as elevated to deity), but not relationships; dependency; hierarchy; meaning; how-does-this-fit-into-the-architecture; etc, which is formally defined [a][b][c].
Post by y***@gmail.com1. I have noticed you were a great supporter of 5NF/6NF until 2015 circa. Your thought then changed in favour of 3NF. Is this still the case in mid 2020? I think Codd only defined up to 3NF, and any higher normal form is complete "academic" deceive, fraud, trick or rebrand of the original normal forms by DR. E. F. Codd i.e. to my understanding, 3NF is the fullest form of the Only and One original Relational Model defined by Codd, is that correct?
Good that you picked that up.
The history is this. I went from being a senior engineer with a pre-relational platform supplier, into consulting for Relational, with their high-end customers. There were no textbooks or books, just the RM. Thus I managed to stay away from the degradation in the indoctrination system, and the filth that all the textbooks provide. When I came onto the low-end public scene, I was shocked to find that the RM had been perverted, by the very people who were supposed to be progressing it after Codd left the scene. At first I engaged with them, innocently, because I thought they were the authority, and I could correct them here or there. After three years of hard labour at the Date & Darwen Gulag, aka /The Third Manifesto/, aka /Tutorial D/, I formed the conclusion that they were actively suppressing the RM, and running an asylum, in which people engaged in endless argument without resolution. Eg. 30 years and still no replacement for SQL, which they propose is broken, but which is not, it is the perfect data sublanguage for the RM, which Codd defined.
Thus up to a certain point, I accepted that 4NF/5NF/6NF were valid, if only for fools who did not understand the RM, and who were implementing Record ID based filing systems. Note that there are many [17 thus far ?] additional abnormal "normal forms". And after that certain point I concluded that they were wrong because they were intended to KEEP people in the RFS, and avoid the real RM.
You may notice that the freaks have even changed 1NF, to suit their stupid shotgun-in-the-dark approach to temporal data. Truth does not change, if it does, it is not truth. The original 1NF stands, the hell with Date & Darwen's changing "1NF".
So the bottom line for Codd's 3NF is, if understood properly, it eliminates the need for 4NF/5NF, which are only needed for anti-relational systems.
6NF is simply not even an NF, with no definition, no dependency (eg. does it have to be in 3NF first ?). The freaks promote it as the "ultimate NF" (a non-NF is their ultimate NF, like their best woman to spend the night with is a non-woman, a hairy sow lying in pig-swill). Nevertheless, it is important to understand it correctly in the context of Codd's 3NF, whence it is a minor Method, required for a table that requires Pivoting, etc. Again, unworthy of a "NF" categorisation.
Note also that there are *additional* and real NFs in the RM, which the freaks suppress, and never mention.
Codd's 3NF, properly understood, and the Relational Model, properly implemented, actually provides full constraints of all kinds. Not merely DKNF, which even the author (whose head is stuck in the rear crevices of a lactating sow) states as being impossible to achieve. I regularly achieve, via science, that which is impossible by non-science. Full Domain; full Relational Key; and full Other Constraint, Normal Form. That was obviously Codd's intent, as per his RM, thus I do not declare it as separate, but an ordinary progression of his 3NF.
For these and other reasons, I have a complete set of NFs, the definitions of which never change, and that apply across the entire database (the NFs and abnormal "NFs" apply to fragments: single files or tables). No, I will not publish them without the book (that explains the elimination of false NFs and lays out the real NFs). Yes, you can email me directly and I might provide a preview.
Post by y***@gmail.com2. Could you please explain a bit more the Open Architecture standard from 1984 perfected in 1990 from a non-hardware point of view? (http://www.softwaregems.com.au/Documents/Article/Application%20Architecture/Open%20Architecture.pdf)
I shall open a new thread for that.
Post by y***@gmail.comI literally found zero results on Google.
You expect to learn something of value from Google ? From the ever-changing cesspool ? That the uneducated masses excrete on a daily basis ? That is biased towards imbeciles and those who submit to the propaganda machine ? That is heavily maintained by those who push the intellectual enslavement agenda ?
It is a great place to look, if you want (a) confusion, (b) definitions that change as often as underwear in a brothel, and (c) guaranteed sub-standard answers from idiots who are clueless that they are ignorant, or from purposeful saboteurs, such as Date; Darwen; Fagin; and their minions.
As confirmed by the fact that you come to me, not Google, for the one and only authoritative answer.
Cheers
Derek