Re: Hierarchical Model and its Relevance in the Relational Model
Date: Fri, 30 Jan 2015 22:53:36 -0800 (PST)
Message-ID: <62d243c9-7cab-4dc2-b36b-f26998bf6cd9_at_googlegroups.com>
Jan
> On Thursday, 29 January 2015 22:39:37 UTC+11, Jan Hidders wrote:
> On Friday, 30 January 2015 22:35:51 UTC+11, Jan Hidders wrote:
> On Friday, 30 January 2015 23:03:48 UTC+11, Jan Hidders wrote:
> On Saturday, 31 January 2015 02:12:27 UTC+11, Jan Hidders wrote:
> > Op vrijdag 30 januari 2015 12:29:48 UTC+1 schreef Derek Asirvadem:
I accept that you two may be in "diametrically opposite" positions. But I had placed you two in the same basket wrt the subject of the thread. I will answer your each of your posts separately, name at the top. But if I answer them /fully/ I will end up with duplication, which is onerous for me, and offensive to the readership. Therefore, please read all my responses, noting that it applies directly to the named person (answering an identified post), but applies to both of you in general.
> > A. ____the Hierarchical Model rests on a theoretical void____
> > B. ____the HM is dead, it has no relevance wrt the Relational Model___
So what. There are no papers re the Network model either. So what. It was all proprietary in those days. It is silly to expect papers for such.
There are no papers for what Sybase has been doing in the last twenty years, re (eg) caching data buffers in their server. So what. Any Sybase professional knows most of the content fairly well. The absence of papers is not a proof that the server does not cache buffers, or that the caching does or does not do something, or that it does use this or that algorithm. The fact that the buffers are cached, that a specific algorithm is operating, and that the algorithm is more advanced than the 2006 algorithm, is easily proved.
Unless one denies science, evidenced reality, one does not need a paper.
I did not say that.
I said that:
- the 99% THEN understood it perfectly - no idea what the 1% did THEN - that in the forty five years since, the 1% have written about 100 papers that deny various parts of the RM, hierarchies being one of them - that those papers (specifically those authors) are responsible for the diminution of the RM, in general - specifically, re this thread, that hierarchies and the RM are mutually eclusive - that hierarchies do not exist in the RM - that a very important part of the data and referential integrity in the RM, that is dependent on hierarchies, is lost when those hierarchies are ignored or suppressed(you appear to understand that integrity is contained in the RM, but only the non-hierarchical part of it) - So I am saying, NOW, based on the evidence, the 1% does not understand the RM, and specifically, does not understand that Hierachies, the HM, exists within the RM. - all that results in a widely-understood notion among the 99% NOW, of an RM that is legless, fragmented, fractional, devoid of its core principles
That the evidence is, the papers, those authors, have little knowledge of the RM. To wit, it WAS widely understood, but as a result of the mathematical papers, it is now NOT widely understood. It is starting to look like your understanding of the RM is based on papers that preceded yours, and not on the RM itself. Which categorically means, you should not be making statements about the RM.
They are not in a position to agree or disagree. Refer my comments above re proprietary systems.
That is my point, agreed. It is silly to argue that the HM is not known, or that the model doesn't exist as a model, etc.
(This does appear to contradict your initial points, Concerning A.)
Sure. And there was THEN a well-established implementation foundation, evidence that it worked well, really well, for what it was designed to do: DSS (inferential systems). (The market was moving and it had new demands, namely from Batch TP to OLTP, and the HM did not work well for those new demands. Network came along with instantaneous OLTP, and started wiping the floor with IMS. That did not mean that OLTP lives and HM died, it was not an exclusive issue. IBM, their customers, and others wanted OLTP /and/ DSS, in the one data bank.
Therefore this:
is false for another reason. Anyone in academia or the industry at the time, who did not know the basic issues; the "philosophy"; the implementation; the reality; the market demands, is simply not credible, he is passing himself off as someone knowledge-able in a subject while providing evidence that he is clueless.
That is not accurate. Codd defined a specific Relational algebra using first-order logic, for the purpose concerned. There is no suggestion that the RM contains the whole of FOL, or that the RA contain the whole of any RA that existed at that time.
Come on, Jan, higher-order logic does not apply. Please stay on the point.
Good. I am very happy that that is clear to you, that that is your position. I hope and pray that you have a similar position on the RM.
No.
> then, yes, that is pretty much dead.
Agreed. In 1984.
> But most now understand the relevance of data independence.
I suppose I have to trust that you mean that in the fullness of the data integrity as prescribed in the RM.
Er ...
- I have already stated in the Topological thread:
2. I have already stated in this thread:
> > Hah! you have to be kidding ...
> > Please provide a single link to a paper written by a mathematician in the RDB space, that asserts that.
> > For each that you do provide, I will provide one that asserts the opposite. I would have thought that you are familiar with some of them.
> > For the few that do assert database independence, note that it has taken them forty five years to reverse their position. Practitioners who followed Codd, have never left that position.
> > That is precisely part of the problem I am saying, that contaminated and damages the RDB space.
> > And then invent tiny fragments of integrity, to patch up a few of the gaping holes, what they have lost.
>
> I have not idea what you are referring to here.
Let's cut to the chase.
One of those papers I refer to in [1] is yours.
Actually, one of the many papers I would produce as evidence that mathematicians in this space assert the opposite [2], could be the same paper.
And quickly skimming it again, it contains one perfect example for [3].
It contains many of the issues that I raise and wish to address in this thread:
a. the absence of understanding data, as data, independently, including that all constraints on data should be located in the RDb b. the absence of understanding (perceiving ? observing? recognising ?) that the data is an hierarchy, which obviously is a result of the suppression of hierarchies that are contained in the RM c. the proposition that, while data integrity as implemented in the program space has failed, yet more complexity and programming in the program space will solve the problem, complete with a mathematical proof to that effect d. ignorance [of the relevant parts] of the RM, although it is referenced e. ignorance of the fact that the RM handles hierarchies beautifully f. ignorance of the fact that if the data were implemented as per the RM, the problem would not exist, therefore the proposal is totally without merit g. that your statement now, re the value of data independence is in direct contradiction to papers that you have written in the past h. re [f], I am not suggesting that you have based the paper on a Straw Man concept, but on ignorance, of the RM, such as mathematicians in this space commonly exhibit. A point that I assert in this thread. So it serves as a proof of that, too.
Of course, the great danger, the evidenced damage to the industry, is that people use such papers, believing the propositions to be true, the proofs to be valid, and they implement systems on that basis. People like me know that they were false from the beginning, that data independence, etc are irrefutable facts, a science, since 1970, clearly demonstrated in systems since 1985. However, people who implemented systems based on such papers don't know that, they are influenced to implement systems that contradict science. And they keep implementing them, because the papers have not been retracted, all we have is a statement from the author in an unrelated post on c_d_t stating that "most /now/ understand the relevance of data independence". Uhuh, twenty years later (1995 to 2015), after twenty years of causing damage, propositions and proofs that were made in denial of the RM; science, that existed in the previous twenty five years (1970 to 1995).
Therefore we can use that paper, which is still being used today, to deal with many of the issues raised in this thread, to prove (evidentially, not by mathematical proof) many of my declarations in this thread. As well as giving you an example of what I mean *the head of this section).
Rather than diving into it, I would like to ask your permission to do so. Of course, your credibility will be diminished if you do not accede, but that is a consequence of your actions, historically, and now. It is not a consequence of my actions.
> > 5.
> > > > [Rule: Circular references are not permitted]
> > >
> > > The German page is more explicit and does mention this rule.
> > >
> > > > Which appears to be a perfectly reasonable rule to me ...
> > > > Is it perfectly evident that this requirement must be enforced, since a
> > > > model with cyclic dependencies is plain "spaghetti", maybe even
> > > > violating some normal form?
> > >
> > > It has nothing to do with normalization in the classical relational
> > > theory sense of the word.
> >
> > I declare, it has a direct and definitive relationship with Normalisation, as defined in the RM.
>
> Really? Which of the classical normal forms (1NF, 2NF, 3NF, BCNF, 4NF, 5NF [in its different variants you find in text books]) forbids cyclic dependencies?
Please stick to the statements made, please read again. I did not refer to the NFs (which I have severally stated are abnormal; tiny fragments of Normalisation), I have specifically stated "Normalisation, as defined in the RM". I am under no obligation to defend statements that I did not make.
What do you have to say in response my statement, in the posted context, refuting your statement, without reference to the NFs ?
But of course, you raise an important point, which I will not avoid. And it is a great point because it drags us back to the scope of this thread, my initial post:
> > The first [obstacle] is, mathematicians are scared of The Hierarchy. And they are in a state of denial re the evidenced fact, that the Relational Model is a progression of, not a substitute of, the Hierarchical Model. Note the comments of J Hidders and J K Lowden in recent threads. Of the two Normal Forms defined in the RM (which btw remain undefined by mathematicians after 44 years), the first is, if we were to name it without psychological impediments, the
> ____ Hierarchical Normal Form____
> - It destroys many of the problems that mathematicians, even today, are grappling with.
> - It deems many of the proposals of Date, Fagin, and Darwen *non-relational*, which is why they suppress the Hierarchical issue, in order that people won't identify their proposals as massive breaches of the RM.
The Hierarchical Normal Form, as defined in the RM, forbids cyclic dependencies. It demands a tree without circular references.
The definition in the RM is informal, but scientific, and easily understood by the 99%.
Since the mathematicians in this space have failed, miserably, to (a) understand the RM, and (b) to provide /formal/ definitions for the two Normal Forms in the RM, they may not be known to the 1%. Which again, is additional evidence that the 1% are clueless about the RM.
Separately, anyone who is programming in any language, or designing databases in any platform, who does not know that cyclic dependencies and references are wrong, wrong, wrong, proves themselves to be an un-scientific ignoramus, executing work in denial of science. Because science (many proofs in other, related sciences) tells us that. Because anyone working with an hierarchy (recognised as an hierarchy) knows that: what would happen if a directory structure, such as James mentioned, were to have a circular reference ?
Circular references are invalid, period. Everywhere. Constraints. Programs. Database definitions. Data. Newspapers. Electricity grids. Puzzles, mazes. Marital bliss. Gender "orientation". Transit systems. Any set of manuals. Hospital porcedures. In any system (using the word in the loosest sense) of related articles. The list is endless, and only necessary if one does not understand that Circular references are invalid, period.
It is not invalid /only/ because they are invalid in the HM, or /only/ because Codd defined them as invalid in the RM. It is a law of nature, established in every science (except the 1% in our field).
Unless one is in a state of denial re evidenced reality (schizophrenic), *and* ignorant of other sciences, one does not need a mathematical proof to prove that circular references are invalid, plain stupid. And the corollary is, in order to accept circular references, one must drink the Kool-Aid.
Now it must be said, separate to the fact that they are prohibited by the RM, in the mathematical desert that is the database space, idiots and schizophrenics write papers, with mathematical proofs, which deny reality and other sciences. The hilariously side-splitting evidenced fact is, there are papers here that provide mathematical proofs that circular references are "valid". (Here the mathematical proof which is final to mathematics, and its non-finality to scientists, is revealed. But let us not get distracted.) And the same freaks provide examples using circular references at every opportunity, in order to re-inforce the totally invalid proposal, the daily does of Kool-Aid.
Of course, I am rather attached to science, and I expose such nonsense as being in denial of science, of reality (hence schizophrenic), and I dismiss all of it. But again, many people, also ignorant of science, believe that putrid garbage to be "valid", "proved", and they implement systems. Another example of the cancer in the industry, and the cancer-caused agents are easily identified. I would shoot the lot of them.
To finish this point, it must be said, this particular cancer is so wide-spread, that not only implementers implement it, but DBMS implementers implement DBMSs with it built in. PostgresNonSql and others are a sick joke, they legalise the insanity ("deferred constraint checking" which is only required if the cancer of circular references is deemed valid), they support schizophrenia and thus normalise it.
That is why DB2 and Sybase do not have support for "deferred constraint checking", for decades. We do not need it. Only idiots, who listen to and believe schizophrenics, need it. And their "systems", their "databases", are massively inefficient and problematic.
> > 4.
> > The HM is a fundamental part of the RM. (To be clear, "fundamental" means "founded on".) Removing the HM from the RM would remove much of the integrity, power, and speed from the RM.
>
> As you know I agree with that to some extent, but not completely. So I wonder what arguments and observation you have to offer to support that position.
No. It is the other way around. You made statements that the HM has nothing to do with the RM, etc, etc. Contrary to the standing evidence (the RM). So it is up to you to produce evidence of that which contradicts established science. That route, addressing the points you might raise, would be the smallest piece of work.
In the event that you can't do so, sure, I will provide chapter and verse from the RM. Chapter and verse is required, because the HM is unknown these days, whereas the HM was the only model in the 70's and 80's, and very well know. It provides the historical context in which the RM came into existence (the RM was not written in isolation, in a vacuum, or on the fourth moon of Jupiter). Some terms in the RM need exposition, if it to be understood today. That is a larger piece of work.
Third, again not avoiding your request, if we proceed with dealing with your paper, this issue [4] would be addressed and closed fully. So I await your permission.
> > I love my profession.
> >
> > It is easy for me to help others, either furthering their knowledge or implementing something. It is easy for me to correct mistakes in my professional and to protect it from damage. That is the spirit and intent with which I started this thread. I am hoping that you love your profession, at least half as much as I do.
>
> Yes, I do, and recognise this is in you.
Thank you.
You misunderstand me. I am not here for debates. I am here for resolution.
It is not a conversation, it is a demand for evidence. Or "why are you writing this, in denial of established science".
If you consider my direct communication style "combative", then I am very sorry. And please observe that I did not create the combat. The person writing papers that are in denial of science did, my demand is merely the consequence.
> > I hope I do not have to explain why, how, in any profession, and to humanity itself, falsity is damaging. A lot of unnecessary argument can be eliminated if we stick to the truth. There is no such thing as private definitions or private truths, [they] exist only for people who are severely isolated from society, people who do not have an authority. In the professions, we have authorities, standards, laws. If we observed them, there would be no conflict within each profession.
>
> That's a bit too optimistic for me. :-) But, sure, commonly agreed definitions help and lack of them can make it hard to have meaningful discussions. But what can be fare more damaging is the unwillingness to listen to the other side, even if their definitions are not exactly yours.
I am not interested in any definitions other than the scientific ones that are established. There is no point in listening to someone who has a different definition, because the basis of such, is a rebellion against science, against established definitions.
Practitioners couldn't care less about the debates re the ever-changing definitions that mathematicians seem to enjoy. At best (the most politeness I can muster), they are irrelevant, precisely because they are not fixed, and cannot be relied upon to build anything. That is not to say that we do not allow theorists to do engage in it, but even then, the relevance depends on whether it is scientific or not.
I don't know what the slang term is in Holland. The German one is incisive, but too rude to print. The Australian one for that sort of exercise is, mental masturbation. Getting others involved in the conversation makes it mutual mental masturbation.
> > Which leaves us with conflict between professions, as we have here. This skirmish is about mathematicians published falsity about my profession, causing damage to it (as well as to themselves, refer my comments re Norbert's thread).
>
> Which falsities would that be?
The list is endless. We have already started dealing with a few specific falsities, they will progress and close (as decidedly false) as the thread progresses. Eg. your paper. I trust that the short list on the table suffices for now.
> > This does not happen when I deal with the banking industry or the car manufacturing industry: both sides are well aware of the sciences involved. It only happens with mathematicians who declare themselves to be the theorists in my industry, the database implementation profession.
>
> When you say database implementation, do you mean indeed database implementation or DBMS implementation?
Both, and separately.
Jan, you do make me laugh.
It is not the modesty in the paper that is the operative issue. It is that the papers, with proofs, devoid of science, get used, to implement falsity, the result of which is garbage. And twenty years later, when the mathematicians finally accept that the said proofs are garbage, they do not have the professionalism to retract them, they just carry on with new proofs.
In other cases, eg. MVCC, they deny evidenced reality, that there previous proofs were wrong, and write new proofs about a new Wonderland. Vociferous denial.
Third, and this is probably the most important, these same freaks write books, and propagate those false theories, those cancers, to the masses. That is the wilful bombardment of previously undamaged humanity with disease.
Heaps. Let's deal with the ones identified in this post thus far. And again, the specific ones in your paper.
> > Therefore, I requested that any discussion that may be had, be limited to the scientific realm.
>
> That's not so easy, since most issues at hand are actually engineering questions rather then scientific questions.
I do not split those hairs.
Thank you. And in that case, it /should/ be easy.
> On Saturday, 31 January 2015 00:23:48 UTC+11, Derek Asirvadem wrote:
> > You also seem to imply that there is no theoretical research on the Nested Relational Model. I know form direct experience that this is false, and have no idea why you would think that.
>
> I have read five papers. Two are very good. Three are very poor.
>
> The two that are good, are ignorant of hierarchies, and they re-invent the wheel, in the front of the cart, and sideways.
>
> The great problem with ALL the mathematical papers these days is that they are written is staggering ignorant of other sciences; with a narrow focus on their tiny area; in ignorance of the real world, where the thing that they are "researching" already exists, and can be readily observed.
Ie. ALL mathematical papers in the RDB space. The car industry; manufacturing; computer hardware, industries have no such problem. The software industry suffers a fair amount, and within that, the database industry is the worst, pathetic.
To be clear, from where I sit, there are three easily identified categories of mathematicians or theoreticians.
- Knowledge Science, theory (including mathematics, which is a fraction) and implementation, that is based on all the sciences that apply; all the theories that are scientific; all the established rules in the field (eg. the RM).
- Ignorance Theory (including mathematics, which is a fraction) and implementation, that is based on one science, in ignorance of related sciences (eg. laws of physics, architectural principles); ignorance of the established rules in the field (eg. the RM, software architecture); and of the reality (current state) in the field. Un-scientific.
Here, the mathematical theory is asserting itself over everything else. Dangerous and stupid.
One of the two good papers (according to my informal review) is the Wok paper on Nested Normal forms. It is clearly in [2]. To some extent, that is forgive-able, and I have forgiven them, as described above, because the field is damaged; they have been taught cancer; they are not creating it; they are unaware that they have cancer; they are spreading it out of ignorance.
3. Cancer Causing Agent
Theory (limited to the mathematical fraction) and implementation, in violation of the science in the field;in violation of related sciences; in violation of the established rules in the field; and of the reality in the field. This is Anti-scientific. A state of denial, and if they keep doing it after it is pointed out to them it is a pathological state of denial (which is why I use the term schizophrenic). These are the causative agents.
Here, the isolated mathematical proof is god, untouchable. New private definitions are welcomed with glee. Whether the proof is true or false, whether it violates the sciences is suppressed. The only way to do that is to suppress the other sciences, and to demand that all should think like these demented mathematicians (that is the point where you guys triggered me to confront you). That is the Kool-aid: in short, accept isolated and ignorant proofs, and deny the sciences.
Date, Darwe, Fagin, Fowler, Ambler, OMG, Abitebuol, Hull, Vianu, *all* fit squarely into category [3].
And as detailed further above, to the extent that they write books, or teach in universities, they are bombarding us with cancer.
> > So here is a small task ...
>
> Usually it's me who's handing out home work. :-)
It isn't homework.
It is a challenge to prove the extent of your understanding of the RM, re declaration [F], which based on the evidence that I have detailed, I am calling into question. It will also, secondly, deal with and close other issues that have been raised in this thread.
It has nothing to do with nested relations. It has everything to do with whether hierarchies, the HM, exists within the RM or not. And that if you are unaware that it does, then the tables that you create will have less than integrity than an implementation that is not in denial of hierarchies in the RM. Ie. my initial post.
I will give you another chance to deal with the challenge or decline.
I don't mind, either way. If you decline, you will be declining an opportunity to assert your knowledge per stated. In which case, you have no business telling anyone anything about the RM or the HM or about Hierarchies.
Understood.
> > > As you know I agree with that to some extent, but not completely. So I wonder what arguments and observation you have to offer to support that position. And to avoid misunderstandings: I'm interpreting your position as that you think that the Nested Relational Model will lead to more effective DBMSs then the Flat Relational Model. Would that be fair?
> >
> > [.. snip ..]
> >
> > No. My initial post, this thread, is about real hierarchies.
>
> And are these essentially different from nested relations?
No, I have specifically stated that they are the same. That only mathematicians working in isolation, and in ignorance, assert that they are different.
> Can you define how exactly? Just so it's clear what you are talking about. I'm guessing
Please do not guess. I am quite literal in my posts.
> it is that there is a notion of logical relationships between parts of hierarchies.
Yes and no, depends what you mean. In any case, that is going far too much ahead of the current state of play of this discussion. Yes, it is important, but no, it is not relevant /now/, and when we get further, the question will not exist (it would have been resolved by the progression.)
I repeat, nested sets or relations or whatever they are variously called, are not relevant to the thread. Please do not get hung up about it.
Hierarchies (whether acknowledged from the HM or not; whether nested sets or not), are relevant. They are relevant conceptually at concept time; logically at logical time; and physically at physical time. The logical and the physical would not happen unless they were conceptual first. I am saying, hierarchies as they are prescribed in the RM, and as they exist in naturally in data, are suppressed, by the mathematicians in this industry.
> > Not about the various papers that propose to implement data in some nested form (a view that is partially hierarchical) without understanding that the Relational Model has Hierarchies. ignorant people are forever re-inventing the wheel. and forever coming up short and square.
>
> I don't know papers that talk about "implementing" in a nested form, but I do know papers about "representing" it that way. That is of course an essential difference if we take data independence into account. Why did you use the word "implementing" and which papers are doing that?
- Representing where ? In the back of the eyelids ???
Representing for what purpose ? If it has a purpose beyond MMM, then an implementation is is next. Check Norbert's thread, he is trying to cross over from the from eye-closed universe to the eyes-open universe, check the chasm between them.
In any case, as long as the paper is published in the public domain, it will get used. By implementers.
2. Fine. Strike "implemented'. Replace it with "mathematically defined and proofed". It is still garbage, executed in ignorance. I would be pleased in you would address that point, rather than the distinction between representation and implementation.
When I say "invent" I mean propose a concept, as if it were new (novel, novelty), and provide a proof. Ignorant that the concept is not new, it already exists, under a different name, within the subject space. And they should be aware of that.
> Any paper in particular that you can point to that does this in a way you find unacceptable?
So the way they do it is not the problem, all the papers in this space do it. Even Norbert's Topological Space paper does it.
The unacceptable part is that they are ignorant of (a) the science and established rules in this field, and (b) the related sciences. As detailed above, so I won't repeat the detail again.
The result is two-fold, first others, as well as I, laugh our heads off because what is new to the darling author, is twenty or thirty-year-old hat to us. Second, the damage, others not so knowledgeable use the paper and implement the novel concept, in the novel location, which is wrong, because the science tells us the correct location. Thus it usurps existing science. Thus it is offensive.
All the papers in this field do that, including yours, which is why I am saying it is a great example, and dealing with it will answer many of your questions here. I don ot wish to avoid your questions.
All the papers, including the Fagin, Darwen, Pascal papers on the NFs. Don't get me started on that one, please, let's concentrate on this thread and close it.
Most of the books. The Alice book. All the Date and Darwen books.
Ok, I have answered pretty much each of your points in detail. You do not need to do the same in return. If you read the whole of this and only respond to the points that are worthy, only the points that are within the declared scope, that are relevant to you, that would be fine with me.
If you wish to leave this post unanswered, and to proceed directly to dealing with your paper, which will expose specific examples of many of the issues in this thread, and address them to closure quickly, that would be the highest rate of progress.
Cheers
Derek
Received on Sat Jan 31 2015 - 07:53:36 CET