Re: A Topological Relational Algebra in Lisp

Norbert

> On Tuesday, 20 January 2015 20:36:48 UTC+11, Derek Asirvadem wrote:

A couple of clarifications.

> Relational or Not
> -----------------

> Relational (IDEF1X) vs OO (UML)
> -------------------------------

> Therefore, I would strongly recommend that:

> c. Ensure that [a] and [b] are fully Relational. This assures the reader that the constructs; the tables; the queries, enjoy all the integrity, power and speed of the RM

I forgot to mention, it goes without saying, the purpose of my detailed comments under this heading, if you wish to make the claim that the Topology you are publishing is Relational, it had (a) better be Relational, as (b) demonstrated in every instance that you expose it or parts of it.

Claim

i. I might repeat, it is not possible, not feasible, to determine the Relational-ness (qualification against the Relational Model), for your "relations" (terse text strings), unless they are provided in toto and with complete definitions (eg. Foreign Keys; constraints). And especially not if the names of the "relations" and attributes are single character or terse. If you do not understand this, you will get into silly arguments like "how can a relation be non-relational".

Claims re whether such "relations" are Relational cannot be confirmed, and they are unsafe to make. Of course, if the claim is made *only* to other mathematicians, and *not* to implementers, since they are under the same limits, they can argue equally well, that the claim is true, or that it is false. Without resolution one way or the other.

"Relations" are an abstraction. I cannot evaluate the thought that you are thinking, and determine if it is a sin or not.

Therefore, the place where the 'rubber hits the road', where the "relations" become defined in real terms, is the place where one can determine Relational-ness. Pieces of paper, with ink marks on them, that define *all* the "relations" in set, and all the relevant constraints. Now argument is avoided or at least minimised, and resolution is possible.

ii. Until 1985, the marks defining the "relations" were ASCII characters, and the statements were SQL DDL. That engages 4% of ones brain. Determination of whether such "relations" are Relational or not, and indeed whether the database as a whole is Relational or not, can now be made. Engaging the same 4% of the brain. Time-consuming and prone to error.

The claim can now be confirmed, with some difficulty.

The abstract "relations" are no longer abstract, they are real. I can see that you are looking at my daughter with a naughty intention.

iii. Since 1985, we use Data Modelling diagrams, rendered in the IDEF1X Standard. That engages 96% plus the 4% of ones brain. No DDL is required (if implementation was planned, sure, that is the next step). Determination of whether such "relations" are Relational or not, and indeed whether the database as a whole is Relational or not, can now be made. Engaging the whole brain. Far less time-consuming and far less prone to error.

The claim can now be confirmed, with ease.

I can now see all your mannerisms, your glances, when you are around my daughter, and that they are different when she is not there.

Frivolous Claim

For your understanding. Note that it is by the exact same protocol, that:

a. the pseudo-scientists, the boffins who write books allegedly about the RM, are able to make claims that some "relation" (either theirs or some other's) is Relational or not
b. that the RM itself is "relationally incomplete"
c. and such claims are idiotic, frivolous, because they can be neither confirmed nor denied (as detailed above)
d. therefore scientists do not accept such claims
e. only frauds and thieves make unfounded and unprove-able claims
f. they do so petulantly, and continuously (scientists know that a truth needs to be proved, and proved just once)
g. why they steer clear of Data Models and diagrams
h. why they avoid SQL DDL (the denigration of SQL being a separate matter, but nonetheless servicing this need as well: "can't define the 'welation' in SQL because it is bwoken")
i. and why they restrict themselves to describing (not defining) their "relations" in ASCII characters, text strings, sans DDL, sans DM, at the pre-1985 level of capability
j. why that is their permanent cave.
k. why scientists who can erect DMs call [i], their "art", Degenerate Art

Their language, their mantra, is (i) mathematical definition, (ii) only mathematical definitions, and (iii) nothing but mathematical definitions. Now, I have every respect for (i). But given the range of developments in the Relational arena since 1985, given the English language for technical use, given the ease of DDL and Data Models, (ii) and (iii) are for sub-humans, backward cavemen. Unfortunately they write books ! Even worse, some of those books are used as textbooks !!!

Which is why, in order to "hear" the music, to "see" the kaleidoscope, to sing and dance at the private orgy, first you must speak the language, and second, you need the Kool-Aid.

Back to you. The same protocol as I have given you, applies to them. The difference is, you are honest, and willing to engage in a discussion with a view to resolution. They desperately need their continued petulance, the non-resolution.

(I might have to publish my Relational Scorecard. Which thus far, has been for customers only. Just to kill the argument that I expect will develop from the sewer rats.)

Relational Algebra

I think you can see from my previous post, that in the context of whole scheme, it is probably not as relevant as you thought at first. It remains absolutely relevant, in that it forms the foundation, an essential part (but not the whole) of the theory that the package (either standalone monolith or components plus doco) is built upon.

Depending on your target market, with consideration to [1] to [111] above, if you limit yourself to mathematicians who understand Topology, then you need to: a. document all the "relations" together, along with all their constraints, and b. avoid the claim. Perhaps declare that you created them with the RM in mind.

But if you want to reach everyone who can use your packaged tables plus SQL queries plus <language> queries (or at least a large portion of that market), *or* if you wish to make the Relational claim, then in addition to [a], you need to: c. produce a Data Model, in IDEF1X.
d. make the Relational claim. And [c] will be used to prove the claim, or to deny it.

> Versioning
> ----------

> In TDAS 4.4.8, you state:
> > Since the spatial planning also works with versions of rooms,
> > it makes sense to extend this concept of version control on topological spaces.
>
> Doesn't make sense to me.

Mistake, incomplete quotation. Sorry, that does make sense.

> In TDAS 4.4.8, you state:
> > In UNIX, there are the commands diff and patch to compare and update text files ...
> > RCS, CVS, Subversion ...
> > Since the spatial planning also works with versions of rooms,
> > it makes sense to extend this concept of version control on topological spaces ...
> > ... [and to use diff, patch, RCS, CVS, Subversion]

That doesn't make sense to me.

(And then, my comments follow, and make sense.)

Cheers
Derek Received on Tue Jan 20 2015 - 14:06:44 CET