Re: RM VERY STRONG SUGGESTION 4: TRANSITION CONSTRAINTS

On 08/09/2010 7:41 PM, Brian wrote:
> On Sep 7, 8:23 pm, paul c<toledobythe..._at_oohay.ac> wrote:
>> On 07/09/2010 3:14 PM, Reinier Post wrote:
>>
>>> Brian Seltzer wrote:
>>
>>> [...]
>>
>>>> There is a mapping from every tuple in a database to something in the
>>>> microworld that is being modeled, but more importantly, every tuple
>>>> maps to something that has a location in time if not also space.
>>
>>> Where exctly is 'there'?
>>
>> That doesn't matter. There is no useful mapping unless it is recorded.
>> Eg., there is not 'time' in a db unless it is recorded. Brian S has a
>> mental block about this. (For the sake of newcomers, this needs to be
>> repeated once in a while. We all have various mental blocks, but most
>> of us stop harping about them after a while.)
>
> When I'm wrong, I admit that I'm wrong. I don't think I'm wrong, and
> nobody has offerred an irrefutable argument that proves me wrong;
> therefore, I have to conclude that I'm right.
>
> As Codd has written, and as Bob Badour continually asserts every time
> the subject of surrogate keys comes up: Every key is a surrogate. A
> surrogate for what? For an object in the microworld being modeled.
> Every tuple has a superkey consisting of all prime components; every
> key is a surrogate; therefore every tuple has a surrogate. A
> surrogate for what? For an object in the microworld being modeled.
> If that doesn't constitute a mapping, then perhaps you can enlighten
> me as to what does.

Heh, fat chance, you'd have to give up your mutating tuples and other sci-fi like microworlds and unnecessary mapping notions first.

Actually, surrogate isn't an important term for RT since it means merely that every key is just a surrogate for other keys, ie., the term has not much practical leverage. I'd guess the stackoverflow crowd who relish obfuscation might get shrill about that, maybe you too, but I can't help it if my first language is English.

It's too bad about Codd's early phrase, how 'time-varying' makes some people imagine that his theory depends on time. I think it was just a metaphor to help his explanation. Even Date perseveres with 'update' which seems to make people think of time. Replace would have been a better word.

Whether in art or computer science, every medium introduces its own restrictions, Codd's intended medium depends on extreme abstraction, the elimination of as many concepts as possible for its leverage. The leverage means wider application but it's not universal, eg., not a very good medium for poetry.

Date has said that a db language without assignment is a kind of cheat but he tempered that by acknowledging that his chosen medium involves 'imperative' languages (same as procedural I think). I think it's best to keep assignment out of Codd's theory. The reason is none other than it's not necessary to understand his theory. Implementation devices could just as easily include one-time variable assignments as they could Date's repetitive assignments. Too bad I wrote 'one-time' but the English medium has its restrictions too. Received on Thu Sep 09 2010 - 15:54:37 CEST