Re: Sensible and NonsenSQL Aspects of the NoSQL Hoopla

From: Eric <eric_at_deptj.eu>
Date: Sun, 1 Sep 2013 13:22:54 +0100
Message-ID: <slrnl26ccu.qt7.eric_at_teckel.deptj.eu>


On 2013-08-31, karl.scheurer_at_o2online.de <karl.scheurer_at_o2online.de> wrote:
> Am Donnerstag, 29. August 2013 06:30:58 UTC+2 schrieb com..._at_hotmail.com:
>
>>
>> A tuple/row of a relation/table is a point in a multidimensional
>> space. The attribute/column names are the dimension names. They are the
>> points in the space that satisfy the relation's predicate.
>>
> ...
>>
>> To think of a candidate key as an ordinate is, again, a non-relational
>> way of >thinking and reduces a relation to a mere data structure instead
>> of the >extension of the predicate of the variable or query that it is
>> the value of.
>
> No you got it! To access or to distinguish points in a multidimensional space
> you need ...

... to know all the coordinates of any point, i.e. a set of values with some way of knowing which axis each refers to. Sounds like a tuple in which each element belongs to a specific domain. So for all the points you know about, you have a set of tuples. Sounds like a relation.

> (need) additional information not provided with the relational model.

So what am I missing here? What information do you need that is not provided in the relational model?

> Only the tacit assumption of grouping information into a
> tuples/rows... the missing information can calculated from the column
> data.

Again, what missing information?

> Intepreting this grouping as records relations or statements is pretty
> irrelevant.

On the assumption that there is missing information.

Eric

-- 
ms fnd in a lbry
Received on Sun Sep 01 2013 - 14:22:54 CEST

Original text of this message