Re: Sensible and NonsenSQL Aspects of the NoSQL Hoopla

On 2013-09-01, karl.scheurer_at_o2online.de <karl.scheurer_at_o2online.de> wrote:
> Am Mittwoch, 28. August 2013 21:53:02 UTC+2 schrieb Eric:
>> That's because you can't get hold of the idea that physical storage is
>> irrelevant (as long as it works and is efficient enough).
>
> Physical storage can become irrelevant when the logical model is
> sufficiently complete. The Relational Model is not, proved by the
> fact of SQL extensions already starting with DB2.

In what way is the relational model insufficiently complete?

SQL is not the relational model, merely an imperfect attempt at implementing it. With suitable precautions it can be used as part of producing an application that views its data relationally. The fact that SQL has been (and is still being) extended is irrelevant. Some of the extensions take it further away from being relational.

> Codd's requirement of row order insignificance excludes grouping and
> grouped aggregats from his model. Grouping is pretty easy to comprehend,
> when comparisons between adjacent tuples can be formulated. SQL 1999
> with generalized grouping is the next step, but not relational anymore.

Row order insignificance is required because you may want to order by and/or group by any part of the data, and merely means that the system will not impose an order for you. Ordering and grouping are not part of the data definition, they are operations.

> To repeat it once more: every theoretical model based on cursors is a
> valid solution. Cursors are lists of database objects with some sort
> order and comply with some criteria. Cursors with uniform objects can
> be viewed as "relations". Directly working with cursors is at least
> a last resort option when everthing else fail.

That paragraph seems a little muddled, but in any case it does not seem to be any sort of argument about the relational model or its adequacy.

Eric

-- 
ms fnd in a lbry