Re: Hashing for DISTINCT or GROUP BY in SQL

On 16/10/2010 8:56 AM, Cimode wrote:
> If relational logical theorists had put as much effort into*defining*
> an relational compatible mathematical model for optimizing relational
> operations and structure*physical* representations as they did into
> continuously defining new logical operators of higher abstraction,
> we'd probably not be in the situation we are today.
>

My own background probably gives me a bias that wouldn't be typical but for me this has to do with what I think of as optimization. But that's probably different from the optimization that most other IT people think of. My attitude is probably more general than theirs, encompassing everything to do with what they might call implementation.

For example, I've often wondered 'where or what is the implementation theory of constraints?'. Is there 'theory' to tell an implementator how to recognize something as elementary as a way to decide that a certain constraint applies only to a single tuple (or if you like a relation of cardinality one) as opposed to one that depends on more? A very common example of this would be the presence of a so-called 'insert' of a single tuple with a candidate key. When presented with an 'input' relation, and 'output' relation and a 'tuple to be inserted', what 'theory' is available to the author of an implementation that tells him the constraint applies either to the 'output' or to the 'combination' of the 'input' and the 'insert tuple'?

Of course all old-timers know 'how' to implement this, they test for the 'presence' of the key and if it is not 'present', they proceed with the 'insert'. But where is the 'theory'? I'm guessing that this is part of your point. Received on Mon Oct 18 2010 - 03:03:49 CEST