Re: Storing data and code in a Db with LISP-like interface

Marshall Spight wrote:
> Alvin Ryder wrote:
> >
> > For certain advancements changes need to occur at the RM level,
> > relational algebra and if necessary even at the set theory level (I
> > doubt that it is static).
>
> Okay, I'm pretty sure that's false. Can you back it up at all?

When I worked on this stuff during my post-graduate work I wasn't alone, there were many people, lots of papers, lots of research and now there are some respectable books too ... that's pretty much a mountain of back up.

CJ Date and co have written "Temporal Data & the Relational Model" - this will show you in detail the kind of stuff I'm talking about. Mind you, I haven't read this book but I've read the foundational papers.

CJ Date explicitly talks about the "kind of building on top of the RM" I'm talking about in other places.

"The Third Manifesto", with its waves in the object world should be interesting too.

If you still don't accept my position, then fine, what can I say?

> So far you provided one example which Bob knocked down
> trivially. Is there more?
>

No, he didn't solve anything. Just because I chose not to flame him for the wrong answer doesn't mean he gave the right one ;-) It wasn't a trivial distance calculation it was a problem in topology, you are suppose to take the walls and boundaries into account.

Spatial object model theory draws from set theory, predicate logic, graph theory and the RM.

"List all people in my district" is not a naive zip code query, the district is a polygon and my location is a point.

I don't think we need more, I've seen enough.

> The thing is, I think you're not clear on the difference between
> operations on sets and operations on used-defined types.
> Clearly, set theory doesn't include all possible functions
> over user-defined types. Nor is it supposed to. So, for
> example, when you mention multimedia, it just looks to me
> like you're pointing out a lack of, say, digital sound resampling
> functions, which are properly part of the used-defined type
> for digital sound, not part of the relational algebra.
>

I agree it can sound like user-defined functions but what I think shocks people is the notion of adding operators at the relational algebra hence relational operator levels. Nothing in the RM prohibits it, it is indeed open-ended in this respect, you /can/ add relational operators!

What you cannot change are things like the way data is represented it must be in one and only one way, namely as values in attributes, in tuples, in relations.

>
> > Codd and others have not been idle in this pursuit, though of course
> > it'll be decades before it all becomes common knowledge.
>
Well I don't mean him in recent years but other researchers are active.

> I am eagerly awaiting to see what Codd will do in the next decade.
>
>
> Marshall
>
> PS. See if you can spot the ironic sentence.
Yes.

Cheers. Received on Wed May 03 2006 - 14:01:30 CEST