Re: Comments on Norbert's topological extension of relational algebra

Op vrijdag 18 december 2015 17:15:53 UTC+1 schreef Nicola:
> On 2015-12-15 16:14:14 +0000, Jan Hidders said:
>
> > What *I* am interested in is the connection with this work:
> >
> > http://alpha.uhasselt.be/~lucp1080/queries_reals.pdf
>
> Do you mean, with the constraint database model (by Kanellakis et al.)
> in general or with the results in that specific paper?

Both, really.

In my mind (but I have not fully wrapped my head around it yet) Norbert's proposal amounts to proposing a special new domain (or rather a class of domains) with some special predicates such as "is contained in" and "is a border of", and operators that seem similar to those of the relational algebra. In addition he has a special new aggregation operators (such as generalized union and generalized intersection) that can aggregate bags of values of that domain again into a value of that domain. The results of the references paper help us understanding to what extent such operators would actually be computable.

Btw. I'm not sure why it would be such a big deal that the operators of this domain are similar to relational algebra operators. What practical benefit would come from that? Would that make things easier for an optimizer? For human beings trying to specify / understand a query? I don't see that at the moment.

About constraint databases: when allowing in this new (class of) domain(s) any set of points that can be defined in some appropriate constraint language, the idea would actually start to make more sense to me. Of course the relations defined by constraints would the be domain values, and not necessarily top-level relations as they usually are in the literature, but that I do not consider a fundamental problem.

• Jan Hidders