Re: deductive databases

Jan Hidders wrote:
> Mikito Harakiri wrote:
> >
> > Well, "connect by" is [admittedly not especially elegant] way to
> > express transitive closure. Now, are there any queries that can be
> > expressed with "Recursive With" but not with transitive closure?
>
> I think there are. It is known that on an ordered domain FO(TC)
(first
> order logic plus transitive closure) can express exactly all
NSPACE[log
> n] queries and FO(LFP) (first order logic plus a least fixpoint
> operator) exactly all PTIME queries. The latter class is known to be
> strictly greater than the first.

How about FO+aggregation? Actually, I was expecting that somebody from DB2 camp would jump in and exibit such a query. Well, database wars are over.

There is no question that any big database vendor that has no problems with budget is capable implementing a feature X, especially that it's a part of the standard. It seems, however, like swallowing some accounting company has more priority than getting database issues straight. Received on Fri May 13 2005 - 19:29:41 CEST