- From: Sandro Hawke <sandro@w3.org>
- Date: Sun, 17 Dec 2006 17:58:30 -0500
- To: Gary Hallmark <GARY.HALLMARK@ORACLE.COM>
- Cc: Michael Kifer <kifer@cs.sunysb.edu>, W3C RIF WG <public-rif-wg@w3.org>
> Well, given that both Oracle and IBM are represented in the working group, if
> the only interest comes from a lone academic, I think we can safely rule SQL
> -RIF out of scope.
I don't think anyone is actually suggesting that RIF-WG define a level
of conformance which covers only the interchange of relational tables.
I suppose we could, but I wouldn't want that called "RIF Core". That
might be "RIF Data" or something. (And RDF already covers that, more or
less.)
The point is that each dialect (or other kind of unit of conformance)
should correspond to a set of features users will want and be able to
buy (or download, or whatever). The sets of features we're chartered to
work on are the ones commonly associated with "rule systems". In phase
one, the single set of feature should be "essentially Horn" -- but Gary
has put forward a compelling-to-me argument that it should be only
non-recursive Horn, because recursive Horn is not commonly available in
products.
The analysis and decision making process here has to be done in terms of
users, products, and markets -- not just technical elegance.
-- Sandro
Received on Sunday, 17 December 2006 22:59:50 UTC