- From: Gary Hallmark <gary.hallmark@oracle.com>
- Date: Fri, 28 Sep 2007 12:05:59 -0700
- To: Dave Reynolds <der@hplb.hpl.hp.com>
- CC: RIF WG <public-rif-wg@w3.org>
+1 Dave Reynolds wrote: > > So we had a majority in favour of having the whole triangle. > > But we have problems agreeing on how to write down that AS and the > associated mappings. ASN seemed to a reasonable job to me but people > had problems with it and it is clearly not yet working as a single > point of maintenance. The aebnf form is rather verbose and baroque and > makes the spec hard to read. > > A suggestion ... based partly on the way OWL 1.1 folks have gone about > things ... > > o Introduce the structure of the language using UML diagrams, OWL 1.1 > call this a "structural model". We include in those diagrams the > ordering annotations. I would be happy to adopt the <<set>> and > <<list>> stereotypes that OWL 1.1 use. > > o We then provide a presentation syntax that is minimalist, least > punctuation, clearly close to the structural model. What Michael > described as 2/3 of the way from the current human readable syntax to > the aebnf. Let me call it MPS (minimalist presentation syntax) for now. > > o We express the semantics in terms of MPS. > > o We provide an injective (and thus invertible on the covered set) > mapping from MPS to XML data. We express this as a simple table of > structural mappings. > > o We also provide an XML Schema such that any XML data generated by > the mapping will conform to the schema. > > o We put in the document that it is our desire to be able to use the > structural model as a single point of maintenance in the future and > provide a textual notation for it and mappings between that and both > MPS and XML. However, doing so in an agreed way is work for the future. > > o We get a early sketch of PRD together in a similar style. We use > that as a concrete example to understand the issues of extensibility, > single point of maintenance at all that. However, we do that well > after the next BLD working draft. > > This seems to me to keep the spec relatively simple, just one mapping > to explain. The UML structural model aids understanding and gives a > good guide to implementers. > > I don't want to reopen a mega discussion on this nor do I want to push > for a compromise that people will later regret. If this approach > resonates with people great, if not then consider it withdrawn. > > Dave -- Oracle <http://www.oracle.com> Gary Hallmark | Architect | +1.503.525.8043 Oracle Server Technologies 1211 SW 5th Avenue, Suite 800 Portland, OR 97204
Received on Friday, 28 September 2007 19:05:25 UTC