- From: Tim Berners-Lee <timbl@w3.org>
- Date: Mon, 3 Mar 2003 20:50:49 -0500
- To: Brian McBride <bwm@hplb.hpl.hp.com>
- Cc: www-rdf-comments@w3.org
(with apologies, a message delayed for a few days, while my laptop was repaired) On Friday, Feb 14, 2003, at 08:27 US/Eastern, Brian McBride wrote: > I'm following up on this from a process perspective. In my initial > response I forgot to ask whether that response resolved your comment. > > We've had some offlist discussion from which I think I've learnt that > I might reinterpret your question as: > > Should the syntax that triggers generatation of reified statements, > i.e. rdf:bagID and some instances of rdf:ID be removed from the > specification. > > Would that be a fair characterization? Yes > My quick response to this question is that the WG was chartered to > clarify M&S. It wasn't chartered to remove bits of the syntax (though > it has done so in other instance where clarification was difficult). > That's fair. > The WG might have been more willing to consider this question if it > had been brought up earlier, but we are now in last call so it is more > difficult to make such changes. I must indeed apologize for not bringing this up earlier. The question of the semantics of these triples only arose practically for me in relation to a need for reification in some very recent and incomplete work on ontologies for expressing proof. > As I noted earlier, there are now several implementations of the > features you suggest be removed. > If reification has been used in existing applications that will I am sure shed light on the semantics of the triples. A reasonable path then, in keeping with the charter, would seem to be to clarify the semantics of the generated triples (inferences which may be made from them, etc) , and leave in those things which are clear in the spec. One could perhaps look at a future plan for profiles of RDF which re-asses what is included in the very core layer. > I doubt you will find this response satisfies you, but my question to > you is whether you are persuaded to withdraw the comment or whether > you would like this comment recorded and considered by the WG. It really depends on whether the group feels that the semantics can be clarified. You continue in another message, > This is the process policeman again. > > I think the second question you have raised in this thread concerns > the semantics of reification. There has been some discussion and I'd > like to clarify whether this discussion has satisfied your concerns or > whether we should get a clear statement of the issue to go on the WG's > last call comments list. If reification has been used in existing applications that will I am sure shed light on the semantics of the triples. A reasonable path then, in keeping with the charter, would seem to be to clarify the semantics of the generated triples (inferences which may be made from them, etc) , and leave in those things which are clear in the spec. I am not personally familiar with the de re interpretation of reification so I can't really judge how it would be used. I would trust the group though if they felt it was defined. One could perhaps look at a future plan for profiles of RDF which re-asses what is included in the very core layer. I think that is important for the simplicity of RDF as a foundation for many things. > Brian > > > > At 09:35 12/02/2003 +0000, Brian McBride wrote: > >> This response is to "a developer" the signature on the original >> posting. >> >> At 14:15 11/02/2003 -0800, Tim Berners-Lee wrote: >> >>> I have a question about implementation of the bagid feature. >>> My parser doesn't currently implement it, and so fails the RDF >>> tests which involve it. The same seems to apply to Ids on >> >> >> [...] >> >>> Is this feature then worth implementing? What does the group think? >> >> In an ideal world the answer is "no". There are several parsers, for >> different languages, that pass all the WG testcases. Have you >> considered integrating one of those rather than building your own? >> >> If you really need to build your own, the answer in general is yes, >> if you want to be sure of interoperating with other implementations. >> >> Brian
Received on Monday, 3 March 2003 20:51:53 UTC