Re: [RIF] homework for 10/17 telecon

On Mon, 2006-10-16 at 15:40 +0100, Dave Reynolds wrote:
> Dan Connolly wrote:
> 
> > I see...
> > 
> > "In the present version, variables are not sorted and thus can range
> > over all constants, Data or Ind."
> >  -- http://www.w3.org/2005/rules/wg/wiki/A.1_Basis%
> > 253A_Positive_Conditions
> > 
> > Dave, I suggest that <Ind> is what you're after if you want to
> > use URIs for names.
> 
> Well Inds aren't defined on that page but I can guess what they are :-)
> 
> Having variables range over things identified by URIs is one thing but 
> surely if we supposed to be constructing a web based exchange language 
> we start with a presumption that all potentially clashing or sharable 
> symbols like constants, function and relations are identified by URIs?

Yes, that's the Web Architecture way to deal with potential clashes...

Good practice: Identify with URIs
http://www.w3.org/TR/2004/REC-webarch-20041215/#pr-use-uris

If somebody worked out the design details that way, I'd be happy.

> I raised this back in April[*] when the syntax was first proposed and 
> saw no push back at that time.
> 
> > I find it a little awkward that not all names
> > are URIs, but I can perhaps live with that; I'll probably treat
> > the non-uri names as local fragment identifiers or something.
> 
> Do you mean you can "live with it" from the point of view of 
> implementing some sort of N3 translator or do you mean you would defend 
> it if this syntax came to be proposed as a Candidate Rec?

I wouldn't defend it... I mean "live with it" as in abstain
rather than object.

> > As to URIs for functions or relations, I'm not sure; so far,
> > I have only found a need for URIs as constant symbols.
> 
> Would you say the same about builtins?

I'm not sure... in the code I have written so far, builtins
show up as <Ind>s. But if they show up as <Fun> or <Rel>,
then yes, I would need URIs for those too.

> Are functions, relations to be purely local to the XML source file?

Yes... er... no... the "holds" relation and "list" function that I'm
using are intended to be globally defined. So I guess I should have
made URIs for those.

> If so how is merging of rulesets supposed to work?

Indeed, that's the critical question. We need the syntax to allow
URIs in enough places to support merging.

>  Or is that simply out 
> of foreseeable scope?

I would think merging is in scope for the 1st phase... I don't
see it among the current requirements, though. Darn.
http://www.w3.org/TR/rif-ucr/#Requirements

> Dave
> 
> [*] Point 3 in:
> http://lists.w3.org/Archives/Public/public-rif-wg/2006Apr/0091.html
> 
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Monday, 16 October 2006 15:14:55 UTC