Re: (proposal) was Re: defn of Named Graph

On 09/26/2013 02:18 PM, Peter Patel-Schneider wrote:
>
>
>
> On Thu, Sep 26, 2013 at 10:32 AM, Sandro Hawke <sandro@w3.org 
> <mailto:sandro@w3.org>> wrote:
>
>     On 09/26/2013 12:00 PM, Peter Patel-Schneider wrote:
>>     Umm,  doesn't a lot of this have normative consequences?   If
>>     not, then is there any reason to have it?  If so, then both
>>     Concepts and Semantics will have to have changes to support it,
>>     and there has to be a design that the WG can vote on before
>>     proceeding.
>>
>
>     There's a conceptual part (that doesn't affect running code) and
>     an optional part that would just be in a WG Note.
>
>     The conceptual part is (1) telling people what a "Named Graph" is,
>     and maybe (2) shifting the terminology around "RDF Graph".    My
>     sense is this would clear up a lot of confusion, but perhaps I'm
>     just arguing about angels on the head of a pin and this would
>     cause more confusion. I have no interest in pushing this if people
>     are okay with the status quo.   We claim RDF Graphs are
>     mathematical sets, but that doesn't seem to be how the real world
>     works with them.   Does that matter? *shrug*
>
>     The optional part is one RDF class, with a standard name (eg
>     rdf:NGDataset), which would allow people to indicate in a dataset
>     that it has certain semantics (that seem obvious to me, and I
>     think most people use without it being formally specified).   I
>     also have no interest in pushing this, if no one needs that kind
>     of interoperability.
>
>
> How does this optional-but-normative semantics get specified?   I 
> would think that it would require wording in both Concepts and 
> Semantics, not in a WG note.
>

We have two levels of specs, right?  There are the ones that are 
thoroughly reviewed and proven to be implentable, which we call 
Recommendations.  And other ones that we still think are useful, but not 
particularly reviewed or tested, which we can publish as Notes.   
Remember LBase?

Notes with normative content are a step up from W3C Submission, two 
steps up from a self-published spec, and three steps up from a blog 
post, but a big step down from a Recommendation.

In terms of normativity, my understanding is it would only need to be in 
Concepts or Semantics, or normatively referred to from them, if 
implementing it was required in order to be a conformant RDF system 
(however that's defined in Concepts and Semantics -- that's another 
question).     And I don't think that's the case here -- we've decided 
you can have a conformant RDF system, with datasets, that uses only the 
semantics for Datasets as currently specified in RDF Concepts.

       -- Sandro



> Or maybe you are proposing that there be an entailment regime for 
> this.  However, if this is the case, then there is no need for the WG 
> to specify the entailment regime, as specifying entailment regimes can 
> be done by anyone.
>
>
>
>     If the WG passes on both of these, they can be handled later,
>     except that it'll be harder to shift the terminology.
>
>            -- Sandro
>
>
>
> peter
>