Re: Normative vs. non-normative in JSON-LD (Re: JSON-LD skipping CR?)

On Mar 29, 2013, at 13:06 , "Markus Lanthaler" <markus.lanthaler@gmx.net> wrote:

> On Friday, March 29, 2013 11:46 AM, Ivan Herman wrote:
> 
>> I was looking at the document:
>> 
>> http://json-ld.org/spec/latest/json-ld/
>> 
>> And I realized that almost the normative parts in the document are
>> Appendix A, B, C, and E. Which is fine with me. However, Appendix C
>> (Relationship to RDF) refers to RDF11-CONCEPTS non-normatively which I
>> do not think is right.
> 
> It describes the differences to RDF. We do not rely on anything from
> RDF11-CONCEPTS, so why do you think we need to reference it normatively in
> this instance?

The differences are normative, so the document you compare with should also be normative in my view. But it does not change the conclusions below...

(To be clear, I express an opinion with my activity lead's hat put down.)

> 
> 
>> It also refers to RDF-SCHEMA which should, at
>> some point, be changed to RDF11-SCHEMA as a normative reference.
> 
> There doesn't even exist a FPWD for RDF11-SCHEMA, that's the reason why we
> still reference the old version. AFAIK there are no differences between the
> two version that would matter for JSON-LD <-> RDF so it shouldn't be a
> problem.
> 
> 

I fully understand that you refer the existing spec. But the RDF WG will issue a new version and it would look strange if the same working group issued a Rec that does not refer to its own Recs, so to say... And the reference should also be normative I believe.

>> RDF11-CONCEPTS is also referenced from appendix E and again,
>> it should be normative.
> 
> Yeah, in this instance I agree that it should be normative. 
> 
> 
>> Bottom line: I believe JSON-LD should normatively refer to RDF11
>> CONCEPTS and the upcoming 1.1 version of RDF Schemas. That, however,
>> does not make it possible to turn it into a PR before the other
>> documents are at least in CR. We can, of course, skip CR for JSON-LD
>> and leave it on hold as a LC, but that would not buy us any time.
>> (Although it would make it easier to process because we would have one
>> admin hurdle with a transition call less.)
> 
> Hmm.. that's true. Maybe a viable alternative would be to specify how
> fragment identifiers are used in appendix E which would allows us to keep
> non-normative references!? Or do you think that's not an option? 

I am a bit uneasy with any hack just to speed up the publication by a few months. If the document goes to, say, CR, and then it is clear that the spec is ready to go and will be synchronized with the rest, it would not hurt any deployment (unless the other documents get into a very long delay, but I trust the WG chairs to avoid that...)

Cheers

Ivan 


> 
> 
> Cheers,
> Markus
> 
> 
> --
> Markus Lanthaler
> @markuslanthaler
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
FOAF: http://www.ivan-herman.net/foaf.rdf

Received on Friday, 29 March 2013 12:16:11 UTC