Re: [JSON] PROPOSAL: Syntax structure should be object-based

On 16/03/11 03:49, Lee Feigenbaum wrote:
> On 3/15/2011 9:11 PM, Manu Sporny wrote:
>
>> PROPOSAL: The RDF Working Group should design the RDF in JSON syntax
>> structure to reflect the object-based data model that is in wide use in
>> the Web developer community. The group recognizes that both the
>> triple-based and iterative-reduction based approaches are useful and
>> have a purpose to serve, but the time it would take to standardize two
>> RDF in JSON syntaxes may impact the ability for the Working Group to
>> meet its tight 1-year deadline.
>
> As I've said before, my organization would uses triple-based JSON
> serializations of RDF, but does not feel strongly that it needs
> standardization. We do not and would not expect to use an object-based
> RDF-in-JSON syntax. I do not expect that we would object to the work.
>
> My main concern would be whether this Working Group has the right
> composition to effectively enact this proposal if a core goal is to
> increase adoption by the JSON developer community.
>
> Lee

While I can support a data-object style (providing a document is 
self-contained and matters of coverage), the more important question to 
me is whether we are designing to API access or direct datastructure 
access, and within the latter whether there is translation between 
on-the-wire and applications forms.

There is the API for RDF in RDFa.  How does fit in?  If not, aren't we 
getting close to two approaches to much the same thing?

I have seen the suggestion:

   var obj = parseAsRDF(json)

then access via "obj" so even if it's data structure access the 
on-the-wire format could be very different to the object structure 
created.  Indeed, it's almost an argument that Turtle/N-triples is quite 
sufficient for the on-the-wire format and it does not need to be JSON at 
all.  Only speed seems an issue (this is not RDF googles - that would be 
"treatAsRDF(json)")

A style of

   var obj = eval(string)

then get into different issues about different ways to write the same 
thing - e.g. prefixed names vs IRIs - and round-trip implications.

 Andy

Received on Wednesday, 16 March 2011 14:07:37 UTC