Re: [JSON] User segments update

Manu,

On 23 Mar 2011, at 14:58, Manu Sporny wrote:
>>> http://www.w3.org/2011/rdf-wg/wiki/JSON_User_Segments#Potentials_that_Manu_Sees
>> 
>> A-3/4/5/6 --- why do column A people, who don't need or want RDF, need a
>> way to transform JSON to Triples?
> 
> They don't, but the publishers that want to reach the audience that
> wants RDF /could/ encode the information in a way that doesn't bother
> Column A people, but also meets the needs of column B people.

I think you are misreading Sandro's diagram.

The diagram is not about: “Would these groups of developers use our new format?”

The diagram is about: “Do these groups of developer have unmet needs that this WG can address by defining a new format?”

Sandro wrote about A-3/4/5/6: “Nothing for us to do, they are fine already. We just need to avoid making things more difficult for them.”

These are happy people. They have no needs that any new format could address. Hence, gray box in A-3/4/5/6.

> That is -
> one solution is applicable to both Column A and Column B. That is, the
> markup looks like this to both columns:
> 
> {
>   "#": { ... MAPPINGS ... }
>   "name": "Sandro Hawke"
> }
> 
> Column A people would just ignore the '#' key.

Of course. But they still don't need a new format. A new format doesn't improve their lives and doesn't make their jobs easier. They don't care about the '#' stuff and we can't do anything for them (except making the '#' stuff easy to ignore). Hence, gray box.

>> B-3/4/5 --- do you really think this standard method can be simple
>> enough that people will use it without libraries?  That seems impossible
>> to me.
> 
> Yes, it can be, you don't need a library to use this:
> 
> {
>   "#": { ... MAPPINGS ... }
>   "name": "Sandro Hawke"
> }

This is not true. You defined Group B as: “Data consumers willing to use RDF to avoid custom coding per data source, no libraries/APIs.”

Now I believe that in the format-to-be-standardized, in addition to the above, each of the following are legal ways to write down the same RDF:

{
   "#" : { ... OTHER MAPPINGS ... }
   "foaf:name": "Sandro Hawke"
}

{
   "#" : { ... YET ANOTHER MAPPING ... }
   "http://xmlns.com/foaf/0.1/name": "Sandro Hawke"
}

If I want to avoid custom coding per data source, I want to write only *one* piece of code to get Sandro's name out of the JSON structure, and that piece of code should work with *every* data source, regardless of the { ... MAPPINGS ... } that the data source chose.

I don't see how that would be possible *without a library*.

As Sandro said in the diagram about these rows of the Group B column: “Nothing for us to do, no solution is possible in this space.”

If that is true, then: gray box in B-3/4/5.

>> I kind of like that plan, but my current gut feeling is more towards
>> making the yellow box take over the world, via a sparql-json solution.
> 
> I don't think the yellow box would take over the world...

I agree.

>> But a working greenbox solution would be great, too.
> 
> Agreed.

+1, with emphasis on “working”.

I wonder to myself, what evidence would I need to see to believe that a proposal “works”. I'll have a think about that.

Richard

Received on Wednesday, 23 March 2011 19:00:56 UTC