Re: Media type for JSON-LD: is it really a good idea?

On Dec 10, 2012, at 09:46 , Markus Lanthaler wrote:

> Hi Ivan,
> 
> 
>> Markus made a change on the pyRdfa processor the other day (which I
>> propagated to the pyRdfa service on the W3C) to change the media type
>> when distilling json-ld from RDFa. I realized yesterday that this
>> 'broke' a feature I used: I have a JSON viewer/displayer extension in
>> Firefox that did not work any more. The reason is obvious: that
>> extension works on a /json media type but does not on /jsonld+json
>> (probably ignores any YYYY+json in general).
> 
> Yes, I'm aware of that problem. The thing is that currently Firefox doesn't
> provide an API to register extensions for MIME type ranges such as
> application/*+json. If you are using JSONView I can send you a patched
> version that works also with application/ld+json.
> 

Actually, yes, that is the one, so a fixed version would indeed be welcome...

> 
>> Hence my question: do we have an idea how widely the '+' mechanism is
>> implemented among tools that rely on media type? My _hunch_ is that
>> this is not widely done (this was/is an issue with the YYYY+xml stuff
>> as well).
> 
> I would say it is widely used but library support is still quite limited.
> There's an ongoing effort to standardize the +json convention [1]. I do hope
> that changes the situation to the better once it's out.
> 

Ok. I did not know about [1]. This may answer my question.

Thanks

Ivan

> 
>> Because the major value of JSON-LD is that existing JSON
>> tools can happily work with it, possibly ignoring the 'LD' aspects, I
>> wonder whether, in practice, it is not better not to introduce a new
>> media type... Of course, if the experience is that YYYY+jsonld works in
>> general, and only my extension has a bug then, well, o.k.
>> 
>> I am just musing here, not raising some sort of an objection...
> 
> That would just fix a specific symptom but not the underlying problem and
> has some serious downsides. If we would do that, than there wouldn't be any
> way to detect if the response was JSON-LD or not without content sniffing.
> It would also make it impossible for clients to specifically request JSON-LD
> in content negotiation. Furthermore it become a problem if JSON-LD specific
> development tools became available.
> 
> I have been working on something related [2] recently but haven't had time
> yet to transform that into a browser plugin. It's definitely on my todo
> list.
> 
> Just go to [2] and click on the link at the bottom to load a JSON-LD
> document. When you move your mouse over any property or value a tooltip
> showing you the IRI it would expand to appears. The IRI is also dereferenced
> in the background on the corresponding documentation is shown. This is still
> an early prototype - so don't expect too much :-)
> 
> 
> Let me know if you want the patched JSONView version in the meantime.
> 
> 
> [1]
> http://tools.ietf.org/html/draft-ietf-appsawg-media-type-suffix-regs-08#sect
> ion-3.1
> [2] https://m.lanthi.com/HydraConsole
> 
> 
> 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 Monday, 10 December 2012 15:31:19 UTC