JSON, HTTP and data links
In late 2011, Mark Nottingham, whom I very much admire on a personal and professional level, posted ‘Linking in JSON‘ which triggered quite some discussion (see the comments there).
Back then already I sensed that the community at large is ready for the next aspect of the Web. A scalable, machine-targeted way to realise a global dataspace. No, I’m not talking about the Semantic Web. I’m talking about something real. And it’s happening as we speak.
Take JSON and HTTP (some use REST for marketing purposes) and add the capability of following (typed) links that lead you to more data (context, definitions, related stuff, whatever).
And here are the three current contenders in this space (in the order of stage appearance) – Microsoft’s OData JSON Format, The Object Network: Linking up our APIs, and – as I learned from Charl van Niekerk on #whatwg IRC channel tonite – A Convention for HTTP Access to JSON Resources.
What they all have in common is that they define ways to read, create, update and delete data objects, in the Web, based on JSON, using HTTP.
Totally-objective-and-unbiased-verdict: around for some years now, great community, backed by big bucks, heavy-weight (they squeezed friggin APP into it), rather RESTful and becoming more and more a shadow Semantic Web.
The Object Network
Totally-objective-and-unbiased-verdict: too early to tell, really. Seems like a one-man-show, nice idea in theory, time will tell uptake. Many things, incl. link semantic seem half-baked and unclear. Good motivation and marketing but little ‘apps’ or demos to be of any interest.
A Convention for HTTP Access to JSON Resources
Totally-objective-and-unbiased-verdict: just learned about it, but seems to be influenced by CouchDB developments and experiences which means it can’t be that bad, can it? Yeah, I guess I’ll have a closer look at this one.
Now, which one is your favorite? Did I forget any? Before you shout out JSON-LD or the likes now … hold your breath – my #1 requirement is that it does the Full Monty: I want to be able to CRUD, to follow my nose through the data and all this over HTTP. Anyone?
Filed under: FYI, IETF, Linked Data, NoSQL