Re: [xdi] Exciting breakthrough: the JSON Model

Animesh, I do very much agree with you. It's the kind of breakthrough that makes it even more exciting to be working on XDI.

And I should thank you and Joseph for some of the questions and discussion on last Friday's telecon that helped inspire it. In particular, it was your question about "the programming object model for XDI" that got me thinking that, while the ContextFunctions model was pretty good, it would be better if the programmatic model for processing XDI graphs was even simpler. That's what got me focused in further on the programming model for processing JSON data.

I talked to Markus about this and he agreed we should produce, even if it is non-normative, a reference programming object model for XDI (Markus did a full UML model back for our original RDF graph model). In fact I think we should publish it as soon as we can as one of our Informative wiki pages.

I think that the model is as simple as the four XDI context functions on the JsonModel page:

Root context objects.
Member context objects.
Element context objects.
Value context objects.

Obviously it's the value context objects that contain (or reference) the actual values (XDI literals) in the graph. All the other objects just define the graph structure.

Is that how you see it? Who would like to volunteer to do the UML model for this?

=Drummond

On Wed, Mar 27, 2013 at 7:49 AM, Chowdhury, Animesh <Animesh.Chowdhury@neustar.biz> wrote:

Hi Drummond,

I think the syntax is much more intuitive, concise and clean.

Thanks,

Animesh

From: xdi@lists.oasis-open.org [mailto:xdi@lists.oasis-open.org] On Behalf Of Drummond Reed
Sent: Wednesday, March 27, 2013 4:20 AM
To: OASIS - XDI TC
Subject: [xdi] Exciting breakthrough: the JSON Model

After deep cogitation on the feedback from the last TC meeting -- and triggered by the message Markus sent to the list this weekend -- a way occured to even further simplify XDI syntax and the graph model itself by fully aligning it with the very elegant and universal JSON data model.

This is now written up at:

https://wiki.oasis-open.org/xdi/JsonModel

Test: go straight to the examples (in yellow) and see how quickly you get the basic pattern.

Then read the text to see the underlying logic.

Ironically, I believe this approach is actually very close to what Bill has been proposing for some time, only with a solution to the "collection/singleton" disambiguation problem that I was never able to see before.

And one bonus feature: it also frees up curly brackets so we can use them for XDI variable syntax!

One last note: it uses the term "member" differently than we have been using it so far. Instead it uses "member", "element", and "value" exactly as defined in the JSON RFC.

Happy reading - please post any thoughts back to the list. I plan to work on the ABNF (perhaps with Joseph if he's available) tomorrow and Thursday in order to have everything ready to review on Friday's TC call. I want to put these syntax changes to bed once and for all!

=Drummond

xdi message