prototype and consequent feedback on XDI primitives

[Giovanni Bartolomeo <giovanni.bartolomeo@uniroma2.it>]

Dear All,

with the I-WORLD (formerly CATERINA) project proposal still in 
preparation (thanks Drummond for the helpful feedback, I just added 
those sections to the full proposal), I'm focusing now on some 
implementation issues for the E2R-II project 
(http://e2r2.motlabs.com/) demonstrator that will be shown in the 
forthcoming IST-Mobile Summit meeting (Budapest, june 2007 - very 
strict time!!!).

We would introduce here an XDI based data model, the problem they 
expect XDI will solve is a simple synonym issue: they've some xml 
documents (reflecting some given schema) and a relational DB 
containing some data which should be synchronized with these 
documents. Currently I'm aware of a document ("XDI Data Modeling: 
modeling data in XDI and mapping into legacy schemas") which just 
address this issue (with a special regard to RDBMS data model).

Some technical questions:
1. For what concerns the XDI prototype implementations, we've two 
possibilities: use code from svn.ootao.org, or a new implementation 
I've heard about (actually I don't know further details about it 
:-(). Which one do you think it will be possible to use for our demo? 
Are there specific license-restrictions - (E2R-II is a no-profit 
research project)?

2. Supposing the old implementation is used, for what concerns XDI 
functional interface, the XDI service as defined in svn.ootao.org has 
the following (taken from "IDcommons, Prototype XDI-based, Identity 
service, functional interface, phase 0, rel 1.0):
	-Graph State-Change Operations: set,setData,delete,addSynonym, 
removeSynonym,setSignedContract
	-Graph Retrieval Operations: get, head, getContract, post
We argue here that a further interface might be of help: 
NOTIFICATION. We'll try to implement something like that, with an 
underlying support mechanism which could be web based or not (we're 
looking at SIP for this purpose, but there are also solutions inside 
the web, e.g. I'm thinking at using RSS mechanisms); in any case 
we'll try and define an abstract primitive which could be mapped into 
different bindings, like it has been done with other XDI primitives.
If someone of you already though at it, his/her ideas would be of 
great help for us.

3. A further consideration is that this work might go into the 
direction of integrating XDI into Higgins IDAS. As by Andy 
suggestion, I've subscribed to their ML and following their work. 
Andy, to this purpose, do you think it makes sense to start with IDAS 
primitives instead of use directly XDI primitives (and, in case of 
positive, answer, what's the current status of integration)?

Sorry for the long mail :-)
Best Regards,
Giovanni