XDI TC Minutes

Following are the minutes of the unofficial telecon of the XDI TC at:

Date: Friday, 30 January 2015 USA
Time: 09:00AM - 10:30AM Pacific Time (16:00-17:30 UTC)

ATTENDING

Les Chasen
Joseph Boyle
Peter Davis
Drummond Reed
Markus Sabadello

GUESTS

Steve Greenberg

REGRETS

Phil Windley

NEWS & UPDATES

Report on XDI Integration for Personal Data Stores

Steve Greenberg explained that the U.S. NPO (National Program Office) has asked Privo to explain the case for integrating with PDS (personal data stores). Steve would like help from TC members in explaining how XDI can enable interoperable PDS. Les volunteered to help him, and Drummond said that he’d be glad to share drafts of the new “How XDI Builds on RDF” white paper that he’s working on.

PRESENTATIONS/DISCUSSIONS

Report from XDI Editors Subcommittee

https://wiki.oasis-open.org/xdi/XdiOneSpecs
https://github.com/OASIS-XDI-Technical-Committee/xdi-spec-docbook

Due to insufficient attendance, the XDI Editors call did not take place.

JSON Triples Serialization Format

We reviewed a new proposed serialization format posted by Drummond that combines features of the flat and the tree formats.

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

Drummond’s message to the TC list, introducing the new format:

https://lists.oasis-open.org/archives/xdi/201501/msg00012.html

Joseph’s thoughts in reply:

https://lists.oasis-open.org/archives/xdi/201502/msg00000.html

Steve said that it’s very clear where the subject breaks are. And the inner graph makes sense.

Steve advocates “round-tripping” in and out of a canonical format.

Steve asked: “Is the expectation that every developer is going to learn XDI addressing syntax in order to use a typical XDI library or not?”

Joseph: “If you are working with addresses, some of the XDI addressing syntax is going to show through.”

We discussed the question of whether any JSON format obviates the need for the single-line display format. Drummond said that he does not think that is the case; the single-line display format always has its value because it fully exposes XDI addressing. Markus said he always liked the display format, because it maintains the rule that you have exactly one line per arc in the graph.

Different formats may be more or less efficient depending on the kind of query that should be performed, e.g. walk the graph structure step-by-step, vs. retrieving only one literal at a fully known address, vs. searching all nodes by a certain type.

Drummond said that what all the XDI JSON formats have in common is that the XDI common root is represented by an outer JSON object, and the XDI subjects are keys in that outer JSON object.

Peter said that in the XDI RDF profile that he’s working on, he’s using N-Quads and the RDF SPARQL-over-HTTP profile. A typical query would be, “Show me all the email addresses I can get out of this graph.” Very often a developer just wants to query for types.

Steve made many good points about the usability of XDI from a developer’s perspective. He said that the concepts reflection, discovery, and “generators” are important, i.e. the ability for a developer to get something out of an XDI object as easily as possible. The question is, would a developer be able to work directly with XDI data in its serialized JSON form via a standard JSON API, or would they always need a specialized XDI library?

Steve also brought up JQ, a tool for querying JSON. He suggested that he’d like to have a tool like that for XDI. For example, “For this person, show me all their email addresses, and all of their types.” All of the addressing and link contracts and permissioning should be behind the scenes.

Steve used the analogy of a sculptor working with clay—they need to be able to work with big broad lumps before getting down to finer features.

Visual Graph Notation Format

We reviewed Drummond’s suggested revision.

https://lists.oasis-open.org/archives/xdi/201501/msg00013.html

No objections were raised to adopting this graph notation format.

In addition to talking about the graph notation format itself, we also had a productive discussion around the graph shown in the specific diagram, especially about how roots in the graph enable peer-to-peer discovery.

Websocket Binding

We did not discuss this topic.

Clarification of XDI Root Node Syntax