XDI TC Notes
Following are the notes of the unofficial telecon of the XDI TC held on:
Date: Monday, 09 October 2017 USA
Time: 9:00AM - 10:00AM Pacific Time (16:00-17:00 UTC)
The TC operates under a standing rule approved 17 July 2008 under which the TC does not hold regular official meetings and conducts all business by electronic ballot only. Unofficial weekly meetings are held to enable discussion among members but no business is conducted nor actions taken.
ATTENDING
Drummond Reed
Markus Sabadello
Phil Windley
Joseph Boyle
NOTES
Positioning XDI as a simpler way to do RDF
Markus reviewed a recent insight he had about how to explain XDI’s relationship to RDF. He said that we have discussed many times how XDI can be considered an extension or “profile” of RDF, and that an XDI graph can be treated as a valid RDF graph. He also referred to a slide that Drummond has used (below) to explain XDI in light of the evolution of other similar standards.
What Markus realized is there is a different way to describe the relationship that would be of greater interest to developers: position XDI as an easier-to-use form of RDF. He suggested a table that would look like this:
Original technology | Easier-to-use version |
_javascript_ | CoffeeScript & TypeScript |
_javascript_ queries | JQuery |
Java | Scala (& others) |
CSS | LESS & SASS |
Markus explained that, in each case what has evolved is simpler versions of the more powerful language. These simpler versions cannot do anything beyond what the original language can do, but they make it all easier to use. They are shortcuts for using a more powerful technology. For example, there are situations where you need 100 lines of code on the left side, and only 20 lines on the right side. And it is often more intuitive to understand the simpler language.
So this same analogy can be used in comparing XDI to RDF. Drummond asked if that was true even for the XDI concept of contexts. Markus felt that yes, even context arcs can be modeled as RDF relational arcs of a certain type. The XDI profile of RDF would then add the constraint that context arcs must form an acyclic DAG (a tree).
Markus felt that this approach could not only help in explaining XDI to RDF users, but it could be attractive for some specific use cases for RDF graphs that use JSON-LD and thus naturally use a tree structure. We can advocate that RDF graphs that follow the XDI profile can become much simpler and more useful for developers to work with. It would also be like JSON-LD in that it defines a new serialization format. It also adds a simple ontology language.
Joseph likes this approach in it makes it simpler to understand the purpose of XDI as simplifying RDF. Phil likes positioning XDI as the new, simpler version of RDF.
Drummond also noted the analogy with XML and JSON. Even though JSON is not a strict subset of XML, it is much simpler, and as a consequence has seem much faster adoption.
XDI Binary Serialization and Signatures
Markus and Drummond want to propose a solution to the longstanding debate over the best way to serialize XDI graphs—both for the over-the-wire protocol and for XDI signatures. The proposal is:
Define a canonical format for XDI statement serialization (relatively easy).
Define a canonical binary compression format (also relatively easy).
The resulting compressed binary format can be: a) digitally signed, and b) used as the protocol payload.
We talked about Linked Data Signatures and their requirements for canonicalizing an RDF graph before signing it. Drummond explained how very much easier it is to canonicalize.
Phil said he likes the approach used with HTTP, which is that you specify how you do compression. But he is leery of requiring a binary format or a compressed binary format. The standard textual format should always be an option. Phil describes it as one format, with options to compress it. In that case, it makes a simple progression:
Start with the standard graph.
Canonicalize it.
Add signatures if needed.
Add binary format if needed.
Add compression if needed.
Link Contracts as Capabilities
Markus and Drummond both attended the Rebooting the Web of Trust conference in Boston last week, and one of the highlights was an in-depth discussion of the object capabilities security model with Mark Miller, the world expert on the topic. Although we did not have time to go into it beyond a short discussion with Mark, we would very much like to explore XDI link contracts as capabilities, and see how much of the overall model that Mark has been advocating either has been or can be implemented by link contracts.
We agreed that:
We would try to hold a session on this topic at Internet Identity Workshop next week in Mountain View.
We would make it a topic of our next call (which, due to IIW, will be in 2 weeks).
NEXT REGULAR CALL
We will NOT hold a call next week due to Internet Identity Workshop. The next call will be the following week at the usual time (Monday 9AM PT). The link where agenda items can be posted for the next meeting is: https://docs.google.com/document/d/19oDl0lbb56Grehx2a5flZnhrgnua5l8cVvC_dJ8fTXk/edit?usp=sharing