Minutes: XDI TC Telecon Friday 2012-04-27

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

Date: Friday, 27 April 2012 USA

Time: 9:00AM - 10:30AM Pacific Time (16:00-17:30 UTC)

ATTENDING

Cameron Hunt

Kari Lippert

Mike Schwartz

Drummond Reed

REGRETS

Joseph Boyle

THE ETHERPAD LINK FOR TODAY IS:

http://xdi.idearpad.com/62

1) WELCOME TO BITWORLD

Bitworld has joined OASIS as a non-profit member, and Cameron Hunt, Kari Lippert, and Phil Windley will be participating in the XDI TC under the Bitworld membership.

Cam explained that Bitworld is using FUNF from MIT, and wants to be able to run it on mobile devices like Android. FUNF uses SQL Lite, and thus Bitworld wants to look at writing XDI statements directly into the SQL Lite database.

2) TEST TOOL

Mike gave a demo of the test tool developed by the OpenXDI Project.

http://seed.gluu.org/oxTestTool

More tests are coming next week to show the following features:

http://ox.gluu.org/doku.php?id=oxserver:listoffeatures&s[]=features

Mike suggests that we run the tests on the OASIS-sponsored server. This coming week at Internet Identity Workshop Mike and Markus and other TC members that are attending will be able to discuss it.

3) XDI QUERY

Mike said he was still working on putting the requirements in priority order. This may be a subject we discuss at IIW next week.

4) MODELING DEVICE IDENTITY

Drummond shared a graph pattern for device identity that Bitworld needs. This brought up a number of questions related to device identity, device ownership, and graph organization.

Cam mentioned that in his experience there are three key factors involved with graph modelling:

how to model so that persistence and searching is optimized

how to model so the presentation layer is optimized

how to model so other automated processing (e.g., inferencing) is optimized

The goal is to minimize the model remapping between persistence, processing, and display.

Mike explained that the reason the trend in LDAP has been towards "flat" is that you don't want developers to need to know where there data is, but rather to make it easy to reference.

Drummond agreed that flat is better because that way each entity can be as independent.

Mike then explained that the OpenXDI implementation was less concerned about keeping the graph flat because XDI makes it easier to make extensive use of indexing.

With regard to Bitworld's design choices, Mike also suggested looking at the oxPlus design:

https://svn.gluu.info/repository/openxdi/code/docs

specifically https://svn.gluu.info/repository/openxdi/code/docs/oxPlus/seedOverview.ppt

Mike also said that the location of the link contracts is critical.

5) NEXT CALL

Even though Internet Identity Workshop is next week, we will still hold the next call next week at the regular time. One agenda item will be a summary of the XDI discussions at IIW.

xdi message