Minutes: XDI TC Telecon Friday 2012-09-21

xdi message

THE ETHERPAD LINK FOR TODAY IS:

--- Giovanni gave an update about his general work in the Linked Data field. He said that he is currently working on equivalence mining in Linked Data (see http://www.w3.org/wiki/TaskForces/CommunityProjects/LinkingOpenData/EquivalenceMining for an explanation). He presented a paper at LDOW 2012 http://events.linkeddata.org/ldow2012/

Markus demonstrated an example of the XDI connector architecture being developed in the XDI2 project. An XDI connector connects an XDI server to a native data source -- in this case Facebook. Markus showed the XDI map that has been created between the XDI community dictionary and the field elements in the native data source. The connector architecture also handles obtaining the necessary authorization token, in this case using OAuth 2.0.

What makes this so powerful is that the map itself is an XDI document, so any number of connectors can be written in any number of languages and all of them can use a single XDI map to access the native data source. And only this map needs to be updated when there are changes to the native data source.

***** DECISION POINTS FOR THIS CALL *****

Markus has made one final change per discussion on the list. See the Discussion page at:

https://wiki.oasis-open.org/xdi/XdiMultiplicity/Discussion

We did one final review of the proposal, with an explanation from Markus about the minor change so that attribute singletons used the same pattern whether expressed standalone or as members of a collection.

All TC members present on the call were in favor of the proposal. Although Bill Barnhill was not able to attend, prior to the call he sent a message indicating he too felt it should move to last call.

# DECISION: This proposal is moved to Last Call.

NOTE: Subsequent to the call, in preparing the minutes, Drummond realized another issue that needs to be addressed, so he is to wait to move it to Last Call until that issue has been addressed. He will send a separate email to the TC about this issue.

Markus walked us through the major elements of this proposal, which breaks a $add operation into each individual use case and clarifies the exact specifications for each one.

One key issue we discussed is whether the proposal should allow both the "short form" and the "long form" of $add statements when the $add operation is adding a new node. For example, the short form is:

.../$add/=markus+first$!(+name)

In this example, this single operation would:

Add the =markus context, the =markus+first context, and the =markus+first$!(+name) to the graph if not present

Not produce an error if any of these nodes were already present.

The long form would add a full XDI statement. For example:

.../$add/(=markus+first/()/$!(+name))

It was discussed whether the long form should be more granular and precise, i.e., whether it should produce an error if the node to be added already exists, or if any of the subject nodes do not yet exist.

However Markus suggested that it might be easier if a $add operation did not produce an error, but simply provided information in the return statement about whether the state of the graph actually changed.

This was preferably to doing a $get prior to a $add, because the state of the graph could change between the $get message and the $add message.

Another option is to add a new $set message that never produced an error, but always did the write to the graph regardless of whether the node(s) already exist.

Giovanni suggested another approach, which is to mimic what the HTTP protocol did by adding support for conditional requests. An example is a request to change a resource only if it has not been modified since a previous request.

Markus and Drummond both said this approach could use the same $if pattern we established for link contract policies.

We ran out of time and agreed this needs further discussion.

***** DECISION POINT QUEUE REVIEW *****

The decision queue stack is shown on the following three auto-generated Category pages:

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

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

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

Bill and Markus also added a number of proposals-to-do on:

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

We reviewed this list and agreed that overall it is a good roadmap to the priority of upcoming proposals. We also agreed that the next priority should be all the proposals related to XDI messaging and XDI message operations.

The highest priority should be the overall message format.

# MARKUS AND DRUMMOND agree to convert the current XdiMessagePatterns page into a formal proposal, including support for multiplicity syntax.

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

Following that should be proposals for each Individual message operation, such as the $add proposal discussed above.