Minutes: XDI TC Telecon Friday 2013-03-22

[Drummond Reed <drummond.reed@xdi.org>]

XDI TC Minutes

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

Date:  Friday, 22 March 2013 USA
Time:  09:00AM - 10:30AM Pacific Time (16:00-17:30 UTC)

ATTENDING

Joseph Boyle

Animesh Chowdhury

Markus Sabadello

Drummond Reed

Les Chasen

REGRETS

Phil Windley
Bill Barnhill

NEWS & UPDATES

IIW PLANNING

Internet Identity Workshop is coming up starting May 7. We anticipate a high degree of XDI discussion there, so we agreed we should coordinate planning XDI sessions among those of us who will be attending.

PRESENTATIONS/DISCUSSIONS

SPEC DRAFTING MECHANICS DISCUSSION

Drummond, Markus, and Joseph reported on the excellent conversation they had about spec drafting mechanics with Chet Ensign of OASIS. He explained how the overall process has been revised, and also the template choices made by other TCs. Microsoft Word and Docbook are the two most common choices.

With regard to multi-part specifications, Chet shared this link to the AMQP specifications for a good example of how they work:

  http://docs.oasis-open.org/amqp/core/v1.0/

Chet volunteered to set up a call with Ken Holman, co-chair of the UBL TC, for use to discuss the Docbook publishing system they have set up.

# DRUMMOND - send email to Chet to set up a call with Ken Holman.

DECISION POINTS FOR THIS CALL

This week's decision queue is the following set of proposals:

ABNF AND CONTEXT FUNCTIONS

Once again, this has been the subject of much work over the past week.

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

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

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

The main focus of the call was the 7 open issues listed at the bottom of the page:

 https://wiki.oasis-open.org/xdi/ContextFunctions#OPENISSUES

Following are notes on our discussions of each.

#1 LITERALS

Joseph suggested that the literal encodiing format be the same as JSON. Then there is no need for an XDI endpoint to do any parsing of an XDI literal string. Joseph felt this would be much better than data URIs.

This broadened into a discussion of whether all of the JSON datatypes should be supported natively, i.e., whether in the XDI display format the JSON datatype (e.g., object, array, string, number, true, false, null) should just appear after the second forward slash.

This includes JSON null, since again it would standardize representing null values the way JSON does it.

Present consensus: we should adopt using all the JSON datatypes natively, as it would be the most intuitive to developers.

Drummond noted that we would want to define $words (or $letter abbreviations like $n and $b) for all the JSON datatypes except string.

Joseph noted that we should still discuss a recommendation on canonical encoding for JSON strings.

Joseph and Drummond both suggested we drop the term “literal” in the ABNF when referring to characters and not an XDI literal node.

On the issue of the predicate for a literal, Drummond suggested using <> because: a) it is parallel to using () for contextual arcs, b) it is better than using a blank predicate, which invites too much confusion, and also may conflict with RDF.

Markus proposed using <> as the XDI object for a literal reference.

Present consensus: we will use <> as the predicate for a literal arc and <> for the object when the statement is a literal reference.

#2 MEMBER COLLECTIONS

Present consensus: we should support collections of collections.

NOTE FROM DRUMMOND: In starting to implement this after the call, I ran into issues. I will send a separate message to the email list.

#3 MEMBER ORDERING RULES

First, Markus pointed out that in the Option A and Option B work with the $deref parameter.

We then discussed that both options are supported by the proposal as written, and both options have use cases. Markus and Drummond share some concern around Option A because it may be deceptive to developers in that it makes it look like you can easily add, delete, and move members within an array, when it does not work that way. (However there is the possibility such functions could be supported in the future.)

Present consensus: proceed with the current proposal, where [digit] is an ordered member, and may have a value or a $ref to a value.

#4 ORDERING CONTEXT RULE

Markus prefers the *digit format. Drummond does slightly, and Joseph. Animesh however made a strong case for using a plain digit. Phil also strongly recommended using a plain digit in his email to the TC.

Present consensus: proceed with the current proposal to use plain digits.

#5 MEMBER TYPES

On this question, Markus and Animesh leaned towards the second option (typed collections and untyped members). Phil also favored that option in his email. Joseph did not have a strong view. Drummond prefers either the second or third option.

Present consensus: proceed with option 2 -- typed collections and untyped members.

Animesh raised a question about the difference between XDI entities and XDI attributes, given that from a programming perspective, both are classes and only data objects have literal values.

Drummond said the key difference is that attribute objects can be associated with data objects and entity objects cannot be associated directly with data objects.

#6 VARIABLES

Markus had to leave the call (as we passed the 2 hour mark), but was in favor of the first option - the $(variable) syntax.

Animesh suggested %variable% if we are willing to use % as a delimiter.

Drummond said that both suggestions are reasonably distinctive. Examples:

  …$do$if/$true/($($from)/$is/=example)

  …$do$if/$true/(%$from%/$is/=example)

Joseph asked if we wanted to consider the option to declare variables using a $word for this purpose, and then use them in standard syntax. To his knowledge, only Perl uses prefix characters for variables -- @ for an array, $ for a scalar, % for associative arrays, and another symbol for a hash.

# JOSEPH will write up his thoughts on how to do the variable syntax by variable declarations.

Between the two options above, Animesh thinks that the % option is easier to read. Drummond agrees.

No consensus yet.

# DRUMMOND to post this in the minutes and check with Markus and Phil.

#7 PATTERN VARIABLES

We had very little time left to discuss this, but Markus annotated the page with the warning that we should not use our own variable syntax, since we will have template patterns that incorporate variables. Drummond agrees.

So we should adopt a variable syntax for template patterns that is not the same as our XDI variable syntax.

Drummond suggests we look at using double chevrons, e.g.:

  <<from-peer>>/$add/<<from>>{$msg}[<<id>>]

No consensus yet.

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

See also this list of proposals that need to be developed:

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

NEXT CALL

The next call is next week at the regular time.