Minutes: XRI TC Telecon 2-3PM PT Thursday 2009-08-27

["Drummond Reed" <drummond.reed@cordance.net>]

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

Date:  Thursday, 27 August 2009 USA
Time:  2:00PM - 3:00PM Pacific Time (21:00-22:00 UTC)

ATTENDING

Will Norris 
Nat Sakimura
Scott Cantor 
Drummond Reed 
John Bradley
Joseph Holsten

REGRETS
Eran Hammer-Lahav


1) PUNCHLIST OF XRD 1.0 ISSUES

See the updated punchlist of remaining open issues for XRD 1.0 WD05 (you can
disregard the list of closed issues - they are just there for reference as
we complete the punchlist process): 

	http://wiki.oasis-open.org/xri/XrdOne/SpecHome 

The call was spent going over these issues. ALL DECISIONS NOTED BELOW ARE
RECORDED ON THE WIKI PAGE.


1A) DROP MATCH ATTRIBUTE FOR SUBJECT AND ALIAS?

We discussed the history of the need for matching a subject in a set of
subjects vs. the need to use templates.

John and Drummond confirmed that this should not affect XRI matching.

DECISION: Yes, we will drop the "match" attribute on both Subject and Alias.


1B) HOST-META SUBJECTS

Will clarified that host-meta XRDs will need to have some kind of URI
representation, but that's an issue for the host-meta spec, not for the XRD
spec.


1C) XSD SCHEMA PROPOSAL

Scott explained that we want both wildcards (elements from other namespaces)
and optional elements from included namespaces. That creates a possible
conflict between the two, and on that problem XSD validators choke.

Solutions: 

a) Wrap extension elements.
b) Wrapping the signature elements.

We discussed both options. The consensus was that the former is preferred -
it keeps the simple case simple (i.e., when you have no extension elements).
This was the solution that SAML chose.

We also discussed ordering of schema elements, and the consensus was that
ordering should be determined by processing optimization (with human
readability not really a factor).

The consensus for ordering in XRD:
- Expires
- Subject
- Signature
- Extensions
- Choice: Alias, Type, Link

The consensus for ordering in Link:
- Subject
- Extensions
- Choice: Rel, MediaType, URI, URITemplate, KeyInfo

Scott also asked: do we want to say anything about empty elements? He
suggests we not allow them, and say so explicitly in the spec.

There was agreement that we should specify that up front, along with any
other general processing rules (such as XML datetime format restrictions)
that apply both to XRD elements and any extensions.

#ACTION: WILL - add up front section with overall processing rules for
elements.

We also confirmed that TypeType is okay.


1D) TYPE SELECTION

Drummond pointed out Eran's email to the list a few weeks ago on this
question.

#ACTION: WILL to add wording in the Processing section for Linked XRD
processing to talk about looking for XRDs with a particular Type value.


1E) DEWITT'S FEEDBACK

The general consensus was that this had mostly already been answered by Eran
on the list. However one specific item Dewitt did touch on is the currently
required round-robin behaviour for elements of equal priority. 

We discussed this at length and agreed that it was no longer necessary to
require round-robin behavior. In fact the discussion led to the proposal
that we eliminate priority and process everything in document order. 

Scott pointed out that it is only data bindings that do not preserve
document order, and even these can be made to do so if necessary.

#DECISION: The reduced complexity is worth removing the priority attribute
and just using document order -- this will make implementations considerably
simpler.

#ACTION: DRUMMOND to send a special message on this to the list after
posting the minutes.

#ACTION: SCOTT to go ahead and reflect this in the schema he posts.

#ACTION: WILL to see if there is any backlash on the list discussion,
otherwise to reflect this in WD05.


2) XRI SYNTAX 3.0 WORKING DRAFT 03 STATUS/ISSUES

Drummond reported he is working steadily on WD03; no issues yet.


3) NEXT CALL

Next week; regular time.