Minutes of XRI TC Telecon Thursday 2006/2/16

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

Following are the minutes of the unofficial XRI TC telecom on: 

Date:  Thursday, 16 February 2006 USA (Friday morning Asia)
Time:  4:00PM - 5:30PM PT

Attending:

Gabe
Drummond
Steve
Wil
Nat

This call focused entirely on closing issues for Editor's Draft 05 of
Working Draft 10 of XRI Resolution 2.0.

Key to the minutes:

*  = discussion
** = decision
#  = action item

1) SIMPLIFIED SERVICE ENDPOINT SELECTION LOGIC PROPOSAL

See:

http://www.oasis-open.org/apps/org/workgroup/xri/download.php/16727/resoluti
on-v2.0-wd-10-ed-05-flowcharts-v2.pdf

** After a long and excellent discussion, there was consensus that the best
way to implement the service endpoint selection logic to account for all use
cases was to specify an optional "match" attribute for the Type, Path, and
MediaType elements. This attribute would take four values as shown below.
The default would be "content" and this would be the assumed value if the
attribute was not present.

Semantics                          Representation
________________________          ___________________
        
Match anything but null           <Type match="non-null">

Match null                        <Type match="null">

Match any including null          <Type match="any">

Match content of element          <Type match="content">foo</Type>

Match content or null             <Type match="content">foo</Type>
                                  <Type match="null">

# Drummond to add to Editor's Draft 05.


2) APPEND ATTRIBUTE PROPOSAL

This is also discussed in the document above, and addresses the issue of how
to handle service types that do not have a specified URI construction
algorithm.

** It was agreed to use the "append" attribute but place it at the URI
level, not the Service level.
** The default value is "local".
# Drummond to add to Editor's Draft 05.


3) NAMING OF ABSOLUTEXRI AND RELATIVEXRI ELEMENTS

It turns out these are not good names because the criteria is not strictly
absolute and relative. New proposed names are CanonicalID and LocalID.

** There was consensus on the names "Canonical" and "Local" but not
necessarily to add ID.
# Drummond to use editorial judgement in Editor's Draft 05 and then get
feedback.


4) ADDITION OF IDREF ATTRIBUTE TO XRD ELEMENT

This attribute t allows the same XRD to be included in an nested XRDS
document without needing a different xml:id attribute value.

** Agreed.
# Drummond to add to Editor's Draft 05.


5) RECOMMENDED/REQUIRED VALUE OF XML:ID ATTRIBUTE

It has been recommended that the xml:id attribute value be a
UUID because that's a legal xml:id value. We need to close on this, as well
as determine what might be necessary to ensure a UUID will work as the
xml:id value (since it must be an NCName).

** Specifying UUID is the easiest solution.
# Drummond to include in Editor's Draft 05.


6) CONSTRUCTION OF URIS WHEN SEP URI INCLUDES A QUERY PARAMETER

Les suggests we just use raw concatenation on all cases.

** After considerable discussion, there was consensus on Les's
recommendation.
** Resolution parameters will always be stripped off.
** No escaping will be applied -- an attribute to control escaping can be
added if needed in a future version of the spec.
# Drummond to include in Editor's Draft 05.


7) HTTP X-HEADER INPUT PARAMETERS

Open issue as to whether we still need these.

** Agreed we do not need it due to the fact that they can be specified in
the QXRI.
# Drummond to remove from Editor's Draft 05.


8) THREE DIGIT ERROR CODES

Gabe suggests we adopt the same overall model as HTTP, i.e., three digit
error codes with first digit indicates major category of error, second is
more specific category, third is finest grain. This allows different levels
of error processing.

** Agreed.
# Wil to draft a new proposed set of error codes by updating the error code
table in Editor's Draft 04 and sending to Drummond w/in 48 hours.