Minutes of SPECIAL XRI TELECON 2-3PM PT Tuesday 2009-01-13

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

Following are the minutes of a special XRI TC telecon held:

	2-3PM PT (22:00-23:00 UTC) Tuesday 13 January 2009

ATTENDING

John Bradley
Les Chasen 
Drummond Reed

REGRETS

Bill Barnhill 
Markus Sabadello 
Paul Trevithick 


DISCUSSION

We discussed the following three main drivers for the proposed XRI 3.0
syntax (http://wiki.oasis-open.org/xri/XriThree/SyntaxAbnf): 

1) The GCSDelimiter proposal...

	http://wiki.oasis-open.org/xri/XriThree/GcsDelimiter

...is required by the XDI addressing rule based on the XDI metagraph
predicate $has as documented in section 5.3 of...

	http://wiki.oasis-open.org/xdi/XdiOne/RdfGraphModel

We had a long discussion about the $has examples, which included the
suggestion that the +a+b+c+d example needs to be clarified that it is only
specific combinations of the XDI RDF statements shown that produce the XRI
+a+b+c+d (Drummond will fix that for the XDI TC).

Drummond summarized the requirement this way: the XDI RDF graph needs to be
able to identify a subset of the graph constituting a path through the
directed graph as a single XRI subsegment. $has is the XDI RDF metagraph
predicate that permits an XDI RDF statement to decribe the directed arc
between two nodes.


2) The XrefDelimiter proposal...

	http://wiki.oasis-open.org/xri/XriThree/XrefDelimiter

...is required by the XDI reification addressing rule documented in section
4.2 of the same page linked above. This allows any XDI statement -- which
itself is a directed graph but which must include forward slashes in the XDI
address to separate the subject, predicate, and object -- to itself be
treated as an addressable node.


3) Further advantages of both these proposals as reflected in the proposed
XRI 3.0 ABNF (http://wiki.oasis-open.org/xri/XriThree/SyntaxAbnf) include
the following:

	* It produces a shorter, cleaner, clearer ABNF with fewer
exceptions and irregularities that the XRI 2.0 ABNF (also shown at
http://wiki.oasis-open.org/xri/XriThree/SyntaxAbnf). 

	* In this ABNF, all XRI delimiters are now "atomic", i.e., they each
can be used independently to delimit an XRI subsegment anyplace within an
XRI segment, rather than the set of irregular limitations and exceptions in
XRI 2.0 ABNF.

We have long recognized that there are also some disadvantages -- that there
is some cost in backwards compatability, which should be minimized due to
the transition to new XRI resolution service endpoints with
new URIs, and that there are differing opinions on human-readability of the
proposed 3.0 syntax.

Following this discussion, Les was still unsure of the XDI requirements,
however he will defer to the assessment those actively working on XDI.

NEXT STEPS

# DRUMMOND to do a final check-in with Nick Nicholas, who was not able to
attend this call.
# DRUMMOND AND JOHN to begin preparation of the first Working Draft.
# JOHN to start preparing an analysis of XRI 3.0 parsing and any affect it
might have on XRI 3.0 resolution.

=Drummond