Minutes: XDI TC Telecon Thursday 1-2PM PT 2009-02-12

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

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

Date:  Thursday, 12 February 2009 USA
Time:  1:00PM - 2:00PM Pacific Time (21:00-22:00 UTC)

ATTENDING

John Bradley 
Markus Sabadello
Nick Nicholas
Tatsuki Sakushima 
Giovanni Bartolomeo
Drummond Reed


1) CONTINUED DISCUSSION ON $HAS SEMANTICS AND +X/+Y/+X+Y

See Giovanni's message posted here:

	http://lists.oasis-open.org/archives/xdi/200902/msg00017.html

The topic was the list discussion of the "famous" +x/+y/+x+y statement, and
Giovanni's feeling is that it might be a bit "dangerous" in the XDI RDF
model.

We discussed that $has semantics apply to the combination of ANY two valid
XRI subsegments, including, at the very end of the spectrum, the binding of
a delimiter to a literal. For example, =/$has/drummond produces =drummond,
and +x/$has/*y produces +x*y. The ABNF for an XDI address on the Addressing
and RDF Graph Model page
(http://wiki.oasis-open.org/xdi/XdiOne/RdfGraphModel) allows any valid XRI
segment to serve as an XDI subject, predicate, or object.

We next discussed another concern of Giovanni's with some "strange"
statements that can be made with $has. This is related to this email:

	http://lists.oasis-open.org/archives/xdi/200901/msg00054.html

It is the example about balls and their colors, where +x is +ball and +y is
+color. This produces the XDI RDF statement +ball/+color/+ball+color. But in
the same document we have also +ball/+color/+red. So you have two different
objects of the same predicate, which is fine in RDF, but one (=red) is an
instance of color and the other (+ball+color) is a subclass of color.

After some discussion, we realized that what was deceiving about these
examples is that they confuse the dictionary/class space (Bill would say
"T-box") from the individual/instance space ("A-box"). For example, in some
email threads we had talked about having +ball/+color/+ball+color and
+ball/+color/+red in the same graph. While this is technically possible, it
doesn't make sense because the first statement is appropriate to a
dictionary and the second would be more appropriate to an instance, except
that +ball is not a typical instance. More typical would be
=jbradley+ball/+color/+red.

We talked about order in $has statements, and the fact that, due to the
left-to-right structure of RDF itself, with $has statements the subject is
always the context and the object is always the class being subclassed. For
example, +x/$has/+y means +x+y/$is$a/+y, not +x+y/$is$a/+x.

Drummond noted that we won't be ready to complete a full first draft of the
first spec (XDI 1.0 Addressing and RDF Graph Model) until we finish the RDF
transformation rules. He's going to try to concentrate on finishing out that
portion of the wiki page
(http://wiki.oasis-open.org/xdi/XdiOne/RdfGraphModel) as the next step.