Re: [xdi] Minutes: XDI TC Telecon Thursday 1-2PM PT 2010-01-07

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

Giovanni,

I apologize for any misunderstandings - in my attempts to take notes about what are often long discussions on our calls (in which I am an active participant), I am not always able to faithfully capture exactly what was said. I try to do it afterwards, but I am sure I make mistakes.

What makes it even harder is that sometimes we discuss and appear to reach agreement on a certain topic, only to later realize (sometimes hours later, sometimes days later, sometimes weeks later) that there are problems with our solution, and that we need to change something.

Some of the references you provide below are examples. We reached a number of conclusions last spring that, to my understanding, we later realized there were problems. I highlight one of those in particular inline below. Look for [=Drummond]

On Wed, Jan 13, 2010 at 12:21 AM, Giovanni Bartolomeo <giovanni.bartolomeo@uniroma2.it> wrote:

Dear Drummond,

In order to be very clear, could you please distinguish between proposals and agreements? Some suggested changes to your notes are below.

Kind Regards,
Giovanni

1) $HAS ASSOCIATIVITY

See the thread started by Giovanni at:

 http://lists.oasis-open.org/archives/xdi/201001/msg00011.html<http://www.oasis-open.org/committees/download.php/35421/xdi-rdf-box-graphs-v1.pdf>

and the response from Drummond at:

 http://lists.oasis-open.org/archives/xdi/201001/msg00012.html

For the sake of completeness, please check also the following of this thread:

http://lists.oasis-open.org/archives/xdi/201001/msg00014.html
http://lists.oasis-open.org/archives/xdi/201001/msg00019.html
http://lists.oasis-open.org/archives/xdi/201001/msg00022.html

We began discussion by revisiting some of the decisions we made about the
$has operator last spring.



# ACTION: Drummond will look for an email or minutes that he sent last
spring about deciding that $has$a still had a role.



We then progressed into talking about the nature of $has relationships, and
Drummond's assertion that from an RDF standpoint, a $has statement does not
actually assert that an RDF graph node has an outgoing RDF graph arc. That's
the job of a $has$a statement, and the reason that only $has$a statements,
and not $has statements, can have cardinality.

So what Drummond explained is that $has statements do not assert that the
$has object is an outgoing arc from the $has subject, only that the $has
object is a valid outgoing arc for the $has subject. The reification of the
relationship between the $has subject and the $has object, for example
reifying +a/$has/+b into (+a/+b), creates a new XDI RDF subject +a+b.

Please note that during the conference we DID NOT AGREE on this concept. Current specs (http://wiki.oasis-open.org/xdi/XdiOne/RdfGraphModel) says:

"$has describes the relationship between a node and an outgoing arc in the XDI RDF graph. The subject of a $has statement is the XRI of the node from which the arc originates, and the object is the XRI of the arc."

Bill pointed out that +a/$has/+b implies +a/+b and viceversa, i.e. they are logically equivalent statements.

[=Drummond]

While it is correct that Bill said that in the last call, I pointed out immediately after he said it that this was not the agreement that the TC had reached last spring (at least as I understood and documented it in the wiki spec page). In other words, +a/$has/+b does NOT imply +a/+b. Instead +a/$has/+b implies (+a/+b).

The parentheses, for reification, are absolutely necessary or XDI semantics falls apart. We captured this intent in the wiki spec page where we say: "When used alone as a metagraph predicate, $has is an assertion that reifies this subject/predicate relationship so that this reification can serve as a new XDI RDF subject node."

(However I believe that our discussions over the last month prove that we must provide much more detail in the spec about $has relationships. So this is an action item for all of us.)

 

This led us into a discussion of the relationship of XDI RDF graphs and
conventional RDF graphs. For example, the XDI statement +a/$has/+b diagrams
as a normal subject/predicate/object triple in an XDI RDF graph. But when
you convert that statement into what it describes in a conventional RDF
graph, it translates into a single RDF subject node identifier, +a+b. This RDF node "contains" the XDI RDF graph +a/$has/+b, but that is transparent to
conventional RDF, because conventional RDF does not have the notion of
contexts, i.e., graphs as objects of other graphs.

Please note that during the conference we DID NOT AGREE on any notion of "containment" above suggested.

What we agreed during last Spring (reported in http://lists.oasis-open.org/archives/xdi/200904/msg00021.html):

#1:        +x/$has/+y        INFERS:

           +y/$is$has/+x
           +x+y/$is$a/+y
           +y/$a/+x+y
           +x+y
           +x/+y

The corollary is also true, i.e.:

#1A:     +x+y                 INFERS

           +x/$has/+y
           +y/$is$has/+x
           +x+y/$is$a/+y
           +y/$a/+x+y
           +x/+y

[=Drummond]

Again, the problem here is that what I wrote in that message (dated April 30 2009) was what I believed at the time, but that was only a discussion thread, not a final TC decision. What we later decided (again, this is my understanding) was that there was a mistake in the above examples, because the final statement in each set of examples need to be REIFIED (i.e., in parentheses).

In other words, +x/$has/+y  INFERS  (+x/+y)  INFERS  +x+y

By contrast, +x/$has$a/+y  INFERS +x/+y   (notice, no parentheses)

Again, this is both my current understanding, and what we documented in the wiki spec at http://wiki.oasis-open.org/xdi/XdiOne/RdfGraphModel. See the examples (in yellow) listed in the

* $has and $is$has
* $has$a and $is$has$a

sections. In particular, the example in the $has and $is$has section says:

       +x/$has/+y     ==>     (+x/+y)     ==>     +x+y


I just checked and that page has not been modifed (other than the ABNF change I made two weeks ago) 
since May 14 2009. So this was a documented decision of the TC at that time.

I will start a separate thread on this specific topic so we can discuss it in advance of this week's call (and

of course we will discuss it on the call).

=Drummond