Minutes: XDI TC Telecon Thursday 1-2PM PT 2009-12-15

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

Date:  Thursday, 15 December 2009 USA

Time:  1:00PM - 2:00PM Pacific Time (21:00-22:00 UTC)

ATTENDING

 

Markus Sabadello

Giovanni Bartolomeo

Drummond Reed 

John Bradley 

AGENDA

1) XDI RDF BOX GRAPHING 

Drummond posted a PDF file of a new version of the XDI RDF box graphing document
with much more detail:

 

      http://www.oasis-open.org/committees/download.php/35590/xdi-rdf-box-graphs-v2.pdf

 

We reviewed it and the following suggestions were made:

 

      a) Change "XDI
Statement" to "XDI Address" in the titles

      b) Add a note about
literals and XDI addresses.

 

On the second point, we were reminded that we have never fully
closed on the question of literals and XDI addresses. Markus reminded us that
there was a proposal that there be a full XRI version of X3 where literals were
encoded as XRIs, and the XDI4J library's XDI Converter currently supports this.

 

#ACTION: Drummond to add this to the 2010 agenda items.

2) $HAS SEMANTICS AND ASSOCIATIVITY

 

The box graphing discussion then led us back to the topic that
brought it up, which was Giovanni's raising of an issue with the current link
contract examples where the semantics does not appear correct.

 

We focused on associativity of $has as an "operator".
Giovanni pointed out that it could be "associative" or
"non-associative", and if associative, it could be "left
associative" or "right associative". The definitions we were
referencing were from two different pages on Wikipedia:

 

      http://en.wikipedia.org/wiki/Associativity

      http://en.wikipedia.org/wiki/Operator_associativity

 

We agreed that the XDI statement:

 

      +a+b  INFERS      +a+b/$is$a/+b

 

Giovanni's question was then whether

 

      a+b+c INFERS      +a+b+c/$is$a/+c
OR +a+b+c/$is$a/+b+c

 

Drummond's answer was that it infers both. In English:

 

      house door keyhole

      A house door keyhole
IS A keyhole AND

      A house door keyhole
IS A door keyhole

      A car door keyhole is
a door keyhole but NOT a house door keyhole

 

We then talked about the link contract example:

 

      $contract=markus+signature

 

Giovanni pointed out that the semantics are wrong because:

 

      $contract=markus  DOES NOT INFER    $contract=markus/$is$a/=markus

 

He felt that this was not correct semantics, and that this was a
problem since XDI semantics should be mathematically expressable so it can be
used as a reasoning language.

 

Drummond agreed on both counts. He felt the example should be
revised to:

 

      $contract(=markus+signature)

 

This shows how $has remains "left associative" and how
cross-references can be used so that every $is$a statement inferred by a $has
statement is true, enabling semantic correctness.

 

Drummond noted that even though $has operations are left
associative, they are not commutative, i.e., you cannot change the ORDER of any
subsegments in the sequence.

 

#ACTION: Giovanni to confirm the correct associativity term on
the Wikipedia pages.

 

(Post-meeting note: This is DONE – Giovanni sent an email to the
list confirming that the correct term is that $has is "left
associative", meaning that in the absence of parentheses (cross-references),
the $has "operator" should be processed from left to right.

3) NEXT MEETING

The members agreed to take a holiday break. The next XDI TC telecon will be Thursday January 7 2010.

HAPPY HOLIDAYS!