2 key proposals coming out of the last TC telecon & subsequent discussions

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

I just send the minutes of last Friday's telecon (sorry for the delay—things really backed up this weekend).

On it we discussed two key questions which arose from discussions Joseph and I had on finalizing the ABNF. Over the weekend I thought about them further and had a chance to discuss them with Markus and here is where we came out on each.

#1: USING THE EMPTY ADDRESS FOR THE OUTER ROOT

Despite our initial reaction, both Markus and I now agree with Joseph that this is a better approach than our previous approach of representing the outer root as (). Animesh also concurs. So it is now a formal proposal to make this change (and I have indicated that on https://wiki.oasis-open.org/xdi/GraphModelStructure).

Note that this means direct contextual statements will now use double slashes, e.g.:

  //=markus

  =markus//<+name>

It also means that "I am" statements will end in an empty address, e.g.:

  =markus/$ref/

  /$is$ref/=markus

#2: INVERSE CONTEXTUAL STATEMENTS

On last Friday's TC call we discussed that going to an empty path to represent the outer root created a problem for inverse contextual statements, since the inverse contextual predicate would be come just $is, i.e., an identity equivalence statement. Markus and I explored this and decided that it did in fact work semantically if the object of the inverse contextual statement was the complete context. For example:

  =markus//<+name>

  <+name>/$is/=markus<+name>

However after much discussion we felt that this was a different semantic that our current inverse contextual predicate, which only requires the parent context as the object. And we decided that we still needed this semantic. Thus our conclusion was that we should to keep $is() as the inverse contextual predicate. Our logic is that the parentheses become necessary simply because they enclose the actual XDI address of the outer root (which is empty) as a cross-reference so that it can be described by the inverse predicate $is.

So that means no change in our current syntax for inverse contextual predicates, e.g.:

  =markus//<+name>

  <+name>/$is()/=markus

Please post if you have any further thoughts about either of these two conclusions.

=Drummond