feedback on k - Self References

xri-editors message

I'm going to give you more feedback on the k version today, but I wanted to get this started off because I'm not going to be able to get to it until much later today. This email is on "Self References".

This terminology and concept is still unclear to me. You say in section 1.1.1 that self references are equivalent to expressing the intent that an identifier is not to be resolved. And then you give examples (like the concept of love, the planet jupiter, etc).

I don't understand how the term "self reference" is related to this concept of not-to-be-resolved-ness. The modal case is a cross reference. The corner condition is the "top level cross reference" where the parentheses are "outside" the xri (btw, is this in the BNF? I don't remember seeing it.) -- the so-called xri(..) form. It seems the term "self reference" has marginal correlation to the "top level cross reference" situation and no relationship to the modal cross reference use.

In section 2.1.1.5, you talk about a use of XRIs where you want to talk about the identifier rather than the thing the identifiers "points to". Most identifiers schemes I know of don't attempt to do this through syntax, but rather through context, if at all. (ie the XML BASE spec says how relative URIs are interpreted based on the context in which they appear).

We've created a special syntax for this desire to refer to the identifier rather than the thing the identifier points to (this could lead to some interesting uses where you encode data in an xri:(..) form, but thats another topic). I don't see how this interpretation (referring to the identifier rather than the identified resource) extends to any other situation other than the special top-level cross reference (ie xri:(..) form) we've defined.

Furthermore, I wanted to continue talking about the (.foo.bar) form that loses me. In other email, we've discussed how to put a human name under a gcs namespace. Ie =JohnDoe, =John.Doe, =(.John.Doe), etc. The last example was given as a solution which appears to use the "self-reference" form. I don't understand this use at all. Can it be explained in terms of self-rerferences, or are we overloading the () construct?