RE: [xri-editors] feedback on k - Self References

["Drummond Reed" <drummond.reed@onename.com>]

Answers inline marked with ##.

-----Original Message-----
From: Wachob, Gabe [mailto:gwachob@visa.com]
Sent: Tuesday, November 04, 2003 3:01 PM
To: Drummond Reed; Wachob, Gabe; xri-editors@lists.oasis-open.org
Subject: RE: [xri-editors] feedback on k - Self References

Drummond-

OK, this is a little clearer. I was getting some concepts confused I
think.

Let me respond inline.


> session to drill through examples. I'm getting together in person with
> Dave M on Thursday at least an hour before the 3pm Editor's
> call - maybe
> then?

Yeah, we can do that.

> 1) When the cross-reference is the top-level authority segment, e.g.,
> xri:(urn:abc:123/456):2:3:45.
> 
> 2) When the cross-reference is the ENTIRE XRI, e.g.,
> xri:(urn:abc:123/456)

> Case #2 is simply case #1 with no additional delegation below
> the global
> cross-reference. This form of a cross-reference is thus a
> self-reference
> because it doesn't refer to any other resource representation - it IS
> the resource representation.

OK, I get the first sentence here (and I see how the BNF supports it),
losing me on the second.

questions:

1) "self-reference" is a subclass of "cross-reference"? That is, all
self-references are a special form of cross reference? Does this also
mean that "self references" are a specialized form (semantically) of
cross references?

##Yes, self-reference is a subclass of cross-reference. The only thing
that makes them a "specialized form" is that they stand alone, with no
parent context and no children. I'm not sure if this makes them special
"semantically" or not.


2) I get what you mean by saying that the identifier *is* the resource
representation (its also, basically, the resource itself). What I don't
get (still) is how this relates to the concept of "self-reference".
Seems more like we should call this "self-representation" or something.
Ugh. I don't know. See below

##Self-representation is also a good term, but (IMHO) so is
self-reference. It goes like this. Normally identifier A references -->
resource (representation X). In the case of a self-reference, identifier
A references --> resource (identifier A). Which is to say, itself. In
the context of identifiers (vs. resources), that's why I like
self-reference a little more than self-representation.



> As for the relative cross-reference syntax, the example I gave of
> xri:=(.John.Doe) is not a self-reference. It's a cross-reference under
> the "=" space. A self-reference would be xri:(=(.John.Doe)).

OK, that was confusing - i thought they were the same thing. My question
about =(.John.Doe) still stands, but its unrelated to self references
then, I guess.

## I'm not sure I remember your question about =(.John.Doe). My other
msg. was explaining that, IMO, this was a very precise way of capturing
all the information that the identifier being registered for this person
under the = authority was a concatenation of two words. Your mail said
that =JohnDoe, using MixedCase, did the same. I agree, so both are valid
ways of capturing the information (although I'd point out that the
former would survive a normalization transformation while the latter
wouldn't). But in the end, both are legal, so the issue of which one to
use seems arbitrary (at least at the level of the spec). Was that your
question?



> Does that answer your question?

Answers the self reference question, not the =(.John.Doe) question.

        -Gabe