[xri-comment] Comments on the requirements document

[Lars Marius Garshol <larsga@garshol.priv.no>]

Hi everyone,

I'm one of the people working on topic maps standardization, both the
published subjects work here in OASIS as well as the core standards
work in ISO.

I've read through the requirements document and also been introduced
to this work on IRC by Gabe Wachob. Gabe was quite helpful, but I have
to say the document itself does not give a very clear picture of what
the TC intends to do.

The first thing that would be useful would be a clearer list in
section 1 of the deliverables. That is, names (or even numbers) for
the deliverables, and a bit more detail on what is actually going to
be in each. The first is quite clear, but what's in the second? A
protocol? A data model? An interchange format? Anything else?

In section 2 the definition of "directory" appears to be missing. It
may also be useful to add that "resource" as defined there appears to
be identical to the "resource" of RFC 2396 and the term "subject" in
topic maps. Also, a "version", is that truly a state of the resource
or just a state of the data about it?

In section 3 it would help a lot of the requirements were reorganized
so that it was clear which requirements applied to which deliverable.
Also, in 3.1 points 2-4 are very vague. I can't really make out what
they are saying. Number 9 seems to imply that there will be a data
model, but nothing has been said about any such model earlier. To see
a list of the technologies the TC intends to create would make
evaluating these requirements much easier.

In section 3.2, point number 2 seems to say that it should be possible
to encapsulate URIs from other URI schemes within XRI URIs, almost
like with the tdb: URIs proposed by Larry Masinter. Is that really
what it is saying? If so, why?

Point 4 (of 3.2) talks about a "root", but I have no idea what that
means.

Points 6-7 (of 3.2) seem internally contradictory. Is there going to
be a single XRI URI scheme, or multiple ones?

Point 8 (of 3.2) talks about "parent resources", but what they are is
not clear. Is the XRI data model hierarchical? How does that fit with
attributes? And are attribute types themselves resources, BTW?

At the end of section 3.2 there is also talk of an XRID namespace.
What is that? Does it have any relation to the URI scheme(s)?

In section 3.3, point 3, should it say "use" rather than "support"?
One missing requirement in this section seems to be an interchange
format for the data model one presumes is going to be part of this.

In section 3.5, point 2, there is talk of "completely decentralized
resolution". What is that?

In section 3.6, point 2, it is said that the specifications must be
language-independent. What does "language" mean in this context?
Surely not natural language?


Lots of questions, and rather less advice. I hope it is useful even
so. 

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >