Re: URI documents

Al Gilman (asgilman@access.digex.net)
Tue, 6 Jan 1998 10:07:25 -0500 (EST)


From: Al Gilman <asgilman@access.digex.net>
Message-Id: <199801061507.KAA21828@access5.digex.net>
Subject: Re: URI documents
To: Harald.Alvestrand@maxware.no (Harald Tveit Alvestrand)
Date: Tue, 6 Jan 1998 10:07:25 -0500 (EST)
Cc: connolly@w3.org, masinter@parc.xerox.com, leslie@bunyip.com,
In-Reply-To: <199801061202.NAA16175@dokka.kvatro.no> from Harald Tveit Alvestrand at "Jan 6, 98 12:59:34 pm"

to follow up on what Harald Tveit Alvestrand said:

Dan Connolly:
> >Since Larry asked, I'll (re-)state the W3C opinion: we're
> >heavily invested in the notion of a single, extensible universal
> >address space:

Al Gilman:
That's one of the sources of the problem:  too much emphasis on
singleness.  If you frame the problem as understanding a class
of text strings that function as identifiers, in the context of
a web of string and name classes, you will get unstuck from the
impasses.

Harald Alvestrand:
> 
> The problem, to my mind, is that we really have two deep axioms
> here:
> 
> - The class of identifiers that, roughly speaking, start with
>   a short string and a colon, and go on in a charset-limited way.
>   All the URI axioms you cite are axioms of that class.
> - The class of identifiers that, in addtion to being of the first
>   class, obey certain additional rules, such as hierarchy,
>   hostname representation and so on.
>   None of this is necessary for the URI axioms; they are vitally
>   necessary for today's day-to-day usage of the World Wide Web.
> 
> (Everyone with me so far?)

Al Gilman:

Almost.  You are very close.  But you are not getting quite deep
enough.  It is not clear that everything that gets called an URN
needs to have all the armor of your first class.

There is a level of abstraction for names where the relative and
absolute URLs that retrieve the same resource are viewed as
variant forms of one name.  We may need to link the URI documents
to documents which capture this view of the naming agenda in
order to have adequate documentation of what is going on.

Harald A:
> 
> If separation is not the Right Way, the issues are of course slightly
> different....

Al Gilman:

The Right Way is a web of sub-documents specifying a web of
classes.  So long as the semantics of the sub-documents is strong
enough, the grouping into documents becomes a non-problem.  The
IETF is being materially hampered in working this situation by
its failure to adopt more powerful linking and semantic modeling
norms for its working documents.

The canonical sub-document web for this topic is:

Ancestor classes [standards track]:

	a string class that survives a variety of environments

	a name class that guarantees certain semantic properties

Resource classes [BCP track]:

	addressing imported from the Internet platform

	generic hierarchy as has been used in multiple URL schemes
	
	intra-document references to named subdocuments or locations
	[via #fragment construct]

	[I may not have got them all]

Scheme specifications [three tracks per URLreg plans]:

	Draw on the above two sets of class definitions.

The URN development 
	
	- should be free to define names as polymorphic
	and containing some forms that do not meet
	the hardiness requirements of the HTML/HTTP cycle.

	- should get with the program of scheme extension
	so we have one program of scheme vocabulary 
	definition which will preserve the low cost of
	distinguishing things syntactically across a
	broad range of contexts.

-- Al Gilman