The UR* scheme registry, Citing URL/URI specs

Dan Connolly (connolly@w3.org)
Thu, 23 Oct 1997 18:10:11 -0500


Message-ID: <344FD953.2895@w3.org>
Date: Thu, 23 Oct 1997 18:10:11 -0500
From: Dan Connolly <connolly@w3.org>
To: timbl@w3.org, fielding@ics.uci.edu, masinter@parc.xerox.com,
CC: uri@bunyip.com, lassila@w3.org, swick@w3.org, tbray@textuality.com,
Subject: The UR* scheme registry, Citing URL/URI specs

Larry, Roy, Keith, and Harald,

Several W3C documents (HTML, XML, RDF) need to cite
the UR* specs. We're trying to figure out whether
to use the term "URL" or "URI" and in turn, what
specs to cite.

Choices include RFC1630 (informational), RFC1738/1808 (proposed
standard) and the syntax/process drafts (in progress).

The picture I have long had in my mind is:

==============
http://www.w3.org/Addressing/#terms

         _______________________________________________________
        |        _________________                              |
        |       |  ftp:          |                              |
        |       |  gopher:       |                              |
        |       |  http:       _______________                  |
        |       |  etc        |  |  urn:      |                 |
        |       |_____________|__|  fpi: ?    |                 |
        |               URLs  |               |                 |
        |                     |_______________|                 |
        |                             URNs                      |
        |_______________________________________________________|
                               URIs
==============

So URI seems like the right term, but the ratified specs
don't reflect that. The IANA registry[1] claims to be
a URL scheme registry, rather than a URI scheme
registry. So I wonder... is it expected that urn: will
appear in that registry?

[1] ftp://ftp.isi.edu/in-notes/iana/assignments/url-schemes

I think the popular perception looks more like:

         _______________________________________________________
        |                                                       |
        |          ftp:                                         |
        |          gopher:                                      |
        |          http:                                        |
        |          etc                                          |
        |                                                       |
        |                                                       |
        |                                                       |
        |                                                       |
        |_______________________________________________________|
                               URLs

And the terms URN and URI either invoke warm fuzzies, cold
pricklies, or blank stares, but no hands-on understanding.

I don't think there is only one sort of generic UR* implementation,
so for implementation purposes, it doesn't matter what the
list of schemes is called.

I go back and forth between
my "suggestion" (which corresponds to the second picutre)
and my "2nd choice" (which makes picture #1 the official
ratified IETF position) below:

==========
http://www.w3.org/Addressing/9710-uri-vs-url.html

URI vs. URL 

     Concrete question: Which term should HTML 4 spec
     use? How about the XML specs? and the RDF specs? 

     Iana maintains a registry of schemes which includes mid:
     and cid:. If those are URL schemes, then everything is a
     URL. Else, what IANA maintains is a UR*I* registry 

     Does the process draft cover all URIs or just some?
     ("URLs" as of Aug 97) (See Roy's page for latest
     process/syntax draft) 

     Does the syntax draft cover all URIs or just some? (
     "URLs" as of Aug 97) 

     Syntax draft says "URIs are covered by other specs".
     What other specs? 

     HTTP 1.1 (RFC2068, revised draft) uses URI and URL.
     Cites RFC1630 (informative, by TimBL) and 1737
     (informative, requirements doc) for defintion of URI.
     Discusses URI comparison (section 3.2.3), says "host
     names MUST be compared case-insensitive" -- does that
     overlap/conflict with syntax draft? I think it should be
     SHOULD, i.e. that strcmp() should be a conforming
     implementation of URI comparison. 

My suggestion: the distinction isn't useful in any of these cases;
the public knows them as URLs and URL schemes.
So we should do a global s/URI/URL/ everywhere. 

2nd choice: do a s/URL/URI/ in all the formal specs,
begin to educate the public that the list of schemes is a
list of URI schemes. 
==========

I can live with either choice, but I can't live with the
outstanding tension: the current situation is:

The W3C HTML working group is big on readability
	of the spec, and they're firm on using URL.
	We've got a weasle-word footnote in the
	spec about URLs, URIs, and URNs, and we
	cite RFC1738 and RFC1808 normatively,
	and RFC1630 informatively.
http://www.w3.org/TR/WD-html40-970917/

The XML spec currently says URL all the way, but there
	are a lot of folks who are interested in URN
	developments who think this prohibits the
	use of URNs in XML. They call for s/URL/URI/
	in the XML specs. But then... what to cite?
	RFC1630, even with its informational status
	and bugs?
http://www.w3.org/TR/WD-xml-970807.html

The RDF spec says URI in most places, but URL in one.
	It's a pretty early draft, and they haven't
	gotten around to details like what to cite,
	but I get a sense from the mailing list
	that they like the term URI.
http://www.w3.org/TR/WD-rdf-syntax-971002


============
http://www.ics.uci.edu/pub/ietf/uri/draft-fielding-url-syntax-09.txt

1.2. URL, URN, and URI

   URLs are a subset of Uniform Resource Identifiers (URI), which also
   includes the notion of Uniform Resource Names (URN).  A URN differs
   from a URL in that it identifies a resource in a
   location-independent fashion (see [RFC1737]). This specification
   restricts its discussion to URLs. The syntax and semantics of other
   URIs are defined by a separate set of specifications, although
   it is expected that any URI notation would have a compatible syntax.
============

All of these W3C forums intend to be consistent with the IETF UR*
specs, but we're finding it difficult to do so.

I request clear guidance on whether
	(1) we can expect the IANA registry to become
	known as a URI registry, with the corresponding
	change in the syntax and process documents.

or
	(2) we can expect the IANA registry to continue
	to be known as a URL registry, in which case I
	request that the term URI be declared historical
	in the syntax/process drafts,
	and that any mention of URN set an expectation
	that the urn scheme (or schemes) will go in
	the URL registry.



-- 
Dan Connolly, W3C Architecture Domain Lead
http://www.w3.org/People/Connolly/