W3C home > Mailing lists > Public > www-ws-arch@w3.org > March 2002

Re: URI's

From: Mark Baker <distobj@acm.org>
Date: Fri, 1 Mar 2002 16:50:20 -0500 (EST)
Message-Id: <200203012150.QAA04075@markbaker.ca>
To: RogerCutler@chevrontexaco.com ("Cutler, Roger (RogerCutler)")
Cc: jhui@digisle.net ('Joseph Hui'), hugo@w3.org (Hugo Haas), www-ws-arch@w3.org
> It seems to me, then, that if http://12.34.56.78 is indeed a URI, then the
> global IP addresses can be put into one-to-one correspondance with URI's by
> a trivial relationship.

This sounds like a slam dunk, but it's a fair bit more complicated than
just slapping "http://" on the front.  An incredibly important axiom of
Web architecture is one of opacity[1].  In the context of this issue, it
says that you can't say that http://12.34.56.78 identifies; only the
publisher of that resource can.  For example, http://207.236.3.141
currently identifies me by virtue of www.markbaker.ca resolving to
207.236.3.141.  So for somebody to come along and state that
"http://[IP]" identifies an IP address, would conflict with what's
already out there today.

From a Web architecture friendly perspective (and off the top of my
head), there's two ways to make IP addresses identifiable.

1.  Define a new URI scheme

An "ip" URI scheme could be defined, so URIs would look like
ip:12.34.56.78 or perhaps ip://12/13/56/78.

2.  Have the relevant authority publish the URIs in the HTTP scheme

Here, IANA could publish URIs such as;

http://www.iana.org/ip/12.34.56.78

along with a less opaque version for querying;

http://www.iana.org/ip?addr=12.34.56.78

> The reason I am pursuing this is that I am wondering whether it will make
> sense in the architecture to say that participants in web services must be
> identifiable by URI's (including in the sense above).  This would exclude
> perverse things like telephone numbers,

Telephone numbers have the "tel:" URI scheme, e.g. tel:+1-613-789-1818
Also, the HTTP URI scheme can be used to identify phones[2] as well.

> street addresses,

People have thought about this;

http://infomesh.net/2001/03/address/

> and so on, and it
> seems to me something like this is pretty much what people have in mind when
> they are talking about web services.

I agree.  To get an idea for the diversity of things that URI have been
used to identify, see [3].

But note that TimBL and others (myself included) believe that defining a
new URI scheme is a very serious and expensive thing to do, and that the
HTTP URI scheme should be used wherever possible (which, believe it or
not, is practically all of the time).

 [1] http://www.w3.org/DesignIssues/Axioms.html#opaque
 [2] http://dev.w3.org/cvsweb/2001/telagent/
 [3] http://www.w3.org/Addressing/schemes

MB
-- 
Mark Baker, Chief Science Officer, Planetfred, Inc.
Ottawa, Ontario, CANADA.      mbaker@planetfred.com
http://www.markbaker.ca   http://www.planetfred.com
Received on Friday, 1 March 2002 16:47:45 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:24:55 GMT