W3C home > Mailing lists > Public > uri@w3.org > March 2001

Proposal for an "address:" URI Scheme

From: Sean B. Palmer <sean@mysterylights.com>
Date: Fri, 16 Mar 2001 00:57:22 -0000
Message-ID: <004d01c0adb4$946c06c0$52fe7ad5@z5n9x1>
To: <uri@w3.org>
Cc: <www-talk@w3.org>
One of the few things that do not currently have standardized URIs are
generic postal addresses ("snail-mail addresses"). This is a proposal
for a new URI scheme that will enable people to identify postal
addresses as a part of URI (and hence WWW) space.

Two significant and highly relevant architectural axioms of Uniform
Resource Indicators are:-

     "Any resource anywhere can be given a URI" - [1]
     "Any new name space or address space can be represented as
      a subset of URI space." - ibid.

Therefore, it should be quite easy to represent physical postal
address as a subset of URI space. I'd start by suggesting that the new
scheme be named "address" as a postal address is such a widely known
generic object of the word "address". A possible alternative to
"address" could be "post", although I prefer the former due to its
non-ambiguity as being a noun rather than "post", which could also be
a verb ("to post"... hmm, you could address something I suppose, "to
address a letter", but it is less likely to be misinterpreted in this

Secondly, in keeping with how URLs work (but not using exactly the
same format), we would want to list the most general,
internationalized resource (i.e. the country, or possibly continent in
the case of Antarctic research stations, etc.) first, separating each
main fragment with a slash ("/") until we get to the most localized
resource, such as a person, or whatever the final object of the postal
address is.

For example, the World Wide Web Consortium's postal address could be
represented as:-


What's the point? Well, I like to think that URIs can point to just
about anything, and this is a good test of that. It was suggested to
me at an early stage that using latitude and longitude would be a
better idea, at which point I noted that it is rather difficult to
send mail to a latitude and longitude... postal systems are quite neat
and efficient, and their postal schemes well regulated.

Pragmatically speaking, an "address:" URI scheme could have many uses
(and implications):-

   1) For administrative purposes; for anything that needs to
     store an address in a Web-ized format.
   2) As a subject/object in RDF; to refer to addresses in a
     proper format without having to resort to string literals.
   3) For use in HTML; clicking an "address:" link could
     activate the users address book and store the entry.

These are simple examples; there could be many others, some of which
could be better arguments for the scheme.

If I were to pursue this URI scheme any further, I would include
references to known and respected works on postal addressing schemes
and they currently stand. Some best practises will have to be
administered, for some of you may have noticed that there are many
variations on postal address; for example the W3C address example
listed above could also be represented as:-


I would expect the American postal service to deliver that to the same
address even though the strings are slightly dissimilar.

Other considerations: relative addresses? Well, the WWW is a truly
international medium, so I would expect that although in practise
laving off the country in your own country is fine, doing this on the
Web would be very bad practise indeed, and so relative addresses
should not be permitted. Unfortunately, this is impossible to
constrain, for the "creator" of the URI obviously decides which is the
most appropriate internationalized part of a postal address. One
again, referring to standard works on postal addresses should help to
resolve this issue.

In summation, the (proprietary) RFC-ized syntax for the "address:"
scheme is as follows:-

       addressuri   := "address:" addressdata
       data     := *(*urlchar "/")

   where "urlchar" is imported from [2].

Has this suggestion or something like it been raised before? I would
be very shocked indeed if it hadn't, but a quick Google search [3]
didn't reveal anything. If there is widespread support (and no
conflicts) for this proposed scheme, I may well push it through to the
RFC standards track.

[1] http://www.w3.org/DesignIssues/Axioms
[2] http://www.ietf.org/rfc/rfc2396.txt
   - Berners-Lee, T., Fielding, R., and L. Masinter,
     "Uniform Resource Identifiers (URI): Generic Syntax", RFC
     2396, August 1998.

Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://infomesh.net/sbp/> .
Received on Thursday, 15 March 2001 19:59:00 UTC

This archive was generated by hypermail 2.4.0 : Sunday, 10 October 2021 22:17:38 UTC