Yes, I assert that it is risky to make statements about something when you
don't know how good your identifier is. (Although it is possible to state
that "I think this resource might disappear", or "statements made about this
resource at date XXXX are no longer valid, but do apply to resource YYYY" -
the latter is one of the things I am trying to work out how to do in EARL,
where we need to do it for  different reasons)

But I don't understand how a publisher of URNs guaranteeing that they remain
stable is any better than a publisher of URIs making the same guarantee. The
problem to be solved is a human one, not a technical one, as I see it.

From another perspective, the semantic web seems like a very good use case
that encourages publishers of URIs not to break them. Thereby making the idea
of introducing URNs less and less interesting...



On Thu, 12 Apr 2001, Lee Jonas wrote:

  >From what you say, use of URLs in RDF must be treated with extreme caution.
  Unless a publisher guarantees that they won't change the fundamental nature
  of the resource identified by a URL, you cannot rely on it to identify what
  you intend it to.

  This seems like the most compelling argument for using URNs for identifying
  anything other than representation mappings yet!



Charles McCathieNevile  phone: +61 409 134 136
W3C Web Accessibility Initiative    fax: +1 617 258 5999
Location: 21 Mitchell street FOOTSCRAY Vic 3011, Australia
(or W3C INRIA, Route des Lucioles, BP 93, 06902 Sophia Antipolis Cedex, France)

Received on Thursday, 12 April 2001 09:52:21 UTC