Re: What to do about namespace derived URI refs... (long)

On 07 Jun 2001 16:00:19 -0700, Seth Russell wrote:
> [Hypothetical dialog]
> Tim: A subset of a URI is a URL.
> Seth:  Fine, How is is distinguished from URI.
> Tim: An application can use a  URL to retrieve a bit stream.

I don't think this is right :
  mailto:champin@bat710.univ-lyon1.fr
  smtp://bat710.univ-lyon1.fr/
  tel:+358-555-1234567
are all URLs, and do not allow to retrieve (directly) a bit stream.

As a matter of fact, resources must be IMHO though of as "natural
resources" or "services",
that is : they provide something useful. That "something" could be a bit
stream, but it could also be something else:
the possibility to send an e-mail to someone,
the possibility to effactively send a mail over the internet,
the possibility to reach somebody by a phone call.

As I understand them, URLs have the particularity of "locating" a
resource, that is, they "describe the location of a resource" [RFC1736].

This notion of "describing the location" is very relative:
is "bat710.univ-lyon1.fr" a location, or only an opaque name requiring
DNS to be changed into a more "locating" IP address ?
I guess a URI scheme is not opaque as long as it is provided with a well
specified resolution mechanism (HTTP protocol, e-mail address, phone
network...).


> Seth:  How can an internet application distinguish between what the
> URI#fragment retrieves and what it names?

Here again, I slightly disagree with you Seth -- though I agree that the
use of fragments raises a lot of other problems...
Given a URI-ref, there is a clear difference between the base URI and
the fragment ID. I see no reason why the resource identified by the
latter should have the same nature as the resource identified by the
former.

  Pierre-Antoine

Received on Friday, 8 June 2001 04:21:04 UTC