RE: Using URIs to identify non-information resources

> Can you rephrase this all without using words like "resource" and
> "connected", which are just as nebulous as the rest of this, at this
> point?    

It's hard to talk about what a "Uniform Resource Identifier" identifies
(uniformly) without using the word "resource".

> I am certainly "a resource" and I am certainly "connected"
> to "http://www.w3.org/People/Sandro".   I'm pretty sure that's not the
> sense of "resource" and "connected" you had in mind, but I can't tell
> what senses you do have in mind.

I'm not sure whether you are "a resource", at least in the sense of
the "R" in "URI", but I am certain that "http://www.w3.org/People/Sandro" 
identifies a web page served by the computer at "www.w3.org" and not you.


> Let's just pick a 404 example, "http://example.com/foo".  What is that
> particular URI the name of?  What does it identify?  If I talk about
> the thing called http://example.com/foo, what am I talking about?

I think "name" is more dangerous than "resource" and "identify" in
having too much baggage. URIs identify resources (by definition).
The resource identified by the URI "http://example.com/foo" is the
resource reached by using the HTTP protocol to connect to "example.com"
(using DNS) and using the "/foo" path. Currently, the nice people
who run "example.com" give a helpful message at "http://example.com"
but just a 404 at "http://example.com/foo", but they could change
it to give a nice history of "foo" at some point later.

> It seems to me there are two sensible answers:
> 
>    1.  That's not the name of anything.  It's like talking about the
>        governor of the 51st state of the United States.  There's no
>        such state, let alone such a governor.  Sure, you could imagine
>        there might be, and you make some statements that would be true
>        by definition if there were such a thing -- that the governor
>        would be the chief executive officer of the state government --
>        but there's still no such thing.

The meaning of "the governor of the 51st state" doesn't change.
The fact that it's currently 404 doesn't change the meaning
of the identifier, or make it synonomous with "the king of
the United States", just because there isn't one of those either.
(morning star/evening star).

>    2.  That's the name of a slot, of a location in the web's enormous
>        address space, which is currently empty.  In this view,
>        "http://www.w3.org/" names a slot which currently holds the
>        W3C's home page document -- it's not the name of the document
>        itself.

Yes, I think this is closer. What's in the slot depends on the
operations of the web and can vary, but the meaning of the identifier
doesn't change.
 
> Personally, I like the second answer.  I find it very comfortable to
> work with, but that's not how web specs have been written over the
> years.  So that puts us in the first camp, I think.  Or do you have
> another view?

IETF documents (and some W3C documents) were written for engineers writing
software to connect to Internet resources, not people trying to use
URIs as the terms for grounding in knowledge representation systems.
So I wouldn’t take as license the stylistic freedom which didn't precisely
distinguish between the "slot" and its content.

Larry
-- 
http://larry.masinter.net

Received on Friday, 12 August 2005 16:05:47 UTC