Re: Do resources have representations?

Norman Walsh wrote:
> | I.e., every time I go to <http://example.org/1434> I see one
> | particular page, and every time I go to <http://example.net/~foo/bar>
> | I see another particular page, yet the two URIs identify the same
> | resource.
> |
> | How do you explain this?
> 
> I don't feel any compelling need to, any more than I feel a need to
> explain why a given URI might return HTML, XHTML, PDF, RDF, plain
> ASCII, GIF, PNG, Microsoft Word, or other representations depending on
> header settings independent of the URI.

The scenario you describe can be trivially explained in the framework "A 
URI only denotes a resource": The URI denotes a single resource, but 
this resource has multiple valid representations at any given point in 
time. No problem.

When I point a browser to that URI, I would consider it correct behavior 
to get any of the HTML, XHTML, PDF, RDF, plain ASCII, and so on.

However, if I pointed my browser to <http://example.org/1434>, I would 
not consider it correct behavior if my browser opened a HTTP connection 
to <http://example.net/~foo/bar> and showed me the content that the web 
server at that address returns -- even if I have proof that these two 
URIs denote one and the same resource!

This isn't explained by the "A URI does nothing but denoting a resource" 
model, as far as I can see.

> I do observe, however, that the assertion that 
> <http://example.org/1434> and <http://example.net/~foo/bar> identify the
> same resource is not one that I would accept without proof.

I assumed that you had proof.

Assume that at <http://example.org/1434>, you find a web page that says,

     The URI of this page (http://example.org/1434) denotes Norman Walsh
     (Norman.Walsh@Sun.COM).

At <http://example.net/~foo/bar>, you find a page that says,

     The URI http://example.net/~foo/bar has been assigned to denote
     Norman Walsh, the person with email address Norman.Walsh@Sun.COM.

Would this be good enough? Of course, we could also assume that you have 
written, notarized documents from the domain name owners.

> In short: it appears that you want to associate a URI with a particular,
> closed set of representations *independent of* the act of retrieving them.

I don't care about the set being single or closed. I do want to be able 
to say, URI X has representation Y; so e.g. my web server can serve Y 
when a representation of X is requested.

> You can't. Representations are ephemeral, they change without warning and
> without changing the resource that the URI identifies.

Well. If I have retrieved a representation, I cannot conclude that it is 
the only representation available at that URI, nor can I conclude that 
another representation will not be available at that URI in the future. 
I can, though, conclude that it is *one* representation available at 
that URI at the time I retrieved the resource.

This also doesn't mean that it is impossible to state that a particular 
URI maps to a particular representation, nor does it mean that it isn't 
useful to make such a statement. For example, when you configure your 
web browser.

- Benja

Received on Monday, 28 July 2003 11:52:37 UTC