W3C home > Mailing lists > Public > public-webarch-comments@w3.org > October to December 2004

Re: KD 004 [was: Comments on Web Arch WD - 2004-07-05]

From: Chris Lilley <chris@w3.org>
Date: Thu, 7 Oct 2004 11:45:09 +0200
Message-ID: <1372979252.20041007114509@w3.org>
To: public-webarch-comments@w3.org

Hello Karl,

> If I understood your prose, is that it's bad to use the same URI for
> things which have a different meaning.
> 
> Example:
> 
>       http://example.org/sting
> 
> is used by organization A to identify the movie "The Sting"
> is used by organization B to identify a discussion about forum about 
> "The Sting".

Yes, this example shows organization A using it to refer to a concrete
object (a film) and the other using it to refer to an Information
resource (something wholly encapsulatable in bits).

> Now I'm a simple Web master of organization A and I write simple Web 
> pages and in my prose, where:
> 
> <p>You will information about the movie
> <cite class="title"><a href="http://example.org/sting">The 
> Sting</a></cite>
> in the following paragraphs.</p>

Yes. A Machine will get confused by that; in particular if someone else
refers to http://films.example.net/sting/lefilm then a machine reasoning
system will not identify these two as the sdame film in the ansence of
other information.

> The organization B has a Web master who is using the same URI for in 
> this piece of html
> 
> <p>There's a very <a href="http://example.org/sting">good forum</a> 
> about the movie
> <cite class="title">The Sting</cite> which gathers many fans.</p>

Organisation B is referring to an information resource. Lots of
assertions can be made about this, all consistent with referring to that
resource:

Here is a <a href="http://example.org/sting">good forum</a>
Here is a <a href="http://example.org/sting">bad forum</a>
<a href="http://example.org/sting">this page</a> does not validate
Wonderful design on <a href="http://example.org/sting">this page</a>
<a href="http://example.org/sting">Poor choice of colors</a> sucks.


These all make very different statements, bur all about the same
resource, used as an example and not as a definition. The problem with
organization A is that it used this URI as a defining instance. Clearly
others can choose different URIs as the defining instance (leading to a
false negative) or use the same URI to define something else (leading to
a false positive).

Since getting the whole universe to agree is impossible, WebArch
introduces the concept of URI ownership. The organization that owns
example.org gets to decide what that URI identifies. So as a real world
example


http://www.imdb.com/title/tt0070735/

can be used to identify the movie 'The Sting' because the owner,
Internet Movie Database, says so.

> Now Webmaster A and B are confused because there are using the same URI 
> to designate a Web page which is related to the movie

No, in your example A is trying to refer to the movie and B refers to a
page about the movie. And the error is that A is actually referring to a
page about the movie.

> but in a 
> different way AND they have just read the Web Arch document.
> 
> For sure, I have been a bit naive and dishonest in this example, 
> because it's not what you are talking about.


No, not really. Its a good example.

>  But that's my point for
> someone who's not a specialist of URIs and their use, out of context, 
> we don't know what URI overloading means.
> 
> So It's why I asked to encourage to add a more detailed explanation of 
> what you mean. A browser is a kind of user agent,

Yes, a browser is a kind of user agent

> and a URL is a kind of URI.

There is no definition of URL in RFC 2396bis

> My initial comment talks also about ressources available on the Web who 
> have multiple usage. If you go on the site PHP, and browse the 
> document, a page about a feature of PHP has "two meanings".
> 
>       - The description of the feature
>       - A forum debating of the feature and giving more examples
> 
> So what will be the choice of a user agent to identify this feature. 
> Does that mean it's a bad practice to have different functionnalities 
> in a same Web page?

No; that would lead to very short web pages.

>  That the URIs identifying must be something else,
> for example.
> 
>       http://php.example.org/feature1         The Web page
> 
->> http://cooluri.example.org/feature1/definition  The definition
->> http://cooluri.example.org/feature2/forum  The forum

No, try

   http://php.example.org/feature1#definition
   http://php.example.org/feature1#forum

The primary resource is http://php.example.org/feature1 and there are
two identified secondary resources.

I hope this makes things clearer.
  

-- 
 Chris Lilley                    mailto:chris@w3.org
 Chair, W3C SVG Working Group
 Member, W3C Technical Architecture Group
Received on Thursday, 7 October 2004 09:45:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 6 April 2009 12:37:33 GMT