Re: The range of the HTTP dereference function

On Wed, 2002-03-27 at 12:05, Tim Berners-Lee wrote:
[...]
> I am saying that the HTTP 200 response indicates that
> the body is a representation in bits of the resource requested.
> This works very well do long as the notional resource is some
> information which can be represented in various ways.
> The HTTP headers refer to that information item.
> 
> It doesn't work when you consider the resource to be something
> like the person whose home page it is.
> You can't use the same URI to refer to the person and the
> home page.
> (You *can* use a relation, like "the person whose home page
> is  xxxx").  The the properties of one and the properties
> of the other are quite different.

This is argument by assertion, TimBL. You can
say it as many times as you like, but that
doesn't compel or even convince me to agree.

I can play that game too: no, the properties of my car and
the properties of

	http://dm93.org/y2002/myCar-232

are not different. Everything true of my car is
true of http://dm93.org/2002/myCar-232 and
vice versa; yes, my car is an HTTP resource; yes,
the HTTP resource is a car.

I stipulate that this is not intuitive.
Neither are various quantum phenomena that
I nonetheless accept as an accurate model
of the world.


I also stipulate that it is not advisable in practice.
Usually, I want to be
able to say things about web pages
without also saying them about my car, so
I should choose distinct identifiers for them.

>  So you have to chose one.

No, I do not.


> In combination, they work well.  But don't let's expect HTTP caches
> to have to cache cars.

They cache messages; e.g. replies about cars.


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Wednesday, 27 March 2002 13:35:40 UTC