Re: Two questions about URIs

Pat Hayes wrote:
[...]
> This seems to imply a three-way distinction with two mappings, rather
> than the simple two-way name/entity distinction suggested by the term
> 'resource identifier'. A URI identifies (denotes?) a resource, which
> in turn has some *content* (the entities to which it currently
> corresponds):
> 
> URI ----(identifies)---->resource-----(corresponds to)----->entity
> 
> Apparently the identification mapping is fixed, but the correspondence
> mapping can change with time.  Is this a fair understanding of this
> text?

Yes.

> If so, the entire notion of a URI simply denoting or naming
> something seems to need re-thinking.

Why? The three way relationship is hardly new or novel:

	object identifiers / objects / object state
	scheme variable identifiers / variables / values


> I request clarification of what this is supposed to mean, with
> particular reference to the intended distinction between 'resource'
> and 'entity', and between 'identifies' and 'corresponds to'.

In the HTTP spec 'entity' is used specifically to denote
a sequence of bytes (with a bit of metadata, like its MIME type).
Think of 'value' if you like.

For a longer explanation, see the first part of...
  http://www.w3.org/Architecture/state

Unfortunately, the latter parts sorta ramble on...

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

Received on Tuesday, 16 October 2001 13:25:43 UTC