Re: The range of the HTTP dereference function

On Sunday 31 March 2002 06:07 am, Miles Sabin wrote:
> I more or less agree, but I don't think this goes quite far enough.

I was attempting to not be overly radical...

> What you're saying here is that (without additional information),
>
> 1.  The URI -> resource mapping is unique.
> 2.  The resource -> representation mapping is ambiguous.
>
> and you want some mechanism to resolve the ambiguity in (2). This
> would give us,

Yes. You can't / don't want to always fix the representation though, 
so there needs to be some way of distinguishing *reliable* mappings 
from *variant* mappings. 

> Given that the de facto ambiguities wrt URIs run much deeper than
> the resource -> representation ambiguity it seems to me that we'll
> get much further if we push the problem up to he URI/resource level
> and solve the problem there.

You could have something like this:

  1) The mapping from URI -> resource is unique.
  2) The mapping from resource to representation is unique.
  3) Some returned representations are references to resources. This
      mapping  may vary as a function of the request environment
     (essentially a redirect).
  4) Resources that are returned at (2) that are not references to
      other resources, are are stable.

An alternate is to have some URI mechanism for disambiguation such as 
'all URI's with the ;stable parameter are reliable mappings'.

There's lot's of little edge cases in all this.... I think the 
URI->resource->representation model needs to be clarified outside the 
scope of HTTP before much else is done.

I'd be interested in hearing what you think is needed.

Received on Sunday, 31 March 2002 11:38:45 UTC