Re: Definition of 'resource' not consistent with RFC 3986

On Jan 30, 2009, at 11:21 AM, Julian Reschke wrote:
> 3.2.2 (http URL) and 9.3 (GET)?

3.2.2: The "http" scheme is used to locate network resources via the  
HTTP protocol.
-- In RFC 2616, resource = network resource, so this is not  
problematic. But if these are different, one might ask whether  
anything should be said here about other kinds of resources -- can an  
application use the http: scheme to do anything with 'abstract  
resources'?  Actually once the question is raised the language of the  
whole section becomes 
rather weird -- the scheme "is used" - does that mean used with the  
HTTP protocol? (Yes; anything other than HTTP is out of scope.) Does  
that distinguish http: from any other scheme? (No; any URI can be used  
with HTTP to locate network
resources.) Since the http: scheme can be used in other ways than with  
HTTP, and HTTP is not tied to http:, the scheme's distinguishing  
feature really is that it has no distinguishing features.

9.3: Two cases are covered in the text, URIs identifying network data  
objects, and URIs identifying network services.
(Clarification needed perhaps: the URI doesn't identify the  
information, it identifies something whose 'state' is retrieved; at  
least in the REST formulation, and I don't know to what extent the RFC  
should codify that.) Nothing is said about the case where the URI  
identifies neither of these, but the way it's worded it's difficult to  
tell whether this case is unconstrained.

Because there has been so much confusion around this issue, I think  
there should be sufficient language to enable a reader to understand  
one of the following: it's allowed, it's forbidden, or it's out of  
scope and is to be settled in some other way.


Received on Friday, 30 January 2009 21:14:00 UTC