Re: Call for proposals to amend the "httpRange-14 resolution"

On 1 Mar 2012, at 20:44, David Booth wrote:

> Do you mean that you recommend deprecating the use of 303?  

I'm afraid so. The semantics of a 303 response to GET are unclear whereas the semantics of Link: … rel=describedby are much better understood.

For a situation like http://www.w3.org/TR/cooluris/#r303gendocument:

• http://www.example.com/id/alice represents Alice, but does not describe her. So I'd prefer to configure my origin server to respond to a GET / HEAD with a Link: header and maybe 204 No Content.

• GET http://www.example.com/doc/alice is a different matter; this is a web document and it may have multiple representations. Existing practice is fine.

> And do you mean that you think the Link header should supersede the implicit URI definition of the resource as being an information resource, if 200 HTTP response code is received?  FYI, in response to your comment I've added another issue for the latter question: http://www.w3.org/wiki/UriDefinitionDiscoveryProtocol#3.2.1.1_Link_header 

In my view, if a GET for a URI returns content then it is a web document (or information resource, if you prefer). Using 204 and Link: just fits in better with how I understand the web.

-- 
Tim Bannister – isoma@jellybaby.net