Speaking as someone who is trying to do evangelize this internally, I 
have to say I haven't dared to try to describe all of these 
complications to people.

The Link headers seem like a good approach.

(In the hypothetical example below, would the Content-Type: be 
application/atom+xml?  Or would, or could, it be image/png?)

John


Joe Gregorio wrote on 2/7/2006, 1:27 PM:

 >...
 >
 > Well, one option is to make returning the content body
 > a MUST level requirement. That would ensure the client
 > gets the info it needs. The other is to add some Link: headers
 > ala PaceUseLinkHeaders or PaceUseLinkHeaders2.
 >
 > HTTP/1.1 201 Created
 > Date: Fri, 25 Mar 2005 17:17:11 GMT
 > Content-Length: nnnn
 > Content-Type: application/atom+xml
 > Location: http://example.net/binary/edit/b/129.png
 > Link: <http://example.org/imagemetadata/129.atom>; rel="edit"
 > Link: <http://example.net/binary/images/129.png>; rel="external"
 >
 > Don't get caught up on the exact value of
 > the rel values, the *idea* is that the links you
 > need are returned via Link: headers in the response.
 >
 > Note also that there are two links, one points to the
 > URI of the resource to be used for public consumption "external",
 > and "edit" points to an Atom Entry created in a collection
 > that handles metadata for media entries.
 >
 >   -joe
 >
 > --
 > Joe Gregorio        http://bitworking.org
 >

-- 
John Panzer
Sr. Technical Manager
http://journals.aol.com/panzerjohn/abstractioneer


Reply via email to