On Wed, Jan 14, 2015 at 6:31 AM, Robert Sanderson <[log in to unmask]> wrote:
>> Thingy identifiers (http URIs) SHOULD (not MUST) be bound to descriptions
>> identifiers (http URIs) via HTTP 303 redirects.
> Agreed. Best practice (e.g. SHOULD) is 303. 2NN doesn't have sufficient
> traction to make it through the IETF, and there's enough pushback to the
> redirect adding latency and thereby cost that HTTP-Range-14 is going to be
> with us for a long time yet.
> None of which is any reason not to separately identify the things
In my opinion hash URIs is an alternative that is easier to deploy:
We are using "one (optional) RWO per document" approach with "#this"
as fragment identifier:
Document URI: /container/item
RWO URI: /container/item#this
That way the RWO is not directly accessible over HTTP (the fragment
identifier is invisible to the server), but it still is identified and
described in the document.