Re: LDPR Interaction Model on Create

On 15 Oct 2014, at 17:30, John Arwe <johnarwe@us.ibm.com> wrote:

> > Since everything on an LDP server is an LDPR, seems unnecessary for 
> 
> Progress, then - this is the point of divergence. 
> 
> I believe that ordinarily this will be the case, but I also believe LDP provides no guarantees because LDP does not prohibit the creation of non-LDPRs (via Post, or otherwise), having just re-skimmed 5.2.*  ... this is a consequence of allowing LDP support to be added to existing HTTP 1.1 server implementations. 

I think that is too far fetched.

There are much more usual cases of this: any metadata resource that get created the LDPR is created.
For example on rww-play when any LDPR is created the LDPR contains a link to of type rel=acl to a resource
that describes which actors can execute what types of acts on it. That is not I think an LDPR according to the
spec here ( even though it should accept a PATCH and the resource is a graph ).

Deleting the LDPR deletes theses associated resources just as it removes the ldp:contains relation from the 
containing LDPC.


> 
> A trivial counter-example is where a server creates a resource successfully but chooses not to give it an etag (or no Head support, or...).  The same server is capable of creating LDPRs in other cases, but for reasons known only to the implementers (backward compatibility with existing clients?) in this case some of its resources are not LDPRs.  I cannot think of any LDP violation this causes, even when the creation is accomplished using an LDPC interaction model (i.e. the client's creation request is a POST and its request-URI identifies an LDPC). 
> 
> 
> Best Regards, John
> 
> Voice US 845-435-9470  BluePages 
> z/VM OpenStack Enablement and zKVM 
> 
> 

Social Web Architect
http://bblfish.net/