Re: MKCOL for making collections

On 1/23/13 2:47 AM, Sergio Fernández wrote:
> Hi,
>
> not sure if this contradicts any axiom of the current model, but what 
> about a pragmatical solution by base path?
>
> I mean, from an implementation point of view it could be easier to 
> provide something like:
>
>   {ldp}/resource/...  for ldp-r
>   {ldp}/container/... for ldp-c
>
> Keeping the same HTTP verbs for common operation (i.e., POST for 
> creating).
>
> As you know, I'm new in the WG, so sorry if I missed some previous 
> discussions about this.

That would work too. The important thing is to use a de-referencable URI 
to denote something associated with an LDP compatible data space :-)

Given said URI, a user agent (via HTTP message exchange) should be able 
to deduce what's possible re. CRUD operations for a given LDP data space.

Kingsley
>
> Greetings,
>
> On 22/01/13 21:54, Kingsley Idehen wrote:
>> On 1/22/13 3:27 PM, Wilde, Erik wrote:
>>> how linking is done exactly (i.e., which interactions affordances are
>>> exposed by which resources) still needs quite a bit of work, and in the
>>> end specific URIs don't matter anyway. but i'd assume that many 
>>> services
>>> might expose a home resource at some URI x, and that there may be a
>>> collection factory that might end up creating x/y collections, but 
>>> all of
>>> that really is up for the implementation to decide. it might also 
>>> decide
>>> that the home resource is x and then a new collection always has a URI
>>> x/collection/y or something along these lines.
>>
>> Why should there be anything like a so called "home resource" ?
>>
>> All you needs is a URI (maybe you could see this as a starter URI or
>> home URI, so to speak) and RESTful interactions (e.g. over HTTP) that
>> enable a user agent deduce what's possible. In the course of such
>> interaction, the server is also able to help user agents understand
>> what's its capabilities are with regards to specific operations.
>>
>> An example of a starter (or home) URI is a WebID. A URI that denotes an
>> Agent (i.e., person, machine, software etc..). Given a URI one should be
>> able to deduce what operations are possible in a given Web accessible
>> (addressable) data space, associated with said WebID.
>>
>> If you go back to Henry's examples, and pretty much his entire
>> narrative, you'll see he is demonstrating how this is achieved using RDF
>> -- within the constraints of existing Web Architecture.
>>
>> RDF enables very precise data definition using an entity relationship
>> model enhanced with explicit (machine comprehensible) entity
>> relationship semantics. Again, none of that has anything to do with a
>> specific notation for expressing subject-predicate-object triples (or
>> 3-tuples).
>>
>


-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen