- From: Dietrich Schulten <ds@escalon.de>
- Date: Mon, 12 Jan 2015 20:27:48 +0100
- To: John Walker <john.walker@semaku.com>, Ruben Verborgh <ruben.verborgh@ugent.be>
- CC: public-hydra@w3.org
- Message-ID: <54B42034.9000603@escalon.de>
Am 12.01.2015 um 13:47 schrieb John Walker: > Hi Ruben, > > On January 12, 2015 at 1:17 PM Ruben Verborgh <ruben.verborgh@ugent.be> wrote: > > > > > > > But you see where this is heading. People will have to learn about > owl and rdfs before they can use hydra for their restful services. > > > > That can certainly not be the goal. > > > > This discussion is foremost about removing hydra:Resource and > hydra:Class; > > those constructs are only used to build the ontology. > > People don't need them to describe an API. > > > I would disagree on that point for hydra:Class, for example in Markus' > issue tracker demo the API documentation [1] makes liberal use of it. > The implication being that all instances of these classes (i.e. the > resources served by the API) are dereferenceable. The same is true for hydra:Resource, at least Markus told me to use hydra:Resource to mark a Link as dereferenceable in-place (embedded, without dereferencing the ApiDocumentation). If used that way, it even appears in the response, not just in the ApiDocumentation. That is probably why I am hesitant to drop it altogether. On the other hand, I found that one could also make it explicit that a link is meant to be dereferenced by defining a GET operation on it. We do not have a semantic operation type for that, obviously because the original concept is to use hydra:operation for state-changing requests and to use :Resource or :Link to give a hint that clients may GET the resource. Since neither hydra nor schema.org define something like a :DereferenceOperation: what about defining one to tell the client that it should expect a proper response and not a 405 for a GET on a URI. If someone finds that helpful or necessary, they could use it. And we could remove :Resource and :Class. How does that sound? Best regards, Dietrich
Received on Monday, 12 January 2015 19:28:38 UTC