- From: Randall Leeds <randall@bleeds.info>
- Date: Wed, 31 Aug 2016 15:04:32 +0000
- To: Shane McCarron <shane@spec-ops.io>, Ivan Herman <ivan@w3.org>
- Cc: W3C Public Annotation List <public-annotation@w3.org>
- Message-ID: <CAAL6JQhHn_AJAt=-fuf2oi1ycRYEHJAXLj1kH1YCKMSj2Z8w-A@mail.gmail.com>
The model, being format agnostic, could specify that it must be absolute, but I wouldn't require that for tests if they're testing JSON-LD. Due to the way JSON-LD defines this it's impossible for an annotation that's part of a collection not to have an absolute URI. I just don't want to force a requirement on an implementer serving JSON-LD annotations to preprocess their responses just to resolve what the client can resolve by the rules of JSON-LD processing. Actually, I step back to a stronger position. I'm tempted to oppose requiring it at all, because a client may often want to assign a unique (with high probability, such as a UUID) identifier to an annotation but not assign it a base URL because it hasn't been published yet. I suppose there's an argument here that there could use a scheme other than http: and thereby have an absolute URI, but... why? What do we gain? On Wed, Aug 31, 2016, 06:32 Shane McCarron <shane@spec-ops.io> wrote: > Okay. I will leave it to the group to debate but depend on it being > absolute in the tests for the nonce. > > Thanks! > > On Wed, Aug 31, 2016 at 8:27 AM, Ivan Herman <ivan@w3.org> wrote: > >> >> On 31 Aug 2016, at 14:58, Shane McCarron <shane@spec-ops.io> wrote: >> >> So.... while I don't think the spec says this, I assume that the ID >> property of an annotation MUST be an absolute IRI. >> >> If that is not the case... I can change the server tests to impute the >> absolute IRI based upon... something. But I don't know what that something >> should be. Probably if the ID starts with a '/' it is relative to the >> scheme + hostname + port we are referencing for the collection, and if it >> doesn't start with a '/' it is relative to the container URL. >> >> >> Hm. >> >> - The RDF 1.1 abstract syntax does say that a URI must be absolute[1]. >> - JSON-LD says "A relative IRI is an IRI that is relative to some >> other absolute IRI. In JSON-LD all relative IRIs are resolved relative to >> the base IRI."[2] and then describes the base IRI-s in [3] referring to >> @base[2] and, in the absence of it, it is relative to the document. >> >> However, the model does not aim at depending on JSON-LD too much; I am >> not sure whether it would be acceptable to use the @base (note that the >> JSON-LD spec says that a @base in an externally referenced context file is >> ignored). What this leaves us is that a relative URI is to be dereferenced >> with the full collection of annotation as a base URI, following the URI >> spec. This is where JSON-LD leads us. >> >> However, I always regarded every annotation as some sort of a >> self-standing entity, ie, relying on a higher level resource URI is shaky. >> Ie, I think it is perfectly fine if we consider the ID property as always >> being an absolute URI. If we agree on this, this clarification should be >> added to the spec… >> >> Ivan >> >> [1] https://www.w3.org/TR/rdf11-concepts/#h3_section-IRIs >> [2] https://www.w3.org/TR/json-ld/#h3_iris >> [3] https://www.w3.org/TR/json-ld/#h3_base-iri >> >> >> Anyway, help? >> >> -- >> Shane McCarron >> Projects Manager, Spec-Ops >> >> >> >> ---- >> Ivan Herman, W3C >> Digital Publishing Lead >> Home: http://www.w3.org/People/Ivan/ >> mobile: +31-641044153 >> ORCID ID: http://orcid.org/0000-0003-0782-2704 >> >> >> >> >> > > > -- > Shane McCarron > Projects Manager, Spec-Ops >
Received on Wednesday, 31 August 2016 15:05:20 UTC