- From: Robert Sanderson <azaroth42@gmail.com>
- Date: Mon, 3 Nov 2014 10:25:40 -0800
- To: James M Snell <jasnell@gmail.com>
- Cc: ☮ elf Pavlik ☮ <perpetual-tripper@wwelves.org>, "public-socialweb@w3.org" <public-socialweb@w3.org>
- Message-ID: <CABevsUGF3fLVue95dLC386acZyLmD2XE92RyzvBCncVgN0W1zw@mail.gmail.com>
On Mon, Nov 3, 2014 at 10:12 AM, James M Snell <jasnell@gmail.com> wrote: > On Mon, Nov 3, 2014 at 9:57 AM, Robert Sanderson <azaroth42@gmail.com> > wrote: > > > -1 on removing the functionality, +1 on exploring how else it might be > > represented > > > "rel" is essentially establishes a qualified relationship. The > predicate can be used just as easily. Yup, +1 to using relationships when you mean a relationship. Though it makes the context document potentially very long if all of the rels are imported, and introduces a source of potential collision as per dret's mail. That could be solved (to a certain extent) by freezing the set of rels that are available whenever the context document is changed, or by using a prefix. { > "preview": { > "@type": "as:Link", > "@id": "http://example.org/foo" > } > } > This still says that the image is a Link (hooray Open World) which is weird, in my opinion. If everything can be a Link, then it doesn't really make any useful assertion. > The other alternative approach is to use a true qualified relationship > model with an intermediate object (using "url" or "href" to identify > the linked resource instead of "@id") > > { > "image": { > "@type": "as:Link", > "rel": "preview", > "href": "http://example.org/foo" > } > } > (note the lack of @id) > Indeed. This is reifying the relationship to a blank node, where subject is the parent object, the predicate is in rel and the object in href. As with any reification ... is it necessary above and beyond a real, simpler, relationship? I'm good with whichever way the WG decides. I just know that the > current definition doesn't work the way it's currently modeled. > Agreed! :) Rob >> On Mon, Nov 3, 2014 at 9:36 AM, ☮ elf Pavlik ☮ > >> <perpetual-tripper@wwelves.org> wrote: > >> > On 11/03/2014 06:14 PM, James M Snell wrote: > >> >> The "rel" property added to the as:Link is problematic from a data > >> >> modeling point of view, and actually isn't as useful as one might > >> >> imagine in practice. > >> >> > >> >> Based on the current definition of as:Link, here's how it would > >> >> currently be used: > >> >> > >> >> { > >> >> "@context": "http://asjsonld.mybluemix.net", > >> >> "image": { > >> >> "@type": "as:Link", > >> >> "@id": "http://example.org/foo", > >> >> "rel": "preview" > >> >> } > >> >> } > >> >> > >> >> Those familiar with the JSON-LD processing model ought to see the > >> >> problem right away. The "rel" is actually supposed to be a qualified > >> >> relation of the containing object, but as it is defined here, it > >> >> becomes a property of the as:Link itself. > >> >> > >> >> In any case, despite the modeling issues, the "rel" just isn't > proving > >> >> to be very valuable in any case I can identify. > >> >> > >> >> My proposal is just to remove it. > >> > +1 > >> > > >> > in this case, do we have any other reason to keep as:Link? > >> > > >> > IMO JSON-LD embedding provides us with all we need to include > statements > >> > about objects used as property values > >> > > >> > http://www.w3.org/TR/json-ld/#embedding > >> > > >> > > > > > > > > -- > > Rob Sanderson > > Technology Collaboration Facilitator > > Digital Library Systems and Services > > Stanford, CA 94305 > -- Rob Sanderson Technology Collaboration Facilitator Digital Library Systems and Services Stanford, CA 94305
Received on Monday, 3 November 2014 18:26:08 UTC