- From: msporny <msporny@digitalbazaar.com>
- Date: Mon, 08 Mar 2010 23:56:46 -0500
- To: RDFa Working Group <public-rdfa-wg@w3.org>
(co-chair hat off) After quite a bit of reflection on the JSON/JSONP thing, I'm finding myself less and less supportive of that direction for the following reasoning: If the whole purpose of the JSON/JSONP encoding is to enable usage of RDFa in browsers that don't support CORS /AND/ don't support the RDFa API, I think we may be making a design mistake based on a short-term browser issue. That issue being the lack of 100% coverage of CORS in browsers. There are currently two ways to solve the "loading remote vocabulary" issue: 1. Use CORS - which is implemented in the latest version of Firefox, WebKit and Internet Explorer. 2. Use the RDFa API - which, in the very worst case, never gets traction. CORS will eventually be in all web browsers. I can't foresee why the web authoring community at large would reject the technology - it allows many great new Javascript mash-ups to occur. If the RDFa API is successful, that will almost ensure that dereferencing vocabulary profiles will be natively supported in browsers. I would argue that the probability that both 1 and 2 will happen with varying degrees of success are highly probable within the next 5 years. If we implement JSONP, that particular technology won't make much sense after #1 or #2 gain widespread use, and worse - we will not be able to get rid of it in RDFa. It'll be a vestigial feature that was in use from 2011-2016 and then hangs on for far longer than we would want it to hang on in the name of backwards compatibility. I also agree with Ivan's take on using vocabularies to provide RDFa processing rules (such as defining prefixes and tokens). We ensured that rdf:XMLLiteral should be processed differently than plain literals, and we're using an RDF vocabulary to provide that hint. I see no compelling reason why other vocabularies couldn't give the RDFa processors rules on which tokens and prefixes should be defined in vocabulary documents. Especially if a vocabulary was designed to augment the RDFa processing algorithm. I certainly understand that this goes against the general rule of not mixing markup with processing instructions, but that ship sailed long ago when we stated that rdf:XMLLiteral should be treated differently. The thing that should concern us is whether or not using rdfa:prefix and rdfa:reserved (or something to that effect) will cause authoring headaches or non-deterministic parser issues. I don't think we're in danger of either one. Conversely, using xmlns: to define prefixes and tokens is not a good design direction. In this case we certainly are doing something that will cause authoring headaches. Namely, everyone that does a good job of annotating their vocabulary with OWL/SKOS/RDFS/etc. will run the risk of those prefixes being leaked from the vocabulary document to the author's document. -- manu -- Manu Sporny President/CEO - Digital Bazaar, Inc. Establishing an Open Digital Media Commerce Standard http://blog.digitalbazaar.com/2009/09/28/a-digital-content-commerce-standard/
Received on Tuesday, 9 March 2010 04:57:27 UTC