- From: Ivan Herman <ivan@w3.org>
- Date: Mon, 21 May 2012 10:28:06 +0200
- To: Niklas Lindström <lindstream@gmail.com>
- Cc: Gregg Kellogg <gregg@greggkellogg.net>, W3C RDFWA WG <public-rdfa-wg@w3.org>
Well... I know how busy you are Niklas, so please do not take what I say as being pushy at all. Just reflections on the general issue, nothing else. Alex's work is brilliant, no doubt about that. But the jury is still out, as far as I am concerned, to see which approach to RDFa is more suited for the Web developers' community. The RDFa API is not final, is half-baked and I heard Manu saying that he was not overly happy with it as a user when he did the play page. The obvious other possibility is to provide a JSON interface to the data that many Web Application users like. If JSON-LD does not really work for that, than we may also have a problem... I think that a Lite++ or even strictly Lite version of a JSON-LD access would be interesting as an real-life experiment on two different ways of accessing that data. So yes, I think that your statement "Since Alex's implementation was completed I considered mine of less immediate importance" is wrong (o.k., you are right about the 'immediate'). Just my two cents... Ivan On May 20, 2012, at 21:52 , Niklas Lindström wrote: > On Fri, May 18, 2012 at 6:08 AM, Ivan Herman <ivan@w3.org> wrote: >> The more the merrier! > > Indeed! > >> Out of curiosity: Niklas had some work on a javascript version. That approach was different from Alex' insofar as, if I remember well, it returned the result in JSON-LD; I found that approach very compelling. Niklas, have stopped that? > > Yes, I'm afraid it's on hold for now. Since Alex's implementation was > completed I considered mine of less immediate importance. It was > mainly an experiment to see how little I need to do to "skim" JSON-LD > directly from RDFa. > > It became apparent (as expected) that I had to do a thorough > implementation to get more coverage, or else define a kind of "Lite++" > to get the basic things out properly. I'm not sure about the value of > that though, it's probably better to build on Alex's code, perhaps by > refactoring it to get an RDFa "walker" which can build up a JSON-LD > tree. > > The other insight was that the produced JSON-LD probably isn't ideal > to give to developers directly, since it directly corresponds to the > lexical form of the input RDFa. It's better to use a proper RDF API > *or* to use a supplied JSON-LD context to build up a "sparse" graph > reduced by the convenience forms defined in that context. > > (The latter thing is something I think is quite ideal as a general > mapper feature in an RDF API, btw. I've been meaning (for over a year > really) to e.g. rewrite my old Oort mapper [1] to use JSON-LD contexts > instead of the class-based approach. And to propose this as a future > path for both our shelved RDF API and for how to "digest" JSON-LD in > various scenarios. If only I had more time...) > > Best regards, > Niklas > > [1]: http://oort.to/ > > >> Ivan >> >> --- >> Ivan Herman >> Tel:+31 641044153 >> http://www.ivan-herman.net >> >> (Written on mobile, sorry for brevity and misspellings...) >> >> >> >> On 18 May 2012, at 04:38, Gregg Kellogg <gregg@greggkellogg.net> wrote: >> >>> Updated again to include Toby's Perl RDF::RDFa::Parser. >>> >>> Gregg >>> >>> On May 7, 2012, at 11:53 AM, Gregg Kellogg wrote: >>> >>>> I updated to include xml+rdfa and xhtml1+rdfa results for Green Turtle (JavaScript). >>>> >>>> Gregg >>>> >>>> On Apr 23, 2012, at 6:02 PM, Gregg Kellogg wrote: >>>> >>>>> I've updated the reports with four processors passing for XHTML+RDFa 1.1 and XML+RDFa 1.1. The processors passing all tests for these are now the following: >>>>> >>>>> clj-rdfa (Clojure) >>>>> librdfa (C) >>>>> pyRdfa (Python) >>>>> RDF::RDFa (Ruby) >>>>> >>>>> All but librdfa also pass for HTML5+RDFa and XHTML5+RDFa in addition to the vocabulary expansion tests. >>>>> >>>>> Gregg >>>>> >>>>> On Apr 22, 2012, at 5:11 PM, Gregg Kellogg wrote: >>>>> >>>>>> I updated the consolodated EARL reports [1]. It shows complete passing for clj-rdfa for basic and vocabulary tests. Librdfa passes most everything. pyRdfa is failing one minor corner case (0114). I also added any23 results, even though they fail a number of tests. >>>>>> >>>>>> Gregg >>>>>> >>>>>> [1] http://rdfa.info/earl-reports/ >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> Gregg >>>>>> >>>>>> >>>>> >>>>> >>>> >>>> >>> >>> > ---- Ivan Herman, W3C Semantic Web Activity Lead Home: http://www.w3.org/People/Ivan/ mobile: +31-641044153 FOAF: http://www.ivan-herman.net/foaf.rdf
Received on Monday, 21 May 2012 08:24:43 UTC