RE: Normative vs Informative

I dont have any strong technical opinion. I do have a rational, though. What I like about the RDFa of today is that it works with the actual web. This was my test (all along); and was the threshold test for webid's viablity in my user community, too. There is no point bleating on about user centric identity, ranting about the evils of CAs, if the whole thing is a research-community foil for vendor-dominated identity, finding an excuse to make yet more tools.  To take an example, the openid community did itself no favors by switching gears from User centric to cloud-vendor-centric identity management. To be fair, it had to (in order to survive, at all). Having survived, it good now to see some of the wiser vendors of cloud solutions re-biasing things back in  favor of users (and less in the interests of mega-IDPs), so there is some *politically-viable* mix of user and vendor and cloud and government (i.e. civil society) interests - just like the world of SSL. This is what you need, if your poltically-dynamite technology (crypto) is going to work for a billion connected systems, from 1000s of vendors, of all denominations (and quality levels). In reality, my "forcing function" implementation of the spec as actually given (which doesnt work with my actual first test case, probably due to underspecification) did impose RDF/XML as the single common format. I tried to use N3 in that f ole, but that failed the end-end, multi-vendor/provider test (at first attempt, anyways). Regalrdss of all that, the pragmatics of "implementation" did not impose RDF/XML on the poor user, particularly one that will struggle to host such a stream in any common or garden blog site. That web based document translation services parse HTML4/XHTML/RDFa and produce RDF/XML seems IDEAL. It can now gateway 1000 varieties of JSON, for all I care, at this point. What I liked all along about the triple model was that its pure logic, and there is nothing in the tools that make any fundamental difference. Webid is 3+ years old, and counting. It will soon be 12 and have just as many users at this rate, to reuse Kingsley's admonition. We have to focus on what it takes to adopt it, NOW, using the tools consumers ACTUALLY have. Otherwise, folks at Microsoft will laugh at me, when I suggest they tune up windows Server for webid, ready for data center level throughputs. They will tell me its on the list (at priority -999). i.e. go away.      
 > From: mo.mcroberts@bbc.co.uk
> Date: Sun, 27 Nov 2011 13:38:20 +0000
> CC: public-xg-webid@w3.org; henry.story@bblfish.net
> To: ddooss@wp.pl
> Subject: Re: Normative vs Informative
> 
> 
> On 27 Nov 2011, at 11:46, Dominik Tomaszuk wrote:
> 
> > On 27.11.2011 03:14, Henry Story wrote:
> >> Great idea. Let's start with a wiki.
> >> Open one page: Formats, and have it point to an example page in every other format
> >> you can think of - so that each format can get the attention it deserves.
> >> 
> >> Here I wrote up the initial page:
> >> 
> >>    http://www.w3.org/2005/Incubator/webid/wiki/Formats
> > OK, but first I think we should discuss in mail list. For example I disagree that json-ld is important now to WebID. I'm very big fun of JSON-LD, but now it's community draft and we have RDF/JSON draft [1] from RDF WG.
> > So wiki isn't good way to do it, because quickly turns out that we have hundreds of formats in wiki.
> 
> I don’t think having lots of example serialisations and related guidance is a problem by any means — quite the opposite.
> 
> However, I am rapidly coming to the conclusion that the WebID spec should mandate RDF/XML and only RDF/XML (with either conneg or <link>) for the time being.
> 
> The rationale is that RDF/XML is the one serialisation — at present — guaranteed to be supported by RDF consumers, and that RDFa still has some way to go yet before it’s properly reconciled with the HTML5 parsing work. Until it is, any notion of error-handling (vitally important in real web pages) is going to be largely undefined.
> 
> What I’d like to see is the RDFa spec build upon the HTML5 parsing rules, so that there’s clarity and consistency in how to deal with RDFa sprinkled into real-world “tag soup” pages, and for validation and other consuming tools to fall into line — AFAIK this hasn’t happened *yet*, and until it does I don’t think RDFa belongs as a MUST for WebID consumers in the spec as it is now.
> 
> (Side-note: JSON-LD and RDF/JSON solve different classes of problem; they’re not competing specifications, they’re just two JSON-based serialisations of RDF).
> 
> M.
> 
> -- 
> Mo McRoberts - Technical Lead - The Space,
> 0141 422 6036 (Internal: 01-26036) - PGP key CEBCF03E,
> Project Office: Room 7083, BBC Television Centre, London W12 7RJ
> 
> 
> 
>