Hi, Reading Jonathan's briefing note [1], particularly about the social contract, highlighted that one of the things that's been bothering me about the httpRange-14 design is the lack of conformance requirements for applications that consume data from the web.
Whatever decisions are made about revising the wording or design, I think it would be very useful for the implications for at least RDF consumers to be spelled out (it's probably harder to do so concretely for other structured data consumers). For example: * MUST/MAY/SHOULD a consumer use the URI of a document as the URI of a RDF graph or MUST/SHOULD they NOT? * MUST/SHOULD a consumer add particular information to the set they glean from a given representation based on the HTTP response and if so what is it? Without these, I think it can be hard for publishers to realise the ways in which consumers will handle their data differently based on how they publish it, which is of course what motivates publishers to be conformant themselves. Also without conformance levels we don't know who to give gold "Good Linked Data Application" stickers to, and who to bang over the head with big sticks until they behave. Cheers, Jeni [1] http://www.w3.org/2001/tag/2012/04/57guide.html -- Jeni Tennison http://www.jenitennison.com
