- From: Sandro Hawke <sandro@w3.org>
- Date: Wed, 19 Oct 2011 08:25:30 -0400
- To: Andy Seaborne <andy.seaborne@epimorphics.com>
- Cc: public-rdf-wg@w3.org
On Wed, 2011-10-19 at 11:32 +0100, Andy Seaborne wrote: > > 2/ (primer) defining "well-formed" > (formally in RDF semantics). The 2004 Primer suggests well-formed-ness is whatever parsetype=Collection would give, which is on the right track, but you can't put literals in parsetype=Collection lists, so that's not the right definition. Thus my reference to Turtle's list syntax. The 2004 Semantics leaves out the notion that you can't have extraneous links (which also means you can't actually *say* rdf:type rdf:List for the nodes) for well-formedness, and it's hardly obvious. I hadn't realized how close 'well-formed' already was to 'simple lists'; so, yeah, maybe no one will have a problem with saying well-formed is what can be serialized in Turtle's list syntax (but I guess in the Semantics we have to say that without reference to Turtle). That'd be good. "Well-formed" has a fair amount of "you should do it this way" in the name. Should we make some WellFormedList NotWellFormedList test cases? Can you think of a way to do this without a new test type? -- Sandro
Received on Wednesday, 19 October 2011 12:25:39 UTC