- From: Dave Beckett <dave@dajobe.org>
- Date: Sat, 08 Sep 2007 17:02:38 -0700
- To: Ivan Herman <ivan@w3.org>
- Cc: Dan Brickley <danbri@danbri.org>, Ben Adida <ben@adida.net>, W3C RDFa task force <public-rdf-in-xhtml-tf@w3.org>
Ivan Herman wrote: > > Dan Brickley wrote: >> Please publish the draft so I can criticise it! >> >> I'm uncomfortable with literal value inspection as the notation for >> indicating type. It makes round-tripping hard. I could have a load of >> data, all of which in my source database used XMLLiteral as a type, even >> if only some of the values (currently) had markup in them. I can >> reasonably compose SPARQL queries that pull out these as XMLLiterals. If >> I push that data out into RDFa, my literals will sometimes end up >> XMLLiteral, and sometimes plain Literal, in the process perhaps >> acquiring language tags from the HTML context. When parsed back into a >> database, I'm left with significantly mangled data. > > Dan, > > this was a long discussion, and fairly passionate one... the precise > process we have now is at: > > http://www.w3.org/2006/07/SWD/wiki/RDFa/LiteralObject > > which reflects the consensus we reached after, as I said, long > discussions. I think the only additional information missing from that > figure is that @datatype="" yields a plain literal. > > I would be interested what would be your alternative... That process looks good to me - it was the kind of thing I was going to suggest, having seen what the various RDFa things of today were generating, in the absence of a working draft to explain. Dave
Received on Sunday, 9 September 2007 00:02:48 UTC