JSON-LD actually allows you chose between the two: you can always copy the entire context into your data, to make it as self-sufficient as an RDF/XML or Turtle file. On 22/06/2021 15:26, Dan Brickley wrote: > > I knew Ralph had said this, but only just stumbled across where he > wrote it down. > > It was when Netscape submitted the XML-ified version of Guha's MCF to > W3C, June 1997. This work was picked up by the RDF Model and Syntax > WG. To put this in context, this was before the XML spec was itself > finalized. > > https://www.w3.org/Submission/1997/8/Comment.html > <https://www.w3.org/Submission/1997/8/Comment.html> > > "An important requirement on the metadata framework is that parsers > must be able to produce a parse tree of the metadata without the > assistance of an auxiliary format description. This corresponds to the > XML /well-formed/criterion" > > I share this since various of us have been banging on about this > design issue on the Signed LD thread, w.r.t. JSON-LD Contexts. > > RDF/XML took the verbose path, such that knowing only the bytes in the > file + base URI, you could get the same triples out, with no extra > info. It's neither better nor worse than other approaches, but it is a > distinctive design constraint and I'm glad I found it written down > somewhere, finally. Turtle, Ntriples, also work the same way; GRDDL, > CSVW RDF mappings don't, as they require multiple files before you > know what triples you'll get. > > DanReceived on Tuesday, 22 June 2021 14:52:33 UTC
This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:46:09 UTC