Meeting minutes
j22: an RDF 1.2 parser can produce different triples to an RDF 1.2 parser.
<gb> Issue 86 Evolving 1.2 triple term features : rdf:parseType, rdf:annotation (by afs) [ms:CR]
<j22> w3c/
AndyS: Write-up of the different issues to be addressed
j22: If the top element has rdf:version or its:version, then its a parse error
https://
which says "attributes == set())"
j22: fail fast if rdf:version on rdf:RDF
gtw: Does anyone know of arbitrary of rdf:parseType="xxxx" in the wild?
<j22> AndyS: suggest a standard used for buildings in europe.
<niklasl> https://
<niklasl> (don't know if anyone ever used that though)
niklasl: rdflib accepts (and ignores) attributes on rdf:RDF
AndyS: Jena raises an error.
niklasl: and rdf:annotation would be property attributes i.e. triples.
<niklasl> Within RDF/XML documents it is not permitted to use XML namespaces whose namespace name is the RDF namespace IRI concatenated with additional characters.
niklasl: also a way is (I do not advocate it)
<niklasl> https://
AndyS: If RDF/XML is not streamable, then we can require rdf:version on the top element. But may be impactful.
tpt: My current feeling is that optional rdf:version is the way to go.
j22: Also the its:version, its:dir
tpt: maybe use the rdf: namespace
<niklasl> <rdf:Description rdf:value="X"/>
<j22> nodeElement =
<j22> element * - ( local:* | rdf:RDF | rdf:ID | rdf:about | rdf:parseType |
<j22> rdf:resource | rdf:nodeID | rdf:datatype | rdf:li |
<j22> rdf:aboutEach | rdf:aboutEachPrefix | rdf:bagID ) {
<j22> (idAttr | nodeIdAttr | aboutAttr )?, xmllang?, xmlbase?, propertyAttr*, propertyEltList
<j22> }
<niklasl> <rdf:Description rdf:annotation="X"/>
AndyS: its is a spec that changes all XML parsing, even if RDF/XML is embedded in a different document.
rdfs: comment "Idiomatic property used for structured values." ;
<niklasl> <rdf:Description>
<niklasl> <ex:something rdf:annotation="xyz">X</ex:something>
<niklasl> </rdf:Description>
<niklasl> <ex:something rdf:annotation="xyz">
<niklasl> <rdf:Description rdf:value="abc"/>
<niklasl> </ex:something>
AndyS: rdflib and Jena produce an error on property attribute + inline text for the value
AndyS: option of last resort is no RDF/XML 1.2
James: talking about protocol tests. new rdf1.2
gtw: new tests in new rdf 1.2 directory, old tests stay where they are
StrawPoll: +d(ont care) +1 0 -1
<TallTed> 0 is usually don't care
require rdf:version (ideally on root element) -- option 2 - assume control fo rdf nnamepace and// require rdf:version (ideally on root element) -- option 2 - assume control of rdf namespace and define as new words , version optional. -- option 3 don't care
option 2
option 2
<lisp> 3
<Tpt> option 2
<gtw> 3
<tl> +1 to option 2
<TallTed> 3 (I don't have a strong enough grasp of these details)
<niklasl> option 2, noting that I'm slightly worried about annotations being used on triples with empty string values (or langstrings), no rdf:version would interpret the data differently even in compliant RDF/XML parsers
j22: ITS may be supported by at the XML level.
its yes or no strawpoll (-1 no) +1 yes
<j22> +1
<AndyS> +1
<Tpt> +1
<niklasl> +1
zakim: end meeting
rrsagent: draft minutes