- From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
- Date: Mon, 08 Nov 2004 05:45:54 -0500 (EST)
- To: andy.seaborne@hp.com
- Cc: public-rdf-dawg-comments@w3.org
From: "Seaborne, Andy" <andy.seaborne@hp.com> Subject: RE: Coments on first working draft of SPARQL Date: Mon, 8 Nov 2004 09:21:53 -0000 > > On the contrary, the syntax of the language permits a URI to be a bNode > > (or at least something of the form of a bnode) via > > > > [43] URI ::= QuotedURI | QName > > [44] QName ::= <QName> > > [48] <QName> ::= (<NCNAME>)? ":" <NCNAME> > > [62] <NCNAME> ::= <NCCHAR1> (<NCCHAR1> | "." | "-" | ["0"-"9"] | > > "\u00B7" )* [61] <NCCHAR1> ::= ["A"-"Z"] | "_" | ... > > > > So a URI can be, for example, _:A > > Interesting - the syntax for qnames is taken from the recommendations > for XML 1.1 and XML 1.1 namespaces. [...] > Qnames are not bNodes. Given that some other syntaxes use _: for a > bNodes it can create confusion - to be balanced against following the > XML token definitions. > > Andy Yes, agreed, interesting. The intent was probably not to allow bnodes via this path, but the divergence between literal in the grammar and literals in RDF serves to really muddy the waters - if a literal can be a URI reference then why can't a QName be a bnode? peter
Received on Monday, 8 November 2004 10:38:04 UTC