- From: Lee Feigenbaum <lee@thefigtrees.net>
- Date: Thu, 14 Apr 2011 09:40:42 -0400
- To: Steve Harris <steve.harris@garlik.com>
- CC: Eric Prud'hommeaux <eric@w3.org>, antoine.zimmermann@insa-lyon.fr, Antoine Zimmermann <antoine.zimmermann@deri.org>, public-rdf-wg <public-rdf-wg@w3.org>
On 4/14/2011 9:33 AM, Steve Harris wrote: > Sorry, thinko, what I meant was: > >> I just think it's simpler and less weird to have datatyped literals and languaged literals and nothing else. > > I disagree, and I think it's very odd that the canonical serialisation would be: > > "chat"^^xsd:string } > "chat" } > "chat"@en-GB } - all xsd:strings > "chat"@fr } > > v's > > "chat" } > "chat"^^xsd:string } > "chat"@en-GB } - all plain literals > "chat"@fr } > > e.g. there's one syntax that's slightly weird if you normalise to plain literal, and three is you normalise to xsd:string. IMHO. I'm still not sure I understand what you're saying here. How do we observe the difference between these two views? I've been picturing that with my view (normalize to xs:string) that datatype("foo") = xs:string It seems like you're mostly interested in the canonical serialization? I don't have any objection to the canonical serialization of "foo"^^xs:string being "foo", if we're worried about bytes. I just think having theses things in the model without a datatype and without a language introduces (unnecessarily) a 3rd thing that you have to deal with when querying, programming, etc. I'm happy for the serialization issues to make it as easy as possible to deal with xs:string's, though. Lee
Received on Thursday, 14 April 2011 13:41:20 UTC