W3C home > Mailing lists > Public > public-rdf-comments@w3.org > June 2013

Turtle recommendation for serialising xsd:string datatype

From: Peter Ansell <ansell.peter@gmail.com>
Date: Sun, 2 Jun 2013 10:41:47 +1000
Message-ID: <CAGYFOCTuy7fNeqwR994KWb3rezf=6CxefEVi9N9_GjTagL60EQ@mail.gmail.com>
To: public-rdf-comments@w3.org
The Turtle Canididate Recommendation (19 February 2013) includes
compatibility with one part of the RDF-1.1 draft in regards to not
recognising literals with neither a language tag or a datatype. It
specifies that if there is not an explicit datatype then xsd:string is
given [1]. However, it does not recommend whether to explicitly use
xsd:string or leave it out when serialising RDF abstract models out to
Turtle.

Option 1: Recommend that serialisers SHOULD leave xsd:string out when they
are serialising an RDF Abstract Model containing literals with xsd:string
as a datatype.

- This would make documents smaller.
- Serialising RDF-1.1 using Turtle Candidate Recommendation, would remove
xsd:string from Literals that may legitimately need it if the passed
through an RDF-1.0 application using another format (including historical
versions of Turtle) that doesn't have the same presumptions for
non-language-tagged-plain-literals, even if they were intended to
explicitly be xsd:string by the original author when they were created
using the RDF-1.1 abstract model.

Option 2: Recommend that serialisers SHOULD explicitly serialise xsd:string
as datatype if the RDF abstract Literal contains xsd:string as its datatype

- This would provide compatibility for RDF-1.1 abstract models serialised
using Turtle, when they are parsed using Turtle parsers based on past
versions of Turtle. It would also provide compatibility with RDF-1.0
abstract model implementations where xsd:string was not always assumed when
there was no datatype.

Using Option 1 would be fine if everyone was required to use the RDF-1.1
model for representing both parsed Turtle Candidate Recommendation
documents and all other common RDF serialisations, but that doesn't seem to
be an option given the current timelines.

However, given that Turtle includes a normative reference to RDF-1.0
Abstract Concepts, presumably only to make it possible to publish the
Turtle specification before RDF-1.1 is published, Option 2 sounds more
useful for a mixed RDF-1.0/RDF-1.1 ecosystem.

Thanks,

Peter

[1] http://www.w3.org/TR/2013/CR-turtle-20130219/#turtle-literals
Received on Sunday, 2 June 2013 00:42:14 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:29:57 UTC