- From: Andy Seaborne <andy.seaborne@talis.com>
- Date: Tue, 09 Mar 2010 09:12:29 +0000
- To: David Booth <david@dbooth.org>
- CC: Dave Beckett <dave@dajobe.org>, semantic-web@w3.org
On 08/03/2010 3:48 PM, David Booth wrote: > On Mon, 2010-01-25 at 08:23 -0800, Dave Beckett wrote: > [discussion of updating Turtle to handle quads / named graphs] >> I was more thinking of allowing something like >> >> <s> <p> <o> <g> . > > Is there a reason why the graph URI would be the last element of the > tuple, rather than the *first*? > <g> <s> <p> <o> . > As the first element: > - the graph URI would bemore like high order bits, so simple text > sorting of a quad file would group all quads of the named graph > together; and > - the graph URI would be easier to strip off when converting the quads > back to triples, because one could simply strip everything through the > first space on each line -- a very easy sed script. I find it easiser to think of quads as GSPO but the n-quads format putting it last means its more naturally optional. When absent, the line is describing a triple for the default graph for an RDF dataset. <s> <p> <o> <g> . <s> <p> <o> . That could be done by counting elements but having all the subjects in the subject column makes sense as well. Andy
Received on Tuesday, 9 March 2010 09:12:59 UTC