- From: Steve Harris <steve.harris@garlik.com>
- Date: Mon, 17 Oct 2011 15:24:08 +0100
- To: Richard Cyganiak <richard@cyganiak.de>
- Cc: RDF Working Group WG <public-rdf-wg@w3.org>
On 2011-10-17, at 15:19, Richard Cyganiak wrote: > On 17 Oct 2011, at 14:56, Steve Harris wrote: >>> The original use case for TriG (this is years before SPARQL existed) was serializing collections of RDF documents in a single document. In that use case there can be no blank nodes shared between documents, so graph scope made sense and had the convenient side effect of simplifying serializer implementation. >> >> It also simplifies the parse process. >> >> Suppose I start parsing a TriG document and see: >> >> <G1> { _:x a <Foo> } >> >> then 100GB later in the file I find: >> >> <G1734> { <x> :seeAlso _:x } >> >> If there was document scope I would have to carry the assigned skolem form* for _:x around in my parser for the entire parse - that's fine if you only have a handful of bNodes, but with billions of them it can be quite resource intensive. > > I'm not sure if I understand. Don't you have a problem already if you import a single large N-Triples file? Yes. - Steve -- Steve Harris, CTO, Garlik Limited 1-3 Halford Road, Richmond, TW10 6AW, UK +44 20 8439 8203 http://www.garlik.com/ Registered in England and Wales 535 7233 VAT # 849 0517 11 Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10 9AD
Received on Monday, 17 October 2011 14:24:40 UTC