- From: Andy Seaborne <andy.seaborne@epimorphics.com>
- Date: Wed, 14 Dec 2011 10:00:20 +0000
- To: sparql Working Group <public-rdf-dawg@w3.org>
(from this week's telecon - this email picks off this one point in the
hope of "divide-and-conquer" progress)
The GET on a graph store URI is defined to return the service
description. We didn't understand why.
The RDF-WG is (likely to) define formats for n-quads and TriG that can
represent an RDF datasets.
GET <uri> returns a representation of the resource at <uri>.
So returning n-quads seems natural (an implementation can 403 it if it's
too big).
A service description describes a service and its endpoints.
It connects a service to a dataset with sd:defaultDataset.
If GET of the graph store URI is the SD then don't we have:
<uri> sd:defaultDataset <uri> .
A graph store is a container of graphs (unnamed and named). It is not a
general RDF processing service.
It might be that the graph store URI is overloaded with ?query= and
?request= functionality but it is overloading the service endpoints with
the resource itself. In this case alone, does returning the SD make
complete sense to me.
I can see why returning a description of the graph store makes sense and
the SD vocabulary has language for that but that's only a part of the
service description.
1/ Should the return at least be a description of the graph store (using
the sd:defaultGraph sd:NamedGraph etc vocab)?
2/ Could we allow GET on a graph store URI return quads?
Andy
Received on Wednesday, 14 December 2011 10:03:18 UTC