- From: Holger Knublauch <holger@topquadrant.com>
- Date: Wed, 26 Nov 2014 16:10:47 +1000
- To: public-data-shapes-wg@w3.org
On 11/26/2014 15:58, Eric Prud'hommeaux wrote:
>
> I'm happy to be pragmatic; let's look at the cost of all of these
> approaches.
>
> If we want to define Resource Shapes, remember that it is a recursive
> language (the valueShape of a Resource Shape is in turn another
> Resource Shape). There is no way to express that in SPARQL without
> hand-waving "and then you call the function again here" or "and then
> you embed this operation here" text.
The "hand-waving" here is that a compliant SPARQL engine needs to
implement a SPARQL function such as
shape:hasShape(?resource, ?shape) : xsd:boolean
which would be part of the spec that this WG produces.
However this is "just another SPARQL function" and doesn't require any
further changes to the language. Most SPARQL engines already have
functions that go beyond the official SPARQL 1.1 spec. The cost of that
sounds manageable to me.
SPIN also has the extension mechanism for user-defined functions, which
is for example already implemented by AllegroGraph. With a W3C logo on
it, I am confident that most interested database vendors would do the same.
Of course no RDF database would have native support for Shapes at this
stage yet, so of course something needs to be implemented.
Holger
Received on Wednesday, 26 November 2014 06:13:34 UTC