Re: [rdf4j-dev] Google Summer of Code ideas

[Jeen Broekstra <jeen.broekstra@xxxxxxxxx>]

On 29 Jan 2017, at 09:37, Håvard M. Ottestad <hmottestad@xxxxxxxxx> wrote:

Implementing SHACL using the least amount of abstraction would be my recommendation. That way you could really tune the performance.

Would not recommend either SPARQL or SPIN.

I think this overlooks the fact that most RDF databases already have a lot of performance tuning in place for SPARQL querying. Bypassing the SPARQL engine means you lose those optimisations - and it is IMHO unlikely that any optimisations you make in your own code can outperform those native database optimisations. 

Regardless of the specific approach, a SHACL implementation definitely sounds like a good idea (though possibly quite challenging for someone new to RDF4J). Could you log a ticket for this with an explanation of what’s involved and possible approaches? I can then reuse that issue description to log it as an idea on the Eclipse GSoC page.

Cheers,

Jeen