From: Rob V. <rv...@do...> - 2014-02-13 11:10:37
|
Hi Milos To be honest we've moved away from SQL backed storage because in the past it hasn't proven that performant. However providing a R2RML adaptor layer does make a lot more sense and provides a better solution than what we've tried in the past (which was based on storing triples natively in a SQL database). I would suggest looking at implementing a IStorageProvider which also implements IQueryableStorage if you are intending to support SPARQL. Since this likely involves some sort of dependency upon Tom's existing R2RML library then this would be best implemented as a separate library in a similar way to the existing Data.Virtuoso library because we want to mimize dependencies in the core library. I'd suggest reading up on the Storage API (https://bitbucket.org/dotnetrdf/dotnetrdf/wiki/UserGuide/Storage%20API) if you haven't already and taking a look at the implementations of existing IStorageProviders. Contribution wise I would suggest you create your own fork, work in there and then submit pull request(s) as and when you have something you thing is sufficiently stable for inclusion in the main project. Please take a look at our Code Style guidelines (https://bitbucket.org/dotnetrdf/dotnetrdf/wiki/DeveloperGuide/Code%20Style) and Test Environment (https://bitbucket.org/dotnetrdf/dotnetrdf/wiki/DeveloperGuide/Test%20Enviro nment) documentation. Any substantial feature contribution like this will need to be accompanied by appropriate tests which I appreciate can be tricky for testing that requires on external resources I.e. SQL servers but you can use the approach we use for existing storage to provide developer configurable settings for test setup and skip tests where the test system is unavailable. And of course if you have questions about anything then please ask on the developer list which I've CC'd on this response, this is a subscription based list which you will need to subscribe to via https://lists.sourceforge.net/lists/listinfo/dotnetrdf-develop Best Regards, Rob Vesse From: Miloš Chaloupka <cha...@ho...> Date: Monday, 10 February 2014 09:18 To: Rob Vesse <rv...@do...> Subject: DotNetRDF > Hi Rob, > > I have found you as an author of DotNetRDF so I hope I am contacting the right > person. I have already written to Tomasz Pluskiewicz, but got no response yet, > so I am not really sure, whether he even got the message. > > I am currently writing my master thesis "Querying RDF graphs stored in a > relational database using SPARQL and R2RML" and I am choosing how I will write > the implementation. I don't want to write a solution that will end as an > academic work with no chance of real use so I think, that it would be nice to > make it as a "part" of dotNetRDF. I want to implement it as a store, that will > be queryable using SPARQL according to R2RML mapping. I can see that you have > already listed several libraries and that is in fact the only thing I want. > Maybe you want to ask why I am asking before I am starting the implementation, > it is because my supervisor wants to make the implementation as a part of > something bigger so I want to know whether it is possible. Are you interested > in such storage implementation? And if so, do you have any demands? > > Thank you for your reply, > best regards, > Miloš Chaloupka |