Re: bigdata bananardf implementation

Hi Anton,

You started something I wanted to do.
I use currently Jena TDB embedded on my Banana based application,
but I'm aware that TDB has a bottleneck for update transactions.

And I participate in a web e-commerce application that uses BigData through
SPARQL HTTP,
and Jena TDB with Banana in the midlle tier server.

So I'm very motivated to contribute to a BigData implementation.

Please read interleaved answers below.

2014-12-09 20:58 GMT+01:00 Anton Kulaga <antonkulaga@gmail.com>:

> Hi all!
>
> I have started to create bananardf implementaion for bigdata ( bigdata.com
> ), is anybody else interested in collaborating?
>
> I already face many challenges there, as many things are different then in
> sesame. Although bigdata supports sesame it provides also native classes
> with some extra features.
>

I couldn't find the API with extra features in the wiki nor the JavaDoc .
What  extra features does it bring ?


> There are several problems.
>
> 1) Many methods in bigdata (all factories and most of query functions)
> require baseURI or namespace. I have no clue how I can write bigdata
> modules that will have configurable baseURI as most of implicit resolutions
> in banana are done from the typeclass that is passed as generic type
> parameter.
>

I tried using named extra arguments; it does not work.
Implicit can work:

class T { def f(i:Int)= 4 }
implicit val baseURI= "http://bla"
class S() (implicit val baseURI:String) extends T { override def f(i:Int )
= baseURI.length }
new S().f(5)


> 2) Bigdata has embeded and remote modes. Bigdata in remote mode returns
> different connection and Result classes than bigdata in embeded mode. Does
> it mean that I have to create two typeclasses: one for bigdataremote and
> one for bigdata embedded?
>

I don't think this is a problem,
At the worst one can use a asInstanceOf to distinguish them in Banana
implementation.

But what is noteworthy is that the issue of implementing a wrapper <: RDF
in Banana for a remote SPARQL connection is a generic issue, not dedending
on the particular endpoint implementation.


>
>
> --
> Best regards,
> Anton Kulaga
>



-- 
Jean-Marc Vanel
Déductions SARL - Consulting, services, training,
Rule-based programming, Semantic Web
http://deductions-software.com/
+33 (0)6 89 16 29 52
Twitter: @jmvanel , @jmvanel_fr ; chat: irc://irc.freenode.net#eulergui