Re: N-Triples Parser for Python

Just out of curiosity, what are you defining as "fast"? On my server, I 
found MySQL peaks at around 2K (simple) transactions a second -- 
sufficient for some purposes, laughably poor for others. Is there 
anything smart you've done to get around this?

Cheers,
Chris

> Hi Kendall, Sean, Ivan, RDF Interest,
>
> While we're doing a bit of show and tell, I've implemented a fast
> mysql database store on top of rdflib for my veudas project. It's
> based (now loosely) around Steve Harris' excellent 3store design, but
> is indexed 6 ways like kowari and written entirely in python using the
> python/mysql drivers.
>
> There's a demo of it running on my site - I've stuck the wordnet data
> in it, along with a subproperty assertion linking wnet:wordforms to
> rdfs:labels. This means you should be able to search for any wordnet
> word via the label search.  (back-chained subproperty inferences on
> the fly)
>
> http://www.phildawes.net/temporary/veudas-0.6pre3/browse.html
>
> (N.B. this is dirt-cheap webspace at 20 quid a year, so don't expect
> it to be super fast!).
> (N.B.2 there's also no access restrictions. Please don't delete the
> data ;-)
>
> There's a query engine supporting a subset of sparql (including
> SOURCE, construct and optional). The parser is just a crappy regex one
> that I knocked up in an afternoon. - I'd love to be able to replace
> that with a proper parser.
>
> If anybody is interested (and brave enough), they can download a
> pre-release cut of the source code (GPL) at:
> http://www.sf.net/projects/veudas
>
> There are installation instructions, but the install is a bit rough
> and ready - it requires knowledge of apache + mysql and you might have
> to fight it a bit.
>
> Also I'd love something similar to sparta[1] in rdflib. Oh, and
> named-graph support. :-)
>
> Cheers,
>
> Phil
>
> [1] http://www.mnot.net/blog/2004/08/21/sparta

Received on Wednesday, 20 October 2004 22:24:31 UTC