SPARQL Update

[Toby A Inkster]

On 8 Feb 2009, at 14:09, Kjetil Kjernsmo wrote:

> I don't have time to play, but I think the idea is good, so I'd  
> vote for
> an upload anyway :-)


Either 0.02 or 0.03 certainly will. I just want to get a handle on  
what people think definitely needs fixing first.

The SPARUL reference is here by the way:
<http://jena.hpl.hp.com/~afs/SPARQL-Update.html>

CREATE GRAPH, DROP GRAPH, LOAD and CLEAR should all work pretty much  
flawlessly. INSERT, DELETE and MODIFY should work for simple cases.  
The intention is for read-only SPARQL operations (ASK, CONSTRUCT,  
DESCRIBE, SELECT) to simply count as a no-op, but in practice it  
currently trips over parsing most of them, so they should not be  
included in SPARUL scripts at all.

I'd especially welcome feedback from kasei, as RDF::Query is  
obviously fairly related, even if I've gone down a different method  
of implementation (i.e. reusing Redland's functionality whenever  
possible).

(I initially thought I'd just be able to parse the SPARUL with  
regexps, break it into bits to pass to query Redland, manipulate  
Redland's query results with a few more regexps and then feed triples  
into Redland again. But then it hit me that any plan that begins with  
using regexps to parse a scripting language is doomed to failure. Two  
days and almost a thousand lines of code later, it ended up as mostly  
a recursive descent parser, with a few regexps thrown in to shortcut  
parsing the bits that I know I can throw to Redland as-is.)

-- 
Toby A Inkster
<mailto:mail at tobyinkster.co.uk>
<http://tobyinkster.co.uk>