jena-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andy Seaborne <andy.seabo...@epimorphics.com>
Subject Re: query support in fuseki
Date Sat, 05 Feb 2011 20:47:17 GMT


On 04/02/11 23:04, Tim Harsch wrote:
> I noticed some code in Fuseki that will throw error if a query
> contains FROM/FROM NAMED clauses.  I think it makes sense given it is
> Sparql 1.1 SOH protocol and since you specify the dataset in your
> call you can't do it in the FROM/FROM NAMED clauses too.  But, does
> that mean that Fuseki will never deprecate Joseki since in order to
> query multiple data sets you would need FROM/FROM NAMED clauses in
> the query and only Joseki would support that?  Or will/does Fuseki
> support standard Sparql endpoint semantics as well?

Fuseki provides the ability to publish and update datasets on the web. 
Mostly it's one dataset (and the current command line only handled one 
dataset but internally it looks up any incoming request and directs it 
to the right dataset).

That's one feature Joseki currently has that Fuseki has not expose.  Yet.

The reason you get an error if you use FROM/FROM NAMED is because it's 
publishing a dataset.  Normally FROM/FROM NAMED means load (using HTTP 
GET) the graphs named, create a dataset and query it. That's what the 
/sparql service (joseki:ProcessorSPARQL) does.  It offers a general 
purpose query service, which is not offering query over a dataset but 
enables the loading of graphs into a dataset.  Some people use 
sparql.org/sparql for this - not that I can guarantee it's up and
running.

Fuseki is running like joseki:ProcessorSPARQL_FixedDS.  This standard 
SPARQl behaviour - in fact, most other systems only offer this behaviour 
because they provide access to the data stored behind them.

It can be a bit expensive and opens up certain issues of reducing 
service to other datasets by hitting the server with work to load data. 
  By the way, Joseki has a built in 50K triple limit on any data it 
loads to reduce accidents.  There's no reason why another service 
couldn't be added to Fuseki to do that.

GRAPH is the SPARQL keyword to pick one graph in the dataset.

It would be possible to enable the other behaviour of FROM/FROM NAMED 
for TDB:

http://openjena.org/wiki/TDB/DynamicDatasets

which is useful for the effect of default graph (select a few from many, 
merge and query, except it does not actually merge the data) and GRAPH ?g.

And you can do the FROM/FROM NAMED thing with the ARQ command line tools.

	Andy

Mime
View raw message