W3C home > Mailing lists > Public > public-rdf-dawg-comments@w3.org > March 2009

SPARQL 1.1 input from ARC folks

From: Benjamin Nowack <bnowack@semsol.com>
Date: Tue, 10 Mar 2009 09:26:03 +0100
To: public-rdf-dawg-comments@w3.org
Cc: "arc-dev" <arc-dev@semsol.org>
Message-ID: <PM-GA.20090310092603.F1BDD.1.1D@semsol.com>

I collected feature requests/priorities on the ARC mailing list[1] and
took some notes during the last months. Here is the merged result, 
loosely ordered by expressed priority (from "essential" to "nice2have"):

* Aggregate functions (esp. COUNT and SUM) + GROUP BY
* assignments / projectable expressions
* bnode referenceability
* Service description when accessing an endpoint w/o a 
  query parameter
* multiple queries per endpoint call (e.g. a DELETE + an INSERT)
* a syntactical equivalent to SQL's "IN", e.g. to check a var against
  a set of values from a separate query (via Andraz Tori, Zemanta CTO)

Aggregates, INSERT, DELETE, and LOAD are already supported by ARC's
SPARQL+[2], an extension that tries to keep the changes to the standard
spec at a minimum (8 grammar tweaks/additions).

BNode referenceability is supported by a <_:bnode> (uri+bnode) notation.

I understand that it won't be possible to incrementally release minor
SPARQL updates due to the REC process, but maybe it could be possible 
to define URIs for the different features (or feature profiles). I am 
pretty sure that ARC won't be able to support all of SPARQL 1.1, so it 
would be nice to be able to formally describe and publish the 
capabilities of an endpoint. Clients could then decide how/if to send
queries. Profiles could possibly also lead to more SPARQL implementors
and endpoint providers.

Some general feedback for SPARQL 1.0 (as I think the group members are
mostly the same): Great work! During the last months, I had the chance
to chat with quite a number of non-RDF experts who discovered SPARQL 
through Web 2.0-ish channels (mostly via semantic crunchbase, but also
via SPARQL-driven demo apps like sparqlbot, dooit, or smesher). I've 
been asked for a LIKE-like equivalent to the FILTER+REGEX pattern a 
couple of times, but in general, people seemed to be very happy with 
the syntax. It apparently makes it easy to get started with SPARQL 
for web developers who know (basic) SQL.


[1] http://bit.ly/mIsj
[2] http://arc.semsol.org/docs/v2/sparql+

Benjamin Nowack

semsol / benjamin nowack
bielefelder str. 5
40468 duesseldorf, germany

fon:   +49.211.7316824
fax:   +49.211.1587107
cell:  +49.177.5241935
skype: bnowack

Received on Tuesday, 10 March 2009 08:32:08 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:52:10 UTC