W3C home > Mailing lists > Public > public-rdf-wg@w3.org > July 2013

Re: RDF-ISSUE-137 (aligning-trig-and-sparql): Should TriG be a subset of SPARQL Quad Pattern? [RDF TriG]

From: Andy Seaborne <andy@apache.org>
Date: Wed, 03 Jul 2013 15:49:09 +0100
Message-ID: <51D439E5.6080300@apache.org>
To: public-rdf-wg@w3.org
On 03/07/13 03:24, RDF Working Group Issue Tracker wrote:
> RDF-ISSUE-137 (aligning-trig-and-sparql): Should TriG be a subset of
> SPARQL Quad Pattern? [RDF TriG]
>
> http://www.w3.org/2011/rdf-wg/track/issues/137
>
> Raised by: Sandro Hawke On product: RDF TriG
>
> (Raised as per discussion at 2013-06-26 meeting)
>
> TriG and SPARQL Quad Patterns *almost* overlap nicely, but there are
> two syntactic differences that make the languages entirely disjoint.

The description includes opinion.  Please include all points of view.

Some people (not me) have said they see it as a positive advantage -- 
evidence: different MIME types.

-1 to the current description with opinion.

> 1.  SPARQL uses the GRAPH keyword before the graph name.  Traditional
> TriG doesn't.   We agreed to include the GRAPH keyword as an "at
> risk" feature, in parallel to PREFIX and BASE being "at risk".   One
> could argue that it automatically follows from accepting PREFIX and
> BASE into Turtle that we accepted GRAPH into TriG.

Disagree.

> At very least
> the spec should include it, marked at risk.
> https://www.w3.org/2013/meeting/rdf-wg/2012-10-03#resolution_7

Noting the wording there

"... the case-insensitive keyword "graph" MUST NOT appear before the 
name, in a name-graph pair ..."

TriG was published April 2013

>
> 2.  SPARQL does not have curly braces around the default graph
> triples (although it kind of looks like it, since it has them around
> the whole quad pattern).
>  We decided that TriG would require them:
> https://www.w3.org/2013/meeting/rdf-wg/2012-10-17#resolution_2 ...
> but I don't think we were thinking about SPARQL compatibility at that
> point.

I was. I have raised this matter before.  It is not new and that is why 
there is this resolution.  (Also raised was N-Quads in TriG.)

> Also, at that point, the same bytes parsed as Turtle and
> TriG had different semantics, but since
> https://www.w3.org/2013/meeting/rdf-wg/2013-06-12#resolution_2 they
> now have the same semantics, so having overlapping syntaxes should be
> okay.

This conflicts with having simple parsing which is also a goal we have 
had since the beginning.

It would require significant changes to many (all?) existing 
implementations of TriG parsers.

significant => structural change to the parsers, not a single rule change.

It is a divergence from Turtle.

> PROPOSED:  Make TriG (not counting prefix & base) be a subset of the
> SPARQL quad pattern language, like Turtle is, by allowing (1) braces
> around the default graph and (2) the GRAPH keyword.  Explain that
> this form is preferred for SPARQL alignment, but isn't handled by
> older TriG parsers.

"preferred" in (2) conflicts with 2012-10-03#resolution_7

	Andy
Received on Wednesday, 3 July 2013 14:49:40 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:02:15 UTC