SPARQL version supported by ARQ

[Frank Budinsky <fr...@ca.ibm.com>]

Hi,

It seems that the latest ARQ drivers are not compatible with any W3C
Working Draft, that is, the current SPARQL 1.1 working draft is version 5
(January 2012) which is using "BINDINGS", not "VALUES", but ARQ seems to be
implementing an as-yet-unpublished editor draft 99. Does anybody know what
the target date is for having a published W3C working draft 6?

Maybe this is a dumb question, but how is an ARQ user supposed to figure
out the exact SPARQL syntax supported by a particular version of ARQ?

Thanks,
Frank.

Re: SPARQL version supported by ARQ

[Paul Gearon <ge...@ieee.org>]

There's a minor change required in the grammar (currently prefixes are
allowed to contain a : character, which shouldn't be permitted), but
as soon as that is done then SPARQL 1.1 Query will be published as a
Last Call Working Draft. I expect it within a week (or so).

ARQ tends to track the latest editors' draft from the SPARQL Working
Group. It is typically one of the first systems to implement and test
changes to the proposed spec. So ARQ is going to support whatever was
the most recent editors' draft when it was released. That's hard (but
not impossible) to track for historical versions, but easy enough to
tell when a version of ARQ comes out.

With the exception of BINDINGS/VALUES there haven't been significant
changes lately, and with the Last Call coming up I don't expect that
there will be any more (it would have to be a major problem to reopen
the process at this point). So users can work with the Last Call
Working Draft with reasonable confidence that this will be the final
version. The current editors' draft is the one that will be published
(with the one issue I already mentioned).

Regards,
Paul Gearon

On Tue, Jul 17, 2012 at 10:53 AM, Frank Budinsky <fr...@ca.ibm.com> wrote:
>
>
> Hi,
>
> It seems that the latest ARQ drivers are not compatible with any W3C
> Working Draft, that is, the current SPARQL 1.1 working draft is version 5
> (January 2012) which is using "BINDINGS", not "VALUES", but ARQ seems to be
> implementing an as-yet-unpublished editor draft 99. Does anybody know what
> the target date is for having a published W3C working draft 6?
>
> Maybe this is a dumb question, but how is an ARQ user supposed to figure
> out the exact SPARQL syntax supported by a particular version of ARQ?
>
> Thanks,
> Frank.

Re: SPARQL version supported by ARQ

[Frank Budinsky <fr...@ca.ibm.com>]

Thanks Andy and Paul for your replies, and great news that SPARQL 1.1 is
going to Last Call Working Draft now.

Thanks,
Frank.


Andy Seaborne <an...@gmail.com> wrote on 07/17/2012 11:18:07
AM:

> From: Andy Seaborne <an...@apache.org>
> To: users@jena.apache.org,
> Date: 07/17/2012 11:19 AM
> Subject: Re: SPARQL version supported by ARQ
> Sent by: Andy Seaborne <an...@gmail.com>
>
> On 17/07/12 15:53, Frank Budinsky wrote:
> >
> >
> > Hi,
> >
> > It seems that the latest ARQ drivers are not compatible with any W3C
> > Working Draft, that is, the current SPARQL 1.1 working draft is version
5
> > (January 2012) which is using "BINDINGS", not "VALUES", but ARQ seems
to be
> > implementing an as-yet-unpublished editor draft 99. Does anybody know
what
> > the target date is for having a published W3C working draft 6?
>
> The semantics of property paths are different as well, although as they
> are now back to what SARQ originally had prior to the working group, not
> many people will have noticed.
>
> The WG has resolved (today) to publish SPARQL Query as a Last Call
> Working Draft, subject to alignment with Turtle over the prefix grammar
> rule that arose overnight.
>
> > Maybe this is a dumb question, but how is an ARQ user supposed to
figure
> > out the exact SPARQL syntax supported by a particular version of ARQ?
>
> Normally, ARQ implements whatever the W3C has made a recommendation.
> The language extensions need to be explicitly requests (parse with a
> different syntax), the default is strict SPARQL.
>
> ARQ does contain a strict SPARQL 1.0 parser separate from 1.1
> developments for complete compatibility.
>
> At the moment, SPARQL 1.1 Query is being finalised.  ARQ has been
> shipping with where ever the WG has got to at the time of publication.
> Most people seem to want that.
>
> The BINDING->VALUES change has been publicly signalled on the sparql dev
> list and the SPARQL comments list and is in the editors working draft.
>
> BINDING->VALUES and removal of {n} property paths are the only
> incompatible changes.  In fact, it's very rare for any incompatible
> change to be made after a working draft publication.
>
> The technical answer is to look in the editors working draft because the
> draft grammar and the ARQ grammar come from the same javacc file.
> For the grammar, ARQ isn't using some implementation of the grammar, it
> is using the grammar itself.  It's how it's produced and gets checked.
>
>
>    Andy
>    Co-editor SPARQL Query.
>
> >
> > Thanks,
> > Frank.
> >
>
>

Re: SPARQL version supported by ARQ

[Andy Seaborne <an...@apache.org>]

On 17/07/12 15:53, Frank Budinsky wrote:
>
>
> Hi,
>
> It seems that the latest ARQ drivers are not compatible with any W3C
> Working Draft, that is, the current SPARQL 1.1 working draft is version 5
> (January 2012) which is using "BINDINGS", not "VALUES", but ARQ seems to be
> implementing an as-yet-unpublished editor draft 99. Does anybody know what
> the target date is for having a published W3C working draft 6?

The semantics of property paths are different as well, although as they 
are now back to what SARQ originally had prior to the working group, not 
many people will have noticed.

The WG has resolved (today) to publish SPARQL Query as a Last Call 
Working Draft, subject to alignment with Turtle over the prefix grammar 
rule that arose overnight.

> Maybe this is a dumb question, but how is an ARQ user supposed to figure
> out the exact SPARQL syntax supported by a particular version of ARQ?

Normally, ARQ implements whatever the W3C has made a recommendation. 
The language extensions need to be explicitly requests (parse with a 
different syntax), the default is strict SPARQL.

ARQ does contain a strict SPARQL 1.0 parser separate from 1.1 
developments for complete compatibility.

At the moment, SPARQL 1.1 Query is being finalised.  ARQ has been 
shipping with where ever the WG has got to at the time of publication. 
Most people seem to want that.

The BINDING->VALUES change has been publicly signalled on the sparql dev 
list and the SPARQL comments list and is in the editors working draft.

BINDING->VALUES and removal of {n} property paths are the only 
incompatible changes.  In fact, it's very rare for any incompatible 
change to be made after a working draft publication.

The technical answer is to look in the editors working draft because the 
draft grammar and the ARQ grammar come from the same javacc file.
For the grammar, ARQ isn't using some implementation of the grammar, it 
is using the grammar itself.  It's how it's produced and gets checked.


	Andy
	Co-editor SPARQL Query.

>
> Thanks,
> Frank.
>