Convene

<DanC> ScribeNick: SerT

<DanC> Scribe: Sergio Tessaris

Accept minutes

<DanC> minutes 26 Jan

<DanC> RESOLVED to approve minutes 26 Jan

<ericP> AndyS___, definitely W3C?

<DanC> next meeting 6 Feb?

<DanC> RESOLVED: to meet again 6 Feb. Lee to scribe.

<libby> 7 feb?

Feb 6 is on Monday

<DanC> RESOLVED: to meet again 7 Feb. Lee to scribe.

Toward updated protocol WD (and results format WD)

<DanC> new protocol WD

<kendallclark> I'd like to thank all the little people...

<DanC> ACTION: KC to work on edited changelog since 2005-09-14 LCWD [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action01]

<DanC> the [SRD] cite

<DanC> ACTION: KC to update protocol references to [SRD] and WSDL [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action02]

<DanC> new results format spec

<DanC> ACTION: EricP to arrange for announcement of results format spec [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action03]

Protocol for syntax errors.

See SPARQL: Error handling in Last Call Comment Status. The problem is whether a SPARQL server is required to issue a MalformedQuery fault message whenever it receives an (syntactically) illegal query. This choice affects conformant servers in two ways:

• servers must implement a full SPARQL parser;
• servers providing SPARQL extensions (affecting the syntax) would not be conformant.

Relevant part of the SPARQL Protocol for RDF is in Section 2.1.4 query Fault Messages:

When the value of the query type is not a legal sequence of characters in the language defined by the SPARQL grammar, the MalformedQuery fault message should be returned. According to the Fault Replaces Message Rule, if a WSDL fault is returned, including MalformedQuery, an Out Message must not be returned.

The decision is on whether the should be returned should be changed into must be returned

<DanC> cases: (1) a syntax error that returns QueryRequestRefused (2) a syntax error (i.e. extension) that returns an Out Message

<kendallclark> I'll do some scribing, then:

<AndyS___> I support case (2) - whether that is a SPARQL service, I don't know

<kendallclark> I prefer a design where a service can support a superset of the language and can return an Out Message even though the request is technically a syntax error.

<kendallclark> As for (1), I'd be happy to say that MalformedQuery fault has precedence over QueryRequestRefused; that is, when a service may return either, it must return MalformedQuery -- but I'm not sure how *this* preference interacts with my other preference.

option (1) requires (possibly) multiple server endpoints when a server implements SPARQL extensions (SPQRQL++)

<kendallclark> yes, and that's also an implementation burden on *clients* to be in some sense syntax checkers...

<DanC> AFS: (1) should be allowed, (2) not

<AndyS___> I also support protocol extensions :-)

<DanC> SH: like AFS

<DanC> EP: yes please (2)...

option (2) allows language extensions

<DanC> EP: yes, (1) is allowed

<ericP> EP again fuzzy on the ramifications of 1

<kendallclark> but what if it's "FRM <...>"

<ericP> SELECT * --THINK_REALLY_HARD WHERE { ?s ?p ?o }

<kendallclark> how about: <strong><strong>should</strong></strong>? :>

<LeeF> works for me !

<AndyS___> "should" means "implementers must think and have good reason not to follow the text"

<kendallclark> So how about this proposal for fixing the spec:

<kendallclark> 1. Leave the language in the spec alone.

<kendallclark> 2. Add a bit about deploying a separate interface if you want to deploy sparql++

<kendallclark> (We only want to extend the protocol, which isn't really at issue here. That's clearly a separate WSDL interface. Done.)

<kendallclark> I don't care about the label on the box, Eric. I just said that. :>

<kendallclark> hmm, +1 to pat's suggestion. it's complex-er, but I think it's reasonable.

<LeeF> All talk of SPARQL Extended interfaces stink of lack of interoperability to me

<kendallclark> this ALL stinks. :>

<LeeF> Yes

<patH> gentlemen, please...

<DanC> I'd like: ACTION KC: add an example of a "SPARQL++" query in the spec, and think about a separate interface

<AndyS___> I have users using http://host/x?query=...&stylesheet=...

<LeeF> Are there syntax differences for OWL entailment?

<SerT> I don't think so

<DanC> ACTION: EricP to provide a "SPARQL++" example query in the accessingCollections neighborhood [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action04]

<DanC> ACTION: KC to add an example of a "SPARQL++" query in the spec, and think about a separate interface [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action05]

<ericP> I note that I am advocating motion towards a slippery slope (and apologize).

<LeeF> Related, but orthogonal.

<ericP> patH, could you give me a notional SPARQL query that would ask for results of OWL entailment (and is not a currently valid SPARQL query)?

<DanC> Please have opinions formed and on the WG list by agenda-building time on Mon 6 Jan

<patH> Eric, I can make one up.

<LeeF> Mon 6 Feb

<DanC> Please have opinions formed and on the WG list by agenda-building time on Mon 6 Feb. [tx]

comments related to rdfSemantics, owlDisjunction

<kendallclark> hmm, eric, FWIW, i'd like 2: one from pat re: owl, one from you re: the list functionality stufff you were talking about earlier re: annotea

<DanC> Draft response to: Re: major technical: blank nodes

<kendallclark> reasonable

<kendallclark> ?

<DanC> Dan's [OK?] msg

Dan's answer is fine

<LeeF> link to PFPS original message?

<DanC> http://www.w3.org/2001/sw/DataAccess/issues#rdfSemantics

<ericP> kendallclark, re 2 examples, yup that was my plan (well, at least those two)

<LeeF> There's also a postponed issue there, isn't there?

<DanC> partial reply to Barstow: http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2005Sep/0008.html

<LeeF> http://www.w3.org/2001/sw/DataAccess/issues#accessingCollections

<kendallclark> (I was just wishing this morning that we'd gone with Versa for more of our syntax. ;>)

<DanC> ACTION: DanC to complete response to Barstow [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action06]

<scribe> ACTION: patH to respond to PFPS comments [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action07]

<DanC> ACTION: AndyS to draft reply to Wood [recorded in http://www.w3.org/2006/01/31-dawg-minutes.html#action08]

<DanC> AndyS: I'm also working on Fred's editorial stuff.

<DanC> ADJOURN

<LeeF> thanks, everyone.