See also: IRC log
<DanC> scribe: EliasT
<DanC> 16 Aug 2005
<DanC> RESOLVED: to accept minutes of 16 Aug
<AndyS> (thanks Elias)
<kendall> I'm only available till about 11:45 my time (68 minutes from now)
<DanC> roger, kendall
Next meeting scheduled for Sept 6, 2005
Kendall will be the scribe.
<scribe> done: Kendall.
KendallC, this is something we need to fix. It's a mistake on the protocol.
KendallC, AndyS proposes it's query.
<kendall> Proposed: to change "sparql-query" to "query"
<DanC> editor is so advised.
<DanC> (or... no advice to the contrary)
KendallC: WSDL lets you redefine the parameter names using a new binding. If we don't allow that, we should expect havoc. We should say somewhere in the spec, you should these bindings if you want to implement a SPARQL service. Although, we have not talked about whether HTTP is the only concrete interface or whether SOAP is also required.
DanC: I'll be happy with a note that says you can change the parameters in your own binding, but it will reduce interoperability.
KendallC, link to email?
<DanC> Subject: Re: agenda: RDF Data Access 30Aug
<DanC> Date: Mon, 29 Aug 2005 12:57:31 -0400 (11:57 CDT)
<DanC> To be a compliant SPARQL Protocol service, a service MUST support the
<DanC> SparqlQuery interface, and it MUST support either the HTTP bindings or the
<DanC> SOAP bindings or both the HTTP and SOAP bindings as described in
<DanC> <sparql-protocol-query.wsdl>. A SPARQL Protocol service MAY support other
<DanC> interfaces as well.
<kendall> the term "SPARQL Protocol" already exists in the document...FWIW
<DanC> (now I forgot what question I was going to ask... oh yeah... "service"...)
<DanC> PROPOSED: that SPARQL services that export HTTP bindings must export the HTTP binding in this SPARQL protocol spec; likewise for SOAP
<DanC> so RESOLVED. Action KC
<kendall> i hate to burn so much time on minutiae, but thx all anyway...
KendallC, the output serialization is too strict in WSDL 2.0. We might need to send an email to their list.
KendallC: We would like WSDL 2.0
to allow us multiple output serialization formats for one
... We either return text or we split our operations to support multiple mime formats.
<scribe> ACTION: EliasT to draft test for WSDL 2.0 mime type restriciton in output serializatin. [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action01]
<DanC> options I see (a) ask WSDL WG to fix this limitation on returned mime type, wait for the fix (b) change our design to meet the constraint, (c) reconsider the requirement on WSDL
<DanC> a 4ish b 0 c 0
<DanC> (Andy seems to be exploring b...)
<DanC> PROPOSED: to weaken return type to application/xml , and note that this is due to a limitation to WSDL that we hope gets relaxed
AndyS: you can't have a web service that's officially a web service that returns non-XML data.
<DanC> PROPOSED: to leave out the whttp:outputSerialization param, note the conflict with the current WSDL spec, and take the risk that we'll have to come back if they say no
<scribe> ACTION: KendallC to add editorial note to protocol spec showing that our WSDL is not kosher and it depends on their final decision. [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action02]
<DanC> ?who: the spec currently requires RDF/XML in 2.a.3
<kendall> "equivalent serialization" language was there but removed because it's a bit loose
<DanC> say "... an RDF graph[RDF concepts] serialized for example in RDF/XML"
<LeeF> Kendall, do you happen to have a pointer to where the WSDL 2.0 draft(s) discuss outputSerialization?
<kendall> Lee: it's in their Part Two spec, the HTTP adjuncts stuff... I think section 6.3ish
<kendall> sad, but true, dan...
<AndyS> Maybe going back to "equivalent serialization" is the best of the choices.
<LeeF> If I'm reading http://www.w3.org/2002/ws/desc/5/lc-issues/type_sort.html#LC304 correctly, the WSDL 2.0 folks haven't yet formally defined what can (and can't) be a "IANA media type token" for the outputSerialization parameter - I think we should make our value for this attribute: */* :)
<DanC> PROPOSED: to relax the query-result type to allow "equivalent serialization" [salt to taste] and leave out the whttp:outputSerialization param, note the conflict with the current WSDL spec, and take the risk that we'll have to come back if they say no
<kendall> LeeF: :>
<AndyS> Prose would be fine
<DanC> so RESOLVED.
<scribe> ACTION: KendallC, to relax the query-result type to allow "equivalent serialization" [salt to taste] and leave out the whttp:outputSerialization param, note the conflict with the current WSDL spec, and take the risk that we'll have to come back if they say no [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action03]
KendallC, regarding working with WSDL WG on moving style. This is about allowing inputSerialization to allow www-form-urlencoded for POST binding. However, WSDL spec says you can't use URI style if using form url-encoded.
<DanC> request for ACTION ?who: assemble materials for a test for POST with application/x-www-urlencoded
<DanC> already there in the spec. k. Very long SELECT query using POST binding
<DanC> KC's ACTION on style continues.
<kendall> Elias: k. very long thingie has an error, actually... the http request line should be: POST /sparql/ HTTP/1.1... FWIW
<DanC> EricP's review http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JulSep/0304.html
kendall, got it.
<kendall> I was imagining I'm better at all of this, but it *hurts*...
<DanC> KC intends to make utf-8 explicit in the coures of doing eric's comments
<kendall> but of course DESCRIBE is under-spec'd, intentionally
<DanC> ah... ericp notes the "Accept: application/sparql-results+xml" in d. ASK with simple RDF dataset and notes this suggests that the client is peeking into the query.
<DanC> ACTION: KC to make conneg explicit in c. CONSTRUCT with simple RDF dataset and take accept: out elsewhere [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action04]
<JanneS> I have to go for tonight.. hoping to get LC version out next week.
<DanC> done... ish... DanC to extract and machine-check examples
<DanC> [right?] ACTION EliasT: establish consistency between protocol examples and tests
<DanC> ACTION: LeeF to draft WSDL 1.1 for SPARQL thingy with AndyS and Elias ETA 9 sep [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action05]
<DanC> ACTION: DanC to ask WSDL WG to review WSDL 1.1 and WSDL 2 SPARQL protocol stuff, once both are available [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action06]
<DanC> Elias is hereby appointed protocol-test czar/maintainer, with ericp's help
<DanC> ACTION: ericP to send [OK?] message to Bjoern. [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action07]
<DanC> PROPOSED: that 1.474 of http://www.w3.org/2001/sw/DataAccess/rq23/ addresses issues#queryMimeType
<AndyS> $Revision: 1.1 $ of $Date: 2005/08/30 16:01:34 $
<DanC> so RESOLVED.
<DanC> ACTION: EricP to change contact email to dawg-comments and call for review in IETF [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action08]
<DanC> ACTION: EricP to respond to http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2004Jul/0001.html [recorded in http://www.w3.org/2005/08/30-dawg-minutes.html#action09]