See also: IRC log
<kendallclark> Scribe: FredZ
<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0140.html
<AndyS> The minutes seem to have carried over DONE actions
PROPOSED: to accept Minutes from last meeting with corrections as a true record of the last meeting
APPROVED
<AndyS> At risk for next time: AndyS (need to be elsewhere from 15:00-15:30 local/UTC+1)
next meeting: probably 14:30Z (10:30 eastern US), scribe: Elias-scribe-for-life-T
<kendallclark> PROPOSED: to change the DAWG telecon time to 14:30 UTC, starting 20 June and from thence on
<kendallclark> RESOLVED
<kendallclark> ACTION: [CLOSED] AndyS to resend a proposal (sent) to solve equality testing over email [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action01]
ACTION: AndyS to resend a proposal to solve equality testing over email, ericP and patH to give eyes. [CONTINUES] [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action02]
ACTION: EliasT to submit mime-type registration for json-results. [CONTINUED] [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action03]
ACTION: EricP to propose text and tests to add {boolean < > [CONTINUED] [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action04]
ACTION: EricP to send mail about 3-part interpretation of optionals [DONE] 3 interpretations for OPTIONALs (and UNIONS)
ACTION: KendallC to tell Martin Soto that we're considering his comments [CONTINUED] [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action06]
ACTION: PatH to kibbitz with EricP about 3-part interpretation of optionals [WITHDRAWN] [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action07]
ACTION: AndyS: to draft of open-world = tests (unknown datatypes) [DONE] [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action08]
ACTION: LeeF to fix mimetype in json-reults [recorded in http://www.w3.org/2006/06/13-dawg-minutes.html#action09]
<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0094.html
<AndyS> http://www.w3.org/2005/xpath-functions/
-> http://www.w3.org/2005/xpath-functions/ XPath namespace document
<AndyS> Test case: op:add => http://www.w3.org/2005/xpath-functions#add ??
<kendallclark> Presumably EricP's fixes include: isIRIis sparql:isIRI
<kendallclark> isURI sparqlisURI
<kendallclark> since they are typo'd
<LeeF> could someone scratch out the current proposal, please?
<kendallclark> would be nice for the record too
note proposal
-> http://www.w3.org/mid/20060612203520.GA16448@w3.org 3 interpretations of OPTIONALs
<EliasT> http://www.w3.org/mid/20060612203520.GA16448@w3.org
<AndyS> It is also http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0158
<LeeF> I think there's an implicit "else" in there.
ericP: if variable is introduced in one OPTIONAL and then appears in a second OPTIONAL, then bindings from one do not constrain the other, they only enhance it
<EliasT> i.e. + :a cc:license "steal this book" .
ericP: this doesn't happen "normally" because OPTIONAL is a binary operator and text near OPTIONAL specifies [left-associativity] for OPTIPONALs without other triples in between
ericP: but adding more triples ends up with the factorial scenario
FredZ: One of the points from the Chile paper is that SPARQL doesn't explicitly consider precedence in main body of text - precedence is implicit in the grammar.
FredZ: <something else about disconnect between semantics and presentation of semantics in document>
ericP: think we should first decide how this SHOULD work before discussing how to present how it works
<AndyS> "Better spec" is bound by the length of the charter
<kendallclark> Yes. But we have to answer and satisfy reasonable, novel claims about the formal semantics of our language. That seems unavoidable.
<AndyS> defn of LEXICAL assumes hard nulls ?
<kendallclark> My org -- not speaking here as chair -- strongly wants a compositional semantics for SPARQL. Lexical order sensitivity is a bad thing in many ways, IMO.
<AndyS> FredZ: reorder of peers should not matter
<AndyS> FredZ: nesting of structures may indicate order
<kendallclark> And we'd really like to be able to optimize RDF queries.
ericP: does the spec need to deal with commutativity and associativity of operators and the equivalences that fall out therefrom?
FredZ: SQL spec does not
ericP: Would like to define eveyrthing in term of lexical order and let lexical indepdence fall out in some cases from logical properties of operators (but the spec says nothing about these clevernesses)
<EliasT> AndyS: Our users today would like queries such as: given a user, give me a name, else, give me an email, else, give me a nick. ie. {A OPT B OPT C OPT D} which depends on ordering.
<kendallclark> Does lexical ordering push us further away from compositional semantics?
<patH> Probably yes, Kendall.
<kendallclark> *gah*
<AndyS> If it does, it is only if a variable is mentioned multiple times.
<patH> Or, to put it Eric's way, compositionality is buried in the clever optimizations.
<kendallclark> I really dislike that intuitively, which is worth very little, but there you go.
<patH> Interesting that the nasty case (use same variable several places) is exactly the case that breaks simpl entailment for OWL.
<FredZ> ?x ?y ?z . OPTIONAL (...) ?a ?b ?c
<EliasT> nay
<FredZ> keep going
<kendallclark> scribe: EliasT
when?
<Zakim> ericP, you wanted to talk about xpath operators
right now?
<AndyS> I have an idea I'd like to try out on you
<kendallclark> PROPOSED: To adjourn
<kendallclark> RESOLVED :>
<ericP> ADJOURNED
<AndyS> OPT( { ?x ?y ?z } , ... ) AND ?a ?b ?c
<LeeF> FredZ, the first param to OPTIONAL is a mandatory triple
<LeeF> the second param is the optional pattern
<patH> see http://www.w3.org/2002/01/UsingZakim
<LeeF> A OPTIONAL { B } translates to OPT(A, B)
<FredZ> x? y? z? OPTIONAL {?A ?B ?C} ?p ?q ?r
<LeeF> this operationally means:
<LeeF> AND(OPT({?x ?y ?z}, {?A ?B ?C}), {?p ?q ?r})
<AndyS> a/b/c
<AndyS> (a/b)/c