See also: IRC log
<kendallclark> ??P2 is AndyS
<kendallclark> Scribe: LeeF
<EliasT> Scribe: EliasT
Title: DAWG Weekly Meeting
yes
<kendallclark> http://www.w3.org/2006/06/27-dawg-minutes
AndyS: seconds
Next Meeting: Aug 08 2006
<LeeF> ACTION: LeeF to make a test case out of the nested GRAPH scenario from DAWG email list [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action01]
<ericP> ACTION: EricP to turn FredZ's test case sketches into tests. http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0171.html CONTINUED [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action02]
yes
<ericP> ACTION: DanC to review PFPS's comments for more test cases CONTINUED [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action03]
<LeeF> he doesn't have a speaker phone
<ericP> ACTION: EliasT to turn C2.38 in PFPS's message into a test case DONE [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action04]
<ericP> action -4
<ericP> ACTION: KendallClark to reopen punctuationSyntax to take up commas in SELECT clause.CONTINUED [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action05]
<ericP> ACTION: take up new issue concatenated nestedOptional CONTINUED [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action06]
<AndyS> http://www.w3.org/2001/sw/DataAccess/issues
FredZ: how does someone raise an issue to the issues list
kendallclark: It's mostly up to
the chair. I have re-opened two issues already in the
past.
... some issues get raised to be postponed for future
iterations of the spec.
... issues can be sent to the mailing list and turned into
testcases.
kendallclark: Could AndyS give us major structural changes to the fork to rq23?
<kendallclark> http://www.w3.org/2001/sw/DataAccess/rq23/rq24.html
<kendallclark> in the agenda too :>
AndyS: It's mostly editorial. It
has an explicit section to making simple queries as
introductional material. Later a section on syntax issues.
Section 4 definitions and then BGP. A little bit of
renumbering. Mostly to give a better idea of what's tutorialish
vs. specification.
... I collapse the three data sections but mostly for
balance.
... Grammar section was reorganized based on an outstanding
message (codepoint escape sequences)
<kendallclark> fwiw, the formal definition extraction stuff is broken :>
AndyS: I haven't touched the definitions and/or the grammer.
kendallclark: Summarizing. Changes were mostly editorial to turn it into more specificational rather than tutorial.
<kendallclark> hmm, no, tha'ts not what I meant to say :)
<kendallclark> or what Andy said
<kendallclark> he more clearly distinguishes between exemplary and specificational stuff
AndyS: ericP had suggested to making a single graph of the example in section, but I haven't gotten around to it.
<kendallclark> :>
fred: ericP suggested fred should reword his comments against rq24.
kendallclark: no opinion
... I would like us to move to more specificational rather than
exemplary. Maybe even a separate document.
... I believe the audience we are targeting doesn't need the
exemplary version. They probably need a more formal one.
... If rq24 distinguishes clearer between exemplary and
specificational, I'd like it better.
... Would you (LeeF) would like to write our primer?
<AndyS> we do have the definitions extraction script - we could make that more explicit.
AndyS: We could work on making
the extracted document better from definitions. We have had
feedback that the audience likes both: definitions and
discursive text.
... we have the ability to extract a definitions only
document.
kendallclark: would we need a more explict structure: like primer section, etc.
AndyS: keeping two documents in sync (primer/spec) is relatively hard.
patH: AndyS's comments are backed up by real examples such as OWL.
<ericP> i'm interesting in hearing PatH's specific experiences in OWL
AndyS: two sets of people read each document and inconsistencies never get pointed out.
<LeeF> Do the definitions give a fully coherent formal specification of SPARQL? I think probably not...
<kendallclark> We have feedback that they don't, I believe.
patH: SPARQL spec style is a bit weird. We need both a good amount of text and definitions.
fred: I'd like more distinction between definitions and examples. I would like to suggest a new section on transformation from grammar to mathematical engine.
kendallclark: Andy, do you think we reached the point where parts of the document (via CSS) are normative and other not?
AndyS: My current suggestion would be to extract a def-only document.
kendallclark: Could the script
also extract the normative part of the spec as part of the
separate document?
... besides the definitions.
AndyS: I think we are probably
very close to that.
... I will go through the document trying to markup what text
is normative and should go along with the definitions.
patH: A compromise could be to add a header that says "Example" used over and over again
<kendallclark> ACTION: LeeF to To review rq24. [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action07]
<ericP> LeeF, if you are thinking about how to mechanically (perhaps XSLT) separate rq24 into tutorial and normative, i can geek with you a bit
<LeeF> thanks, ericP -- will probably review it first and note structural feedbacks at a non-markup level first, and then think about mechanical separation -- at which point I will be sure to bug you relentlessly :-)
kendallclark: we need to have some meetings to get rid of old tests
LeeF: I've been starting to jot down a whole bunch of tests that are either old or unncessary. I'll contribute that as needed.
LeeF: the test is correct and the spec needs changing, but the spec change is substantial and needs review from the WG.
<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0024.html
LeeF: it should fail
ericP: By fail do you mean fail or skip that solution?
fred: filter will not pass anything that errors.
AndyS: I thought we decided that if the impl added understanding of new types it could answer correctly.
ericP: The problem we have is
that = for RDFTerms is the same as for strings.
... sometimes you can pass the test, but with new information
you could fail the test.
<kendallclark> Open world tests for datatypes from Andy: http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0104.html
<kendallclark> yes, it is :)
<scribe> Chair: kendallclark
<AndyS> 90mins?
sorry
<patH> you wish, elias
<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0104.html
<ericP> last time we discussed termEquals
<AndyS> I'm happy to turn a description into a concrete test in the test suite.
scribe: the new = syntax from AndyS.
<kendallclark> ||
<AndyS> || is a special form.
what was the outcome of FILTER("foo" = "foo"^^xsd:string)?
<AndyS> True. Same value.
<kendallclark> ACTION: PatH to To write another open world value test to add to Andy's. [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action08]
<kendallclark> ACTION: EricP to respond to PatH's new test with supplemental info [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action09]
<AndyS> FILTER (datatype(?x)=xsd:string)
<ericP> ACTION: EricP to respond to PatH's new test with a proof of whether it's monotonic [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action10]
<ericP> ACTION: EricP to respond to PatH's new test with a proof of whether it's monotonic to extended datatype support [recorded in http://www.w3.org/2006/08/01-dawg-minutes.html#action11]
<ericP> action -9
<ericP> action -10
RRSAgen, please draft minutes