SV_MEETING_TITLE -- 28 Feb 2005

<Yoshio> hi all

<DanC_> Amy? I don't see OlympiaC in http://www.w3.org/1998/12/bridge/dialout.php3

<amy> hi

<amy> C, yes, ok good

<amy> np

<amy> bye

Convene, take roll, review records and agenda http://www.w3.org/2001/sw/DataAccess/ftf5-bos.html#agenda

we're convened

participants:

* Dan Connolly <connolly@w3.org> (W3C/MIT) : attending monday, tuesday

* Yoshio Fukushige <fuku@w3.org> (Matsushita Electric Industrial Co., Ltd. (MEI)) : attending monday, tuesday

* Hiroyuki Sato <sato.hiroyuki@lab.ntt.co.jp> (Nippon Telegraph & Telephone Corp. (NTT)) : attending monday, tuesday

* Andy Seaborne <andy.seaborne@hp.com> (Hewlett Packard Company) : attending monday, tuesday

* Stephen Harris <swh@ecs.soton.ac.uk> (Southampton, University of) : attending monday, tuesday

* Kendall Clark <kendall@monkeyfist.com> (Maryland Information and Network Dynamics Lab at the University of Maryland) : attending monday, tuesday

* Eric Prud'hommeaux <eric@w3.org> (W3C/MIT) : attending monday, tuesday

* Jos De Roo <jos.deroo@agfa.com> (Agfa-Gevaert N. V.) : attending monday, tuesday

Plus:

some observers:

<Yoshio> regrets for 8th meeting

<DanC_> ACTION: DaveB to consider use of xsi:dataType ala comment from Steer [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action01]

<tlr> Thomas Roessler

Ivan Herman, Jacek K.,

<DaveB> I can dialin for a bit. But that action continues...

hi dave!

<DaveB> hi kendall

observer: Danny Weitzner

issue: nestedOptionals

[what's the latest re: nestedOptionals? latest QL draft 5.5. Been implemented and folks seem happy with it.]

SteveH has same objections. "It's existence is unacceptable."

DaveB has another proposal involving a side condition on the grammar.

DaveB suggested making it not possible to *write* nested optionals, even though they'd still be possible semantically.

SteveH fine with the DaveB proposal to "hide" nested optionals by disallowing them syntactically.

debate about the relative merits of diff implementation strategies to implement nested optionals

how much more difficult are these queries to write if you can't write nested optionals?

<ericP> from [[

<ericP> OPTIONAL { ( ?x vcard:N ?vc )

<ericP> ( ?vc vcard:Given ?gname )

<ericP> OPTIONAL ( ?vc vcard:Family ?fname ) }

<ericP> ]]

Hmm, so we seem to agree that to write the equivalent query, you'll have to repeat some graph patterns

<ericP> to [[

<ericP> OPTIONAL { ( ?x vcard:N ?vc )

<ericP> ( ?vc vcard:Given ?gname ) }

<ericP> OPTIONAL { ( ?x vcard:N ?vc )

<ericP> ( ?vc vcard:Given ?gname )

<ericP> ( ?vc vcard:Family ?fname ) }

<ericP> ]]

<Yoshio> is it that whether lack of appearence of B in the result means the failure of B or not?

Jos suggests that in his system implementing nested optional (or just optional?) is difficult.

<SteveH> just optional

debate about the value or advisability of making open or closed world assumption

i think there are unshared views about the implications of OWA

eric and jos discuss altenate implementation strategies (translating sparql into n3, then filtering results... or something)

Jos:

A optional B has 2 interpretations:

1. A & (B v True)

which creates unhelpful solutions

2. A & (B v ~B)

which, if i follow, implies some closed world assumption, but works

for jos

<Yoshio> so absence of B in the result means, in the 2nd interpretation, the failure of B?

danc suggests making the parts of QL that involve CWAs more explicit

<scribe> ACTION: AndyS to clarify 5.4 w/r/t closed world assumption [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action02]

back to *nested* optionals

<Yoshio> my concern is that the restriction on the use of the variable names is complicated...

<Yoshio> I have to write

<Yoshio> SELECT ?a ?b ?c

<Yoshio> WHERE OPTIONAL (?a foo ?b)

<Yoshio> OPTIONAL (?a bar ?c)

<Yoshio> enen when all I want to know is ?a and the other sole thing

<SteveH> SELECT ?a ?b OPTIONAL (?a ?p ?b) AND ?p=<foo> || ?p=<bar>

<AndyS> Drop OPTIONAL (they were only there to structure the first appraoch)

<DanC_> leaving... SELECT ?a ?b WHERE (?a ?p ?b) AND ?p=<foo> || ?p=<bar>

<Yoshio> hmm,,, I got inclined to draw my concern back... I can't tell which OPTIONAL is met without distinguishing ?b and ?c

<Yoshio> there should be many situations where I wnat to konw which

<scribe> ACTION: danc resume discussion of nestedOptionals [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action03]

<DanC_> inconclusive data

<DanC_> how many to www2005? -JD +AS ~KC +YK +HS +EP +SH -DC

<DanC_> inconclusive data re amsterdam, southampton, florida, bristol offers

<patH> Hey, are we on the phone? What number?

<DanC_> Pat, where should we meet next? 1/2 ;-)

<Yoshio> WWW2005 is 10-14 May

<DanC_> www2005 ~-PH

<Yoshio> oh Dan is NOT coming... we should give up the hosting idea then...

<DanC_> (then resume tomorrow with other stuff)

f2f6 proposals

A one day meeting before (or after?) WWW05 got some interest.

test suite maintenance

<DanC_> SELECT ?PG WHERE ?PG =~ "http://www.w3.org/". # pls add this test case

http://www.w3.org/2001/sw/DataAccess/tests/

<DanC_> SELECT ?PG WHERE ?PG dc:creator ?anybody AND ?PG =~ "http://www.w3.org/". # pls add this test case

<DanC_> should NOT match { <http://www.w3.org/"> dc:creator "anybody" }

steve's action hereabouts continued

(daveb's action re: xsi:datatype also continued earlier, in case that didn't get scribed)

<AndyS> Ivan - http://www.w3.org/2001/sw/DataAccess/tests/data/

<Yoshio> how about (?u stringmatches: "http://www.w3.org/*) (?PG log;uri ?u)

<scribe> ACTION: [CONTINUED] SteveH to to propose tests for str() and =~) [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action04]

<scribe> ACTION: SteveH buy Andy a beer! [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action05]

<AndyS> GRAPH tests: http://www.w3.org/2001/sw/DataAccess/tests/data/source-named/untrusted-graph*.rq

has PatH swapped in the new named graph stuff?

<AndyS> http://www.w3.org/2001/sw/DataAccess/rq23/#rdfDataset

(keyword renaming... -sigh-)

<DanC_> (PatH finds the relevant tests, with andy's help)

DanC withdraws suggesting formal defintions for SOURCE

<DanC_> (break 'till 3:40p Boston)

<ericP> ping pat

<patH> hi guys. We back?

<ericP> yup

<ericP> can you hear dan well (enough)?

<patH> just about.

<ericP> [mutterings while looking for a particuarl test]

<AndyS> http://www.w3.org/2001/sw/DataAccess/tests/data/source-named/manifest.n3

<timbl> "requires a ton of my tools to build it"

<ericP> ACTION: EricP to pair with SteveH on making the HTML test results page [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action06]

valueTesting

<ericP> walking through http://www.w3.org/2001/sw/DataAccess/tests/data/ExprEquals/query-eq-1.rq

<ericP> evaluated against http://www.w3.org/2001/sw/DataAccess/tests/data/ExprEquals/query-eq-1.rq

<timbl> http://www.w3.org/2001/sw/DataAccess/tests/data/ExprEquals

<timbl> is the directory

<ericP> evaluated against http://www.w3.org/2001/sw/DataAccess/tests/data/ExprEquals/data-eq.ttl

<ericP> ivan: the parsers i use read "1" as 1

<ericP> [kendall warns that RDF/XML treatment of datatypes is broken]

<ericP> walking through http://www.w3.org/2001/sw/DataAccess/tests/data/ExprEquals/query-eq2-1.rq

<patH> Pat protests that it may be ugly, but its not broken.

i warned that *rdflib* has no real typed literal support -- v. different issue

<ericP> equivlence test that can bind to one of several operators

<ericP> (answering yoshio)

<patH> Echo Yoshio. We seem to be using '=' for several differnt things.

<ericP> see

<DanC_> SELECT ... AND ?p = dc1:title || ?p = dc2:title

<Yoshio> is it different from "=="?

<DanC_> also, users expect 1 = 1.0

<ericP> XPath overloading of the operators defined in http://www.w3.org/TR/xquery/#mapping

<ericP> tim: re-use the investment in value equiv test cases by copying them to graph test cases

<ericP> eg

<ericP> (?x1 :p ?v1 )

<ericP> (?x2 :p ?v2 )

<ericP> AND ?v1 = ?v2

<ericP> goes to

<ericP> (?x1 :p ?v)

<patH> +q about entailing.

<ericP> (?x2 :p ?v)

<ericP> .

<ericP> ACTION: Andy to add the above graph test cases [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action07]

<Zakim> Yoshio, you wanted to ask if it is not "==" we are talikng about

<DanC_> SELECT ... WHERE (?exp :results ?x. ?exp2 :results ?y) AND ?x != ?y #@@@

<ericP> DanC: we don't have a spec that exactly matches the editor's understanding (cf. the == and eq in section 11)

<AndyS> SELECT ... WHERE (?exp :results ?x) (?exp2 :results ?y) AND ?x != ?y

<DanC_> against { exp1 result "1/1"^^rational. exp2 result "2/2"^^rational)

<DanC_> ericp: the != will fail.

<DanC_> timbl: good

<DanC_> andys: or is it fn:not(?x = ?y) ?

<ericP> http://unagi/2001/sw/DataAccess/rq23/#StandardOperations

<ericP> [[ Expressions generating errors cause the solution to be rejected (not to match). ]]

<DanC_> Andy convinces DanC (and EricP? and others?) that the != will turn into not( ... = ...) and yield true

<ericP> SteveH: propose "when you call the = op on datatypes that you don't understand, it fails"

<DanC_> "abc"^^integer

<ericP> ammend SteveH's proposal with first checking if the two are literally the same and the same datatype

<DanC_> REQUEST FOR TESTCASE: input data has "abc"^^integer

<DanC_> "10"^^byte = "10"^^integer?

<DanC_> "10"^^long = "10"^^integer

<DanC_> ?

<DanC_> "10"^^nonnegative = "10"^^integer?

<DanC_> REQUEST FOR TESTCASE

<DanC_> AndyS: for example, OWL uses nonnegativeinteger for cardianlity constraints

<DanC_> EricP seems convinced

<ericP> sadly so

<DanC_> ACTION: EricP to propose to close valueTesting (bonus points for test cases, to EricP or others) [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action08]

<DanC_> http://www.w3.org/TR/rdf-mt/#DTYPEINTERP

XSCH review

<DanC_> http://www.w3.org/2001/sw/BestPractices/XSCH/xsch-sw-20050127/

<ericP> jacek: i would expect a W3C RDF query language to support all the XML schema datatypes and all of OWL

<ericP> patH: what does it mean to "support"?

<ericP> DanC: this sounds like requirements. take up in hallway discussion?

<SteveH> DaveB, observer, from DERI

<DaveB> I laughed when I read supporting all XSD and all owl.

<SteveH> yes, "laughed" ;)

<patH> Hey, Im listening to a lovely exposition here.

<DanC_> ericp's action is withdrawn

SPARQL punctuation syntax (turtle, N3, ...)

<DanC_> DanC: since this wasn't on the agenda at T-2 weeks, I won't put any questions formally

<DanC_> EricP: but I don't argue to use { ... } as a term

<patH> the phone is dropping out very freqently. Can yall turn up the mike?

<ivan> ivan-

<DanC_> see examples from http://www.w3.org/2001/sw/DataAccess/ftf4.html#item18

<ericP> turtle+variabels = turtle (with ',' and ';') + vars (eg ?x)

<patH> Im happy with the N3 style as descrbed there. BUt Im also happy with turtle style.

well, we've liberally changed keywords at just about every f2f, including helsinki, and that was only a month ago. :>

<patH> +kendall

<patH> Don't we all, Zakim

hey! :>

<DanC_> timbl: for example S

<patH> example S??

<DanC_> timbl: for example SELECT ?who WHERE ?who :brother [ in Army ], :mother [ in Navy].

a bnode in a graph pattern is equivalent to to an unused variable in a select or construct clause

<DanC_> andy: is that like ... ?who :brother ?notUsedElsewhere1. ?notUsedElsewhere1 in Army ...

<DanC_> yes

<DanC_> Jos: in turtle, one can write :bob :brothers ( :pete :sam).

<ericP> turtle+variabels = turtle (with ',' and ';' and '()') + vars (eg ?x)

<DanC_> ... which suggests SELECT ?who WHERE ?who :borthers (:pete :sam).

<timbl> ; , . () [] ?x

<patH> that's a pretty god transcription of what it sounds like on the phone, Tim

<ericP> Yoshio, widely distributed, I believe the only implementations are from AndyS, DaveB and myself.

<DanC_> SteveH: my users are mostly folks dumping databases into RDF/XML

<ericP> and perhaps Ivan

<DanC_> SteveH: I support turtle input and noone uses it

<patH> DO we have to choose one?

we don't if/when we have an xml exchange syntax for sparql

just map any old surface/user syntax into it, and send *that* to the server

<patH> That makes sense, indeed. XML to the rescue once again?

well, you can then do neato schema web services stuff. that's gonna help adoption in areas where SW has little presently.

<AndyS> We need at least one common syntax for encoding into a protocol for web use

<patH> Maybe the issue is how the results get sent back. We need an XML for that too.

we have one for that, pat :>

<patH> Just testing.

<patH> I have to leave shortly. Don't do anything serious while Im gone.

never!

<timbl> (([A-Z_a-z\u00c0-\u00d6\u00d8-\u00f6\u00f8-\u02ff\u0370-\u037d\u037f-\u1fff\u200c-\u200d\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd\U00010000-\U000effff][\\-0-9A-Z_a-z\u00b7\u00c0-\u00d6\u00d8-\u00f6\u00f8-\u037d\u037f-\u1fff\u200c-\u200d\u203f-\u2040\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd\U00010000-\U000effff]*)?:)?[A-Z_a-z\u00c0-\u00d6\u00d8-\u00f6\u00f8-\u02ff\u0370-\u037d\u037f-\u1fff\u200c-\u200d\u2070-\u218f\u2c

<timbl> qname in n3.n3

<AndyS> ?? Dot is UTF-8 002E

<Yoshio> doesn't "." in qname in N3 mean a path expression?

<patH> is that really ((...)?:)?... ?? SO the : is optional?

<DanC_> WHERE { GRAPH ?g { ?g dc:source myPartOfUriSpace:foo.rdf } } # sh, ar like this

<patH> Bye.

<timbl> .BadSyntax: Line 5 of <file:/devel/WWW/2000/10/swap/test/syntax/trailing-dot-in-qname.n3>: Bad syntax (expected directive or statement) at ^ in:

<timbl> "...foo: <http://example.com/>.

<timbl> foo:fred foo:loves foo:sushi. ^.

<timbl> Bad syntax (EOF found in middle of path syntax) at ^ in:

<timbl> "# DAWG meeting 2005-2-28

<timbl> #

<timbl> @prefix foo: <http://example.com/>.

<timbl> foo:fred foo:loves foo:sushi.food .

<timbl> ^..."

<timbl> WHERE joe age [ > 5 ]

<timbl> WHERE { joe age [ > 5 ] }

<DanC_> ACTION: EricP to propose results of N3 syntax discussion to DAWG [recorded in http://www.w3.org/2005/02/28-dawg-minutes#action09]

<DanC_> (based on mail from danc)

- DRAFT -

SV_MEETING_TITLE

28 Feb 2005

Attendees

Contents

Convene, take roll, review records and agenda http://www.w3.org/2001/sw/DataAccess/ftf5-bos.html#agenda

issue: nestedOptionals

f2f6 proposals

test suite maintenance

valueTesting

XSCH review

SPARQL punctuation syntax (turtle, N3, ...)

Summary of Action Items

Scribe.perl diagnostic output