W3C

- DRAFT -

Data Access Working Group

31 May 2005

Attendees

Present
Kendall_Clark, Jeen_Broekstra, EricP, Yoshio, DanC, AndyS, DaveB, Kevin, SteveH, JanneS, Souri, JeenB, JosD, PatH
Regrets
Chair
DanC
Scribe
EricP

Contents


Date: 31 May 2005

<scribe> Scribe: EricP

<DaveB> oops

<JanneS> I'm trying to call in... should be there soon enough

Convene, take roll, review records and agenda http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0265.html

<DanC> RESOLVED to accept minutes http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0258.html

<kendall> more the merrier :>

<kendall> it's in last weeks minutes

<kendall> he already had to spell his name once :<

<kendall> (ah, so that's the magic, i haven't done my actions phrase: "such is life" -- now you tell us! :>)

<DanC> regrets Howard

<kendall> I sent regrets, but I hereby unsend them.

<scribe> ACTION: EricP to pair with SteveH on making the HTML test results page

<scribe> ACTION: DaveB to to propose source test to approve

<scribe> ACTION: DanC to follow up re optional test based on op:dateTime triple

<scribe> ACTION: SteveH to prepare test cases for publication as WG Note (no deadline/urgency)

<scribe> ACTION: AndyS to add the above graph test cases (analagous to valueTesting test cases) (don't expect quick delivery)

publishing results format, protocol

<scribe> ACTION: EricP: to publish http://www.w3.org/2001/sw/DataAccess/rf1/ 1.25 + editorial notes from DaveB + check from Kendall [DONE]

<scribe> ACTION: EricP to try out the "2 list of URIs" datasets design from http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0440.html [DONE]

<kendall> sorry for the noise

this is comical

<DanC> KC: tx IanJ for help with the protocol publication

<DaveB> more xmly-lists e.g. xml-dev

<DaveB> & copy the comment lists, so they replyall to comments

Editors are encouraged to announce their specs

<kendall> (god, freaking garbage pickup taking forever today)

issue fromUnionQuery

ericP has written up some note: http://www.w3.org/2005/05/05-algae-graphs/

<DanC> $Date: 2005/05/31 18:31:24 $

<DanC> EricP: can a service have named graphs that don't come from FROM NAMED? AndyS: yes.

<DanC> EricP: a service is never required to have something in a named graph that isn't in a background graph? DanC: right; service gets to pick which datasets to support

<DanC> EricP: likewise, a service is never required to have something in the background that isn't in a name graph? DanC: right, again.

<DaveB> hmm, tricky to do that in the QL since you can only give FROM once, versus FROM NAMED many times

SteveH: may need a test case because we'll give different answers on FROM and FROM NAMED.

DanC: but you won't be able to expose that in a single test.

<kendall> that's the same w/ the protocol spec

DanC: what happens if you don't support FROM?

Andy: no. nothing outside the spec is described.

<kendall> semantic aggravator, indeed

<afs> Jos asked for FROM <u1> <u2> <u3> to be a merge

ooo, me too!

<kendall> i wld support that as well, i think

<JosD> was actually FROM <u1> FROM <u2> FROM <u3>

<DanC> test case: dataset has <a>, <b>, <c>... query has FROM <d> ... hmm...

<kendall> I'd like to be able to say, in my implementation, that it won't execute a query against the specified RDF dataset.

EricP: how about expanding the test case to say <A> <B> and <C> are in the dataset and <D> is lying around reachable. now execute this query (with a FROM in it)

<kendall> hi pat

Souri: Do we have a way to say that some graphs are optional?

<patH> hi kendall. Sorry Im late.

DanC: I don't think we've considered it.

<DanC> (hmm... protocol should perhaps have a "good question, but I don't care to answer it" thingy)

<afs> How about EITHER the dataset is given outside the query OR it is completely specified by FROM/FROM NAMED.

KendallC: I would like, as an implementor, to say no to the client in a way that it can understand.
... distinguish between a broken impl and a "understood, but one execute".

DanC: been thinking about the mailing list search service. if you ask it for graphs outside of a constrained domain, you get a "darned if I know".

AndyS: I think we can define the errors without defining exactly when they occur.

<DanC> [10] IDefaultGraphClause I ::= I'FROM' SourceSelector

<DanC> [9] IDatasetClause I ::= I( DefaultGraphClause )? ( NamedGraphClause )*

AndyS: and there is never more than one IDatasetClause

<kendall> +1 to AFS's point

<DaveB> +1 from me for allowing multiple FROM URI and merging

<kendall> +q to ask about straw poll for a generic "J'refuse" response type

use mention issues on http://www.w3.org/2001/sw/DataAccess/ _

<afs> Will write an email to the list to write what I think I'm hearing

<DanC> PROPOSED: to address fromUnionQuery ala http://www.w3.org/2001/sw/DataAccess/rq23/ 1.363 plus multiple FROMs allowed

DanC: the servers get to say how much entailment they do

<afs> We do not allow merge into NAMED graphs for this reason.

<kendall> i yield to andy in the queue; his comment is relevant here.

Souri: would the user get an explanation of an inferred graph?

DanC: we had that as an issue, serviceDescriptions, but decided we wouldn't reach closure soon enough

<kendall> danc: you should take andy next; i'm out of synch

ba da bum

<DanC> PROPOSED: to address fromUnionQuery ala http://www.w3.org/2001/sw/DataAccess/rq23/ 1.363 plus multiple FROMs allowed

<kendall> where the graphs identified by multiple froms are RDF merged to form teh default graph

<DaveB> I see 1.367 now

<DanC> rather 1.367

<DanC> eric: not ready

<DanC> Steve: unhappy with the whole dataset thing

http://www.w3.org/TR/2005/WD-rdf-sparql-query-20050217/#queryDataset

<DanC> PatH: query can say FROM <a>, <b>, <c>; query is answered w.r.t. a graph that contains at least <a>, <b>, <c>

FROM <x> WHERE { GRAPH <y> { ?s ?p ?o } }

PathH: you are expected to read all of <x> and match it

http://www.w3.org/TR/2004/WD-rdf-sparql-query-20041012/#source

<DanC> 9 Querying the Origin of Statements of 12 Oct draft

<afs> See also: http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2004Nov/0020.html

<kendall> q

<kendall> err

<Zakim> kendall, you wanted to ask about straw poll for a generic "J'refuse" response type

q~ s/foo/bar/

<DaveB> I think swh meant http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0078

<SteveH> problems I found when implementing the default graph: http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0078.html

<DanC> "SPARQL: SOURCE is suboptimal" 0020

<SteveH> yes, thanks DaveB

<DanC> PatH: I expect to be more available for the next month

<afs> So - Steve - this about data maintenance effects?

<SteveH> afs, yes

<DanC> EricP: [...?]

<DanC> AFS: but how do you say "from both of them but no others"?

<kendall> "trust" is a total red herring IMO, which is neither here nor there. :>

<afs> GRAPH ?g {...} . FILTER ?g = x or y . GRAPH ?g { ...} ??

<Zakim> DanC, you wanted to explore how the 12 Oct draft is different

<afs> data maintenance would be about changing UC&R?

<patH> kendall, ignore the T-word, think only of being able to control the sources used for answering. Then it makes sense.

<DanC> EricP: yes, "dataset" is: an RDF graph where each triple is associated with zero or more URIs

<kendall> path: okay, i can do that. and that's *all* i care about. but others... :>

<Zakim> SteveH, you wanted to swap in 0078

<kendall> this seems a bit of an odd way to spend telcon time, if I may be so bold.

<patH> Jos, I notice that Sparql is underneath trust and inference in that picture.

<kendall> http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0078

<JosD> true Pat and was to point to necessity of proof to support trust

<patH> Ah, good point indeed, Jos.

<kendall> but there are *lots* of blockers, it seems to me

<DanC> (I had never thought that named graphs are rigid. I owe the editors a message about this)

PatH: it was the intent of Named Graphs that once a graph is named, it is rigid. you can sign it and whatever.
... But we should allow graphs that are created dynamically, say, from RSS feeds.
... for instance, I trust CNN and BBC but not FOX.

<Zakim> DanC, you wanted to propose taking SOURCE, GRAPH, FROM, FROM NAMED, and all that out, and going back to querying just one RDF graph, ala RDQL.

<Yoshio> difference is the reason for standardization, though

<kendall> WHAT?! :>

<DanC> I disagree, Yoshio

<kendall> carpet biting seems purely a FOX issue ;>

<patH> I see what Yoshio means. Theres a line between everyone all agreeing, and everyone feeling thats its not a research issue.

<kendall> +1 path & yoshio

<patH> If we already agree, theres no need for a standard.

<patH> :-)

<kendall> this isn't a research issue, IMO -- we simply don't have robust consensus.

look for predicated trust in http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0067

<patH> agree, kendall

<DaveB> there doesn't seem to me to be so much of a problem with GRAPH, just the FROM stuff - making the dataset

<kendall> part of the problem is that this design space is very unconstrained, either "naturally" or by UC&R or by our mutual implementation strategies.

<SteveH> kendall, thats certainkly true

<DanC> POLL OPTIONS: (0) take out GRAPH; query against one RDF graph punt SOURCE and fromUnionQuery issues (1) address fromUnionQuery ala http://www.w3.org/2001/sw/DataAccess/rq23/ 1.367 plus multiple FROMs allowed (2) use 12 Oct design

<kendall> yeah, and that makes it hard to reach consensus, since there are lots of different designs, most of them are internally coherent, but no one of them seems to cover everyone's common case.

<DanC> POLL OPTIONS: (0) take out GRAPH; query against one RDF graph punt SOURCE and fromUnionQuery issues (1) address fromUnionQuery ala http://www.w3.org/2001/sw/DataAccess/rq23/ 1.367 plus multiple FROMs allowed (1b) ala 1 but with no FROM/FROM NAMED in the QL, only in the protocol (2) use 12 Oct design

PatH: Is trust the primary reason for having this provenance text?

<Yoshio> will someone elaborate 12 Oct design on IRC, please?

<SteveH> Yoshio, theres a set of graphs, but no background graph

KendallC: i feel that not specifying provenance queries will threaten interop

<Yoshio> thanks, steven

<afs> The background graph is always the merge of the named graphs

<Yoshio> oops, Steeve

<SteveH> Yoshio, its Steve actually, but I'm not bothered :)

<afs> There is a background graph - WHERE { ?s ?p ?o } does have an answer

<Yoshio> sorry (^_^;)

<SteveH> afs, thats a point of view, but I see your point

<DanC> POLL OPTIONS: (0) take out GRAPH; query against one RDF graph punt SOURCE and fromUnionQuery issues (1) address fromUnionQuery ala http://www.w3.org/2001/sw/DataAccess/rq23/ 1.367 plus multiple FROMs allowed (1b) ala 1 but with no FROM/FROM NAMED in the QL, only in the protocol (2) use 12 Oct design (2b) ala PatH: bg graph always contains (vs is) the (er... something)

<afs> As editor, my POV is relevant

<SteveH> afs, true :)

PatH: Oct 12 has the background graph having all the named graphs

<Yoshio> oh, so there's no "no background graph" option?

<Yoshio> which I'd support

yoshio, could you explain your requirements?

<DanC> huh? (0) is pretty close, perhaps, Yoshio

<DanC> [[

<DanC> SELECT ?title

<DanC> WHERE { <http://example.org/book/book1> <http://purl.org/dc/elements/1.1/title> ?title }

<DanC> ]]

<DanC> SELECT ?title FROM <books.rdf> WHERE { ?book dc:title ?title }.

<DanC> SELECT ?title FROM <books.rdf> WHERE GRAPH ?g { ?book dc:title ?title }.

<kendall> I believe Yoshio wants the entirety of the RDF dataset to be explicitly specified in the query. (?)

<DanC> POLL OPTIONS: (0) take out GRAPH; query against one RDF graph punt SOURCE and fromUnionQuery issues (1) address fromUnionQuery ala http://www.w3.org/2001/sw/DataAccess/rq23/ 1.367 plus multiple FROMs allowed (1b) ala 1 but with no FROM/FROM NAMED in the QL, only in the protocol (2) use 12 Oct design (2b) ala PatH: bg graph always contains (vs is) the (er... something) (3) ala Yoshio, no implicit dataset; all data to be queried is specified in the query

<Yoshio> my position is to be as close to REST style, in a sense

i suspect that Yoshio's proposal is orthogonal

KendallC: 1

<DanC> DB: -0 +2

tx

<DanC> EP: +2 umm...

<DanC> Yoshio: +3

7 then

<DanC> AFS: [implicitly, 1]

<DanC> KW: +0

<DanC> ... but I see interest in others.

<DanC> SH: +2

<DanC> JS: +2

<DanC> Souri: just as an observer... this seems important; haven't finished my thinking

<DanC> JB: -0 +1 or +1b ... maybe 2

<DanC> JD: +1

<DanC> PH: +3, if I understand it...

<DanC> ... else 1

<kendall> problem is that yoshio's seems a bit underspecified; otherwise, I like 3, too.

<DanC> JD: I kinda like 3 also

<kendall> so, I'm: +1, maybe 3

<JanneS> (time to run for me, too)

<DanC> (bummer... time has run out)

<scribe> ACTION: DanC to send comment on non-use of http://www.w3.org/2001/XMLSchema-datatypes to schema comments list

<scribe> ACTION: EricP to propose a test re "x"^^mytype != "y"^^mytpe cf. http://www.w3.org/2001/sw/DataAccess/tests/#extendedType-eq-pass

<scribe> ACTION: DaveB to work with EricP to clarify valueTesting proposal

<scribe> ACTION: DanC to noodle on uri() and str() issue from email http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0239.html

<scribe> ACTION: DanC to rethink the use/mention issue w.r.t Tim's email http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0154.html

<scribe> ACTION: DanC to try splitting SparqlQuery interface into SparqlGraph SparqlBindings

<DanC> ADJOURN.

<patH> Yoshio, let me check your idea. Is it that there must be a single (URI or graph name) from which the query is evaluated, but that this might identify an arbitrarily complicated virtual graph, controlled by the answering service?

<Yoshio> well, there may be multiple specification

<patH> Ah, yes, OK, sorry.

<Yoshio> and about virtualness(?), yes

<patH> OK, good, that is what I thought you meant by the REST refernce. We need a notion of a 'graph resource' :-)

<Yoshio> in a nutshell, I want to hold an option of excluding unclear background graph

<patH> OK, I see the point.

<DanC> yoshio, you don't see the value of having a service like the w3c list archive search service, where FROM makes no sense?

<patH> The other side however is the need to allow an answering service which maintains a changing virtual graph.

<afs> Pat - while you are here ...

<Yoshio> dan, sorry, I don't see your point

<patH> DanC, why does FROM make no sense there? If that sedrvice has a URI, then why can that not be used as the source? Its not a GRAPH name, but so what?

<afs> I sent a round of questions (e.g. solution vs substitution terminology) - should I prioritise the questions?

<patH> afs, yes?

<patH> Oh, wait...

<DanC> suppose there's a SPARQL interface to the w3c list archive search service. Its dataset is what it is. It doesn't make sense to do FROM or FROM NAMED.

<patH> DacC, not FROM NAMED, agreed, but FROM seems to make sense to me.

<Yoshio> I agree with patH

<afs> (Sorry to nag)

<JosD> agree Pat, just being explicit about assumptions

<kendall> afs: i'd be interested in that

<patH> DanC, for example, suppose I query some other service and tell IT to use th ew3c archive. That makes sense, surely?

<patH> afc, not nagging but I seem to have missed the questions so am at a loss.

<Yoshio> but I don't want the instructin outside the query

<afs> Will resend.

<patH> afc, thanks.

Summary of Action Items

[NEW] ACTION: AndyS to add the above graph test cases (analagous to valueTesting test cases) (don't expect quick delivery)
[NEW] ACTION: DanC to follow up re optional test based on op:dateTime triple
[NEW] ACTION: DanC to noodle on uri() and str() issue from email http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0239.html
[NEW] ACTION: DanC to rethink the use/mention issue w.r.t Tim's email http://lists.w3.org/Archives/Public/public-rdf-dawg/2005AprJun/0154.html
[NEW] ACTION: DanC to send comment on non-use of http://www.w3.org/2001/XMLSchema-datatypes to schema comments list
[NEW] ACTION: DanC to try splitting SparqlQuery interface into SparqlGraph SparqlBindings
[NEW] ACTION: DaveB to to propose source test to approve
[NEW] ACTION: DaveB to work with EricP to clarify valueTesting proposal
[NEW] ACTION: EricP to pair with SteveH on making the HTML test results page
[NEW] ACTION: EricP to propose a test re "x"^^mytype != "y"^^mytpe cf. http://www.w3.org/2001/sw/DataAccess/tests/#extendedType-eq-pass
[NEW] ACTION: SteveH to prepare test cases for publication as WG Note (no deadline/urgency)
 
[DONE] ACTION: EricP to try out the "2 list of URIs" datasets design from http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0440.html
[DONE] ACTION: EricP: to publish http://www.w3.org/2001/sw/DataAccess/rf1/ 1.25 + editorial notes from DaveB + check from Kendall
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.126 (CVS log)
$Date: 2005/05/31 18:31:24 $