RDF Data Access Weekly meeting

Convene, take roll, approve minutes

<scribe> Scribe: DanC

-> minutes 22 Aug

<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0188.html

<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0187.html

RESOLUTION: to accept 22 Aug minutes as a true record.

<LeeF> next Tues. works for me

PROPOSED: to meet again Tue, 5 Sep 10:30aET

<FredZ> fred can attend

<EliasT> works for me

<LeeF> I'll scribe.

Bijan: regrets for next week, most likely

RESOLUTION: to meet again Tue, 5 Sep 10:30aET; LeeF to scribe; regrets from PatH

<SimonR> I don't object to dealing with process first before the agenda items.

DanC: last publication was 6 Apr, more than 3 months ago; I'd like to talk about that a bit

Issues List updated

<kendallclark> http://www.w3.org/2001/sw/DataAccess/issues

http://www.w3.org/2001/sw/DataAccess/issues v 1.143

KC: I count 7 open issues.

[[

1. punctuationSyntax open

2. nestedOptionals open

3. formsOfDistinct open

4. contradictoryKB open

5. nonliteralValueTesting open

6. openWorldValueTesting open

7. bnodeRef open

]]

KC: 3 of those are re-opened ( punctuationSyntax, bnodeRef, [missed]).
... the others are new.

Fred: did you see my request for 8 issues?

KC: yes... a lot of that was new to me...
... though I didn't look at it again when I did this issue update, though 2 of them (punctuationSyntax, nestedOptionals) are related.

-> FredZ 2 Aug

<scribe> ACTION: KC to review FredZ 2 Aug for issue updates [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action01]

Bijan: Fred, is this still exhaustive?

Fred: it was exhaustive at that time.

KC: on your list of 8, #3 is on the WG list as is #6...
... I consider nestedOptionals to include your #6

FredZ: though even in the non-nested case, the issue comes up

KC: ok, so the name is perhaps off by a bit

FredZ: my #7 looks close to something LeeF recently raised.

KC: indeed; let's look at that presently.

BP: there has been discussion of semantic framework, includin the bug from [a Chilean?]... seems to correspond to Fred's #1

<SimonR> [Chilean == Perez?]

<kendallclark> Yes, "the Chileans" == Perez's comments ;)

<scribe> ACTION: KC to consider a new issue "entailment framework", somewhat like rdfSemantics, though perhaps different [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action02]

<scribe> ACTION: Bijan to review FredZ 2 Aug and relate to WG issue list [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action03]

Bijan: sometimes it helps to have separate messages per issue...

Fred: yes, I did send separate comments and then consolidated them after a "where are we?" request.

<Zakim> DanC, you wanted to ask who has the ball on each issue? (perhaps not worth the time)

Simon: [missed?]

<SimonR> I have an outstanding action to do a second review (in addition to LeeF's) of the r23 vs rq24 revision.

KC: I think punctuationSyntax ball is with Fred/Andy...

<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0145

KC: so I hope to hear from Andy on punctuationSyntax
... on optionals... have you (Fred) and Andy explored that?

FredZ: I've written a series of proposals; I think the latest one is on target...

-> Constructive mapping semantics for SPARQL 18 Aug

^ latest, Fred?

Fred: yes, 0144 is current.

<LeeF> I've read it but haven't internalized it for comment.

<scribe> ACTION: PatH to review FredZ Constructive mapping semantics for SPARQL 18 Aug [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action04]

<bijan> I'll read it

<scribe> ACTION: SimonR to review FredZ Constructive mapping semantics for SPARQL 18 Aug [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action05]

* RRSAgent records action 4

<bijan> Why not?

<SimonR> I'll queue it to read as soon as I can finish rq24

<scribe> ACTION: bijan to review FredZ Constructive mapping semantics for SPARQL 18 Aug [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action06]

KC: I think for formsOfDistinct the ball is with bijan...

bijan: I've done what I expect to do on that one [and on contradictoryKBs]
... design options seem reasonably clear, though I haven't found a consensus; there are differing preferences.

PatH: are those issues lumped together?

KC: no, they're tracked separately

bijan: I've also done what I expect to do on nonliteralValueTesting
... including test cases.
... I think there's consensus about ill-formed literals, for the editors to clean up.
... for non-literals, there are varying preferences.

PatH: I see some connection with contradictoryKB [though I don't mind tracking them separately]

KC: I think the ball is with eric on openWorldValueTesting

<kendallclark> good for the chair to finish the review

KC: on bnodeRef... bijan?

bijan: I'm expecting some feedback from my recent msg

<bijan> Andy feels that they are query scoped

<bijan> or so I understand

heartbeat requirement, time between publications

(the rule is SHOULD publish every spec every 3 months; MUST publish one)

KC: I thought we published the JSON results WD, but evidently not so.

Lee: we were close, and then I asked a javascript security question and I don't recall an answer...

Elias: and there was some discussion of the shortname

<scribe> ACTION: Lee to figure out the next step toward publication of SPARQL results format er.. JSON in particular [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action07]

<kendallclark> hearbeat requirement:

<kendallclark> you have to publish a draft often enough that the loop between commenters and comments and responses remains coherent

[trouble scribing and talking...]

bijan: so one possibility is to republish the CR with a SOTD that says "see also rq24, in progress"

<SimonR> I note that sections 1.2 and 2.5 are still not filled in in rq24 -- it's not in CR condition yet, as a result.

DanC: I have a mild preference to get the SPARQL results format out in days and publish rq24 in due course, as long as the group can do that within, say, 6 weeks

Lee: I think rq24 is headed in the right direction

bijan: I'm prepared to publish rq24 today; I think going to WD and doing another Last Call is in order.
... we haven't signalled to the wider community the level of open issues we're considering.

PatH: I concur with Dan; I prefer rq24 to rq23 as our next publication

bijan: I think it's critical to publish soon.
... sooner than 6 weeks.

Fred: I support going back to WD

<bijan> The other point is that a WD rq24 focuses *us* on that draft.

quite.

<patH> indeed.

<bijan> For example, I tend to ignore it when raising issues because it's just a editor's draft, marked as experimental

<LeeF> bijan, as far as I know the point of SimonR and my reviews are to decide *if* we want to focus on rq24

<bijan> LeeF, great, but I think I'm ready to decide

<bijan> I've heard nothing against it

<bijan> We should be able to decide this by next week

Open World Value Tests

<SimonR> No objection to postponing agendum 3 from me.

<LeeF>

KC: ok to postpone item 3? ok, let's.

Filter evaluation issues

<kendallclark> Lee's message about filter clauses:

<kendallclark> http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0186.html

-> order of FILTERs in a BGP? scope of FILTERs? 28 Aug

KC: I don't like the design where you get different results when you move the FILTER around.

Lee: quite.

<bijan> Good god! I hope Lee is wrong!

[[

Now, we change the position of the FILTER (query #3):

PREFIX: <http://example.org/>

SELECT *

FROM <http://thefigtrees.net/lee/sw/data/g.n3>

{

FILTER (?t = :s2) .

?s :p1 :o1 .

?t :p2 :o2 .

}

and now we get *zero* solutions.

]]

^ that's the surprising one.

^ the zero solutions are from a running instance of ARQ

<bijan> That's just wrong

<LeeF> the data is:

<LeeF> @prefix : <http://example.org/> .

<LeeF> :s1 :p1 :o1 .

<LeeF> :s2 :p2 :o2 .

<SimonR> I have to admit, I always assumed FILTERs were just a conjunction term of the basic graph pattern. This disabuses me of that happy notion....

<bijan> That looks like broken code to me :)

<patH> me too.

http://www.w3.org/2001/sw/DataAccess/tests/#dawg-bound-query-001

not approved.

Lee: by my reading, the FILTER in dawg-bound-query-001 is connected to an empty basic graph pattern, so I don't see how those are the expected results.

<SimonR> It's my longstanding opinion that the nulls generated by OPTIONAL are not unbounds, so I'm not at all surprised to see them making no sense at all once they interact with other constructs. :/

Simon: let's connect this test to the optional issue?

<Zakim> DanC, you wanted to ask Fred what answers his constructive semantics in any of these cases

bijan: ah... then I see an issue separate from optional

http://www.w3.org/2001/sw/DataAccess/tests/#dawg-bound-query-001

<kendallclark> so the issue is the scope of filter?

<kendallclark> damn

<bijan> ouch

Fred: my '
... my 'constructive..." proposal doesn't address mapping of grammar to math constructs... but...

<LeeF> SELECT * { { <s> <p> ?o } FILTER (?s > 5) . }

<LeeF> does that FILTER affect anything?

<LeeF> err ?o > 5

<bijan> Hmm. i suspect the chilean algebra might not directly deal with this

PatH: I recall some discussion of where filters go and being able to move them after their siblings...

others: yes, I recall that discussion but not a clear outcome

<bijan> Oh, they have filters

Simon: if we try to separate the parts of the query... [scribe is not quite following...]

<bijan> +1 to pat's last point

<LeeF> I don't think commutativity is really the (only) issue...

<kendallclark> proposals to radically simplify the language are always in order, IMO

<kendallclark> ACTION: Bijan to see if the Chilean's semantics paper offers any advice re: filters [recorded in http://www.w3.org/2006/08/29-dawg-minutes.html#action08]

bijan: I'm interested to review [which paper?] and ...[?]

<SimonR> [paper = `Semantics & Complexity of SPARQL']

KC: I hope to hear from Andy on whether SPARQLER's behaviour here is as designed.

ADJOURN.