SPARQL Working Group Teleconference -- 09 Aug 2011

<trackbot> Date: 09 August 2011

<LeeF> Scribe: bglimm

<LeeF> Scribenick: bglimm

<AndyS> Hi there

<AndyS> A lot of echo

better?

<kasei> success!

yes

Admin

<LeeF> PROPOSED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2011-08-02

<LeeF> RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2011-08-02

<alexpassant> that should be me

LeeF: Next meeting next week
... AndyS is set to scribe

<LeeF> Next regular meeting: 2011-08-16 @ 15:00 UK / 10:00 EST (scribe: Andy)

LeeF: AndyS, anything from the RDF working group?

I didn't get what Andy said

I have shitty sound :-(

<LeeF> Andy: Nothing new from RDF WG, but they will probably be reaching a decision on language tagged literals soon

Federated Query

LeeF: We have a federated query review. Carlos, will you discuss further changes by email?

greg: The only major issues is that we agreed to not formally specify the endpoint semantics (?), the section is informative, but is wrong

LeeF: I suggest to make the text less formal and keep the section informative

<cbuilara> +q

Axel: As long as the section is informative, I think it is ok. We can change it later

LeeF: I think we should still get the semantics right

Carlos: The section was to address the changes suggested by Andy

LeeF: I think we just have to keep working towards convergence
... Carlos, can you have a look at the comments and further discuss by email?

Carlos: Ok

<axelpolleres> I can take an action to check Greg's review and implications as well...

<axelpolleres> (would make it easier to remember for me ;-))

<LeeF> ACTION: Axel to look at Greg's review of federated query [recorded in http://www.w3.org/2011/08/09-sparql-minutes.html#action01]

<trackbot> Created ACTION-514 - Look at Greg's review of federated query [on Axel Polleres - due 2011-08-16].

<AndyS> What I see in doc is not as Carlos described (from a quick skim) is doc in CVS up to date?

Carlos: I commited a change

Other documents

<kasei> that "ranging over all services" is the part I'm not seeing and have a problem with.

AndyS: I'll read it more carefully then

LeeF: Axel did work on the overview doc and AndyS reviewed it

Axel: I partially adressed the comments
... rest we can to by email
... I will work on it further, but nothing seems critical for publishing

LeeF: Let's see whether we can publish th overview together with fed. query and the protocol oc

Axel: We should probably publish as a FPWD to get some feedback

LeeF: I think it is an important document, but we probably won't get too much feedback on it
... Can anybody else review the overview doc?

(silence)

LeeF: Matt can you review it?

<LeeF> ACTION: Matthew to review the overview document [recorded in http://www.w3.org/2011/08/09-sparql-minutes.html#action02]

<trackbot> Created ACTION-515 - Review the overview document [on Matthew Perry - due 2011-08-16].

Matt: Yes, I can review it

<Zakim> AndyS, you wanted to ask about xmlspec / respec ization

Axel: I asked on the mailing list about the wiki to HTML script

Sandro said it is a lot of work to set the scrpt up

It is only worth doing if we use it several times

<axelpolleres> ok, I can do it manually, just wanted to know whether there's an easy way.

<Zakim> axelpolleres, you wanted to ask one more question in the context off the overview doc

LeeF: I think we can keep working on the wiki for a while and then manually convert to XML

Axel: Andy had a comment about using named graphs
... I can avoid using named graphs, so we can get away without them
... The other thing was the list of all documents, which we now have in all dcs
... I used a different order than the other docs
... I ordered to make a nice story in the overview
... in some of the other documents, we don't have the list. I suggest to link from all other docs to the overview, where we have the list
... so we would only have one list, which is in the overview doc

+1 to Axel's suggestion

AndyS: I personally don't find it useful to be directed to the overview
... I find the order we ended up with appropriate

Axel: I don't have a particular order preference for thenon-overview docs, but I do have one for the overview
... would you be ok to keep the structure of the document, but adjust the list?

AndyS: I think the structure should be major areas first and then the minor areas

<SteveH> I have no preference

<kasei> I tend to favor Andy's approach

LeeF: We have slightly more votes for Andy's sugggestion. Axel, can you restructure?

Axel: Yes and it is anyway a FPWD

<axelpolleres> I can restructure, finding time is more the issue.

AndyS: I can't remember how long it takes overall from FPWD to LC

LeeF: FPWD has no fixed duration
... LC has a minimum time of three weeks I believe

<axelpolleres> Shall we tendentially decide for a short name? proposal: sparql11-overview

LeeF: I am happy with the short name

AndyS: Looks good

LeeF: Let's decide when we decide to publish
... Neither the chairs nor team contacts did make progress on the csv-tsv and json document (scribed correctly?)
... I addressed most of Andy's comments for the protocol doc

<axelpolleres> it should follow the same rules as FROM FROM NAMED, shouldn't it? (didn't check the mails in detail)

LeeF: we still have some discussions
... Andy can you outline the usecase that you had on the mailing list?

Didn't get what Andy said :-(

<AndyS> AndyS: One possible UC is that the dataset for matching is a new, temporary dataset (maybe retrieved from the web)

<axelpolleres> Lee's way sounds good, USING/USING NAMED is described in Table 2 of the update doc

LeeF: Clearly the spec still needs to be improved
... but the question whether the current model is acceptable

<axelpolleres> see http://www.w3.org/TR/sparql11-update/#mappingRequestsToOperations

AndyS: Regardless what we decide, the change will change the update doc

Axel: Sounds to me that removing the parameters is in line with the update document

LeeF: If that

<axelpolleres> Are you sure that we need to change Update, I don't think so.

LeeF: If you specified it like this, then we might not need a change

AndyS: We have time to do a last cal for update in parallel with the protocol doc

LeeF: Are we happy with the semantics of the protocol or should we consider alternatives?
... AndyS, do you need more time to think about it?

<axelpolleres> I understand that protocol says that updateReq with parameters simply means that the requested endpoint needs to answer Tr(updateReq,parameters) where Tr just replaces the USING USINGCLAUSES

AndyS: Yes and we are a small group of people and I want to make sure we address the right problem
... we are not getting enough input

<axelpolleres> this can be defined similarly as the tables in the update document, but it is ok if it defined in the protocol dfocument.

AndyS: Steve, you are another major update implementor. Did you get your head around that?

Steve: It is not a feature that we currently use, so I can't give input

LeeF: Anybody else implementing it?

Axel: How is it for query request now when there are parameters?

<SteveH> Lee's interpretation is the only one that makes sense to me

LeeF: The design for update is different because there are different update requests

<axelpolleres> LeeF: my proposal was that parameters replace any using/using named in any operation part of a request.

LeeF: We don't want to do something that we regret later. Maybe I write up the current design and send it to the list to get feedback

AndyS: That might be a good idea. Any other points, where you need feedback?

LeeF: I think I addressed most points, but I will get back to it

AndyS: Add more example

LeeF: There is still an issue with characters in query strings
... I'll rewrite the text, to make it clearer

<LeeF> ACTION: Lee to email list with proposed design for dataset parameters in protocol for update requests [recorded in http://www.w3.org/2011/08/09-sparql-minutes.html#action03]

<trackbot> Created ACTION-516 - Email list with proposed design for dataset parameters in protocol for update requests [on Lee Feigenbaum - due 2011-08-16].

LeeF: Test case covering
... Axel, can you give an overview of the status

Axel: We had a couple of actions to evaluate coverage
... 492 and following
... for update, the action is done

<axelpolleres> close ACTION-492

<trackbot> ACTION-492 Check coverage of test suite (on Update) closed

493 is create a summary on the wiki

<axelpolleres> http://www.w3.org/2009/sparql/wiki/TestSuiteCoverage

scribe: For each area we wanted to have statements from implementors as to who implements the features
... please have a look and add yourself under implementation or remove yourself as appropriate
... action 494 is query on greg

greg: I still need to finish it

<AndyS> JSON results test suite broken.

Axel: Yes, it is a lot of work
... 495 is protocol test cases coverage, which should also cover how we test protocol at all

LeeF: No progress yet

Axel: Action 496 is on Chime
... for the graph store protocol
... it might need an extension for the manifest structure

<axelpolleres> close ACTION-497

<trackbot> ACTION-497 Check entailment regimes test case coverage closed

Axel: Action 497 is on entailment reg.
... that is completed
... That is the last action on test case coverage

<AndyS> Close ACTION-507

<trackbot> ACTION-507 Draft text for CSV/TSV status para closed

<AndyS> Close ACTION-500

<trackbot> ACTION-500 Review updates in Fed query doc (particularly section 2.4 and 4) for LC readiness closed

Axel: for update, we have not yet everything covered: http://www.w3.org/2009/sparql/wiki/TestSuiteCoverage#Update
... Do we need to test silent?
... In syntax tests that is covered, but sine we cannot test error, it is hard for non-syntax test

<AndyS> Is there not a EvalFail test type?

Axel: I propose to not test that
... any objections?
... silence=agreement?

greg: We could create a negative evaluation test

<SteveH> a test would be DROP SILENT GRAPH <http://nosuchgrah>

<SteveH> I think minimal testing for SILENT is OK

<SteveH> but we can test it to an extent

greg: we could probably add something to make such tests possible

Axel: So the proposal is to add negative evaluation tests

greg: We might need a test for success, not a test for the state of the graph store

AndyS: A test without result, just saying that you somehow got through

Axel: I will look into that

LeeF: Shall we record an action?

<axelpolleres> ACTION: Axel to look into negative evaluation tests and "silent success test" possibility for update tests. [recorded in http://www.w3.org/2011/08/09-sparql-minutes.html#action04]

<trackbot> Created ACTION-517 - Look into negative evaluation tests and "silent success test" possibility for update tests. [on Axel Polleres - due 2011-08-16].

Axel: Andy suggested to move the negative syntax tests to the syntax test folder
... It is probably o t o move them there. Some tests are negative syntax tests because we disallowed bnodes, but they have not been moved after the decision

<LeeF> Suggest just leaving it as is for now then

Axel: Any volunteers to move the tests?
... We anyway need to action somebody to create the missing update tests. Maybe that person can then also move the tests.

Steve: No way I can find the time

LeeF: Let's leave it as it is for now

Axel: Next is entailment
... we should also have negative tests for container membership properties

no tests for axiomatic triples yet

scribe: we could just add an ask query for the triples

http://www.w3.org/2009/sparql/wiki/TestSuiteCoverage#Entailment

Axel: It seems we are out of time

yes

Axel: Birte, can you look into completing the test cases

Bire: Yes

adjourned

<kasei> axelpolleres: got a minute for a csv test question?

<MattPerry> bye

<LeeF> close ACTION-489

<trackbot> ACTION-489 Review outcome of ACTION-462 as soon as it's done closed

<LeeF> close ACTION-509

<trackbot> ACTION-509 Review latest federated query document for Last Call readiness closed

<axelpolleres> ACTION: birte to add missing test cases to improve http://www.w3.org/2009/sparql/wiki/TestSuiteCoverage#Entailment [recorded in http://www.w3.org/2011/08/09-sparql-minutes.html#action05]

<trackbot> Created ACTION-518 - Add missing test cases to improve http://www.w3.org/2009/sparql/wiki/TestSuiteCoverage#Entailment [on Birte Glimm - due 2011-08-16].

- DRAFT -

SPARQL Working Group Teleconference

09 Aug 2011

Attendees

Contents

Admin

Federated Query

Other documents

Summary of Action Items

Scribe.perl diagnostic output