Chatlog 2012-09-11

From SPARQL Working Group
Jump to: navigation, search

See original RRSAgent log and preview nicely formatted version.

Please justify/explain all edits to this page, in your "edit summary" text.

<kasei> Present: kasei, Sandro, AndyS, pgearon, AxelPolleres, chimezie, Carlos, bglimm
13:57:11 <RRSAgent> RRSAgent has joined #sparql
13:57:11 <RRSAgent> logging to
13:57:13 <trackbot> RRSAgent, make logs world
13:57:15 <trackbot> Zakim, this will be 77277
13:57:15 <Zakim> ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start in 3 minutes
13:57:16 <trackbot> Meeting: SPARQL Working Group Teleconference
13:57:16 <trackbot> Date: 11 September 2012
13:58:38 <cbuilara> cbuilara has joined #sparql
13:59:20 <Zakim> SW_(SPARQL)10:00AM has now started
13:59:52 <AxelPolleres> agenda:
14:01:03 <AxelPolleres> chair: sandro
14:02:46 <kasei> scribenick: kasei
14:03:31 <sandro> sandro has changed the topic to: SPARQL WG  :
14:05:32 <kasei> topic: open comments
<kasei> subtopic: JL-4
14:06:04 <kasei> sandro: don't really understand James' comment that well.
14:06:28 <kasei> ... feel some obligation to understand what he's saying more thoroughly
14:06:38 <kasei> ... if others understand it well enough, that's ok, too
14:06:59 <chimezie> Zakim unmute me
14:07:32 <kasei> AxelPolleres: his system has another way to do indirect graph identification
14:07:38 <kasei> ... apart from that he'll implement GSP
14:07:51 <kasei> ... he can't support the indirect identification as written.
14:08:17 <kasei> ... see no reason why we couldn't say what he wants isn't being done this time around, but refinement could be done by future WG or linked data WG
14:08:47 <kasei> sandro: we would say that our approach is *one way* to do it that doesn't preclude future ways?
14:08:50 <kasei> AxelPolleres: I think so
14:09:05 <kasei> sandro: as I read the spec I think I would say other ways are precluded.
14:09:22 <kasei> ... if we aren't precluding other ways, then I think that is a reasonable response.
14:09:41 <kasei> chimezie: so we could say we're not preventing an extension to GSP?
14:09:55 <kasei> ... what he's asking for requires a service description vocabulary
14:10:08 <kasei> sandro: he wants to be able to use indirect URIs of a different form (without ?graph=)
14:10:08 <AxelPolleres> AxelPolleres has joined #sparql
14:10:17 <kasei> ... and also to use service descriptions to find out what that URI pattern is.
14:10:29 <kasei> ... I think that would be OK if we're not precluding that in the future in GSP
14:10:48 <chimezie> chimezie has joined #sparql
14:11:01 <kasei> ... I'm fine with us clarifying the wording to indicate that our approach is *a* way to do the identification.
14:11:18 <sandro> sandro: I'm okay with clarifying that ?graph= is *a* way to do indirection, and leaves the door open for others.
14:11:29 <kasei> chimezie: in James' thread, he describes two alternatives. one where you discover the prefix and another with the purl mechanism which you replace up to a prefix.
14:11:40 <AxelPolleres> If we're not preventing future extensions to GSP that allow a way that JL could support, we should be ok to answer that we postpone other ways toaddress graphs as future work, we could point to his proposal on the future work items page.
14:11:43 <kasei> ... if you leave it open to other approaches, those other ways are underspecified.
14:12:03 <kasei> sandro: yes. they're not specified yet. 3 round stones (James' company) could make a member submission for how to do that.
14:12:20 <kasei> ... that would be for a future work item. want to make sure the door is open for that.
14:12:40 <kasei> chimezie: I think that would be the right way to go.
14:12:59 <sandro> PROPOSAL: We respond to JL-4 saying the door is open for what he wants in the future -- but we're not going to spec that kind of thing in this round.
14:13:31 <AndyS> +1
14:13:41 <AxelPolleres> +1 (we should/could propose to add a future work item to our wiki explicitly in the wiki)
14:13:55 <chimezie> +1
14:14:15 <kasei> sandro: not sure if we need to do a slight editorial wording change to make this clear in the spec.
14:14:24 <kasei> ... chimezie, would you be OK to look at the wording?
14:14:33 <AndyS> Seems more LDP than SPARQL.  That is billed as working with RDF RESTfully (in some quarters)
14:14:35 <sandro> RESOLVED: We respond to JL-4 saying the door is open for what he wants in the future -- but we're not going to spec that kind of thing in this round.
14:14:36 <kasei> chimezie: I can take a look.
14:14:53 <sandro> action: chimezie to draft editial change to GSP to make clear the door is open
14:14:53 <trackbot> Created ACTION-677 - Draft editial change to GSP to make clear the door is open [on Chimezie Ogbuji - due 2012-09-18].
14:14:55 <chimezie> chimezie has joined #sparql
14:15:11 <sandro> subtopic: RC-2
14:15:51 <sandro> kasei: This is a long running thread on Richard wanting to stdize what error responses look like in the protocol
14:16:01 <sandro> ... he wrote this when we were using WSDL
14:16:23 <sandro> ... Lee and I talk about me working out details with Richard
14:16:42 <sandro> ... adding some text, in a non-normative way, suggesting people use HTTP status line, is appropriate.
14:16:56 <sandro> ... Richard wanted it as response body should be that message.
14:17:46 <AndyS> Does anyone have experience of intermediaries changed supplied text?
14:17:53 <sandro> sandro: like in "404 not found" you would be changing the "not found"
14:18:30 <sandro> sandro: does the spec let you change it?
14:18:47 <sandro> kasei: Yes, the spec says this text is meant for humans.
14:19:42 <sandro> sandro: you're planning on putting more specific information in that text?
14:19:52 <sandro> kasei: Yes.   We've got this in the examples.
14:20:01 <kasei>
14:20:12 <kasei> HTTP/1.1 500 Query Request Refused
14:20:23 <kasei> HTTP/1.1 400 Syntax Error
14:21:07 <chimezie_> chimezie_ has joined #sparql
14:21:27 <sandro> sandro: So you want people to be able to put HTML in the body
14:21:37 <sandro> kasei: We don't want to standardize that.
14:22:40 <sandro> sandro: So, how hard is it to get at this in software
14:23:49 <AndyS> q+
14:24:42 <sandro> ack AndyS 
14:25:10 <sandro> AndyS: I do occassionally see HTTP status messages in different languages.
14:25:47 <AxelPolleres> Is this comething we need resolved to progress? (after all, it's a non-normative section, right?)
14:26:08 <sandro> scribe: pgearon 
14:26:23 <pgearon> kasei: my computer seems to have locked up
14:26:31 <sandro>
14:26:43 <pgearon> sandro: what other issues can we progress, and how far?
14:27:05 <sandro> topic: Query
14:27:18 <pgearon> sandro: are we ready to go to PR for Query?
14:27:48 <pgearon> sandro: 2 implementations for every test case. Is there a document that shows that?
14:28:19 <sandro> kasei: bind, datasets
14:28:27 <pgearon> AndyS: where are we on the final errors that came up? Greg?
14:28:54 <AxelPolleres> my question would be: shall we rather wait a week(?) until tests are re-run or progress to CR now? (probably the former?)
14:29:08 <pgearon> AndyS: didn't feel we can change the test suite with tests that are not approved
14:29:17 <AndyS>
14:29:24 <pgearon> kasei: I may have removed test from manifest, but not removed the actual test
14:29:59 <pgearon> AndyS: who's putting in results for tests?
14:30:26 <chimezie>
14:30:27 <sandro>
14:31:00 <pgearon> kasei: between AndyS and I (except these tests) I think we have full implementations
14:31:15 <bglimm> I plan to submit a test report later this week for ent. tests
14:31:24 <pgearon> AndyS: Rob Vesse wants to get dotNetRdf onto the list
14:32:01 <pgearon> kasei: think Jeen is interested, but stopped running regularly while the tests were in flux. He may be interested in running again now that the tests are stable
14:32:09 <pgearon> AndyS: how long is CR?
14:32:19 <Zakim> -AxelPolleres
14:32:22 <pgearon> sandro: ~3-4 weeks
14:33:09 <pgearon> kasei: getting results from people to update the page is indeterminant. I can do my own results quickly
14:33:34 <Zakim> -pgearon
14:33:57 <kasei> scribenick: kasei
14:34:13 <Zakim> +pgearon
14:34:34 <kasei> sandro: seems to me that we should try to go straight to PR as soon as we can get fixed test results
14:35:19 <kasei> sandro: the implementations report looks good for many specs. is that accurate?
14:37:05 <kasei> kasei: everything but Query, Update, and Federation should be good.
14:37:26 <kasei> ... those three were affected by recent test changes, so need to be look at again.
14:37:33 <sandro> sandro: Any GSP tests?
14:37:38 <kasei> chimezie: I'm working on GSP tests.
14:37:48 <sandro> chimezie: Still underway.  Do we need a harness?
14:37:54 <kasei> ... still a question on the test harness, but I've got an action to create the actual tests.
14:37:55 <cbuil> +q
14:38:10 <kasei> cbuil: what's missing from federation?
14:38:57 <kasei> kasei: nothing is missing, but one of the federation tests got affected by recent BIND changes.
14:39:06 <kasei> ... so the results need to be re-done.
14:39:26 <kasei> sandro: if people are able to move forward quickly, in 1-3 weeks we should have updated test results
14:39:45 <AndyS> kasei - please check
14:39:48 <kasei> ... so things looking good in the implementations results should go to PR. CR with others (GSP, Entailment)
14:40:06 <kasei> AndyS - will do
14:40:15 <cbuil> I will work on the test results
14:41:16 <kasei> sandro: are there other implementors that know we're asking for test results?
14:42:00 <bglimm> I asked Kendall Clark, Jeff Pan and Michael Schneider
14:42:20 <bglimm> Chime is talking to the SILK implementors
14:42:30 <kasei> kasei: virtuoso is nominally on the WG
14:42:45 <kasei> AndyS: didn't we send out an email to the sparql list(s)?
14:43:45 <kasei> sandro: looking on public-sparql-dev...
14:43:50 <kasei> AndyS: I can't see anything.
14:44:04 <sandro>
14:44:09 <sandro> Public implementations of SPARQL 1.1 draft features?
14:45:26 <kasei> sandro: worst case is we try to do PR in 2 weeks and we can't make the case we've done wide enough outreach.
14:45:49 <kasei> ... then we do CR for a month and come back. hopefully we would get the reamining documents on board at that point.
14:45:59 <kasei> ... or we don't have to try for PR right now.
14:46:16 <kasei> AndyS: I think we need LeeF or Axel's opinions on this.
14:46:45 <kasei> sandro: our deadline is the end of the year (?)
14:47:40 <kasei> sandro: working backwards, PR needs to go out first week of november.
14:47:53 <kasei> ... november is busy with TPAC and ISWC and publishing moritorium.
14:48:02 <kasei> ... not a lot of time to kill.
14:48:12 <kasei> ... have time for a quick CR for the things that need it.
14:48:37 <kasei> ... what implementations do we have that do GSP? two that might pass the coming tests?
14:48:49 <kasei> chimezie: Andy has mentioned Fuseki before, and I have one.
14:49:02 <kasei> sandro: so that could concievable by ready in a few weeks?
14:49:04 <kasei> chimezie: yes.
14:49:28 <kasei> sandro: aob?
14:49:48 <kasei> chimezie: is there a precedent for putting text in the spec that referrs to future work?
14:50:10 <kasei> ... we have places for saying "at risk," but don't see anything that says "this could change in the future".
14:50:16 <kasei> sandro: for things that might change after REC?
14:50:40 <kasei> ... no so much that it would change, but that it's left open for future standardization work.
14:50:46 <kasei> ... can't think of a specific case.
14:52:05 <kasei> sandro: adjourned