Chatlog 2012-04-10

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.

14:02:31 <RRSAgent> RRSAgent has joined #sparql
14:02:31 <RRSAgent> logging to http://www.w3.org/2012/04/10-sparql-irc
14:02:33 <trackbot> RRSAgent, make logs world
14:02:35 <trackbot> Zakim, this will be 77277
14:02:35 <Zakim> ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start 2 minutes ago
14:02:36 <trackbot> Meeting: SPARQL Working Group Teleconference
14:02:36 <trackbot> Date: 10 April 2012
14:03:24 <AxelPolleres> zakim, who is on the phone?
14:03:30 <Zakim> I notice SW_(SPARQL)10:00AM has restarted
14:03:31 <Zakim> On the phone I see +43.664.801.173.aaaa, kasei, Olivier, SteveH, bglimm, LeeF, +1.603.897.aadd
14:03:44 <AxelPolleres> Zakim, aaaa is me
14:03:44 <MattPerry> zakim, aadd is me
14:03:46 <Zakim> +AxelPolleres; got it
14:03:47 <Zakim> +MattPerry; got it
14:04:03 <SteveH> scribenick: SteveH
14:04:04 <AxelPolleres> scribe: SteveH
14:04:08 <SteveH> scribe: SteveH
14:04:20 <AxelPolleres> agenda:  http://www.w3.org/2009/sparql/wiki/Agenda-2012-04-10
14:04:30 <AxelPolleres> topic: admin
14:04:37 <AxelPolleres> PROPOSED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2012-04-03
14:05:08 <AxelPolleres> RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2012-04-03
14:05:08 <cbuilara> cbuilara has joined #sparql
14:05:24 <AxelPolleres> Next regular meeting: 2012-04-17 @ 15:00 UK / 10:00 EST (scribe: cf. scribe_list) 
14:05:35 <Olivier> I do
14:05:46 <SteveH> AxelPolleres: anyone planning to attend WWW
14:05:51 <AxelPolleres> who's attending WWW in Lyon? Olivier, Axel
14:06:10 <Zakim> +??P25
14:06:14 <AxelPolleres> regrets for next week
14:06:18 <cbuilara> zakim, ??P25 is me
14:06:18 <Zakim> +cbuilara; got it
14:06:23 <Olivier> regrets
14:06:37 <AxelPolleres> topic: RDF WG liaison
14:07:04 <AxelPolleres> no news from RDF
14:07:13 <AxelPolleres> topic: Publications
14:07:29 <SteveH> … overview document, any comments?
14:07:41 <AxelPolleres> continued
14:07:42 <SteveH> … leave til next week
14:08:11 <SteveH> … we should publish the docs all together, maybe without Query
14:08:21 <SteveH> … so will probably try to publish existing sooner
14:08:29 <AxelPolleres> topic: Query/PP
14:09:18 <AxelPolleres> http://lists.w3.org/Archives/Public/www-archive/2012Apr/0013.html
14:09:32 <SteveH> … link to origninal mail, several replies from commentors
14:10:01 <AxelPolleres> Lee, would you mind to summarize?
14:10:15 <AxelPolleres> q?
14:10:17 <kasei> Wim responded, but not to the archive list, I think.
14:10:34 <AndyS> Apologies - will be 5 mins
14:11:18 <SteveH> LeeF: summary: nobody really had much to say about whether the design addresses the perf. concerns
14:11:29 <AndyS> AndyS has joined #sparql
14:11:29 <SteveH> … people would need to do research, but they think it's ok
14:11:51 <SteveH> … negative reaction to the actual choices, e.g. that / is counting, but * and + are not
14:12:00 <SteveH> … and removal of {} part of language
14:12:22 <SteveH> … my take is that different people will differ on best design, but both are well-motivated
14:12:35 <SteveH> AxelPolleres: any opinions on that
14:12:37 <SteveH> ?
14:12:37 <Zakim> +??P26
14:12:41 <AndyS> zakim, ??P26 is me
14:12:41 <Zakim> +AndyS; got it
14:13:33 <AxelPolleres> Opinions on Lee's proposal to go forward with what was decided last week?
14:13:44 <AndyS> zakim, who is on the phone?
14:13:44 <Zakim> On the phone I see AxelPolleres, kasei, Olivier, SteveH, bglimm, LeeF, MattPerry, cbuilara, AndyS
14:14:08 <AxelPolleres> silence is agreement?
14:15:09 <SteveH> AxelPolleres: commenters want an official response
14:15:28 <kasei> I think we should push forward with option 6. The commenters seem to have a rather narrow view on what paths are for, and the required balancing act.
14:15:42 <SteveH> … I fear the point that the commenters have, I see the concern
14:16:00 <SteveH> … but the RDF list usecase makes sense with the proposed semantivs
14:16:13 <SteveH> .. two different views of naturallity
14:16:42 <SteveH> … if we think we don't block ourselves to future use-cases, question for me is whether it's unintuative for other usecases
14:17:01 <SteveH> … ones with distinct paths can be modelled with DISTINCT subqueries
14:17:21 <Zakim> + +1.781.899.aaee
14:17:37 <SteveH> … we could reply that we've come to descision via use-cases, lists and counting [something]
14:17:55 <SteveH> … and we have scope for future operations, for e.g. counting
14:18:16 <AxelPolleres> any further opinions on that?
14:18:20 <MacTed> MacTed has joined #sparql
14:18:44 <SteveH> bglimm: one point in advantage of what we have now is that / reduces to normal BGP, so can rewrite
14:18:57 <SteveH> … might be difficult if you had non-counting semantics
14:18:57 <kasei> you can't always rewrite... (:a/:b)*
14:19:18 <SteveH> … you can rewrite the inner pattern to normal
14:19:44 <SteveH> kasei: rewrite has to happen inside the exavuation
14:19:46 <AndyS> Non-rdf-list UC: SELECT (sum(?cost) AS ?total) { :order :hasItem/:price ?cost } 
14:20:57 <Zakim> -cbuilara
14:21:07 <AxelPolleres> Andy: would be non-intuitive otherwise
14:21:14 <SteveH> AndyS: I think the support and education costs of non-counting semantics would be considerable
14:21:23 <SteveH> LeeF: volunterrs to draft response
14:21:30 <AxelPolleres> ACTION: lee to draft joint official reply
14:21:31 <trackbot> Created ACTION-608 - Draft joint official reply [on Lee Feigenbaum - due 2012-04-17].
14:21:39 <AndyS> (for "/" in the example -- i.e. on its own)
14:22:39 <SteveH> LeeF: intending to draft one response to all commenters
14:23:46 <SteveH> AxelPolleres: can you suggest date when we'll be at 2nd LC
14:24:02 <SteveH> AndyS: I've not gone back to the document yet, but don't think it's a huge task
14:24:18 <Zakim> +??P3
14:24:19 <AndyS> Regrets for next week.
14:24:24 <SteveH> AxelPolleres: can you update next week (on email)
14:24:34 <AxelPolleres> topic: Recent comments 
14:24:38 <cbuilara> zakim, ??P3 is me
14:24:38 <Zakim> +cbuilara; got it
14:24:41 <AxelPolleres> http://www.w3.org/2009/sparql/wiki/Comments
14:25:39 <SteveH> AxelPolleres, one comment by Alexander Dutton, two by Dave Beckett
14:25:45 <SteveH> AxelPolleres: one comment by Alexander Dutton, two by Dave Beckett
14:26:29 <SteveH> … any volunteers to take ownership?
14:26:47 <AxelPolleres> anybody volunteering to take ownerhip for AD-1 and AH-1?
14:26:59 <AxelPolleres> Zakim, pick a victim
14:26:59 <Zakim> Not knowing who is chairing or who scribed recently, I propose bglimm
14:29:39 <AxelPolleres> Any volunteers for DBeckett-4?
14:29:47 <AxelPolleres> Zakim, pick a victim
14:29:47 <Zakim> Not knowing who is chairing or who scribed recently, I propose LeeF
14:29:58 <kasei> is dajobe asking for a new test type in the manifests?
14:30:39 <SteveH> LeeF: I think it would make more sense for someone who's looked into it
14:30:44 <SteveH> … I'm not sure what he's asking for
14:31:15 <SteveH> … we have these evaluation tests that really test syntax
14:31:58 <bglimm> q+ to ask about the comments I got assigned
14:32:43 <SteveH> [discussion about test types]
14:33:26 <AxelPolleres> q?
14:35:05 <kasei>   mf:CSVResultFormatTest
14:35:07 <AxelPolleres> something like CSV_Result_Test?
14:35:19 <LeeF> I'm ok with that, kasei
14:35:43 <SteveH> bglimm: the comments I got assigned suggest changing the semantics of BIND - is that even an option at this point?
14:35:53 <AxelPolleres> q?
14:36:30 <SteveH> AndyS: Dutton's comment suggests multiple BINDINGS, doesn't change semantics, but there's a gotcha
14:36:44 <SteveH> … could be made to work, but it's not simple
14:37:04 <AxelPolleres> Dutton's comment looks more like a future-work-items comment, if anything?
14:37:15 <SteveH> … Hogan's is quite different to current semanitcs
14:37:57 <SteveH> bglimm: he wants some examples added to current spec, I can't do that myself
14:38:01 <SteveH> q+
14:38:16 <bglimm> Zakim, ack me
14:38:16 <Zakim> bglimm, you wanted to ask about the comments I got assigned
14:38:18 <Zakim> I see SteveH on the speaker queue
14:38:45 <SteveH> SteveH: I don't think we should add more examples to the spec text - it's too big already
14:39:03 <AndyS> (Just for the record, I prefer loop semantics so FILTERS work with BINDINGS)
14:39:21 <AxelPolleres> Way forward on examples would be to check whether we have test cases for those example and refer to it.
14:39:31 <SteveH> bglimm: so I could say we don't add an example, but we've added a test case / there is already a test case XYZ
14:41:35 <kasei> there's big performance implications on doing the loop semantics. they might make sense for a few bindings, but sending in a huge list of bindings could be bad.
14:45:04 <AxelPolleres> AxelPolleres has joined #sparql
14:47:12 <cbuilara> There is a system call FedX that implements SERVICE and implements a BIND JOIN, it gets the results from one SERVICE call and creates a set of queries ot the other remote SPARQL endpoints
14:47:42 <cbuilara> in a remote SPARQL endpoint configuration without result limit, it generates like 20000+ queries
14:47:50 <cbuilara> using this BIND JOIN
14:47:51 <AxelPolleres> we do looping semantics in XSPARQL (SPARQL in XQUERY, where we also have SERVICE)
14:49:11 <Zakim> +EricP
14:49:23 <AxelPolleres> greg: late in the game for changing the semantics
14:49:35 <SteveH> [discussion on semantics of BINDINGS]
14:50:17 <SteveH> AxelPolleres: if we want something like templating, how would we do that with a non-conflicting design?
14:51:13 <SteveH> AndyS: I don't see the federated query issue - if it's loop substitiution semantics, I don't see why then engine can't [rewrite?]
14:51:25 <SteveH> AxelPolleres: let's discuss on mailing lisyt
14:51:30 <AxelPolleres> I guess we need to let this comment simmer for a week or so and hopefully address AH-1 next time
14:51:42 <SteveH> … other issues?
14:51:50 <AxelPolleres> q?
14:51:53 <AndyS> can't not rewrite and send a single SERVICE and the bindings to the other end (hard to explain without a picture)
14:51:58 <SteveH> Zakim, ack me
14:51:58 <Zakim> I see no one on the speaker queue
14:52:38 <SteveH> AndyS: was there a discussion of schedule at start?
14:53:05 <SteveH> AxelPolleres: yes, but didn't discuss schedule for other docs already into LC, into PR
14:53:36 <AxelPolleres> GSP, CSV, Overview should be published LC as soon as possible.
14:53:37 <SteveH> AndyS: Graph store protocol was approved in Feb. we should press on with that
14:54:06 <Zakim> -EricP
14:54:09 <SteveH> AxelPolleres: hopefully we will decide to publish Overview next week
14:54:53 <AxelPolleres> plan for next week, to decide to publish GSP, CSV and Overview LC together asap (short after WWW), and talk about schedule for other docs (as well as test case licensing, etc.
14:55:43 <Zakim> - +1.781.899.aaee
14:55:44 <Zakim> -bglimm
14:55:46 <Zakim> -Olivier
14:55:49 <Zakim> -MattPerry
14:55:51 <Zakim> -cbuilara
14:55:53 <SteveH> adjourned
14:55:56 <Zakim> -kasei
14:55:58 <Zakim> -SteveH
14:56:03 <Zakim> -AxelPolleres
14:56:04 <AxelPolleres> adjourned
14:56:04 <AxelPolleres> rrsagent, make records public
14:56:09 <Zakim> -AndyS
14:56:25 <Zakim> -LeeF
14:56:27 <Zakim> SW_(SPARQL)10:00AM has ended
14:56:27 <Zakim> Attendees were +43.664.801.173.aaaa, kasei, +33.4.92.38.aabb, Olivier, SteveH, bglimm, +1.617.553.aacc, LeeF, +1.603.897.aadd, AxelPolleres, MattPerry, cbuilara, AndyS,
14:56:27 <Zakim> ... +1.781.899.aaee, EricP
15:08:22 <AxelPolleres> AxelPolleres has joined #sparql
# SPECIAL MARKER FOR CHATSYNC.  DO NOT EDIT THIS LINE OR BELOW.  SRCLINESUSED=00000172