From SPARQL Working Group
Revision as of 17:46, 7 July 2010 by Lfeigenb
Please justify/explain all edits to this page, in your "edit summary" text.
13:56:33 <RRSAgent> RRSAgent has joined #sparql 13:56:33 <RRSAgent> logging to http://www.w3.org/2010/07/06-sparql-irc 13:56:35 <trackbot> RRSAgent, make logs world 13:56:35 <Zakim> Zakim has joined #sparql 13:56:37 <trackbot> Zakim, this will be 77277 13:56:37 <Zakim> ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start in 4 minutes 13:56:38 <trackbot> Meeting: SPARQL Working Group Teleconference 13:56:38 <trackbot> Date: 06 July 2010 13:56:38 <SteveH> I used to get a working line like one time in 3 13:56:51 <LeeF> zakim, this will be SPARQL 13:56:51 <Zakim> ok, LeeF; I see SW_(SPARQL)10:00AM scheduled to start in 4 minutes 13:56:52 <AndyS> "adding column" = join, the easy case :-) 13:56:55 <LeeF> Chair: LeeF 13:57:07 <LeeF> Would anyone please volunteer to scribe? 13:57:47 <MattPerry> MattPerry has joined #sparql 13:58:17 <LeeF> Regrets: AlexPassant, AxelPolleres, Souri 13:58:36 <pgearon> sorry, was off getting coffee. I can scribe 13:58:47 <Zakim> SW_(SPARQL)10:00AM has now started 13:58:47 <LeeF> pgearon, that would be great, but didn't you just scribe last week? :) 13:58:54 <Zakim> +[IPcaller] 13:58:57 <Zakim> +Lee_Feigenbaum 13:59:02 <pgearon> yes, and I'm not too good at it either. :-) 13:59:06 <pgearon> so I could use the practice 13:59:10 <AndyS> zakim, +[IPcaller] is me 13:59:10 <Zakim> sorry, AndyS, I do not recognize a party named '+[IPcaller]' 13:59:19 <AndyS> zakim, +IPcaller is me 13:59:19 <Zakim> sorry, AndyS, I do not recognize a party named '+IPcaller' 13:59:20 <LeeF> i will take you up on it then, paul, thanks 13:59:23 <AndyS> zakim, IPcaller is me 13:59:23 <Zakim> +AndyS; got it 13:59:24 <LeeF> Scribe: Paul Gearon 13:59:26 <LeeF> Scribenick: pgearon 13:59:33 <Zakim> +[IPcaller] 13:59:41 <SteveH> Zakim, [IPcaller] is me 13:59:42 <AndyS> zakim,who is on the phone? 13:59:48 <Zakim> +SteveH; got it 13:59:50 <AndyS> zakim, who is on the phone? 13:59:52 <Zakim> On the phone I see AndyS, Lee_Feigenbaum, SteveH 13:59:54 <Zakim> +kasei 13:59:56 <Zakim> On the phone I see AndyS, Lee_Feigenbaum, SteveH, kasei 14:00:14 <Zakim> +pgearon 14:00:50 <Zakim> +MattPerry 14:00:56 <Zakim> + +18.104.22.168.aaaa 14:01:13 <OlivierCorby> zakim, 38.aaaa is me 14:01:22 <Zakim> sorry, OlivierCorby, I do not recognize a party named '38.aaaa' 14:01:41 <SteveH> I'm going to dial in again, skype is a bit ill 14:01:48 <kasei> procmail can sort that out, AndyS 14:01:49 <LeeF> zakim, aaaa is OlivierCorby 14:01:50 <Zakim> -SteveH 14:01:51 <kasei> :) 14:01:52 <Zakim> +OlivierCorby; got it 14:02:17 <bglimm> bglimm has joined #sparql 14:02:50 <Zakim> +[IPcaller] 14:02:54 <SteveH> Zakim, [IPcaller] is me 14:02:58 <Zakim> +SteveH; got it 14:03:18 <Zakim> +Sandro 14:03:41 <SteveH> seems not 14:03:44 <NicholasH> I can't get the UK number to work 14:03:53 <SteveH> me neither 14:03:53 <NicholasH> just phoning BBC Operator to connect me to US number 14:03:56 <Zakim> +bglimm 14:03:58 <Zakim> +??P45 14:04:09 <bglimm> Zakim, mute me 14:04:10 <Zakim> bglimm should now be muted 14:04:12 <Zakim> -??P45 14:04:16 <ivan> ivan has joined #sparql 14:04:29 <LeeF> topic: Admin 14:04:31 <Zakim> +??P1 14:04:38 <ivan> ivan has joined #sparql 14:04:40 <LeeF> PROPOSED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-06-29 14:04:41 <NicholasH> zakim, ??P1 is me 14:04:42 <Zakim> +NicholasH; got it 14:05:01 <LeeF> zakim, who's on the phone? 14:05:01 <Zakim> On the phone I see AndyS, Lee_Feigenbaum, kasei, pgearon, MattPerry, OlivierCorby, SteveH, Sandro, bglimm (muted), NicholasH 14:05:15 <chimezie> chimezie has joined #sparql 14:05:23 <chimezie> Zakim, what is the code? 14:05:23 <Zakim> the conference code is 77277 (tel:+1.617.761.6200 tel:+33.4.89.06.34.99 tel:+44.117.370.6152), chimezie 14:05:40 <LeeF> RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-06-29 14:05:41 <pgearon> LeeF: any issues from last week? 14:05:59 <LeeF> Next meeting: 2010-07-13 @ 15:00 UK / 10:00 EDT 14:06:05 <Zakim> +Chimezie_Ogbuji 14:06:17 <LeeF> regrets next week: Sandro 14:06:24 <sandro> yep 14:06:27 <sandro> :-) 14:06:55 <pgearon> LeeF: any other admin business to take care of? 14:07:03 <LeeF> topic: LET/Assignment 14:07:09 <ivan> ivan has joined #sparql 14:07:16 <LeeF> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010AprJun/0397.html 14:07:23 <chimezie> Zakim, mute me 14:07:23 <Zakim> Chimezie_Ogbuji should now be muted 14:07:49 <ivan> ivan has joined #sparql 14:08:14 <pgearon> LeeF: LET syntactically easy way to bind values, esp for use in a FILTER 14:08:23 <ivan> ivan has joined #sparql 14:08:44 <pgearon> LeeF: one of the most popular features to miss the cut when we decided on features for SPARQL 1.1 14:08:44 <ivan> ivan has joined #sparql 14:09:34 <pgearon> LeeF: strong response from a few organizations, such as TopQuadrant. Also from working group members like pgearon and AndyS that it ought to be done, and can be easily implemented 14:10:09 <pgearon> LeeF: alternative of subquery with projected expression will work, but is complex 14:10:37 <pgearon> LeeF: last F2F, some people thought it would be a bad idea, some thought it should be included 14:10:38 <ivan> ivan has joined #sparql 14:10:59 <ivan> zakim, dial ivan-voip 14:10:59 <Zakim> ok, ivan; the call is being made 14:11:01 <Zakim> +Ivan 14:11:08 <pgearon> LeeF: AndyS as an editor said that semantically this is easy, because it fit into algebra easily 14:11:43 <pgearon> LeeF: several implementations now support LET 14:12:04 <AndyS> Yes - no new algebra - it can be defn'ed just by exposing (extend) used in SELECT expressions directly. 14:12:35 <chimezie> Zakim, unmute me 14:12:35 <Zakim> Chimezie_Ogbuji should no longer be muted 14:12:37 <pgearon> LeeF: anyone want to speak about it? 14:12:41 <SteveH> I think I've spoken about it enough - it looks too much like assignment 14:13:56 <pgearon> Chimezie_Ogbuji: is this just a column extension? AndyS: yes, that's one implementation 14:14:19 <pgearon> SteveH: LET looks like a procedural feature. 14:14:34 <pgearon> SteveH: functionality, it's just like a subselect 14:14:56 <pgearon> SteveH: if it's a popular feature, then why not? IT makes me feel a little queasy :-) 14:15:08 <pgearon> +q 14:15:17 <SteveH> if it wasn't called LET I'd be much less unhappy, FWIW 14:15:23 <pgearon> q- 14:15:23 <chimezie> I sympathize for the fact that it makes the language seem very procedural 14:15:48 <LeeF> pgearon: i think not having LET is a similar mistake to SPARQL 1.0 not having a negation operation 14:15:57 <LeeF> ... possible to do things, but complicated and confusing for users 14:16:14 <SteveH> there are good historical reasons why S1.0 had no MINUS 14:16:29 <sandro> +1 paul it's an important feature that'll be useful to a lot of people 14:17:10 <pgearon> LeeF: do we want to include this at all, before figuring out the details 14:17:55 <chimezie> s/sympathize for/sympathize with 14:17:55 <chimezie> pgearon: we would be making the same mistake as we did in not including MINUS originally in SPARQL 1.0 14:18:02 <pgearon> LeeF: any more comments before straw polling? 14:18:08 <pgearon> no response 14:19:34 <pgearon> LeeF: strawpoll - choice 1: do not include LET/assignment. choice 2: include LET as a keyword, choice 3: include with other keyword (such as BIND) 14:19:40 <LeeF> straw poll: (1) do not include LET/assignment ("do nothing"), (2) include with LET as keyword ("LET"), (3) include with other keyword ("other") 14:19:52 <SteveH> one advantage of a new keyword is that it wont break anyone's implementation 14:20:24 <ivan> 2 14:20:25 <kasei> I imagine 2.5 isn't a valid option? I don't have a preference on the keyword... 14:20:27 <chimezie> 3 14:20:29 <pgearon> 2 14:20:33 <MattPerry> 3 14:20:35 <OlivierCorby> 1 14:20:39 <SteveH> 1 or 3 14:20:41 <sandro> 3 14:20:45 <bglimm> 2 or 3 14:20:50 <AndyS> 2 14:20:54 <NicholasH> 3 14:20:58 <LeeF> 0 14:21:12 <sandro> wow, what an awkward split 14:21:13 <AndyS> Or no keyword : ?x := ?y+1 14:21:41 <pgearon> LeeF: strong feeling that we ought to do *something* (so not 1) 14:22:14 <pgearon> OlivierCorby: SPARQL is a graph matching language. Would prefer a simpler language with simpler principles 14:22:38 <pgearon> LeeF: this is similar to SteveH's concerns 14:22:40 <pgearon> +q 14:22:44 <LeeF> ack pgearon 14:23:20 <NicholasH> LET is very retro :) 14:23:25 <LeeF> q+ 14:23:27 <SteveH> and misleading 14:23:33 <LeeF> ack me 14:24:40 <pgearon> both AndyS and myself seem happy to break LET. In my case this is because LET is not any kind of a standard 14:25:42 <pgearon> chimezie: if we choose a different keyword, then wouldn't it have identical semantics to current implementations of LET? 14:26:03 <SteveH> pgearon, it doesn't seem like any 2 people have implemented it the same way 14:26:15 <SteveH> [sorry, misread] 14:26:16 <AndyS> The most common use of LET is to introduce a new var name. That's the case that matters to me (and, I believe, my users) 14:27:47 <pgearon> LeeF: for each of 3 options, asking if anyone feels strongly enough to register a formal objection if the group goes that way 14:28:45 <pgearon> AndyS: as a principle, I'm uncomfortable with this approach 14:28:56 <LeeF> poll: likely to object if the group decides not to include LET/assignment? 14:29:40 <LeeF> pgearon: potentially, not sure 14:29:43 <ivan> q+ 14:29:46 <chimezie> -1 (not likely to) 14:29:48 <LeeF> ack ivan 14:29:57 <pgearon> LeeF: you don't have to vote here. Just trying to figure out who is likely to object. Want to avoid objections in the future 14:30:15 <pgearon> Ivan: would TopQuadrant object if LET is not there? 14:30:33 <pgearon> LeeF: yes, TQ indicated that they would register a formal objection 14:30:52 <pgearon> Ivan: hope they do not require the keyword LET 14:30:58 <SteveH> the last time I spoke to jeremy abut this, he really wanted actual assignment, FWIW 14:31:03 <SteveH> that was some time ago though 14:31:11 <SteveH> not as in SPARQL LET 14:31:15 <pgearon> LeeF: don't know enough about TQ's formal position 14:31:32 <LeeF> poll: likely to object if the group decides to include the LET keyword (option 2 from above)? 14:31:37 <pgearon> LeeF: who's likely to object if the keyword LET is included? 14:31:51 <LeeF> (no comments) 14:31:52 <SteveH> I might, if I have time on my hands 14:31:56 <SteveH> so, not likely 14:32:15 <LeeF> poll: likely to object if the group decides to include the feature with a keyword other than LET (option 3 from above)? 14:32:33 <pgearon> resounding silence 14:33:02 <OlivierCorby> lambda x ... 14:33:19 <pgearon> LeeF: to repeat, this does NOT preclude anyone from objecting or not objecting in the future. (this was just an indication) 14:33:19 <SteveH> AndyS, not so, he cornered me for several hours at the east coast F2F 14:33:34 <SteveH> EXTEND? 14:33:39 <SteveH> ADD 14:33:47 <SteveH> EXTRA... 14:33:55 <SteveH> VALUE() 14:33:58 <SteveH> whatever really 14:34:00 <pgearon> I liked AndyS's no keyword idea (?x := expr) 14:34:06 <SteveH> erk! 14:34:07 <kasei> ☃ 14:34:26 <pgearon> LeeF: I'm glad the snowman came into it :-) 14:34:47 <pgearon> LeeF: support for each keyword? 14:34:54 <LeeF> LET 14:34:57 <SteveH> -1 14:35:01 <chimezie> -1 14:35:03 <pgearon> +1 14:35:07 <AndyS> +1 14:35:14 <NicholasH> -1 14:35:14 <OlivierCorby> -1 14:35:18 <MattPerry> -1 14:35:30 <LeeF> BIND 14:35:40 <SteveH> +0.5 14:35:47 <OlivierCorby> -1 14:35:49 <NicholasH> heh, yeah +0.5 14:35:49 <MattPerry> +1 14:35:54 <kasei> +1 14:35:55 <bglimm> +1 14:36:15 <LeeF> SET 14:36:17 <pgearon> +1 14:36:20 <OlivierCorby> -1 14:36:21 <SteveH> -1 14:36:23 <chimezie> -1 14:36:25 <MattPerry> -1 14:36:31 <NicholasH> -1 14:36:42 <LeeF> EXTEND 14:36:45 <SteveH> +1 14:36:45 <chimezie> +1 14:36:45 <pgearon> -1 14:36:46 <AndyS> -1 14:36:47 <OlivierCorby> -1 14:36:48 <ivan> -1 14:36:55 <NicholasH> 0 14:37:06 <LeeF> ADD 14:37:10 <AndyS> -1 14:37:11 <SteveH> -1 14:37:11 <pgearon> -1 14:37:12 <OlivierCorby> -1 14:37:13 <ivan> -1 14:37:13 <MattPerry> -1 14:37:13 <chimezie> -1 14:37:20 <NicholasH> -1 14:37:30 <LeeF> EXTRA 14:37:32 <pgearon> -1 14:37:33 <chimezie> -1 14:37:34 <AndyS> -1 14:37:36 <OlivierCorby> -1 14:37:36 <MattPerry> -1 14:37:37 <ivan> -1 14:37:40 <LeeF> VALUE 14:37:43 <chimezie> -1 14:37:45 <MattPerry> -1 14:37:45 <SteveH> +0.5 14:37:46 <OlivierCorby> -1 14:37:47 <AndyS> -1 14:37:49 <pgearon> -1 14:37:54 <NicholasH> What about MAP? 14:38:01 <LeeF> MAP 14:38:04 <pgearon> -1 14:38:06 <SteveH> +1 14:38:09 <AndyS> -1 14:38:10 <OlivierCorby> -1 14:38:12 <MattPerry> -1 14:38:13 <chimezie> +1 14:38:13 <NicholasH> +1 14:38:29 <LeeF> NO KEYWORD, JUST A := OPERATOR 14:38:31 <SteveH> -1 14:38:33 <chimezie> -1 14:38:35 <NicholasH> -1 14:38:36 <pgearon> +1 14:38:36 <OlivierCorby> -1 14:38:36 <MattPerry> -1 14:38:38 <AndyS> +0.5 14:38:57 <sandro> [ please put me down as a +1 for BIND ] 14:39:37 <pgearon> SteveH: noted that we haven't discussed the semantics at all 14:40:16 <SteveH> aliases v's adding another "column" might sway people one way or the other 14:40:29 <SteveH> but it's probably not important 14:40:53 <pgearon> LeeF: hoping to get through whether or not to include the feature before taking on the issues of semantics. Hoping that there aren't any semantics issues that we need to be considering just yet 14:40:54 <SteveH> can we hear from people who didn't like BIND? maybe there's something we've not thought of 14:41:05 <SteveH> it's only been on the table for 20 mins 14:42:03 <SteveH> +1 to concerns over scoping rules 14:42:11 <pgearon> Olivier: doesn't like the feature at all. But if it were included would like LET 14:42:28 <pgearon> s/Olivier/OlivierCorby/ 14:42:47 <pgearon> OlivierCorby: bind is already complex. This would make it moreso 14:42:59 <LeeF> PROPOSED: SPARQL 1.1 Query includes the http://www.w3.org/2009/sparql/wiki/Feature:Assignment feature using a keyword of BIND 14:43:00 <pgearon> s/bind/binding/ 14:43:42 <SteveH> the wording of Feature:Assignment is exactly the kind of thing that's logically nasty 14:43:50 <SteveH> it doesn't really speak to extending the binding table 14:43:56 <chimezie> The more I think about it, it seems that not settling on the semantics before we conclude on a keyword might be dangerous 14:44:04 <OlivierCorby> I mean scoping rules for variables are already complex with different rules for pattern, exists, minus and now let 14:44:18 <pgearon> LeeF: would like to put out a question on this on the mailing list, get comments, and address it next week. Unfortunately, this often results in silence on the mailing list 14:45:42 <AndyS> q+ 14:45:45 <pgearon> LeeF: previously suggested test cases were more about contrasting subselect vs. LET 14:45:45 <LeeF> ack AndyS 14:46:08 <pgearon> AndyS: why don't we get each of the implementors to write up what they've done 14:47:11 <pgearon> LeeF: anyone want to take the lead in trying to create test cases? 14:47:12 <AndyS> I have test cases. I can sort through them to pull out the core ones 14:47:57 <SteveH> I'd be more interested in seeing descriptions of the algebraic operations than testcases 14:47:57 <pgearon> LeeF: get AndyS, pgearon, and LeeF to write up what we do in our implementations 14:48:49 <pgearon> LeeF: leaving the LET/assignment topic for today 14:48:50 <AndyS> http://www.w3.org/2009/sparql/docs/query-1.1/rq25.xml#defn_extend 14:48:51 <LeeF> ACTION: Lee to write-up Glitter treatment of LET 14:48:51 <trackbot> Created ACTION-277 - Write-up Glitter treatment of LET [on Lee Feigenbaum - due 2010-07-13]. 14:49:00 <LeeF> ACTION: Andy to write-up ARQ treatment of LET 14:49:00 <trackbot> Created ACTION-278 - Write-up ARQ treatment of LET [on Andy Seaborne - due 2010-07-13]. 14:49:06 <LeeF> ACTION: Paul to write-up Mulgara treatment of LET 14:49:06 <trackbot> Created ACTION-279 - Write-up Mulgara treatment of LET [on Paul Gearon - due 2010-07-13]. 14:49:50 <pgearon> LeeF: not going to discuss test cases now, as we only have 10 minutes left 14:50:06 <pgearon> LeeF: continue test case discussion on the mailing list 14:50:39 <pgearon> LeeF: want to discuss and hopefully approve test cases next week. Please get familiar with them before the next meeting 14:51:16 <pgearon> AndyS: what are we approving exactly? The last time around this would handled differently with CVS 14:51:57 <pgearon> LeeF: hoping to approving them modulo getting the syntax right, but if that's not good enough then happy to do it more formally 14:52:15 <pgearon> +q 14:52:19 <LeeF> ack pgearon 14:53:59 <pgearon> pgearon: concerned that not having all tests syntactically correct, and checked into CVS in their final form, might be OK in the early stages, but could lead to a procedural mess towards the end 14:54:35 <pgearon> LeeF: OK, let's look at getting them checked into CVS in their final form before approving 14:54:43 <pgearon> LeeF: other business? 14:54:58 <SteveH> bye all 14:54:59 <Zakim> -Chimezie_Ogbuji 14:55:00 <ivan> zakim, drop me 14:55:00 <Zakim> -Lee_Feigenbaum 14:55:00 <Zakim> Ivan is being disconnected 14:55:01 <MattPerry> bye 14:55:01 <Zakim> -bglimm 14:55:01 <Zakim> -Ivan 14:55:04 <Zakim> -OlivierCorby 14:55:05 <NicholasH> bye bye! 14:55:05 <Zakim> -SteveH 14:55:06 <Zakim> -MattPerry 14:55:08 <Zakim> -NicholasH 14:55:11 <LeeF> paul, thanks very much for scribing again 14:55:13 <Zakim> -Sandro 14:55:15 <pgearon> np 14:55:19 <LeeF> RRSAgent, make logs world 14:55:39 <Zakim> -AndyS 14:55:49 <Zakim> -kasei 14:55:51 <Zakim> -pgearon 14:55:51 <Zakim> SW_(SPARQL)10:00AM has ended 14:55:51 <NicholasH> I don't mind doing some scribing 14:55:52 <Zakim> Attendees were Lee_Feigenbaum, AndyS, SteveH, kasei, pgearon, MattPerry, +22.214.171.124.aaaa, OlivierCorby, Sandro, bglimm, NicholasH, Chimezie_Ogbuji, Ivan 14:56:11 <NicholasH> but I don't recognise everyone's voices at the moment # SPECIAL MARKER FOR CHATSYNC. DO NOT EDIT THIS LINE OR BELOW. SRCLINESUSED=00000330