Warning:
This wiki has been archived and is now read-only.
Chatlog 2010-09-28
From SPARQL Working Group
See original RRSAgent log and preview nicely formatted version.
Please justify/explain all edits to this page, in your "edit summary" text.
13:58:55 <RRSAgent> RRSAgent has joined #sparql 13:58:55 <RRSAgent> logging to http://www.w3.org/2010/09/28-sparql-irc 13:58:57 <trackbot> RRSAgent, make logs world 13:58:57 <Zakim> Zakim has joined #sparql 13:58:59 <trackbot> Zakim, this will be 77277 13:58:59 <Zakim> ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start in 2 minutes 13:59:00 <LeeF> zakim, this will be SPARQL 13:59:00 <trackbot> Meeting: SPARQL Working Group Teleconference 13:59:01 <trackbot> Date: 28 September 2010 13:59:02 <Zakim> ok, LeeF, I see SW_(SPARQL)10:00AM already started 13:59:07 <Zakim> +MattPerry 13:59:40 <Zakim> +LeeF 13:59:40 <Zakim> + +1.310.729.aaaa 13:59:40 <bglimm> All circuits are busy now for me too :-( 13:59:49 <SteveH_> yup :( 13:59:53 <kasei> Zakim, aaaa is me 13:59:53 <Zakim> +kasei; got it 14:00:16 <LeeF> I haven't seen any W3C notices, which doesn't really mean much. 14:00:26 <Zakim> +pgearon 14:00:34 <Zakim> +??P26 14:00:35 <LeeF> Regrets: Axel, Chime, Alex 14:00:41 <AndyS> zakim, ??P26 is me 14:00:41 <Zakim> +AndyS; got it 14:01:03 <Zakim> +Sandro 14:01:13 <LeeF> Scribenick: MattPerry 14:01:21 <Zakim> +[IPcaller] 14:01:28 <SteveH_> Zakim, [IPcaller] is me 14:01:28 <Zakim> +SteveH_; got it 14:01:30 <Zakim> +bglimm 14:01:38 <Souri> Souri has joined #sparql 14:01:45 <bglimm> Zakim, mute me 14:01:53 <Zakim> bglimm should now be muted 14:01:56 <Zakim> +??P30 14:02:18 <Zakim> +Souri 14:02:33 <NickH> Zakim, ??P30 is me 14:02:33 <Zakim> +NickH; got it 14:04:05 <Zakim> +Garlik 14:04:10 <Zakim> -SteveH_ 14:04:14 <MattPerry> LeeF: want to go through documents to see status for next round of publication 14:04:18 <SteveH> Zakim, Garlik is temporarily me 14:04:18 <Zakim> +SteveH; got it 14:04:29 <OlivierCorby> Not able to join by tel, system refuses ... 14:04:46 <MattPerry> LeeF: not in a position to go over Let/Bind this week but maybe next week 14:04:47 <LeeF> Agenda: http://www.w3.org/2009/sparql/wiki/Agenda-2010-09-28 14:04:55 <LeeF> PROPOSED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-09-14 14:05:12 <Zakim> +OlivierCorby 14:05:19 <OlivierCorby> Hi 14:05:37 <LeeF> RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-09-14 14:05:57 <MattPerry> topic: query document reviews 14:06:37 <MattPerry> LeeF: issues: BINDING keyword and isNumeric function and errors in aggregates 14:06:48 <LeeF> subtopic: semantics of BINDINGS 14:06:58 <LeeF> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0370.html 14:07:35 <LeeF> """ 14:07:36 <LeeF> for each row: 14:07:36 <LeeF> substitute all occurrences of named variables in the query 14:07:36 <LeeF> execute modified query 14:07:37 <LeeF> """ 14:08:42 <LeeF> """ 14:08:43 <LeeF> SELECT * 14:08:43 <LeeF> { ?s ?p ?o . FILTER ( ?o < ?v ) } 14:08:43 <LeeF> BINDINGS ?v { (1) (2) } 14:08:44 <LeeF> """ 14:09:26 <LeeF> this bindings list has two rows, each with one value 14:09:29 <MattPerry> AndyS: for each binding set, substitute into query and then execute query 14:10:09 <MattPerry> AndyS: in federated doc, turn binding into table and then join with query 14:10:33 <SteveH> q+ 14:10:36 <kasei> is this a better example for this discussion (whether the results have any variables bound)?: select * where {} bindings ?s { (1) (2) } 14:10:47 <LeeF> ack SteveH 14:11:18 <MattPerry> SteveH: our implementation works more like the join description 14:12:09 <LeeF> Not so sure if that's a good example, Greg - wouldn't that turn the same way either way? 14:12:13 <MattPerry> SteveH: for Greg's example we would get ?s bound to 1 and 2 14:12:23 <AndyS> kasei, independent design point - can make ?v visible of * or not for either design 14:13:01 <OlivierCorby> What about variables in minus ? 14:13:30 <kasei> I share SteveH's concern about optimization. 14:13:56 <ericP> ericP has joined #sparql 14:14:31 <ericP> Zakim, please dial ericP-office 14:14:31 <Zakim> ok, ericP; the call is being made 14:14:32 <Zakim> +EricP 14:15:06 <MattPerry> AndyS: this definition is an improvement in federated query 14:15:09 <OlivierCorby> q+ 14:15:21 <LeeF> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0370.html 14:16:44 <LeeF> ack OlivierCorby 14:16:52 <MattPerry> LeeF: issue is that some variables, e.g. those in filter, can't get their values from a join 14:17:31 <AndyS> q+ 14:17:35 <MattPerry> Olivier: what about MINUS, can it be a substitution? 14:17:40 <LeeF> ack AndyS 14:17:45 <ericP> q+ 14:17:57 <MattPerry> AndyS: I would think that substitution would work for MINUS 14:18:12 <SteveH> I don't like the approach of conflating parameterised queries and pre-binding, they're different 14:18:27 <LeeF> ack ericP 14:19:19 <MattPerry> ericP: biggest issue is the FILTER issue 14:21:17 <SteveH> q+ to ask why bindings goes at the "end" 14:21:57 <MattPerry> ... I would go with the same semantics we used with OPTIONAL, the FILTER works on the previous set of variables 14:22:07 <LeeF> ack SteveH 14:22:07 <Zakim> SteveH, you wanted to ask why bindings goes at the "end" 14:23:23 <AndyS> q+ 14:23:28 <MattPerry> ericP: BINDINGS go at the end because you may want to stream the result, and this is only possible if the query is done before you get the bindings 14:24:52 <LeeF> ack AndyS 14:25:35 <MattPerry> AndyS: if you see the query first, you can do a "prepare" on the query 14:26:41 <MattPerry> ericP: optional has a left pattern, right pattern and filter ... the filter works on the left and right pattern 14:27:03 <kasei> I don't think people would ever figure that out. 14:27:05 <MattPerry> ... filters now have a left and right where right comes from the BINDINGS 14:28:29 <MattPerry> ericP: I think that if people are given a mandate for substitution, they will still figure out a way to do it with joins but this may be difficult for people 14:28:44 <SteveH> +1 to ericP 14:29:22 <MattPerry> AndyS: substitution is not a new operation in the algebra, but a new join is a new operation 14:29:31 <SteveH> that's the problem 14:30:09 <MattPerry> SteveH: the problem is implementing a substitution operation 14:31:18 <MattPerry> AndyS: I think the FILTER case is the important one 14:31:45 <MattPerry> LeeF: is the working group motivated to make the FILTER case work 14:31:48 <AndyS> q+ 14:32:27 <kasei> q+ 14:32:29 <LeeF> ack AndyS 14:32:30 <ericP> i'm happy to leave it out for now 14:32:35 <ericP> but i'm not the alpha team contact 14:33:10 <MattPerry> SteveH: BINDINGS keyword does not imply substitution to me 14:33:15 <LeeF> ack, kasei 14:33:31 <LeeF> ack kasei 14:34:11 <MattPerry> kasei: I agree with SteveH on most of the issues ... BINDINGS imply a join to me 14:34:50 <SteveH> well put kasei 14:34:57 <MattPerry> ... BINDINGS should give same execution but just a subset of the results ... substitution could give different results 14:35:06 <pgearon> +1 to kasei's POV 14:36:11 <LeeF> straw poll: should the values in rows in the BINDINGS clause be available within FILTERs etc. within the query a la AndyS's proposal? 14:36:35 <SteveH> that statement is not correct 14:37:34 <MattPerry> SteveH: this is about what happens when you have a variable in Filter but not in the graph pattern 14:38:05 <LeeF> straw poll: should the BINDINGS clause have Join or Substitution semantics? 14:38:21 <kasei> Join 14:38:23 <SteveH> Join 14:38:28 <pgearon> join 14:38:28 <ericP> join 14:38:29 <NickH> Join 14:38:31 <MattPerry> join 14:38:40 <bglimm> 0 (not enough understanding of the issues) 14:38:46 <Souri> +1 to Andy's substitution proposal (if I understand it correctly, substitution of the values from each binding row and then querying would return some results) 14:38:47 <OlivierCorby> 0 too 14:39:08 <LeeF> 0 14:39:12 <ericP> 7 14:39:13 <AndyS> filter visibility 14:39:14 <sandro> 0 14:40:00 <SteveH> q+ to ask about templating 14:40:07 <LeeF> ack SteveH 14:40:07 <Zakim> SteveH, you wanted to ask about templating 14:40:27 <LeeF> The group has general consensus to keep the semantics of BINDINGS as is for now. 14:40:27 <MattPerry> SteveH: is templating on our to-do list? 14:40:45 <MattPerry> LeeF: templating missed the cut for 1.1 14:41:11 <MattPerry> subtopic: isNumeric 14:41:16 <LeeF> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0410.html 14:41:17 <SteveH> +1 to isNumeric() 14:42:13 <LeeF> SUM(?x) 14:42:13 <MattPerry> SteveH: isNumeric is very useful for numeric operations due to the weak typing 14:42:47 <MattPerry> SteveH: right now, we go to xpath numeric-add operation 14:43:02 <LeeF> SUM(IF(isNumeric(?x), ?x, 0)) 14:43:02 <MattPerry> ... a date, for example, will give a type error 14:43:18 <AndyS> sum("a") is legal :-) 14:43:28 <SteveH> AndyS, yeah, semi-deliberate :) 14:43:31 <LeeF> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0460.html 14:44:14 <LeeF> PROPOSED: Include an isNumeric function in SPARQL 1.1 14:44:27 <AndyS> seconded 14:44:45 <LeeF> RESOLVED: Include an isNumeric function in SPARQL 1.1 14:45:13 <AndyS> SteveH, I implemented the more obvious way 14:45:29 <SteveH> AndyS, which is that? :) 14:45:31 <MattPerry> topic: go through document reviews 14:45:40 <AndyS> sum("a") -> error 14:45:48 <MattPerry> subtopic: query 14:45:54 <SteveH> ah, right, that's what I'm going to write in when I have time 14:46:06 <AndyS> like sum("a"+0) or sum("a", 0) -> error 14:46:12 <SteveH> right 14:46:20 <ericP> q+ to ask if federation is rolled in 14:46:48 <LeeF> ack, ericP 14:46:51 <LeeF> ack ericP 14:46:51 <Zakim> ericP, you wanted to ask if federation is rolled in 14:47:13 <MattPerry> ericP: who is editor of federation? 14:48:45 <MattPerry> ericP: I can take a look at the comments for federation 14:49:21 <MattPerry> I fine with it 14:49:24 <kasei> me too 14:49:28 <MattPerry> s/I/I'm 14:49:33 <SteveH> actually, I'd like to make a couple of edits, based on reviews 14:49:38 <SteveH> if that wont gum up the process 14:49:41 <SteveH> otherwise its fine 14:50:45 <MattPerry> SteveH: some non-algebraic changes I would like to make 14:51:23 <MattPerry> LeeF: should be fine if you do it by Friday 14:51:47 <LeeF> PROPOSED: Publish http://www.w3.org/2009/sparql/docs/query-1.1/rq25.xml + Steve's wording changes in response to Greg's comments as Working Draft 14:52:37 <LeeF> PROPOSED: Publish http://www.w3.org/2009/sparql/docs/query-1.1/rq25.xml + editorial changes in response to Greg's comments as Working Draft 14:52:45 <AndyS> 0.5 14:52:50 <SteveH> 1 14:53:43 <MattPerry> AndyS: would like to process all the comments first 14:54:30 <SteveH> I'm also not swapped in on that 14:54:32 <LeeF> http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0454.html 14:55:12 <MattPerry> LeeF: lets hold off on publishing for now and try to go through all docs next week 14:55:44 <AndyS> q+ to ask about isNumeric (get in if possible?) 14:56:18 <SteveH> and fix sum("a") on same basis 14:56:18 <LeeF> ack AndyS 14:56:18 <Zakim> AndyS, you wanted to ask about isNumeric (get in if possible?) 14:56:20 <MattPerry> AndyS: I can try to put in isNumeric 14:57:14 <bglimm> Zakim, unmute me 14:57:14 <Zakim> bglimm should no longer be muted 14:57:48 <MattPerry> kasei: will summarize what changes went into service description 14:58:08 <MattPerry> bglimm: service description is ready from my point of view 14:58:20 <MattPerry> ... entailment is awaiting LeeF's review 14:58:45 <MattPerry> Olivier: entilment is ok with me 14:58:54 <MattPerry> s/entilment/entailment 14:59:02 <SteveH> bye all 14:59:03 <MattPerry> bye 14:59:06 <Zakim> -LeeF 14:59:08 <Zakim> -pgearon 14:59:08 <NickH> bye! 14:59:10 <Zakim> -SteveH 14:59:11 <LeeF> Matt, thanks very much for scribingg 14:59:12 <Zakim> -Sandro 14:59:14 <Zakim> -bglimm 14:59:16 <Zakim> -kasei 14:59:18 <Zakim> -MattPerry 14:59:20 <Zakim> -EricP 14:59:22 <Zakim> -NickH 14:59:24 <Zakim> -Souri 14:59:30 <Zakim> -OlivierCorby 14:59:32 <Zakim> -AndyS 14:59:36 <Zakim> SW_(SPARQL)10:00AM has ended 14:59:38 <Zakim> Attendees were MattPerry, LeeF, +1.310.729.aaaa, kasei, pgearon, AndyS, Sandro, SteveH_, bglimm, Souri, NickH, SteveH, OlivierCorby, EricP 15:03:19 <OlivierCorby> OlivierCorby has left #sparql 17:05:23 <Zakim> Zakim has left #sparql 17:33:19 <AndyS> Defining what it means to be numeric. Reusing some of ericP's fine words. 17:51:34 <AndyS> isNumeric added to rq25. Note: the lexical form of the term must be valid ("1220"^^xsd:byte is not) to make it consistent with numeric operations. And RDF-MT. 19:34:05 <AndyS> AndyS has joined #sparql 20:00:19 <SteveH> SteveH has joined #sparql 21:43:53 <LeeF> LeeF has joined #sparql 22:00:36 <karl> karl has joined #sparql # SPECIAL MARKER FOR CHATSYNC. DO NOT EDIT THIS LINE OR BELOW. SRCLINESUSED=00000226