IRC log of sparql on 2010-11-02
Timestamps are in UTC.
- 13:53:34 [RRSAgent]
- RRSAgent has joined #sparql
- 13:53:34 [RRSAgent]
- logging to http://www.w3.org/2010/11/02-sparql-irc
- 13:53:36 [trackbot]
- RRSAgent, make logs world
- 13:53:38 [trackbot]
- Zakim, this will be 77277
- 13:53:38 [Zakim]
- ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start in 7 minutes
- 13:53:39 [trackbot]
- Meeting: SPARQL Working Group Teleconference
- 13:53:39 [trackbot]
- Date: 02 November 2010
- 13:53:41 [LeeF]
- zakim, this will be SPARQL
- 13:53:41 [Zakim]
- ok, LeeF; I see SW_(SPARQL)10:00AM scheduled to start in 7 minutes
- 13:58:06 [Zakim]
- SW_(SPARQL)10:00AM has now started
- 13:58:13 [Zakim]
- + +1.617.245.aaaa
- 13:58:20 [LeeF]
- AlexPassant, AndyS, AxelPolleres, bglimm, ivan, NicoM -- the call starts in 5 minutes today :)
- 13:58:26 [LeeF]
- zakim, aaaa is me
- 13:58:26 [Zakim]
- +LeeF; got it
- 13:58:36 [LeeF]
- Chair: Lee Feigenbaum
- 13:58:38 [Zakim]
- +??P17
- 13:58:45 [AndyS]
- zakim, ??P17 is me
- 13:58:45 [Zakim]
- +AndyS; got it
- 13:59:01 [ivan]
- zakim, dial ivan-voip
- 13:59:01 [Zakim]
- ok, ivan; the call is being made
- 13:59:02 [Zakim]
- +Ivan
- 13:59:07 [LeeF]
- Regrets: Carlos, Souri, Sandro
- 13:59:10 [Zakim]
- +NicoM
- 13:59:23 [LeeF]
- Agenda: http://www.w3.org/2009/sparql/wiki/Agenda-2010-11-02
- 13:59:43 [Zakim]
- +bglimm
- 13:59:59 [kasei]
- hrm. cambridge zakim just dropped my call.
- 14:00:11 [LeeF]
- zakim, please be nicer to kasei
- 14:00:11 [Zakim]
- I don't understand 'please be nicer to kasei', LeeF
- 14:00:24 [bglimm]
- Zakim, mute me
- 14:00:24 [Zakim]
- bglimm should now be muted
- 14:00:27 [Zakim]
- + +1.310.729.aabb
- 14:00:37 [kasei]
- Zakim, aabb is me
- 14:00:37 [Zakim]
- +kasei; got it
- 14:00:38 [LeeF]
- zakim, thanks
- 14:00:38 [Zakim]
- you are very welcome, LeeF
- 14:01:35 [kasei]
- Zakim, mute me
- 14:01:35 [Zakim]
- kasei should now be muted
- 14:01:40 [Zakim]
- +MattPerry
- 14:02:42 [LeeF]
- zakim, who's on the phone?
- 14:02:42 [Zakim]
- On the phone I see LeeF, AndyS, Ivan, NicoM, bglimm (muted), kasei (muted), MattPerry
- 14:03:07 [LeeF]
- topic: Admin
- 14:03:16 [LeeF]
- PROPOSED: Approve minutes from http://www.w3.org/2009/sparql/meeting/2010-10-26
- 14:03:23 [Zakim]
- + +3539149aacc
- 14:03:39 [AlexPassant]
- Zakim, +3539149aacc is me
- 14:03:39 [Zakim]
- +AlexPassant; got it
- 14:04:15 [LeeF]
- Scribenick: AlexPassant
- 14:04:22 [Zakim]
- +pgearon
- 14:05:45 [Zakim]
- +[IPcaller]
- 14:06:17 [MattPerry]
- MattPerry has joined #sparql
- 14:06:57 [pgearon]
- Hi Lee
- 14:07:34 [LeeF]
- zakim, IPcaller is SteveH
- 14:07:34 [Zakim]
- +SteveH; got it
- 14:07:43 [LeeF]
- zakim, who's on the phone?
- 14:07:43 [Zakim]
- On the phone I see LeeF, AndyS, Ivan, NicoM, bglimm (muted), kasei (muted), MattPerry, AlexPassant, pgearon, SteveH
- 14:07:56 [SteveH]
- SteveH has joined #sparql
- 14:08:11 [LeeF]
- RESOLVED: Approve minutes at http://www.w3.org/2009/sparql/meeting/2010-10-26
- 14:08:25 [LeeF]
- Next regular meeting: 2010-11-09 @ 15:00 UK / 10:00 EST (scribe: Chime Ogbuji)
- 14:08:30 [SteveH]
- Zakim, who's on the phone?
- 14:08:30 [Zakim]
- On the phone I see LeeF, AndyS, Ivan, NicoM, bglimm (muted), kasei (muted), MattPerry, AlexPassant, pgearon, SteveH
- 14:08:37 [AlexPassant]
- LeeF: Any changes on last week minutes ?
- 14:08:39 [bglimm]
- Next week is ISWC
- 14:08:50 [AlexPassant]
- ... some people at ISWC next week
- 14:08:54 [ivan]
- I will be at iswc
- 14:08:55 [SteveH]
- will be at ISWC
- 14:08:56 [bglimm]
- I'll be there
- 14:08:58 [AlexPassant]
- ... but meeting back at normal time
- 14:09:00 [AlexPassant]
- I'll be at ISWC
- 14:09:39 [AlexPassant]
- ... ISWC attendees will probably won't join
- 14:10:02 [AlexPassant]
- ... but regular call with people that can attend
- 14:11:04 [AlexPassant]
- ... agenda for next week: shortcuts for updates
- 14:11:17 [AlexPassant]
- ... one of the full remaining thing to formalise grammar in update / query
- 14:11:29 [AndyS]
- yes - getting the grammar sorted would help me.
- 14:12:54 [NicoM]
- LeeF, got a glass roof?
- 14:13:13 [AlexPassant]
- topic: SELECT * behavior
- 14:14:01 [AlexPassant]
- LeeF: need to agree on the select * behavior
- 14:14:20 [AlexPassant]
- ... in-scope variables (bound variables)
- 14:14:25 [AlexPassant]
- ... consensus around that
- 14:14:43 [AlexPassant]
- ... shouldn't select variables inside the subquery
- 14:14:58 [LeeF]
- seems that SELECT * { { SELECT ?x { ... ?x ?y ... } } } just means ?x
- 14:15:19 [AndyS]
- agree
- 14:15:20 [pgearon]
- +1 absolutely
- 14:15:22 [bglimm]
- +1
- 14:15:31 [NicoM]
- +1
- 14:15:31 [ivan]
- 1
- 14:16:15 [AlexPassant]
- ... what about variables that are hidden as being part of aggregate query but not in a GROUP BY
- 14:16:22 [LeeF]
- SELECT * { ... ?x ?y ... } GROUP BY ?x
- 14:16:43 [SteveH]
- that's the same as SELECT DISTINCT ?x { ... ?x ?y ... }
- 14:17:21 [LeeF]
- 1/ * stands for ?x and ?y, so this query is an error because you can't select ?y
- 14:17:42 [LeeF]
- 2/ * stands for only the variables that are legally select'able at this point - which means just ?x because of the aggregating
- 14:17:49 [SteveH]
- ... if you limit it to just "bindable" ones
- 14:18:28 [SteveH]
- I think we're in a position to descide
- 14:18:32 [AlexPassant]
- ... is there any option besides these 2 ones ?
- 14:18:32 [AlexPassant]
- ... if not, can we decide between both ?
- 14:18:41 [SteveH]
- q+
- 14:19:21 [LeeF]
- ack SteveH
- 14:19:31 [pgearon]
- AndyS: not in my case, but it wouldn't be too hard to use that machinery instead
- 14:20:08 [Zakim]
- +AxelPolleres
- 14:20:30 [AndyS]
- I find that it is useful
- 14:20:47 [SteveH]
- where?
- 14:20:49 [kasei]
- I think it's quite useful
- 14:21:16 [SteveH]
- what about the semantic conflict with COUNT(*)
- 14:21:19 [bglimm]
- Are we discussing whther star is useful at all? Yes, but sound is quite bad
- 14:21:34 [kasei]
- 2
- 14:21:51 [SteveH]
- in 4store it's all
- 14:21:56 [SteveH]
- 1) I think
- 14:21:58 [AlexPassant]
- LeeF: any implementation doing * and aggregate ?
- 14:22:25 [kasei]
- I think it would be rather odd to design the language such that using * is guaranteed to produce an error in lots of queries.
- 14:22:28 [AxelPolleres]
- Zakim, who is on the phone?
- 14:22:28 [Zakim]
- On the phone I see LeeF, AndyS, Ivan, NicoM, bglimm (muted), kasei (muted), MattPerry, AlexPassant, pgearon, SteveH, AxelPolleres
- 14:22:44 [SteveH]
- kasei, well SQL works that way, it's never bothered me
- 14:23:30 [AlexPassant]
- ... strawpoll
- 14:23:35 [LeeF]
- straw poll: 1 "all", 2 "legal", 0 "don't really care too much"
- 14:23:39 [bglimm]
- 1
- 14:23:39 [SteveH]
- 1
- 14:23:40 [kasei]
- 2
- 14:23:42 [ivan]
- 0
- 14:23:43 [AndyS]
- 2
- 14:23:44 [LeeF]
- 0
- 14:23:44 [MattPerry]
- 1
- 14:23:46 [pgearon]
- 0
- 14:23:46 [AlexPassant]
- 0
- 14:23:47 [NicoM]
- 0
- 14:23:48 [AxelPolleres]
- 2 (mildly)
- 14:24:48 [AlexPassant]
- LeeF: which decision will be easier to change if we've done the wrong one
- 14:24:55 [AlexPassant]
- ... or if people implementing the other way
- 14:24:58 [OlivierCorby]
- OlivierCorby has joined #sparql
- 14:25:01 [AlexPassant]
- ... any thoughts ?
- 14:25:03 [NicoM]
- legal can be extended later :-9
- 14:25:16 [SteveH]
- AndyS, I just meant the dramatically different meanings
- 14:25:34 [pgearon]
- 2 makes more sense to me if I were a user. 1 will be easier as an implementor. I can live with either
- 14:25:45 [AxelPolleres]
- I would find different meaning of * in COUNT and SELECT weird... that's one point for 2
- 14:25:48 [AxelPolleres]
- or no?
- 14:25:50 [AlexPassant]
- ... slight personal preference for 1
- 14:26:06 [SteveH]
- AxelPolleres, no, because then COUNT(*) and SLEECT(*) would bare no resemblance
- 14:26:19 [SteveH]
- er, SELECT *
- 14:26:35 [AndyS]
- SteveH, are you proposing no SELECT * at all?
- 14:26:39 [AlexPassant]
- ... if no new suggestion, tempted to say that we'll do 1
- 14:26:44 [AlexPassant]
- ... but happy to hear additional suggestions
- 14:26:49 [SteveH]
- AndyS, no, just that it would mean all in-scope variables
- 14:27:04 [SteveH]
- no just the ones that can legally be used in scalar expressions
- 14:27:08 [SteveH]
- *not...
- 14:27:46 [SteveH]
- I can write SELECT COUNT(?a) WHERE { ... } GROUP BY ?b, so ?a is in scopew
- 14:28:58 [SteveH]
- I can write SELECT (COUNT(?a) AS ?ca) WHERE { ... } GROUP BY ?b, so ?a is in scopew
- 14:29:02 [kasei]
- this is where the "potentially bound" terminology might have been slightly clearer than "in scope"
- 14:29:06 [MattPerry]
- what about: SELECT * WHERE { ... } GROUP BY (?a + ?b / 2)
- 14:29:15 [Zakim]
- + +34.92.38.aadd
- 14:29:41 [OlivierCorby]
- Zakim, aadd is me
- 14:29:41 [Zakim]
- +OlivierCorby; got it
- 14:29:52 [OlivierCorby]
- A little bit late ...
- 14:30:11 [AxelPolleres]
- matt, according to 1) that would be forbidden, yes?
- 14:30:23 [SteveH]
- AxelPolleres, no, it would just be an error
- 14:30:30 [AxelPolleres]
- according to 2) that would be like ASK?
- 14:30:57 [AlexPassant]
- AndyS: problem of semantic equivalence of the 2 expressions
- 14:31:06 [AxelPolleres]
- steveH, what's the diff between "an error" and "forbidden" for you here? ;-)
- 14:31:27 [AlexPassant]
- LeeF: anyone change his mind ?
- 14:31:40 [SteveH]
- would consider objecting
- 14:31:50 [bglimm]
- I really prefer 1, but wouldn't formally object
- 14:31:51 [AlexPassant]
- ... anybody feels strongly about this and would raise objection ?
- 14:31:51 [kasei]
- I'm going to be very unhappy, but probably won't object.
- 14:32:05 [SteveH]
- I might be less upset if anyone had a real use?
- 14:32:20 [SteveH]
- just seems like pointless syntax messing
- 14:32:25 [LeeF]
- No consensus.
- 14:32:35 [kasei]
- SteveH, same as SELECT * in 1.0 -- it's a really useful shortcut.
- 14:32:54 [SteveH]
- kasei, but it's not useful in this case, it's just longhand for SELECT DISTINCT
- 14:32:59 [pgearon]
- I'm happy with 1. As for using it to see what variables are in scope, then using 1 can give you errors to say what *isn't* in scope
- 14:33:46 [LeeF]
- Given the lack of consensus, the chair decides towards option 1 based on potential objections and the chair's belief that option 1 leaves more options open in the future and leads to more interoperability if all implementations do not implement this the same way.
- 14:33:47 [kasei]
- oh, I was under the impression that it could be combined with project exprs also...
- 14:34:01 [SteveH]
- option 1 also gives us to possibility of MAX(*) etc. in the future
- 14:34:06 [AndyS]
- I object.
- 14:34:43 [AlexPassant]
- AndyS: is there any list of objections ?
- 14:34:48 [AlexPassant]
- LeeF: no formal objections so far
- 14:35:43 [LeeF]
- LeeF: Please mail any formal objections to the mailing list so that we have an official record and URI for them. Thanks.
- 14:35:52 [AxelPolleres]
- We still have an open issue which is a bit related (that was the second place where we'd discussed about the definition of "inscope" or "potentially bound"... "The new variable is introduced using the keyword AS; it must not already be potentially bound."
- 14:36:10 [LeeF]
- topic: BIND and FILTER order execution
- 14:36:15 [AlexPassant]
- topic: BIND + FILTER
- 14:36:21 [LeeF]
- http://lists.w3.org/Archives/Public/public-rdf-dawg/2010OctDec/0154.html
- 14:36:33 [AxelPolleres]
- q+ to mention still open issue on prev. topic
- 14:36:42 [LeeF]
- ack AxelPolleres
- 14:36:42 [Zakim]
- AxelPolleres, you wanted to mention still open issue on prev. topic
- 14:37:24 [AndyS]
- What is the issue number?
- 14:38:43 [AxelPolleres]
- no issue number assigned yet, left open in disussion of last time... will ask again in the end whether we should open an issue.
- 14:38:53 [LeeF]
- PREFIX : <http://example.org/>
- 14:38:53 [LeeF]
- SELECT ?s ?p ?o ?z
- 14:38:53 [LeeF]
- {
- 14:38:53 [LeeF]
- ?s ?p ?o .
- 14:38:53 [LeeF]
- FILTER(?z = 3 )
- 14:38:53 [LeeF]
- BIND(?o+1 AS ?z)
- 14:38:55 [LeeF]
- }
- 14:40:01 [AlexPassant]
- AndyS: implementaion acts differently
- 14:40:24 [AlexPassant]
- ... BIND does not have the semantics of LET
- 14:40:34 [AlexPassant]
- ... so it will not act as a filter
- 14:40:43 [AlexPassant]
- ... no error if you assign to an existing variable
- 14:41:51 [kasei]
- LeeF, do you preserve the lexical ordering amongst the BINDs when you "shove" them to the end of the BGP?
- 14:42:54 [kasei]
- q+
- 14:43:39 [pgearon]
- I have a mild preference for lexical ordering.
- 14:44:35 [LeeF]
- ack kasei
- 14:44:37 [kasei]
- Zakim, unmute me
- 14:44:37 [Zakim]
- kasei was not muted, kasei
- 14:45:03 [LeeF]
- kasei: lexical order is much more intuitive when reading a query
- 14:45:12 [LeeF]
- kasei: but would filter floating mean that this is floating order?
- 14:45:58 [AxelPolleres]
- { ?s ?p ?o . FILTER(?z = 3 )
- 14:45:59 [AxelPolleres]
- BIND(?o+1 AS ?z)
- 14:45:59 [AxelPolleres]
- }
- 14:45:59 [AxelPolleres]
- shouldn't that mean the same as
- 14:45:59 [AxelPolleres]
- { { ?s ?p ?o . FILTER(?z = 3 ) }
- 14:46:01 [AxelPolleres]
- BIND(?o+1 AS ?z)
- 14:46:02 [AxelPolleres]
- }
- 14:46:04 [AxelPolleres]
- ?
- 14:46:22 [AlexPassant]
- LeeF: support from Axel, paul and greg that FILTER will fail in that case
- 14:46:22 [AxelPolleres]
- at least that's how I understood the earlier resolution
- 14:46:27 [AlexPassant]
- ... any different POV ?
- 14:47:23 [LeeF]
- Consensus around FILTERs and BINDs happening in lexical order a.k.a BIND is "just outside" a BGP, rather than at the end of it
- 14:47:46 [LeeF]
- topic: + for fn:concat ?
- 14:47:58 [kasei]
- Zakim, mute me
- 14:47:58 [Zakim]
- kasei should now be muted
- 14:48:22 [SteveH]
- STRONG preference for not overloading
- 14:48:40 [AlexPassant]
- LeeF: question about the + operator
- 14:48:44 [AlexPassant]
- ... overloaed for string operation
- 14:48:50 [AlexPassant]
- ... most reponses where against that
- 14:49:04 [AlexPassant]
- ... but need for operator for string concatenation
- 14:49:24 [MattPerry]
- How would that work for SELECT (?a + ?b) AS ?c WHERE ...
- 14:49:47 [AlexPassant]
- AndyS: naturally wrote that in the query
- 14:49:51 [AndyS]
- ARQ implements -- it's a legal extension to expression evaluation anyway.
- 14:50:18 [kasei]
- I think "1" + "2" (plain literals) is going to produce unexpected results for people.
- 14:50:47 [LeeF]
- q?
- 14:51:16 [pgearon]
- +q
- 14:51:29 [AlexPassant]
- LeeF: strawpoll for + in string concatenation
- 14:51:37 [LeeF]
- straw poll: Using + operator for string concatenation
- 14:52:12 [LeeF]
- straw poll: Using + operator for string concatenation (no implicit casting)
- 14:52:17 [SteveH]
- -1
- 14:52:20 [LeeF]
- +1 in favor / 0 don't care / -1 against
- 14:52:23 [kasei]
- -1
- 14:52:26 [AndyS]
- +1
- 14:52:27 [LeeF]
- 0
- 14:52:27 [MattPerry]
- -1
- 14:52:29 [ivan]
- 0
- 14:52:30 [AxelPolleres]
- 0
- 14:52:31 [pgearon]
- +1
- 14:52:32 [NicoM]
- +1
- 14:52:34 [AlexPassant]
- +1
- 14:52:35 [OlivierCorby]
- +1
- 14:52:44 [AxelPolleres]
- meaning 0 + "str" wouldn't work, but "0"+" str" would ?
- 14:52:52 [bglimm]
- 0
- 14:53:01 [LeeF]
- 5 / 3 / 3
- 14:54:20 [SteveH]
- no xs:string?
- 14:54:37 [AxelPolleres]
- ... by example was meant just to clarify that no casting implicit
- 14:54:44 [AlexPassant]
- LeeF: preference to include this
- 14:55:06 [AxelPolleres]
- ?x + "str" worries me a bit still
- 14:55:39 [AxelPolleres]
- but str(?x) + str" would work "
- 14:55:56 [AndyS]
- SteveH, Doing more than the spec requires (e.g. xs:string) is legal as an extension. Can add URI+string if you want :-)
- 14:56:10 [SteveH]
- AndyS, erg!
- 14:56:12 [LeeF]
- ACTION: AxelPolleres to come up with test cases around + for string concatenation
- 14:56:12 [trackbot]
- Sorry, couldn't find user - AxelPolleres
- 14:56:21 [LeeF]
- ACTION: Axel to come up with test cases around + for string concatenation
- 14:56:21 [trackbot]
- Created ACTION-330 - Come up with test cases around + for string concatenation [on Axel Polleres - due 2010-11-09].
- 14:56:22 [AndyS]
- Why? namesapce + local name!
- 14:56:29 [SteveH]
- the difference between plain literals and xsd:strings is already odd enough
- 14:56:55 [pgearon]
- -q
- 14:57:00 [AndyS]
- That is not a SPARQL matter - blame RDF-MT xsd1(a|b)
- 14:58:04 [MattPerry]
- with comparison return type is always boolean
- 14:58:06 [NickH]
- NickH has joined #sparql
- 14:58:22 [SteveH]
- AndyS, sure, not SPARQL's fault, but we don't have to make it worse than it is
- 14:58:29 [MattPerry]
- with overloaded + can be string or number
- 14:58:40 [LeeF]
- "1" * 4
- 14:58:45 [LeeF]
- "aaaa" * 4
- 14:58:55 [AndyS]
- always the case of for ?x * ?y
- 14:59:46 [AxelPolleres]
- q+ to ask about http://www.w3.org/2009/sparql/docs/query-1.1/rq25.xml#selectExpressions ... "The new variable is introduced using the keyword AS; it must not already be potentially bound." is still open, raise an ISSUE?
- 15:00:22 [Zakim]
- +??P21
- 15:00:45 [NickH]
- Zakim, ??P21 is me
- 15:00:45 [Zakim]
- +NickH; got it
- 15:00:49 [SteveH]
- q+
- 15:01:19 [LeeF]
- SteveH: fn:concat casts its arguments to a string, so we'll need to be careful
- 15:01:52 [SteveH]
- q-
- 15:02:21 [LeeF]
- ACTION: Andy to clarify the meaning of "potentially bound" vis a vis what can go on the right hand side of an AS in a SELECT list
- 15:02:21 [trackbot]
- Created ACTION-331 - Clarify the meaning of "potentially bound" vis a vis what can go on the right hand side of an AS in a SELECT list [on Andy Seaborne - due 2010-11-09].
- 15:02:24 [AxelPolleres]
- ack AxelPolleres
- 15:02:24 [Zakim]
- AxelPolleres, you wanted to ask about http://www.w3.org/2009/sparql/docs/query-1.1/rq25.xml#selectExpressions ... "The new variable is introduced using the keyword AS; it must not
- 15:02:27 [Zakim]
- ... already be potentially bound." is still open, raise an ISSUE?
- 15:02:46 [AlexPassant]
- topic: Aggregate over mixed datatypes
- 15:02:58 [AlexPassant]
- ... discuss next week, with status of test cases
- 15:03:30 [SteveH]
- bye all
- 15:03:31 [ivan]
- zakim, drop me
- 15:03:31 [Zakim]
- Ivan is being disconnected
- 15:03:32 [AxelPolleres]
- regrets for next week most likely... will probably collide with poster session at ISWC
- 15:03:33 [Zakim]
- -Ivan
- 15:03:40 [bglimm]
- bye
- 15:03:42 [Zakim]
- -AlexPassant
- 15:03:44 [Zakim]
- -SteveH
- 15:03:45 [MattPerry]
- bye
- 15:03:49 [NickH]
- ARGH to timezones!
- 15:03:53 [Zakim]
- -bglimm
- 15:04:02 [AxelPolleres]
- rrsagent, make records public
- 15:04:11 [Zakim]
- -NicoM
- 15:04:25 [NicoM]
- bye
- 15:04:33 [Zakim]
- -NickH
- 15:04:40 [Zakim]
- -OlivierCorby
- 15:06:06 [Zakim]
- -MattPerry
- 15:10:49 [kasei]
- AndyS?
- 15:10:51 [Zakim]
- -pgearon
- 15:10:53 [Zakim]
- -LeeF
- 15:10:53 [Zakim]
- -AndyS
- 15:10:55 [Zakim]
- -AxelPolleres
- 15:11:01 [AxelPolleres]
- AxelPolleres has left #sparql
- 15:11:01 [Zakim]
- -kasei
- 15:11:03 [Zakim]
- SW_(SPARQL)10:00AM has ended
- 15:11:05 [Zakim]
- Attendees were +1.617.245.aaaa, LeeF, AndyS, Ivan, NicoM, bglimm, +1.310.729.aabb, kasei, MattPerry, AlexPassant, pgearon, SteveH, AxelPolleres, +34.92.38.aadd, OlivierCorby, NickH
- 15:11:07 [AndyS]
- kasei - just finished talking ... test case Q?
- 15:11:17 [kasei]
- bind04 - I think you wrote it?
- 15:11:25 [kasei]
- it has this:
- 15:11:26 [kasei]
- { BIND(?o+1 AS ?z) } UNION { BIND(?o+2 AS ?z) }
- 15:11:44 [kasei]
- but has expected answers
- 15:11:48 [kasei]
- and I can't sort out why
- 15:12:38 [AndyS]
- :-)
- 15:12:44 [AndyS]
- Bottom up we have:
- 15:13:17 [AndyS]
- Eval BIND(?o+1 AS ?z) (one table row) and Eval BIND(?o+2 AS ?z) as another table row from the UNION.
- 15:13:41 [kasei]
- right, but ?o isn't bound there is it? I expect zero results out of the union.
- 15:13:43 [AndyS]
- Ah - OK - I see the Q now. Let me think a bit - looks wrong to me ...
- 15:14:29 [kasei]
- I think it produces the expected answers if the triple pattern is distributed into the union.
- 15:15:05 [AndyS]
- Yes - looks like a bug to me.
- 15:15:23 [kasei]
- ok. good to know i'm not going crazy.
- 15:15:35 [AndyS]
- One of the pain points of SPARQL is the need to duplicate patterns sometimes.
- 15:16:14 [kasei]
- yeah
- 15:17:38 [AndyS]
- LET ($$table := SELECT * { ?s ?p ?o } ) and have table variables.
- 15:26:06 [kasei]
- hahaha
- 15:26:08 [kasei]
- oh god
- 15:26:33 [kasei]
- AndyS, if you're still around. I know selecting * in addition to projexps has been talked about, but do you recall if a decision was ever made?
- 15:30:10 [AndyS]
- Formal DECISION? Don't think so.
- 15:30:26 [AndyS]
- I don't recall one anyway.
- 15:40:29 [kasei]
- ok
- 15:40:40 [AndyS]
- What is your opinion?
- 15:40:50 [kasei]
- it occurred to me just a bit too late that that was the big reason I wanted SELECT * to work with the GROUP BY stuff.
- 15:41:00 [kasei]
- I think it's useful
- 15:42:28 [kasei]
- feel like SELECT * (?qty*?price AS ?total) is a valid thing to want to do.
- 15:44:40 [AndyS]
- I agree - and if you don't like a feature, don't use it.
- 15:45:35 [kasei]
- if we do support it, i fell as if that undermines Steve's point that SELECT * with GROUP BY is just shorthand for DISTINCT
- 16:01:41 [OlivierCorby]
- OlivierCorby has left #sparql
- 16:23:37 [SteveH]
- longhand for DISTINCT, it's more characters
- 17:59:08 [SteveH]
- SteveH has joined #sparql
- 18:41:50 [kasei]
- SteveH, yes, fair enough.
- 19:08:28 [LeeF]
- LeeF has joined #sparql
- 20:33:19 [LeeF]
- trackbot, close ACTION-257
- 20:33:19 [trackbot]
- ACTION-257 Craft a test case for SELECT * ... GROUP BY and solicit implementor, WG, and community feedback closed
- 20:33:26 [LeeF]
- (I suppose today obviates that to a large extent...)
- 20:51:54 [LeeF]
- trackbot, close ACTION-280
- 20:51:54 [trackbot]
- ACTION-280 Mark subquery tests 5-10 approved closed
- 21:04:11 [SteveH]
- SteveH has joined #sparql
- 21:07:01 [LeeF]
- trackbot, close ACTION-317
- 21:07:01 [trackbot]
- ACTION-317 Review entailment closed
- 21:08:20 [LeeF]
- trackbot, close ACTION-308
- 21:08:20 [trackbot]
- ACTION-308 Mark group_concat and projexp tests as approved closed
- 21:12:40 [AndyS]
- AndyS has joined #sparql
- 21:16:15 [bglimm]
- bglimm has joined #sparql
- 21:16:53 [LeeF]
- trackbot, close ACTION-305
- 21:16:53 [trackbot]
- ACTION-305 Start discussion on mailing list about set of functions to include in SPARQL closed