Chatlog 2011-03-01

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.

14:53:30 <RRSAgent> RRSAgent has joined #sparql
14:53:30 <RRSAgent> logging to
14:53:36 <Zakim> Zakim has joined #sparql
14:53:52 <AxelPolleres> Zakim, this will be sparql
14:53:52 <Zakim> ok, AxelPolleres; I see SW_(SPARQL)10:00AM scheduled to start in 7 minutes
14:56:14 <Zakim> SW_(SPARQL)10:00AM has now started
14:56:21 <Zakim> +??P11
14:56:38 <SteveH> Zakim, ??P11 is me
14:56:38 <Zakim> +SteveH; got it
14:56:56 <Zakim> +[IPcaller]
14:57:03 <AndyS> zakim, IPCaller is me.
14:57:03 <Zakim> +AndyS; got it
14:57:25 <Zakim> +??P13
14:57:32 <Zakim> +corby
14:57:38 <SteveH> AndyS, did you just say something?
14:57:39 <NickH> zakim, ??P13 is me
14:57:39 <Zakim> +NickH; got it
14:58:06 <AndyS> Yes - did you hear anytghing?  I can't hear you.  skype testing service worked though.
14:58:21 <SteveH> I'm muted, one sec
14:58:43 <Zakim> +kasei
14:58:45 <cbuilara> cbuilara has joined #sparql
14:58:48 <Zakim> +pgearon
14:58:58 <NickH> getting some echo
15:00:32 <MattPerry> MattPerry has joined #sparql
15:00:48 <Zakim> +[IPcaller]
15:00:49 <AxelPolleres> trackbot, start meeting
15:00:51 <trackbot> RRSAgent, make logs world
15:00:53 <trackbot> Zakim, this will be 77277
15:00:53 <Zakim> ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start now
15:00:54 <trackbot> Meeting: SPARQL Working Group Teleconference
15:00:54 <trackbot> Date: 01 March 2011
15:01:06 <cbuilara> zakim, IPcaller is me
15:01:06 <Zakim> sorry, cbuilara, I do not recognize a party named 'IPcaller'
15:01:29 <AxelPolleres> Zakim, who is on the phone?
15:01:29 <Zakim> I notice SW_(SPARQL)10:00AM has restarted
15:01:30 <Zakim> On the phone I see SteveH, AndyS, NickH, corby, kasei, pgearon, [IPcaller], AxelPolleres, Sandro, MattPerry
15:01:35 <cbuilara> Zakim, IPcaller is me
15:01:35 <Zakim> +cbuilara; got it
15:01:44 <cbuilara> zakim, mute me
15:01:46 <Zakim> cbuilara should now be muted
15:02:07 <SteveH> SteveH has joined #sparql
15:02:14 <AxelPolleres> scribe: steveH
15:02:21 <AxelPolleres> chair: Axel Polleres
15:02:35 <AxelPolleres> agenda:
15:02:53 <AxelPolleres> regrets: LeeF, Chime 
15:03:12 <Zakim> +bglimm
15:03:15 <AxelPolleres> topic: admin
15:03:20 <bglimm> Zakim, mute me
15:03:20 <Zakim> bglimm should now be muted
15:03:24 <AxelPolleres> PROPOSED: Approve minutes at
15:03:59 <AxelPolleres> RESOLVED: Approve minutes at
15:04:10 <AxelPolleres> topic: comments
15:04:32 <SteveH> Axel: I've updated the comments page, picked any additional comments found on list
15:04:48 <SteveH> ... have several things not assigned, 1 from Greg R, from 26th of Jan
15:05:02 <SteveH> ... not sure if there was an overall answer
15:05:10 <SteveH> ... Lee thinks he's ok with the answer
15:05:17 <SteveH> ... Axel to check with Lee
15:05:42 <SteveH> ... next comment from Jeremy C. on Update, graph identification
15:05:52 <SteveH> ... using URIs that are IRI
15:06:03 <SteveH> ... I will ask Chime if noone else wants to comment
15:06:28 <SteveH> next, some comment on AVG() testcase from Jeen B
15:06:43 <AxelPolleres>
15:06:45 <SteveH> Axel: anybody voluneering?
15:07:06 <SteveH> I guess I should, but I'm really busy
15:07:40 <SteveH> Axel: I'll put Steve's name on it
15:07:56 <SteveH> ... we should clear all the comments before going to last call
15:07:56 <NickH> NickH has joined #sparql
15:08:05 <AxelPolleres> ACTION: Steve to draft an answer for JB-4 
15:08:05 <trackbot> Created ACTION-401 - Draft an answer for JB-4  [on Steve Harris - due 2011-03-08].
15:08:24 <alex_> alex_ has joined #sparql
15:08:51 <SteveH> Axel: last comment from ???, I can take a look at that
15:08:59 <AxelPolleres> ACTION: Axel to take care of RV-7
15:08:59 <trackbot> Created ACTION-402 - Take care of RV-7 [on Axel Polleres - due 2011-03-08].
15:09:04 <SteveH> ??? = Rob V.
15:09:04 <AxelPolleres>
15:09:14 <Zakim> +AlexPassant
15:09:38 <AxelPolleres>
15:09:59 <SteveH> Axel: one problem with comment from Jeen which we can't address yet
15:10:07 <SteveH> ... there's a mail about it on list
15:10:40 <SteveH> ... Jeen commented on MINUS v's NOT EXIST
15:10:44 <AxelPolleres>  ... not exists vs. minus
15:10:53 <SteveH> ... don't know how to proceed there
15:11:13 <kasei> that isn't true, though
15:11:16 <SteveH> ... Jeen says that MINUS is equiv. in all cases where it's not redundant
15:11:38 <Zakim> -SteveH
15:12:03 <AxelPolleres>
15:12:16 <Zakim> +??P11
15:12:24 <SteveH> Zakim, ??P11 is me
15:12:24 <Zakim> +SteveH; got it
15:12:52 <SteveH> kasei: Jeen posted it on his weblog before this comment, and I commented on the blog
15:13:28 <AxelPolleres> ACTION: greg to answer to  with counterexample to JB-2
15:13:28 <trackbot> Created ACTION-403 - Answer to  with counterexample to JB-2 [on Gregory Williams - due 2011-03-08].
15:13:58 <AxelPolleres> Next regular meeting: 2011-03-08 @ 15:00 UK / 10:00 EST (scribe: Paul Gearon) 
15:14:18 <AxelPolleres> regrets for next week from me
15:14:30 <SteveH> AxelPolleres: I will not be here but lee will be
15:14:31 <OlivierCorby> regrets also
15:14:40 <SteveH> ... geosparql has asked us for a review
15:14:49 <AxelPolleres> topic: geosparql review
15:15:05 <SteveH> MattPerry: we have a group of about 10, working on geosparql, it is straightforward, vocab + filter functions for SPARQL
15:15:30 <SteveH> ... in OGC they've had std. object models for geometry, takes on and created RDF vocab that does the same thing
15:15:45 <SteveH> ... different classes for polygon,. point, line + dimensions and so on
15:16:07 <SteveH> ... also functions for objects, created analagous filter functions for them
15:16:15 <SteveH> ... union intersection etc.
15:16:24 <SteveH> ... translation of existing standard
15:16:37 <SteveH> .... serialise these geometries in RDF, created datatypes
15:16:40 <AxelPolleres>
15:16:57 <SteveH> ... existing text based standards can be put in literals
15:17:23 <SteveH> ... use these literals as arguments to functions
15:18:21 <cbuilara> I can do it, or ask anybody in my grpup
15:18:25 <cbuilara> group
15:18:34 <cbuilara> zakim, unmunte me
15:18:34 <Zakim> I don't understand 'unmunte me', cbuilara
15:18:44 <cbuilara> zakim, unmunte me
15:18:44 <Zakim> I don't understand 'unmunte me', cbuilara
15:18:54 <cbuilara> zakim, unmute me
15:18:54 <Zakim> cbuilara should no longer be muted
15:19:19 <SteveH> cbuilara: people in my research group working on GML(?)
15:19:25 <SteveH> ... it would be worth them looking at it
15:19:47 <SteveH> ... I will send a review
15:19:54 <AxelPolleres> ACTION: carlos to look in his group� for a review for GeoSPARQL and send it to us
15:19:54 <trackbot> Could not create new action (failed to parse response from server) - please contact sysreq with the details of what happened.
15:19:54 <trackbot> Could not create new action (unparseable data in server response: local variable 'd' referenced before assignment) - please contact sysreq with the details of what happened.
15:20:16 <SteveH> AxelPolleres: to last call...
15:20:23 <SteveH> ... take a look at status of drafts
15:20:24 <AxelPolleres> topic: to last call
15:20:25 <AxelPolleres>
15:20:25 <AxelPolleres> \
15:20:43 <bglimm> Zakim, unmute me
15:20:43 <Zakim> bglimm should no longer be muted
15:21:09 <bglimm> I had a few comments on the revised sections
15:22:43 <SteveH> AndyS: there are 12 markers in doc for things to do, all editorial - Steve can you look at the eval of ToList, not sure it can be removed
15:22:55 <SteveH> ... if people are happy to comment in that state we can go ahead
15:23:04 <AxelPolleres> andy: 12 editorial markers in the� doc at the moment... apart from that, we can go ahead with review
15:23:42 <SteveH> AxelPolleres: we have reviewers assigned, so can turn to BGP matching
15:23:53 <AxelPolleres>  BGP matching extensions rewording suggestions by Birte
15:24:02 <SteveH> ... has rewording suggestions
15:24:31 <SteveH> bglimm: the main problem is that terminology is outdated, uses old terminology
15:24:47 <SteveH> ... uses different word from rest of text, not been updated since defintions were tested
15:24:56 <SteveH> ... would like it updated
15:25:17 <SteveH> ... the only thing I suggest is to require that solutions are uniquely satisfied
15:25:26 <SteveH> ... I think it's quite important
15:25:41 <AxelPolleres> q?
15:25:44 <SteveH> ... doesn't really change any of the semantics
15:26:06 <SteveH> AndyS: looked through the text, it looked pretty good, was planning on putting that text in
15:26:16 <SteveH> ... there's a couple of points I want to check
15:26:36 <SteveH> AxelPolleres: one point, reformulation of cond. 1, sent to list
15:27:08 <SteveH> [ discussion of uniquely specified issues ]
15:27:41 <Zakim> -SteveH
15:27:53 <AxelPolleres> s/uniquely/uniquely (up to RDF graph equivalence)/ ?
15:27:54 <SteveH> sorry, something wrong with the sip connection :(
15:28:13 <Zakim> +??P11
15:28:19 <SteveH> Zakim, ??P11 is me
15:28:19 <Zakim> +SteveH; got it
15:29:02 <SteveH> q+
15:29:44 <SteveH> q-
15:29:46 <pgearon> I can try
15:30:01 <SteveH> AxelPolleres: no action here
15:31:23 <AxelPolleres> ACTION: Birte to draft xml for revised Extending BGP matching section for query
15:31:23 <trackbot> Created ACTION-405 - Draft xml for revised Extending BGP matching section for query [on Birte Glimm - due 2011-03-08].
15:31:38 <SteveH> ... that's all for query
15:31:55 <SteveH> ... for update, any news? lots of discussion, will come to that later
15:32:10 <SteveH> ... service descirption, grega, can you give update
15:32:11 <bglimm> Zakim, mute me
15:32:11 <Zakim> bglimm should now be muted
15:32:29 <SteveH> kasei: I've added a couple of things to the SD section, biggest thing is relationship with dataset protocol
15:33:14 <SteveH> ... said I'd take it to email, suggested simple change to SD vocab which will be sufficient to describe a dataset in terms of the dataset protocol
15:33:32 <SteveH> ... if its goes beyond that I'm going to push back, to avoid getting into out of scope things
15:33:37 <SteveH> ... waiting on more discussionm
15:33:57 <SteveH> ... been discussing conformance language with Andy
15:34:16 <SteveH> ... I'm happy discussing on email
15:35:07 <SteveH> I owe a review on {something}
15:35:10 <bglimm> Zakim, unmute me
15:35:10 <Zakim> bglimm should no longer be muted
15:35:23 <SteveH> bglimm: waiting for a review on entailment
15:35:57 <SteveH> ???: the parital review that was sent out is the final one
15:36:05 <SteveH> bglimm: I saw one review
15:36:11 <AxelPolleres> s/???/Matt/
15:36:34 <SteveH> bglimm: connonicalisation issue
15:36:49 <SteveH> ... only define d-entailment within some limit
15:37:05 <SteveH> ... not happy, but it's what we decided, not sure if we want to revisit
15:37:28 <SteveH> AxelPolleres: I guess at the moment we just go on, what's your suggestion
15:37:46 <SteveH> bglimm: I would just remove the entailment, it's not nice that systems behave differently
15:38:09 <SteveH> ... SPARQL doesn't specify this behaviour, so maybe it's not up to the entailments regimes
15:38:48 <SteveH> ... you can only return ansers that occur in the graph, but that depends on parsing process
15:39:04 <SteveH> ... might have two different values, so might return two answers
15:39:22 <AxelPolleres> issue different answers and also differenct cardinalities implied.
15:39:23 <SteveH> ... can get different cardinality
15:39:40 <SteveH> ... do you want to require canonicalisation - seems to be the only way to solve this problem
15:40:24 <SteveH> bglimm will add a note to the document
15:40:27 <AndyS> zakim, who is on the call?
15:40:27 <Zakim> On the phone I see AndyS, NickH, corby, kasei, pgearon, cbuilara, AxelPolleres, Sandro, MattPerry, bglimm, AlexPassant, SteveH
15:40:32 <AxelPolleres> q?
15:40:56 <SteveH> cbuilara: proposed semantics of service algebra, difficult to understand what I was proposing in email, I will follow your comments
15:41:16 <SteveH> ... it is almost finished I think, will be finished this/next week
15:41:44 <SteveH> formal update semantics]
15:41:51 <AxelPolleres> topic: Update Formal Semantics (incl. bnodes as DELETE-wildcards) 
15:41:53 <SteveH> particularly bnodes as wildcards
15:41:59 <cbuilara> zakim, mute me
15:41:59 <Zakim> cbuilara should now be muted
15:42:25 <SteveH> AxelPolleres: summary - there were concerns about update semantics as in current draft
15:42:28 <AxelPolleres>
15:42:48 <SteveH> ... because it doesn't cover the resolution that we wanted to create blank nodes in delete patters as wildcards
15:42:57 <SteveH> ... there were some mails with suggestions
15:42:59 <AxelPolleres>
15:43:19 <AxelPolleres>
15:43:19 <SteveH> ... three options
15:43:39 <Zakim> -SteveH
15:43:52 <Zakim> +??P0
15:43:59 <SteveH> Zakim, ??P0 is me
15:43:59 <Zakim> +SteveH; got it
15:44:11 <SteveH> ... several coinflip descisions on what it means
15:44:37 <SteveH> ... do the bns match all resources which are available, or whether they are [something]
15:44:43 <SteveH> ... people not so convinced
15:44:55 <SteveH> ... third option is to revisit resolution. no blank nodes as wildcards
15:44:57 <SteveH> q+
15:46:52 <SteveH> q-
15:46:55 <AndyS> q+
15:47:10 <AxelPolleres> Option 1: Bnode match all resources in the graph (essentially rewriting 1 in the mails)
15:48:12 <AxelPolleres> Option 2 : bnodes match whatever they match if treated as "moficy_tmeplate copied tyo body" 
15:48:39 <AxelPolleres> (essentially rewriting 2)
15:49:14 <SteveH> AndyS: are there usecases where 1 & 2 are different
15:49:24 <AxelPolleres> Option 1 and 2 differ �,  see
15:51:12 <SteveH> AxelPolleres: at end of mail there's a simple example, so prolem with unbound is there
15:51:21 <AxelPolleres>  Option 3: treat bnodes as in CONSTRUCT/INSERT, i.e. as new bnodes... would mean they don't match anything.
15:51:38 <AxelPolleres> (that is against our earlier resolution)
15:51:55 <AxelPolleres>  Option 4: is to forbid bnodes in DELETE
15:51:55 <AndyS> q-
15:52:03 <SteveH> SteveH: rather to make it an error
15:52:18 <SteveH> AxelPolleres: one mail from birte, not happy with rewritings
15:52:26 <bglimm> Zakim, unmute me
15:52:26 <Zakim> bglimm was not muted, bglimm
15:52:50 <SteveH> bglimm: you could just define that you wanted all things deleteed that are an instance of the graph
15:53:16 <SteveH> ... not specifiy how you want the deletion, reqriting is one way of doing the delete, would like a more abstract definition
15:53:35 <SteveH> ... would have to do some graph matching to decide which triples are deleted
15:53:39 <SteveH> ... re. option 2
15:53:42 <AndyS> IIUC version 1 does not reflect pattern of bNode use -> makes me v nervious
15:54:21 <SteveH> ... equiv. to option 1, but not specified with UNIONs
15:54:31 <SteveH> AxelPolleres: think it could be done but not sure anyone has time
15:55:12 <SteveH> ... need some kind of skolemisation/deskolemisation
15:55:25 <AndyS> Who was a supporter of the original point of making bNodes variables at all?  Do we need to check with them?
15:55:34 <SteveH> bglimm: they are like variables in normal evaluation, don't have to skolemise
15:56:20 <SteveH> bglimm: think it's more difficult to delete lists
15:56:31 <SteveH> AndyS: do any of the mechanisms actually help?
15:57:01 <AxelPolleres> Can someone draft a Testcase that does list deletion?
15:57:05 <SteveH> bglimm: I think they do
15:57:12 <SteveH> AndyS: only works where you know the links
15:57:14 <SteveH> q+
15:57:25 <SteveH> bglimm: that's the main usecase
15:57:34 <SteveH> AxelPolleres: only works with fixed length lists
15:59:19 <Zakim> -SteveH
15:59:37 <Zakim> +??P0
15:59:41 <SteveH> Zakim, ??P0 is me
15:59:41 <Zakim> +SteveH; got it
16:00:23 <AndyS> Hmm - something like    DELETE { ?x rdf:first ?f ; rdf:rest ?y } WHERE { ?list rdf:rest* ?x . ?x rdf:first ?f ; rdf:rest ?y }
16:00:33 <SteveH> 4, where forbidden = causes an error, rather than bannedby syntax
16:00:34 <bglimm> 3 or 4
16:00:41 <pgearon> option 4 (followed by 3)
16:00:43 <AndyS> which works by finding the list element one at a time
16:00:48 <kasei> 0
16:00:50 <sandro> (sorry,  undecided.)
16:00:54 <AndyS> no opinion - need to see details.
16:01:02 <OlivierCorby> don't know yet
16:01:12 <Zakim> -pgearon
16:01:15 <AxelPolleres> Strawpoll Option1-4 ?
16:01:16 <AxelPolleres> prefer 1 over 2 , but can live with 3/4
16:01:39 <AxelPolleres> let's take it to email