14:57:48 RRSAgent has joined #rdf-wg 14:57:48 logging to http://www.w3.org/2011/07/20-rdf-wg-irc 14:58:19 zakim, this is rdfwg 14:58:19 ok, gavinc; that matches SW_RDFWG()11:00AM 14:58:29 Zakim, who is on the phone? 14:58:29 On the phone I see +1.707.861.aaaa, +20598aabb, ??P9 14:58:37 zakim, aaaa is me 14:58:37 +gavinc; got it 14:58:45 Zakim, ??P9 is me 14:58:45 +yvesr; got it 14:58:53 AndyS has joined #rdf-wg 14:59:16 +Tony 14:59:33 Zakim, Tony is me 14:59:33 +Scott_Bauer; got it 15:00:25 zakim, who is here? 15:00:25 On the phone I see gavinc, +20598aabb, yvesr, Scott_Bauer 15:00:26 Zakim, code? 15:00:26 the conference code is 73394 (tel:+1.617.761.6200 sip:zakim@voip.w3.org), MacTed 15:00:28 +??P15 15:00:34 +OpenLink_Software 15:00:35 zakim, ??P15 is me 15:00:35 +AndyS; got it 15:00:40 Zakim, OpenLink_Software is temporarily me 15:00:40 +MacTed; got it 15:00:42 Zakim, mute me 15:00:42 MacTed should now be muted 15:01:00 zakim, +20598aabb is me 15:01:00 +Guus; got it 15:01:03 +EricP 15:01:23 + +1.540.898.aacc 15:01:54 + +539149aadd 15:02:04 Zakim, aacc is me 15:02:05 +davidwood; got it 15:02:07 tomayac has joined #rdf-wg 15:02:22 LeeF has joined #rdf-wg 15:02:39 +Sandro 15:02:42 zakim, aadd is me 15:02:42 +cygri; got it 15:03:04 +??P22 15:03:13 + +1.617.553.aaee 15:03:17 zakim, aaee is me 15:03:17 +LeeF; got it 15:03:28 Zakim, ??P22 is me 15:03:28 +SteveH; got it 15:03:41 scribe: cygri 15:03:44 +??P26 15:03:53 zakim, ??P26 is me 15:03:55 +mischat; got it 15:03:57 -SteveH 15:03:59 zakim, mute me 15:04:01 mischat should now be muted 15:04:07 +??P27 15:04:12 Zakim, ??P27 is me 15:04:13 +SteveH; got it 15:04:17 topic: Admin 15:04:35 +??P22 15:04:37 http://www.w3.org/2011/rdf-wg/wiki/Meetings:Telecon2011.07.20 15:04:51 zakim, ??P22 is me 15:04:51 +mbrunati; got it 15:05:06 PROPOSAL: accept last week's minutes http://www.w3.org/2011/rdf-wg/meeting/2011-07-06 15:05:25 RESOLUTION: accept last week's minutes http://www.w3.org/2011/rdf-wg/meeting/2011-07-06 15:05:27 +Bert 15:05:36 topic: Action item review 15:05:49 Zakim, Bert is me 15:05:49 +FabGandon; got it 15:05:55 + +1.404.978.aaff 15:06:06 guus: richard did ACTION-64, it's done, close it 15:06:13 zakim, aaff is me 15:06:13 +tomayac; got it 15:06:32 sandro: didn't do ACTION-52 yet 15:06:56 AZ has joined #rdf-wg 15:07:42 cygri: ACTION-62 done on list (not on wiki) 15:08:06 ACTION-65? 15:08:07 ACTION-65 -- Sandro Hawke to and Pat to consider what words to add to minimal proposal. -- due 2011-06-29 -- OPEN 15:08:07 http://www.w3.org/2011/rdf-wg/track/actions/65 15:08:17 +??P5 15:08:18 sandro: not yet done 15:08:26 zakim, ??P5 is me 15:08:26 +NickH; got it 15:08:31 zakim, mute me 15:08:31 NickH should now be muted 15:08:32 guus: ACTION-66 not yet done, paul on holiday 15:08:39 Zakim, mute me 15:08:39 yvesr should now be muted 15:08:41 pchampin has joined #rdf-wg 15:09:08 ... ACTION-67 and ACTION-68 remain open 15:09:34 ... next telecon is August 3 15:09:47 davidwood: i can chair on that date 15:09:55 topic: Turtle Editors Draft 15:10:04 guus: davidwood made transition request 15:10:11 ... there are still admin things to be done 15:10:30 Turtle is already a reasonably settled serialization of RDF. Many implementations of Turtle already exist, we are hoping for feedback from those existing implementors and other people deciding that now would be a good time to support Turtle. There are still a few rough edges that need polishing, and better alignment with the SPARQL triple patterns. The working group does not expect to make any large changes to the existing syntax. 15:10:40 gavinc: we now have status paragraph that the WG needs to agree to 15:11:04 davidwood: we need to find words in this telecon that reflect the status as the WG sees it 15:11:23 pchampin has left #rdf-wg 15:11:49 guus: i'm happy with the text 15:11:49 +1 to text 15:11:53 Looks good for WD 15:12:12 PROPOSAL: accept status text as proposed 15:12:24 RESOLUTION: accept turtle status text as proposed 15:13:06 +wcandillon 15:13:11 sandro: technically speaking, the team contacts write this text. discussing it is fine though 15:13:19 Zakim, wcandillon is me 15:13:19 +AZ; got it 15:13:33 pchampin has joined #rdf-wg 15:13:53 guus: i'd like to take brief look on turtle issues http://www.w3.org/2011/rdf-wg/track/products/3 15:14:35 gavinc: most of these issues are already resolved, just not closed 15:14:50 +??P31 15:15:27 ISSUE-1 => ISSUE-19 , ISSUE-20 15:15:54 ericP: (discussion of ISSUE-67) 15:16:16 ericP: SPARQL is changing escaping rules 15:16:34 Why \u in prefix part? What is it escaping? 15:16:35 ... used to escape everywhere; now escapes only certain places; discussion is which places exactly 15:16:44 Souri has joined #RDF-WG 15:16:57 guus: gavin, can you take action to update the issue list? 15:17:04 + +1.603.897.aagg 15:17:06 ACTION: gavinc to update Turtle issue list to reflect current status 15:17:06 Created ACTION-69 - Update Turtle issue list to reflect current status [on Gavin Carothers - due 2011-07-27]. 15:17:09 aagg is me 15:17:18 zakim, aagg is me 15:17:18 +Souri; got it 15:17:32 http://dvcs.w3.org/hg/rdf/raw-file/d74c1b87d47f/rdf-turtle/index.html#sec-diff-ntriples 15:17:45 gavinc: (discusses the section on relationship between turtle and n-triples) 15:18:29 ericP: question to the community is whether they'd want a section that describes the subset of turtle that's n-triples 15:19:05 gavinc: defining n-triples as part of turtle would have implications on utf-8 support 15:19:24 q+ to ask about MIME types 15:19:49 ericP: we could publish n-triples as part of turtle, or as a separate document 15:20:04 AndyS: if we make it a chapter, then how do we handle the media type registration? 15:20:08 +1 to ericP's proposal that we add a section that describes the subset of turtle that's n-triples or as a separate document 15:20:17 current media type for ntriples is text/plain 15:20:19 gavinc: that's a problem. we may be better off making a separate document 15:20:33 ... otoh, it means more boilerplate and more document than we really need 15:20:38 +1 to getting a proper MIME type for n-triples 15:21:13 guus: can you make a proposal? 15:21:31 gavinc: it would have wide-ranging implications. it's defined in a different document that we may not plan to change? 15:21:41 AndyS: it's currently defined in the RDF Test Cases document 15:21:55 gavinc: it's a bit weird at the moment, parts are informative 15:21:56 New MIME type - text/plain is "unhelpful" at best. 15:22:04 guus: i agree. that's something we want to fix 15:22:04 is the plan to make n-triples accept UTF-8? 15:22:36 gavinc: can we have a poll? separate document vs. section in turtle? 15:23:23 ericP: if nt is ASCII only, then that's a bigger change, not quite just subset 15:23:28 POLL: (1) ntriples in turtle document, (2) ntriples in its own document 15:23:47 1 15:23:53 1 15:23:53 1 15:23:56 2 15:23:58 1 15:24:02 Don't care : UTF-8 v important. 15:24:04 1 15:24:06 0 (not decided) 15:24:08 1 15:24:14 0 (not decided) 15:24:25 don't care 15:24:25 ericP: we can go FPWD with turtle now either way 15:24:29 0 15:24:35 0 15:25:08 guus: let's do poll on UTF-8 vs ascii to? 15:25:12 s/to/too/ 15:25:50 I don't prefer ASCII, FWIW 15:25:51 (discussion of performance of utf-8 vs ascii) 15:25:56 I'd rather have UTF-8 15:26:07 q+ 15:26:13 ack me 15:26:13 AndyS, you wanted to ask about MIME types 15:26:18 utf-8++ 15:26:21 +1 on utf-8 15:26:25 +1 on utf-8 15:26:36 +1 on utf-8 15:26:38 utf-8 15:26:38 0 15:26:41 0 15:26:42 would utf-8 mean that \uXXXX are not allowed anymore ? 15:26:43 -.5 to utf-8 15:26:46 +1 on utf-8 15:26:47 (seems slighly weird it's not yet!) 15:26:47 +1 to UTF-8 15:26:48 0 15:26:55 +1 to UTF-8 15:27:01 gavinc: \uXXXX would still be legal 15:27:01 +1 utf-8 15:27:18 so it means that there are several ways to serialize the same graph in n-triples? 15:27:23 ericP: for high throughput, ascii is faster 15:27:33 makes it grep/sed/sort-unfriendly 15:27:35 I dispute "much" as well 15:27:38 AndyS: i don't think it's much faster 15:27:40 AndyS1 has joined #rdf-wg 15:28:04 ericP: only reason for using nt is performance 15:28:13 you can write a much faster parser, if you know to only expect n-triples 15:28:46 AndyS: i got 90k with turtle and 120k with nt, with same character conversion code 15:28:52 NickH, yes, it still need to have its own mimetype 15:28:58 pchampin, modern grep etc. is UTF-8 compatible 15:29:07 (discussion of lexing and escaping) 15:29:56 @steveh: yes, but grep does not know that "\uc3a9" is the same as "é", or does it?? 15:30:11 pchampin, that's not a UTF-8 thing though 15:30:16 q? 15:30:19 yvesr: yes, it needs its own MIME type 15:30:22 ack cygri 15:30:35 Can delay Unicode conversion and work in bytes only in UTF-8 15:30:40 cygri, +1 15:30:47 AndyS: yes! 15:30:53 cygri: serializers would become simpler if utf-8 was allowed 15:31:04 gavinc: last time we had objection from oracle 15:31:14 souri: we use n-triples a lot 15:31:50 ... we'll have to modify our software if we get input that uses utf-8 15:32:00 a new name would make some sense 15:32:06 q+ 15:32:09 ... we'd prefer calling the new thing something else, n-triples prime or so 15:32:13 q- 15:32:18 IRI_REF"<"(([#-;=?-\[\]_a-z~-\x7F]|([\xC2-\xDF][\x80-\xBF])|(\xE0([\xA0-\xBF][\x80-\xBF]))|([\xE1-\xEC][\x80-\xBF][\x80-\xBF])|([\xE1-\xEC][\x80-\xBF][\x80-\xBF])|(\xED([\x80-\x9F][\x80-\xBF]))|([\xEE-\xEF][\x80-\xBF][\x80-\xBF])|(\xF0([\x90-\xBF][\x80-\xBF][\x80-\xBF]))|([\xF1-\xF3][\x80-\xBF][\x80-\xBF][\x80-\xBF])|(\xF4([\x80-\x8E][\x80-\xBF][\x80-\xBF])|(\x8F([\x80-\xBE][\x80-\xBF])|(\xBF[\x80-\xBD])))]))*">" 15:32:22 q+ to ask about multiple n-triples serialization for the same graph 15:32:47 gavinc: is there an issue if the utf-8 version gets a new media type? 15:32:48 (STRING_LITERALs are even longer) 15:32:55 ericP - you are confusing bytes and codepoints. 15:33:05 souri: we need to talk about it internally. might be able to tell you more next week 15:33:16 flex/bison lexing is in terms of bytes 15:33:17 steveh: in our code, the change was not that difficult 15:33:31 ... the problem is deployed software that claims to handle n-triples 15:33:59 ack SteveH 15:34:02 ... there will be confusion if deployed software has to deal with new data that uses utf-8 15:34:04 souri: exactly 15:34:09 yes 15:34:13 in a very loud place 15:34:35 so I'll write it down: 15:34:44 I seem to remember a use case of NT: 15:34:44 ericP - that's one way of doing it. There are others. e.g. Do checking of points elsewhere - lots (and lots!) of impl choices 15:34:52 n-triples 1.1? 15:34:55 handling it with text-based tools, such as grep, sed... 15:35:25 pchampin, recent versions of those text-based tools deal with utf-8 correctly 15:35:45 ericP: we should publish it with an issue that says: we'll make n-triples utf-8, but some people might prefer an ascii version 15:36:13 guus: is everyone ok with proceeding in that way? 15:36:14 pchampin, from grep's man "also work on UTF-8 files and 16-bit wide Unicode files" 15:36:19 souri: yes 15:36:59 gavinc: there's still some text in the document that claims that turtle is a subset of n3 15:37:05 ... that's not strictly true any more 15:37:17 ... i'd like wg feedback 15:37:27 "It is intended to be compatible with, and a subset of, Notation 3." 15:37:45 q+ to ask about motivation of text 15:37:48 ericP: i suspect tim will change n3 according to what we do with turtle 15:38:14 guus: i'd like to have a statement in there that says what the issue is, without formal claim about subset 15:38:15 I agree that Turtle doesn't need to be a format subset of N3 15:38:17 formal 15:38:21 +1 to AndyS 15:38:26 +1 15:38:32 AndyS: the spec currently says, it's a *goal* to be a subset of n3 15:38:41 ... we want to make turtle a free-standing language 15:38:45 ack me 15:38:45 AndyS, you wanted to ask about motivation of text 15:38:47 Perfectly happy to have N3 say "I'm an extension of Turtle" ;) 15:38:50 q- 15:38:52 Yeah. "Turtle is similar to, inspired by, and largely compatible with N3" 15:38:55 ... n3 is not a standard 15:38:56 make provenace clear, nof formal subset statement 15:39:01 q+ 15:39:13 ericP: there are some ppl who come to rdf via n3 15:39:18 +1 - it could make a reference to it though 15:39:21 inspired by? 15:39:25 AndyS: i don't see any point in documenting the differences 15:39:36 ... n3 will change, so this will be updated 15:39:41 +100 to AndyS 15:39:45 +1 to AndyS 15:39:50 +1 to andyS 15:39:50 +1 15:39:52 +1 15:39:53 ack cygri 15:39:54 +1 to AndyS 15:39:54 +infinity 15:40:02 yvesr, why do you ask "inspired by" ? That's a polite way to say 'a blatant ripoff", right? :-) 15:40:16 sandro, heh :) yes, sort of 15:40:21 cygri: "The genesis of Turtle is N3" 15:40:32 ... i don't see value in detailing the differences 15:40:33 i see no reason to talk about n3, +1 AndyS 15:40:39 ack yvesr 15:40:45 ... "N3 is a more powerful predecessor of Turtle." 15:40:57 "While Notation 3 (N3) syntax played a large role in the creation of Turtle they are not strictly compatable." 15:40:58 yvesr: we can leave it to the N3 spec to discuss this in detail 15:41:03 that's in the text already 15:41:11 ... just have historic note mentioning n3 in turtle 15:41:22 gavinc: ok, so i'll do this change before FPWD 15:41:25 gavinc, in that above sentence, s/compatable/compatible :-) 15:41:46 Imagine in +10years, what should the doc say? Do specs record history? Not a conf paper. 15:41:49 Guus, still totally breaking up.... 15:41:52 ah yes, being able to spell 15:42:02 guus: ... ... editorial disgression ... ... 15:42:40 topic: Graph terminology 15:42:58 q+ 15:43:14 guus: trying to summarize the consensus: 15:43:20 ... g-snap = RDF graph 15:43:33 q+ 15:45:51 cygri: we have a proposal on the table, just adopt SPARQL's design. i want to hear clearly articulated what the problem is about that proposal 15:45:58 sandro: sparql misuses the term graph 15:46:18 ... you can't do provenance if you don't know what's mutable and what's not 15:46:24 not the spec, but often the community, and it's ambiguous. 15:46:30 ... (not the sparql spec, but the community misuses the term) 15:47:07 q? 15:47:16 scribenick: pchampin 15:47:26 thx pa 15:47:41 scribenick: ericP 15:47:42 cygri: sandro, are you saying that the spec is not appropriate for doing provenanve? or are just ppl using it the wrong way? 15:47:46 scribenick: pchampin 15:48:19 sandro: the RDF spec should be more useful for doing provenance 15:48:42 cygri: which use case is not currently solved with the SPARQL proposal? 15:49:15 q? 15:49:24 we could just provide terms to document the state of the universe at an instant 15:49:42 ack cygri 15:49:45 so we could just use graph for g-snap and let other invent g-box etc. 15:49:50 ack sandro 15:50:18 sandro: (some distinction btw g-snap and g-box, unmutable vs. mutable graph) 15:50:35 In SPARQL, graphs are immutable values. Graph store is slots holding values. 15:50:39 cygri: the current RDF model does not take time into account; a graph is a snapshot view of the world 15:51:31 sandro: agreed, RDF is about snapshots, but computers often deal with changing things 15:51:36 "RDF graph" in SPARQL doc = g-box, in RDF docs = g-snap? 15:52:04 cygri: you can make an RDF statement like "this graph is the state of that thing", "this graph is valid from time t0 to time t1" 15:52:26 Guus, RDF graph = g-snap. 15:52:30 Guus, RDF graph is a set of triples in RDF specs *and* in SPARQL specs 15:52:37 ... provided the correct vocabulary; but this group does not have to provide such a vocabulary, only the underlying data model 15:53:07 q+ to ask Sandro about spec text he thinks is unhelpful. 15:53:08 Guus, but SPARQL use the word "graph" in the phrase "named graph", in which case it's mutable 15:53:11 sandro: either us or the provenance WG has to do it 15:53:17 ... or a joint task force 15:53:30 q+ to ask a question to Richard 15:54:12 cygri: we should ask the provenance WG whether they have use cases that the SPARQL model does not solve 15:54:20 ack AndyS 15:54:20 AndyS, you wanted to ask Sandro about spec text he thinks is unhelpful. 15:54:54 Zakim, unmute me 15:54:54 MacTed should no longer be muted 15:54:55 q+ 15:55:55 sandro: for provenance use cases, you need a tight association between the triples in a graph, and the URI we make provenance statements about 15:56:18 danbri_ has joined #rdf-wg 15:56:46 @Andy, this is related: what is named in SPARQL-update ? (imutable) graphs or slots? 15:57:01 containers and contained is the issue I see.... trying to respect vocal queue.... 15:57:02 gavinc: is there a way in SPARQL 1.1 to name a graph and freeze that graph? 15:57:24 AndyS: no 15:57:33 sandro: that's the kind of functionality that would be needed for provenance 15:57:41 ack pchampin 15:57:41 pchampin, you wanted to ask a question to Richard 15:57:45 ack me 15:57:46 my question to Richard: 15:58:06 if we define precisely a data model 15:58:16 what is the cost of defining the associated vocabulary? 15:58:38 :-) 15:58:45 Freeze a value is HTTP GET :-) 15:58:56 (I think you stated that we need to define a data model, but not the vocabulary for representing it) 15:59:00 But that doesn't freeze the NAME 15:59:01 yes, I'm suggesting that it is low 15:59:18 I think that's really all we need? 15:59:29 cygri: provenance is a complex issue 15:59:46 ... ppl have been working for quite a while on an implicit named graph model 16:00:00 Name is the serialization if you want to use with a store with change-of-slot value. Or use store that uses naming how you want. 16:00:05 ... we need to provide an explicit model for them (charter of the groupe) 16:00:09 s/named graph/multi-graph/ 16:00:20 q+ there are two kinds of vocabularies. 16:00:25 ... but we don't need to enter into detail provenance use cases 16:00:28 q? 16:00:54 MacTed: i see a lack of coherent account of containers and contained structures 16:01:09 ... in the SQL world, there's a concept of a database 16:01:13 q+ to say there are two kinds of vocabularies. 16:01:17 .. that's somewhat equivalent to RDF store 16:01:29 ... a database contains catalogs, which contains schemas, which contains tables 16:01:41 ... RDF stores contain graphs, which are somewhat equivalent to catalogs 16:01:49 ... (or that's how people use them) 16:02:11 ... the term graph here is horribly overused 16:02:27 ... it can mean catalog, snapshot, etc etc 16:02:30 Also amusing issue with Turtle in HTML: http://dvcs.w3.org/hg/rdf/raw-file/tip/rdf-turtle/index.html is a set of RDF Graphs, Dataset? ;) 16:02:37 ... we need more specific terms for more specific things 16:02:59 ... without more specific terms, provenance cannot succeed 16:03:04 kinda +1 to MacTed 16:03:09 Zakim, mute me 16:03:09 MacTed should now be muted 16:03:16 +1 to MacTed 16:03:17 ack 16:03:27 q? 16:03:30 ack me 16:03:34 gavinc: i had multiple bits of turtle in one html file. what should i call these things? 16:03:40 Zakim, mute me 16:03:40 MacTed should now be muted 16:03:43 q- 16:03:44 davidwood: how about dataset? 16:03:56 gavinc: well but that's a SPARQL term 16:04:24 FabGandon: for me there are two kinds of vocabularies, domain dependent and domain independent 16:04:29 ack FabGandon 16:04:30 FabGandon, you wanted to say there are two kinds of vocabularies. 16:04:54 ... for the provenance WG, something like attaching the date to a graph should be defined not by the RDF WG, but by the prov WG 16:04:57 an RDF dataset is {G,(URIi, Gi)} -- Gavinc seems to want {Gi} (set of graphs? set of g-texts?) 16:05:10 date of g-text falls into the "metadata" about that g-text ... but g-text needs that name in order to have such metadata... 16:05:18 Yeah, how do those graphs in the Turtle HTML get named? 16:05:28 ... but stating that something is a g-text should be done by the RDF WG 16:05:29 +1 we should be doing the vocab for talking about RDF (eg the replacement for RDF reification) 16:06:22 gavinc: my problem was that i have a single HTML document that contains multiple ... g-texts? g-snaps? that are not particularly named ... 16:06:23 they get named via HTML id's, I think. 16:06:36 ... so it's multiple graphs in one document 16:06:50 AndyS: there are other collections of RDF graphs that are not RDF datasets 16:07:06 gavinc: yes and we need more specific terms 16:07:20 sandro: this should say SPARQL dataset, not RDF dataset 16:07:28 AndyS: SPARQL term is RDF dataset 16:07:52 I would object to redefining the term "RDF dataset" 16:09:11 sandro: dataset is such a good term, it's a shame to use it for such a peculiar thing as a SPARQL dataset 16:09:23 graphset ? 16:10:12 sandro: so maybe the turtle texts each get their html ID, forming named graphs in the dataset, and the RDF or un-id'd turtle goes into the default graph. 16:10:22 guus: i think we made some progress here. is it useful to involve the provenance wg? 16:10:31