14:26:05 RRSAgent has joined #rdf-wg 14:26:05 logging to http://www.w3.org/2012/03/28-rdf-wg-irc 14:26:07 RRSAgent, make logs world 14:26:07 Zakim has joined #rdf-wg 14:26:09 Zakim, this will be 73394 14:26:09 ok, trackbot; I see SW_RDFWG()11:00AM scheduled to start in 34 minutes 14:26:10 Meeting: RDF Working Group Teleconference 14:26:10 Date: 28 March 2012 14:30:15 mischat_ has joined #rdf-wg 14:43:53 mischat has joined #rdf-wg 14:53:22 danbri_ has joined #rdf-wg 14:55:42 pchampin has joined #rdf-wg 14:57:29 SW_RDFWG()11:00AM has now started 14:57:36 +??P5 14:57:41 zakim, ??P5 is me 14:57:41 +AndyS; got it 14:58:59 cgreer has joined #rdf-wg 14:59:34 +Guus_Schreiber 14:59:46 +cgreer 15:00:13 +Arnaud 15:00:30 +pfps 15:01:12 +??P12 15:01:23 zakim, dial ivan-voip 15:01:23 ok, ivan; the call is being made 15:01:25 +Ivan 15:01:25 zakim, ??P12 is me 15:01:25 +pchampin; got it 15:01:33 +gavinc 15:01:35 +[OpenLink] 15:01:41 pfps has joined #rdf-wg 15:01:50 Zakim, [OpenLink] is temporarily me 15:01:50 +MacTed; got it 15:01:54 Zakim, mute me 15:01:54 MacTed should now be muted 15:02:24 +sandro 15:02:45 +davidwood 15:02:53 +mhausenblas 15:03:02 +??P21 15:03:08 cygri has joined #rdf-wg 15:03:41 zakim, who is on the phone? 15:03:41 On the phone I see AndyS, Guus_Schreiber, cgreer, Arnaud (muted), pfps, pchampin, Ivan, gavinc, MacTed (muted), sandro, davidwood, mhausenblas, ??P21 15:03:42 Zakim, ??P21 is me 15:03:44 +yvesr; got it 15:03:50 zakim, mhausenblas is temporarily me 15:03:50 +cygri; got it 15:04:08 +LeeF 15:05:14 scribe: sandro 15:05:20 proposed: accept minutes from last week 15:05:28 resolved: accept minutes from last week 15:05:34 topic: action item review 15:06:09 + +1.603.897.aaaa 15:06:14 https://www.w3.org/2005/06/tracker/users/my 15:06:28 davidwood: Anyone want to own up to progress on any actions? 15:06:42 AndyS: I'm waiting for word from the editors that the draft is ready for review. 15:06:43 Souri has joined #rdf-wg 15:06:51 scribe: LeeF 15:06:58 I claim victory on 150, and ...mostly... 155 15:07:04 I had the same understanding 15:07:15 http://www.w3.org/2011/rdf-wg/track/actions/150 15:07:20 ACTION-150? 15:07:20 ACTION-150 -- Gavin Carothers to fail to update Turtle/N-Triples grammar to not use \u escaping in the local part of prefix names -- due 2012-03-21 -- OPEN 15:07:20 http://www.w3.org/2011/rdf-wg/track/actions/150 15:07:31 Zakim, NickH is with yvesr 15:07:31 +NickH; got it 15:07:44 close ACTION-150 15:07:44 ACTION-150 Fail to update Turtle/N-Triples grammar to not use \u escaping in the local part of prefix names closed 15:07:45 gavinc: it was really easy and it did work 15:07:54 AndyS: Is the doc ready for review? 15:08:01 gavinc: for my part, yes. ask eric too 15:08:19 AndyS: Excellent. OK if there are still bits to do 15:08:34 gavinc: right, let's see if Eric is happy with the collections in turtle bits. in section 4 15:08:48 FabGandon has joined #rdf-wg 15:08:54 Topic: Turtle 15:09:02 +ericP 15:09:38 davidwood: tried to close all actions and issues from last week that were already closed and left ones that need further discussion. gavin, eric, andy, pierre-antoine -- should we go through actions or just pick off select ones? 15:10:23 15:10:37 scribenick: LeeF 15:10:45 +FabGandon 15:11:43 ISSUE-74? 15:11:43 ISSUE-74 -- Prefixed names and slashes -- open 15:11:43 http://www.w3.org/2011/rdf-wg/track/issues/74 15:12:02 ACTION-151? 15:12:02 ACTION-151 -- Gavin Carothers to update issue 74 with new escaping rules -- due 2012-03-21 -- OPEN 15:12:02 http://www.w3.org/2011/rdf-wg/track/actions/151 15:12:11 davidwood: status of 151? 15:12:19 gavinc: ok to close, but should also close ISSUE-74 15:12:26 ... no longer valid 15:12:28 close ACTION-151 15:12:28 ACTION-151 Update issue 74 with new escaping rules closed 15:13:41 ACTION-152? 15:13:41 ACTION-152 -- Gavin Carothers to create new issue for :'s in the local part of prefix names -- due 2012-03-21 -- OPEN 15:13:41 http://www.w3.org/2011/rdf-wg/track/actions/152 15:13:46 Closed ISSURE-74 15:13:53 Closed ISSUE-74 15:14:13 gavinc: issue is FB Open Graph Protocol, which have URIs with colons in local names, particularly in RDFa representation. When Facebook serializes it, you either get invalid Turtle (includes colons), or you get back full IRIs 15:14:22 close ISSUE-74 15:14:22 ISSUE-74 Prefixed names and slashes closed 15:14:43 ericP: if we had influence, we could persuade them to put a backslash before the colon and all would be ok 15:14:50 davidwood: shoudl we really be changing the standard based on this? 15:15:01 gavinc: problem is that RDFa is OK with it 15:15:19 ... facebook is confused because the 2 specs are confusing 15:15:23 davidwood: 15:16:09 davidwood: does anyone object to this change in turtle? 15:16:35 q+ how does this affect SPARQL? 15:16:39 q+ to ask how does this affect SPARQL? 15:16:42 looking at http://ogp.me/ seeing og:locale:alternate and og:image:url and music:album:track etc... 15:16:46 ericP: seems like a safe change 15:17:04 Open Graph Protocol changes regularly... 15:17:33 +1 allowing colons in local parts 15:17:54 +1 to sandro, i.e., transitively, to allow colons in local parts 15:18:01 ericP: 2 languages that are similar, one syntax difference that pepole are tripping over. other differences which haven't been come against. make our lives easier by allowing colons? 15:18:12 q? 15:18:18 ack LeeF 15:18:18 LeeF, you wanted to ask how does this affect SPARQL? 15:18:22 having colons in the local part is a bit confusing but it does look nice 15:18:26 d'oh 15:18:28 like a URN ;-) 15:18:31 forgot about SPARQL 15:18:42 LeeF: Uhh guys, what about SPARQL? 15:18:42 Just on ":" (and not "/"). Technically, can be done (I think, not checked), but it breaks a lot of existing software in a confusing way. 15:18:53 Everyone: Oh. 15:19:14 ivan: not a question of RDFa or SPARQL. there are millions of pages that already do this 15:19:34 sandro: this isn't just a gorilla, it's sensible design 15:19:49 ivan: would it trigger a new last call for SPARQL? 15:20:17 lee: There might be a new LC for SPARQL, so there MAY be a window here. 15:20:54 q+ 15:21:02 lee: If you want to do this, ask the SPARQL WG *today* please 15:21:13 q+ 15:21:21 ack AndyS 15:21:25 AndyS: think it's technically possible, i could check 15:21:33 ... i don't think it's just the 2 WGs involved 15:21:39 ... convincing implementers of this change shouldn't be dismissed 15:21:44 ... turtle has been this way for a very long time 15:22:00 ... initial reaction will certainly be pushback (active and passive) 15:22:07 q? 15:22:13 q+ 15:22:18 ack Arnaud 15:22:23 ack Arnaud 15:22:26 PatH has joined #rdf-wg 15:22:38 q+ 15:22:41 Sorry Im late (again). 15:22:42 Arnaud: we could consider asking facebook if they'd be willing to change. even though they are really big, they have a couple of lines of code to change in one place which might make the change everywhere 15:22:53 ack sandro 15:22:55 gavinc: not enough for FB to change their parser, everyone who has created a web page using this has to change also 15:23:10 +PatH 15:23:21 q- 15:23:26 sandro: this is OK in RDFa, so the issue is only when a OGP doc is expressed in turtle. why can't it be done as a qname type thing with a backslash after the prefix? 15:23:42 me:sandro foaf:likes music:album\:disc 15:24:10 gavinc: yes 15:24:24 sandro: that seems like a one line code change somewhere, in facebooks Turtle serializer. 15:24:33 gavinc: not as big an issue now that we have reserved character escaping 15:24:45 ... now it's merely an issue that people will make, rather than an unavoidable error 15:25:02 sandro: easy error for turtle parser to catch and fix / help with 15:25:18 q+ 15:26:06 ack AndyS 15:26:23 What about a:b:c:d which is currently legal (as "a:b :c :d ." ) 15:26:28 AndyS: there is a corner case (see above) 15:26:42 AndyS: this changes turtle if you've been taking advantage of whitespace being optional 15:26:56 i'm willing to let that case go 15:26:57 sandro: off with their heads! 15:27:22 AndyS: probably only a potential issue with machine outputted turtle 15:27:28 (luddites emerge) 15:27:40 ceci n'est pas un pipe 15:28:01 ceci n'est pas une pipe 15:28:04 ACTION: Andy to see if allowing colons in prefix names is an ambiguity 15:28:04 Created ACTION-160 - See if allowing colons in prefix names is an ambiguity [on Andy Seaborne - due 2012-04-04]. 15:28:41 ericP: is that action in SPARQL? 15:28:46 AndyS: if it works in SPARQL, it will work in Turtle 15:29:20 sandro: what about asking the SPARQL WG? 15:30:21 LeeF: will the RDF WG's decision be predicated on SPARQL WG's willingness? 15:30:26 sandro: would rather not have to make that decision now 15:30:27 LeeF: OK 15:30:28 oh no a Lethal Turtle.... 15:31:24 gavinc: they're already generating legal turtle (with full IRIs) 15:31:32 AndyS: that de-motivates the change, right? 15:31:43 sandro: we'd be accommodating everyone out in the world who is writing OGP data 15:31:52 gavinc: issue is more complicated than.... 15:31:53 s/writing/working with/ 15:32:23 davidwood: can ask FB to change, or ask SPARQL WG to work with us to align RDF + RDFa + SPARQL 15:32:40 ... i think that taking this opportunity to align w3c syntaxes is a useful and valuable thing to do 15:32:48 sandro: we should look at all characters that RDFa allows there, then 15:32:59 davidwood: perfect is the enemy of good 15:33:18 q+ to say we get to make exactly one step 15:33:41 sandro: i'm happy with adding colon to sparql. andy said we should ask FB to change. we'd be asking them to change their predicate names, and there's no way they'll do that 15:33:50 AndyS: that's not the case - just asking them to output legal turtle 15:33:54 sandro: but that's not actually a problem 15:34:03 sandro: they'd have to change predicate names to solve this problem 15:34:09 AndyS: they can continue to use existing predicate names 15:34:45 ericP: right now, to use existing pred names they can output full URIs - not as pleasant to use - or can output prefix names with backslashes. with changes, they can output prefix names without backslashes. it's a nice practice in general to have local names without characters that need to be escaped 15:34:57 swh has joined #rdf-wg 15:35:09 ericP: we only get to make one step here - backslashes are new, so there arent a lot of implementations to stay compatible with. 15:35:17 FB can currently either use , and will be able to use a:b\:c (future turtle) so it's style, not technical. 15:35:59 CURIEs which post-date SPARQL, choose to differ. 15:36:22 ericP: tried to get CURIEs into SPARQL, but couldn't be done because of rest of the grammar. property paths make this worse 15:36:32 ... don't think we'll ever have CURIEs fitting in turtle and sparql 15:36:32 Agreed, AndyS. I think our motivation is just to make FB's vocab easier to work with in Turtle and Sparql (no need for \ ) 15:36:35 AndyS: agree 15:36:42 q? 15:36:50 ack me 15:36:50 ericP, you wanted to say we get to make exactly one step 15:36:55 ack ericP 15:37:08 davidwood: gavin, will you write up this issue? 15:37:11 gavinc: ayup 15:37:21 sandro: can we just make a decision? 15:38:39 PROPOSED: Change the Turtle syntax to allow colons in local parts, IFF the SPARQL WG agrees to change SPARQL to match, in order to make it easier for people to work with facebook's style of property names. 15:38:54 +1 15:38:56 +1 15:38:59 +1 15:39:00 +1 15:39:01 +1 15:39:01 +0 15:39:04 +1 15:39:05 abstain 15:39:06 0 (-1 if there is any grammar ambiguity) 15:39:10 +1 15:39:27 (assuming no grammar ambiguity of course) 15:39:32 0 15:39:47 (other than whitespace. a:b:c:d ) 15:39:57 +1 15:40:10 ericP: that backwards incompatibility is something that I'm ok with 15:40:54 PROPOSED: Change the Turtle syntax to allow colons in local parts, IFF the SPARQL WG agrees to change SPARQL to match, in order to make it easier for people to work with facebook's style of property names. We understand this will change how turtle lines like a:b:c:d are parsed; we don't need to support documents written like that. 15:41:04 +1 15:41:06 (this seems like solving a problem that probably doesn't exist, but it may someday make it easier for me to write turtle, so... *shrug*. The non-WG-member implementers may howl.) 15:41:21 abstain (this is not a problem) 15:41:21 s/(this/LeeF: (this/ 15:41:37 abstain (this is not a problem) 15:42:24 sandro: It's not nearly as big a problem as it sounded at the begging of this problem. 15:43:22 AndyS: there's not a technical problem, we're asking implementers and others to do something 15:43:30 sandro: agree that it's a style issue at this point 15:43:46 davidwood: we could solve this just by acknowledging it's not a problem, closing the issue, etc. 15:43:55 sandro: the people who would be unhappy are those who would have to write lots of backslashes 15:44:43 # is also a contentious character for the same reasons. 15:45:18 +1 15:45:28 davidwood: I propose that we close ACTION-152, take no action, and open no new issue 15:45:35 sandro: I won't object, but I don't think this is a good idea 15:45:46 -0 I think we'll be very sad writing a lot of backslashes for a lot of years 15:46:14 i note that sandro and i differ only in the degree to which we are concearned about facebook users. we both agree that a change is the best choice 15:46:17 but whatever. I tend to be perfectionist about things like that. 15:46:23 backslashes have rights too! 15:46:24 -0 I think we'll be writing parsers that accept invalid Turtle in a year or two 15:46:38 Oh, YES gavin. 15:46:56 zakim, mute me 15:46:56 PatH should now be muted 15:47:24 If we do this, then do it properly : not rush a proposal through ATM. Ask FB + check the grammars. 15:47:43 +1 Andy 15:48:00 gavinc: other issue: turtle does not specify how errors are handled 15:48:09 ... so it isn't clear to me that if i parsed this that i would not be conforming 15:48:24 ericP: we have a language, and that's not the language 15:48:32 gavinc: but conformance is expressed in terms of positive parsing 15:48:46 I suggest that editors make a proposal with "feature at risk" note, and ask explicit feedback from parties concerned 15:49:13 Zakim, unmute me 15:49:13 MacTed should no longer be muted 15:49:33 +1 Guus. Lets be strictly proper. diplomacy and all. 15:50:09 +1 Guus! 15:50:19 davidwood: let's accept guus's suggestion 15:50:32 Zakim, mute me 15:50:32 MacTed should now be muted 15:50:50 davidwood: andy, will you ask SPARQL WG about this? 15:50:52 AndyS: ok 15:50:56 davidwood: what about asking FB? 15:50:59 sandro: there's nothing to ask FB! 15:51:17 not ask, but inform facebook. diplomacy. 15:51:25 FB is now generating valid Turtle. would whatever change we're contemplating *break* that? if not, nothing to ask/inform about... 15:51:40 /invite mr_facebook #rdf-wg 15:52:17 +1 to sandro's point 15:52:22 is there any example of that in the wild? 15:52:34 q+ 15:52:35 like for example maybe fbk are planning a change.... 15:52:40 noting that fb has some influece on how people expect to use their data 15:52:42 ack ivan 15:52:45 (of people writing broken Turtle following Facebook's OGP?) 15:53:20 sandro: So the question for facebook is whether they would prefer people to have to use backslashes in working with their data in turtle and SPARQL. 15:53:21 zakim, unmute me 15:53:21 PatH should no longer be muted 15:53:49 MarkZ has joined #rdf-wg 15:54:23 i'm all down with these colons and stuff 15:54:39 it does no actual harm to just *talk* to North Korea 15:54:47 q+ 15:54:54 ack AndyS 15:55:13 AndyS: if we do this, I think # will be the next character to come up - to handle fragments 15:56:14 (discussion of how that would conflict with comments) 15:56:45 what about non-breaking zero-width trailing combining character whitespace? 15:57:12 ACTION: Andy to ask SPARQL WG about Colon-gate 2012 15:57:12 Created ACTION-161 - Ask SPARQL WG about Colon-gate 2012 [on Andy Seaborne - due 2012-04-04]. 15:57:45 So -- the big motive for allowing colons in local names is because people will be writing parsers to accept them ANYWAY, since people will be writing them anyway. 15:57:46 davidwood: now for an easy topic 15:58:19 zakim, mute me 15:58:27 +Arnaud.a 15:58:34 -Arnaud 15:58:37 PatH should now be muted 15:58:47 crackle, buzz, beep, rattle, smack 15:59:13 let's pls finish the Turtle discussion, we have used enough time on Named Graphs to wait another week 15:59:26 -ericP 15:59:30 sandro - reasonable claim - do we have evidence (seems likely but you said "will be" which is absolute) 15:59:57 +ericP 16:00:15 http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/index.html#n-triples-compatibility 16:01:02 gavin: should a backwards incompatible escape sequence be an issue? 16:01:08 davidwood: won't other changes break b.c.? 16:01:10 gavinc: yes 16:01:15 davidwood: why is this a special issue? 16:01:18 AndyS, I have no evidence people will forget the backslashes, but I think I heard some parser writers on the call saying they expected they'd allow colons without backslashes, because they thought users would forget the backslashes. 16:01:19 gavinc: others are just with encoding 16:02:04 -Arnaud.a 16:02:08 AndyS: i've never seen any data with \b \f in turtle, so... 16:02:32 davidwood: there could be old data around with \b or \f 16:02:37 AndyS: would change from illegal to legal 16:02:41 davidwood: non issue for me 16:02:42 gavinc: me too 16:02:46 \b***** this for a topic of conversation. 16:03:00 +Arnaud 16:03:15 AndyS: i haven't seen anything that indicates this is anything more than a question 16:03:20 Zhe brought this up to Gavin off list. Gavin sent it to the list. 16:03:24 gavinc: offlist indications to the contrary 16:03:39 I suggest this is editorial discretion of the editors, seems small enough, but pls make an explicit note in the document 16:03:39 -Arnaud 16:03:42 s/gavinc: me too/gavinc: me too. Zhe brought it up 16:03:51 davidwood: let's leave this to the editors and move on 16:03:56 ISSUE-4? 16:03:56 ISSUE-4 -- Do we depecate N-Triples and use Turtle instead? -- open 16:03:56 http://www.w3.org/2011/rdf-wg/track/issues/4 16:04:05 +Arnaud 16:04:58 AndyS: people use n-triples, it's faster for loading 16:05:01 +1 16:05:03 +1 16:05:04 davidwood: should we resolve issue-4 and keep n-triples? 16:05:07 AndyS: yes 16:05:09 +1 16:05:20 +1 to not deprecating N-Triples 16:05:26 ivan: what about discussion with Oracle? 16:05:32 davidwood: Oracle wants to keep n-triples 16:05:44 1 16:05:46 (excellentuse of the word "apoplectic") 16:06:22 +1 to closing issue 4. 16:06:36 +1 16:06:41 +1 16:06:42 +1 to closing it with "not deprecating N-Triples" 16:06:58 RESOLVED: Close ISSUE-4, keeping N-triples as not deprecated 16:07:10 close ISSUE-4 16:07:10 ISSUE-4 Do we depecate N-Triples and use Turtle instead? closed 16:07:25 davidwood: all done with turtle! 16:07:32 davidwood: no time for named graph semantivcs 16:07:48 davidwood: let's put solution design .1 on the queue for next week 16:07:54 davidwood: AOB? 16:07:58 Guus: how about clarifying questions? 16:08:15 zakim, unmute me 16:08:15 PatH should no longer be muted 16:08:16 Guus: the rdf:Graph construct is not used in a use case. is that by accident? 16:08:21 +q 16:08:29 ... they used GraphStateContainer construct 16:08:45 sandro: i assume he'd have to use it for the inference case (separation of inference and one of the two ways you'd do endorsement) 16:10:09 -gavinc 16:10:31 Zakim, unmute me 16:10:31 MacTed should no longer be muted 16:10:37 Guus: from use cases, we want to see that if you fetch something at a particular time, there's no guarantee that it will be there at a future time 16:11:04 sandro: quick strawpoll here 16:11:11 ... i intentionally left g-boxes out of 6.1 design 16:11:17 ... because they seem like an extension point 16:11:22 ... should i put them in properly? 16:11:29 PatH: NO. 16:11:36 ... let's get it clear without the g-boxes 16:11:38 sandro: i agree 16:12:27 PatH: is the association, the rdf:graph connection, holds between a resource and a graph? 16:12:28 sandro: yes 16:12:39 PatH: is that intended to extend beyond the document? is it asserted by the document? 16:12:41 sandro: yes 16:12:47 PatH: consider 2 docs with no default graph. 16:12:50 u1 { a b c} 16:12:51 and 16:12:54 u1 { d e f} 16:13:08 do those conclude: u1 {a b c . d e f } ? 16:13:15 sandro: no, those are logically inconsistent 16:13:34 PatH: i think there is something that needs to be fixed then 16:14:47 mischat has joined #rdf-wg 16:16:47 PatH: above case says it's not ok to use same label in mutiple documents -- in ??? semantics, it's perfectly fine and entails the merge 16:17:04 sandro: if you're going to do that you need to be explicit about that fact 16:17:18 +1 to PatH 16:17:19 PatH: a lot of our discussion turns on people having different intuitions about this point 16:17:29 {...} is not automatically a graph, it can be part of a graph but all graph must be in same doc. 16:17:43 davidwood: more next week 16:17:48 ( "this point" being what happens when you have u1 { a b c} in one document and u1 { d e f} in another. 16:17:49 - +1.603.897.aaaa 16:17:50 davidwood: Adjourned. 16:17:52 -PatH 16:17:54 -Arnaud 16:17:56 -sandro 16:17:57 -ericP 16:17:58 -davidwood 16:17:59 -Ivan 16:18:01 -cgreer 16:18:02 -Guus_Schreiber 16:18:02 -yvesr 16:18:04 -MacTed 16:18:05 -LeeF 16:18:09 danbri has joined #rdf-wg 16:18:14 -cygri 16:18:17 MarkZ has joined #rdf-wg 16:18:17 -pchampin 16:18:20 -AndyS 16:18:29 c'mon common scribe, get moving 16:18:52 -FabGandon 16:18:56 sandro - right - hence the requirement seems to be whole graph in doc -- may be fragmented in concrete syntax, not in abstract syntax. 16:25:13 -pfps 16:25:15 SW_RDFWG()11:00AM has ended 16:25:15 Attendees were AndyS, Guus_Schreiber, cgreer, Arnaud, pfps, Ivan, pchampin, gavinc, MacTed, sandro, davidwood, yvesr, cygri, LeeF, +1.603.897.aaaa, NickH, ericP, FabGandon, PatH 16:59:43 manu1_ has joined #rdf-wg 17:28:46 Is CURIE 1.0 the current definitive definition? Is there work in progress anywhere to change it? 17:29:03 (gavinc) 17:32:39 gavinc: was there a formal response to http://lists.w3.org/Archives/Public/public-rdfa-wg/2012Jan/0025.html ? 17:34:55 Hmm - curie ::= [ [ prefix ] ':' ] irelative-ref (as defined in [RFC3987]) but ... 17:36:38 irelative-ref => irelative-part => "//" iauthority ipath-abempty OR ipath-absolute ... but ipath-absolute starts with "/" so irelative-ref always starts with a "/" or "//". 17:44:16 LeeF has joined #rdf-wg 17:46:01 irelative-part also allows ipath-noscheme but that forbids "a:b" -- I see a bug. 18:20:57 Zakim has left #rdf-wg 20:04:04 AndyS, sorry had to work on displaying court cases with Javascript for a few hours ;) Yes, there was http://lists.w3.org/Archives/Public/public-rdf-wg/2012Jan/0091.html