14:58:27 <RRSAgent> RRSAgent has joined #rdf-wg
14:58:27 <RRSAgent> logging to http://www.w3.org/2013/05/29-rdf-wg-irc
14:58:29 <trackbot> RRSAgent, make logs world
14:58:29 <Zakim> Zakim has joined #rdf-wg
14:58:31 <trackbot> Zakim, this will be 73394
14:58:31 <Zakim> ok, trackbot; I see SW_RDFWG()11:00AM scheduled to start in 2 minutes
14:58:32 <trackbot> Meeting: RDF Working Group Teleconference
14:58:32 <trackbot> Date: 29 May 2013
14:58:45 <AndyS> zakim, this is 73394
14:58:45 <Zakim> AndyS, I see SW_RDFWG()11:00AM in the schedule but not yet started.  Perhaps you mean "this will be 73394".
14:59:24 <Zakim> SW_RDFWG()11:00AM has now started
14:59:30 <Zakim> +[GVoice]
14:59:42 <pfps> zakim, gvoice is me
14:59:42 <Zakim> +pfps; got it
15:00:04 <Zakim> +Sandro
15:00:05 <Zakim> +[IPcaller]
15:00:06 <AndyS> zakim, IPCaller is me
15:00:07 <Zakim> +AndyS; got it
15:00:23 <AndyS> zakim, who is making noise?
15:00:29 <TallTed> TallTed has joined #rdf-wg
15:00:34 <Zakim> AndyS, listening for 10 seconds I heard sound from the following: Sandro (47%)
15:00:37 <Zakim> +davidwood
15:00:42 <ivan> zakim, dial ivan-voip
15:00:43 <Zakim> ok, ivan; the call is being made
15:00:44 <Zakim> +Ivan
15:00:56 <pfps> zakim, mute me
15:00:57 <Zakim> pfps should now be muted
15:00:59 <ivan> zakim, mute me
15:00:59 <Zakim> Ivan should now be muted
15:01:00 <Zakim> +OpenLink_Software
15:01:51 <TallTed> Zakim, OpenLink_Software is temporarily me
15:01:51 <Zakim> +TallTed; got it
15:01:53 <TallTed> Zakim, mute me
15:01:53 <Zakim> TallTed should now be muted
15:01:55 <TallTed> TallTed has changed the topic to: RDF WG -- current agenda: http://www.w3.org/2011/rdf-wg/wiki/Meetings:Telecon2013.05.29
15:02:33 <davidwood> Zakim, who is here?
15:02:33 <Zakim> On the phone I see pfps (muted), Sandro, AndyS, davidwood, Ivan (muted), TallTed (muted)
15:02:35 <zwu2> zwu2 has joined #rdf-wg
15:02:36 <Zakim> On IRC I see TallTed, Zakim, RRSAgent, AndyS, pfps, Guus, gkellogg, SteveH, ivan, davidwood, manu, sandro, trackbot, yvesr, ericP, mischat
15:02:42 <Zakim> +??P36
15:02:47 <SteveH> Zakim, ??P36 is me
15:02:47 <Zakim> +SteveH; got it
15:03:01 <pfps> Ivan's comments are benign
15:03:27 <Zakim> +Guus_Schreiber
15:03:30 <Zakim> + +1.619.663.aaaa
15:04:07 <Zakim> +[GVoice]
15:04:16 <Zakim> +??P41
15:04:20 <zwu2> zakim, +1.619.663.aaaa is me
15:04:20 <Zakim> +zwu2; got it
15:04:55 <yvesr> Zakim, ??P41 is me
15:04:55 <Zakim> +yvesr; got it
15:04:57 <markus> markus has joined #rdf-wg
15:05:02 <zwu2> zakim, mute me
15:05:02 <Zakim> zwu2 should now be muted
15:05:20 <davidwood> Zakim, pick a victim
15:05:20 <Zakim> Not knowing who is chairing or who scribed recently, I propose Guus_Schreiber
15:05:33 <davidwood> Zakim, pick a victim
15:05:33 <Zakim> Not knowing who is chairing or who scribed recently, I propose AndyS
15:05:47 <davidwood> Zakim, pick a victim
15:05:47 <Zakim> Not knowing who is chairing or who scribed recently, I propose yvesr
15:05:50 <markus> zakmi, code?
15:05:54 <markus> zakim, code?
15:05:54 <Zakim> the conference code is 73394 (tel:+1.617.761.6200 sip:zakim@voip.w3.org), markus
15:06:00 <yvesr> sorry, very noisy around here
15:06:04 <davidwood> Zakim, pick a victim
15:06:04 <Zakim> Not knowing who is chairing or who scribed recently, I propose davidwood
15:06:13 <davidwood> Zakim, pick a victim
15:06:14 <Zakim> Not knowing who is chairing or who scribed recently, I propose AndyS
15:06:26 <Zakim> +??P9
15:06:26 <davidwood> Zakim, who is here?
15:06:28 <Zakim> On the phone I see pfps (muted), Sandro, AndyS, davidwood, Ivan (muted), TallTed (muted), SteveH, Guus_Schreiber, zwu2 (muted), ericP, yvesr, ??P9
15:06:28 <Zakim> On IRC I see markus, zwu2, TallTed, Zakim, RRSAgent, AndyS, pfps, Guus, gkellogg, SteveH, ivan, davidwood, manu, sandro, trackbot, yvesr, ericP, mischat
15:06:31 <markus> zakim, ??P9 is me
15:06:31 <Zakim> +markus; got it
15:06:39 <Guus> chair: davidwood
15:06:41 <yvesr> davidwood: i can give it a shot, but will need some help in case it gets too noisy here
15:06:42 <Arnaud> Arnaud has joined #rdf-wg
15:07:09 <davidwood> Topic: Admin
15:07:16 <davidwood> PROPOSED to accept the minutes of the 22 May telecon: https://www.w3.org/2013/meeting/rdf-wg/2013-05-22
15:07:26 <Zakim> +Arnaud
15:07:54 <markus> RESOLVED: accept the minutes of the 22 May telecon: https://www.w3.org/2013/meeting/rdf-wg/2013-05-22
15:08:02 <davidwood> Review of action items
15:08:02 <davidwood> http://www.w3.org/2011/rdf-wg/track/actions/pendingreview
15:08:02 <davidwood> http://www.w3.org/2011/rdf-wg/track/actions/open
15:08:05 <tbaker> tbaker has joined #rdf-wg
15:08:22 <pfps> I closed the items that were discussed at the last telecon.
15:08:24 <markus> davidwood: does anyone want claim some action items?
15:09:36 <markus> ... what about blank nodes as graph names?
15:09:44 <markus> andys: I did nothing
15:09:51 <markus> sandro: that should be on hold, right?
15:09:53 <davidwood> Topic: Turtle
15:10:04 <davidwood> Features at risk: proposal by Gavin:
15:10:04 <davidwood>  http://lists.w3.org/Archives/Public/public-rdf-wg/2013May/0173.html
15:10:12 <markus> davidwood: we have some features at risk
15:10:23 <davidwood> Poll at https://www.w3.org/2002/09/wbs/46168/at-sign/
15:10:28 <davidwood> PROPOSED: Allow PREFIX and BASE in Turtle, and thus keep the feature in the document. This is based on the poll results at https://www.w3.org/2002/09/wbs/46168/at-sign/results
15:10:29 <markus> ... we had a poll & based on the results of that poll we have a proposal
15:10:53 <Guus> zakim, who is talking?
15:11:04 <Zakim> Guus, listening for 10 seconds I heard sound from the following: AndyS (4%), davidwood (19%)
15:11:12 <sandro> zakim, mute andys
15:11:12 <Zakim> AndyS should now be muted
15:11:30 <sandro> zakim, unmute andys
15:11:30 <Zakim> AndyS should no longer be muted
15:11:32 <davidwood> Zakim, mute me
15:11:32 <Zakim> davidwood should now be muted
15:11:34 <sandro> zakim, who is on the call?
15:11:34 <Zakim> On the phone I see pfps (muted), Sandro, AndyS, davidwood (muted), Ivan (muted), TallTed (muted), SteveH, Guus_Schreiber, zwu2 (muted), ericP, yvesr, markus, Arnaud
15:11:41 <davidwood> Zakim, unmute me
15:11:41 <Zakim> davidwood should no longer be muted
15:11:44 <sandro> zakim, mute me
15:11:44 <Zakim> Sandro should now be muted
15:11:45 <sandro> zakim mute steveh
15:11:49 <sandro> zakim, mute steveh
15:11:49 <Zakim> SteveH should now be muted
15:11:49 <markus> zakim, mute me
15:11:50 <pfps> zakim, who is talking?
15:11:50 <Zakim> markus should now be muted
15:11:52 <AndyS> zakim, who is making noise
15:11:52 <Zakim> I don't understand 'who is making noise', AndyS
15:11:53 <sandro> zakim, mute ericP
15:11:54 <Zakim> ericP should now be muted
15:11:54 <zwu2> zakim, mute me
15:11:54 <Zakim> zwu2 was already muted, zwu2
15:11:57 <Guus> zakim, mute me
15:11:57 <Zakim> Guus_Schreiber should now be muted
15:12:00 <sandro> zakim, mute yvesr
15:12:00 <Zakim> yvesr should now be muted
15:12:03 <AndyS> zakim, who is noisy
15:12:03 <Zakim> I don't understand 'who is noisy', AndyS
15:12:04 <Zakim> pfps, listening for 12 seconds I heard sound from the following: davidwood (3%)
15:12:04 <Guus> zakim, unmute me
15:12:04 <Zakim> Guus_Schreiber should no longer be muted
15:12:05 <sandro> zakim, who is on the call?
15:12:05 <Zakim> On the phone I see pfps (muted), Sandro (muted), AndyS, davidwood, Ivan (muted), TallTed (muted), SteveH (muted), Guus_Schreiber, zwu2 (muted), ericP (muted), yvesr (muted), markus
15:12:05 <Zakim> ... (muted), Arnaud
15:12:10 <sandro> zakim, mute markus
15:12:10 <Zakim> markus was already muted, sandro
15:12:15 <AndyS> zakim, who is making noise?
15:12:20 <sandro> zakim, mute andys
15:12:20 <Zakim> AndyS should now be muted
15:12:25 <Zakim> AndyS, listening for 10 seconds I heard sound from the following: davidwood (27%)
15:12:28 <sandro> zakim, unmute me
15:12:28 <Zakim> Sandro should no longer be muted
15:12:31 <markus> zakim, unmute me
15:12:31 <Zakim> markus should no longer be muted
15:12:37 <ericP> ack me
15:12:47 <Zakim> -AndyS
15:13:04 <PatH> PatH has joined #rdf-wg
15:13:17 <Zakim> +[IPcaller]
15:13:24 <davidwood> PROPOSED: Allow PREFIX and BASE in Turtle, and thus keep the feature in the document. This is based on the poll results at https://www.w3.org/2002/09/wbs/46168/at-sign/results
15:13:27 <AndyS> zakim, ipcaller is me
15:13:29 <Zakim> +AndyS; got it
15:13:32 <ericP> +1
15:13:48 <ivan> +1
15:13:58 <SteveH> "You're not allowed to see the results of this questionnaire."
15:14:02 <AndyS> Make public?
15:14:02 <SteveH> from the link
15:14:04 <TallTed> +1
15:14:09 <Guus> +1
15:14:10 <Zakim> +PatH
15:14:12 <zwu2> +1
15:14:28 <davidwood> No opinion	2
15:14:28 <davidwood> Slight preference for allowing PREFIX and BASE	5
15:14:28 <davidwood> Slight preference for disallowing PREFIX and BASE	4
15:14:28 <davidwood> Strong preference for allowing PREFIX and BASE, but can live with disallowing	4
15:14:28 <davidwood> Strong preference for disallowing PREFIX and BASE, but can live with allowing	1
15:14:28 <davidwood> We must allow PREFIX and BASE; I cannot live with disallowing.
15:14:28 <davidwood> We must disallow PREFIX and BASE; I cannot live with allowing.
15:14:35 <SteveH> tat's fine
15:14:44 <markus> davidwood: the last two are 0
15:15:17 <markus> ... 9 with preference for keeping PREFIX and BASE vs. 5 for disallowing them
15:15:57 <davidwood> RESOLVED: Allow PREFIX and BASE in Turtle, and thus keep the feature in the document. This is based on the poll results at https://www.w3.org/2002/09/wbs/46168/at-sign/results
15:15:57 <davidwood> 
15:16:08 <AndyS> send email to comments?
15:16:21 <davidwood> Zakim, who is here?
15:16:21 <Zakim> On the phone I see pfps (muted), Sandro, davidwood, Ivan (muted), TallTed (muted), SteveH (muted), Guus_Schreiber, zwu2 (muted), ericP, yvesr (muted), markus, Arnaud (muted),
15:16:24 <Zakim> ... AndyS, PatH
15:16:24 <Zakim> On IRC I see PatH, tbaker, Arnaud, markus, zwu2, TallTed, Zakim, RRSAgent, AndyS, pfps, Guus, gkellogg, SteveH, ivan, davidwood, manu, sandro, trackbot, yvesr, ericP, mischat
15:16:45 <markus> ... Andy, do you have some comments regarding the Turtle test suite?
15:17:13 <markus> AndyS: no, not really. We need to freeze it so that we can start asking for official conformance reports
15:17:25 <markus> davidwood: I'm not sure what the procedure here is
15:17:45 <sandro> q+
15:18:01 <davidwood> ack sandro
15:18:03 <markus> ???: we basically just need to decide that they are done. Some people asked for changes
15:18:20 <markus> ... we need to check whether all the requests have been handled
15:18:55 <markus> sandro: editorial changes are required. We need to ensure that all links point to the same test suite
15:19:10 <markus> davidwood: eric, can you make those editorial changes?
15:19:14 <markus> s/???/eric/
15:19:25 <markus> eric: I can do it in 2 weeks.. busy before that
15:20:08 <markus> ... there are some issues regarding the base. Is a README file sufficient?
15:20:13 <markus> sandro: fine for me
15:20:34 <AndyS> https://dvcs.w3.org/hg/rdf/raw-file/040e24cdacf2/rdf-turtle/reports/index.html
15:21:27 <AndyS> where does the reports/index.html go?
15:21:51 <sandro> please fix: http://www.w3.org/2011/rdf-wg/wiki/Turtle_Test_Suite and http://www.w3.org/2011/rdf-wg/wiki/Turtle-CR-Request#Implementation_Information and http://www.w3.org/2011/rdf-wg/wiki/Main_Page
15:22:20 <markus> eric: another question, should we add an archive (tar/zip)?
15:22:28 <markus> sandro: that would be great I think
15:23:33 <davidwood> Topic: Other documents
15:23:58 <markus> davidwood: does anyone has any comments for Concepts, Schema, or JSON-LD?
15:24:08 <ericP> ACTION: ericP to add tarball to http://www.w3.org/2013/TurtleTests/
15:24:08 <trackbot> Created ACTION-266 - Add tarball to http://www.w3.org/2013/TurtleTests/ [on Eric Prud'hommeaux - due 2013-06-05].
15:24:24 <markus> ... I think we came to the conclusion to leave RDF/XML alone and not re-publish it
15:24:28 <ericP> ACTION: ericP to update http://www.w3.org/2011/rdf-wg/wiki/Turtle_Test_Suite http://www.w3.org/2011/rdf-wg/wiki/Turtle-CR-Request#Implementation_Information http://www.w3.org/2011/rdf-wg/wiki/Main_Page to point to http://www.w3.org/2013/TurtleTests/
15:24:28 <trackbot> Created ACTION-267 - Update http://www.w3.org/2011/rdf-wg/wiki/Turtle_Test_Suite http://www.w3.org/2011/rdf-wg/wiki/Turtle-CR-Request#Implementation_Information http://www.w3.org/2011/rdf-wg/wiki/Main_Page to point to http://www.w3.org/2013/TurtleTests/ [on Eric Prud'hommeaux - due 2013-06-05].
15:25:05 <markus> arnaud: I can work on it the second part of June
15:25:14 <yvesr> s/arnaud/Guus
15:25:15 <davidwood> s/arnaud/guus/
15:25:19 <ericP> ACTION: ericP to review tests comments in http://www.w3.org/2011/rdf-wg/wiki/Turtle_Candidate_Recommendation_Comments to make sure all are addressed
15:25:19 <trackbot> Created ACTION-268 - Review tests comments in http://www.w3.org/2011/rdf-wg/wiki/Turtle_Candidate_Recommendation_Comments to make sure all are addressed [on Eric Prud'hommeaux - due 2013-06-05].
15:25:31 <markus> ... there's actually not that much to do. So we should still be able to get a version out for summer (primer)
15:25:46 <Zakim> -PatH
15:26:11 <Zakim> +PatH
15:26:24 <davidwood> Topic: JSON-LD LC2
15:26:39 <ericP> davidwood, can i get a second to help with ACTION-268? it's kinda big
15:26:43 <markus> http://www.w3.org/2011/rdf-wg/wiki/JSON-LD_Features_at_Risk
15:27:04 <davidwood> ericP: ack
15:27:25 <PatH> sandro's echo is arguing with him.
15:27:33 <davidwood> AndyS, would you be willing to help ericP with ACTION-268?
15:27:36 <markus> sandro: we didn't have a chance to talk about the round-tripping issue and the use of futures
15:27:48 <markus> ... so we may wanna leave them as feature as risk throughout CR
15:28:05 <ericP> AndyS, it involves reading through the comments and responding to the commenters
15:28:21 <sandro> sandro: Use of futures SHOULD remain at risk, as per Director
15:28:24 <ericP> http://www.w3.org/2011/rdf-wg/wiki/Turtle_Candidate_Recommendation_Comments has a place where one can sign up to "own" an issue
15:28:41 <markus> davidwood: I think we should spend the rest of our time with concepts and semantics
15:28:43 <ericP> s/an issue/a comment/
15:28:44 <sandro> sandro: but it would be nice to resolve others.
15:28:54 <davidwood> Topic: LC Drafts of Concepts and Semantics
15:29:16 <markus> davidwood: I think the big issue to knock out today is ISSUE-131
15:29:20 <davidwood> ISSUE-131?
15:29:20 <trackbot> ISSUE-131 -- How can one create an RDF dataset without being a web server? -- open
15:29:20 <trackbot> http://www.w3.org/2011/rdf-wg/track/issues/131
15:29:34 <AndyS> is the wiki up to date?
15:29:44 <markus> ... we have some proposals
15:29:44 <AndyS> (about turtle test comments)
15:29:56 <davidwood> PROPOSED-A: Close ISSUE-131 with rdf-concepts saying that RDF syntaxes and toolkits MAY support RDF (graph and/or dataset) serializations which use relative URIs and a base that is only determined by the receiver.
15:29:56 <davidwood> 
15:29:56 <davidwood> PROPOSED-B: Close ISSUE-131 by allowing blank nodes as graph names.
15:29:56 <davidwood> 
15:29:56 <davidwood> PROPOSED-C: Close ISSUE-131 without adding any text on this subject to our specifications.
15:31:04 <markus> sandro: last week I talked about the difficulty I came across when creating a dataset when you are not a web server (as you can't make up good URIs)
15:31:23 <markus> ... I proposed to allow bnodes as graph names
15:31:37 <markus> ... that caused some pushback
15:31:40 <davidwood> q?
15:31:53 <markus> ... thus I created those three proposals
15:32:00 <SteveH> I think A is more-or-less what we do in this situation
15:32:13 <ericP> q+ to ask what these relative URIs look like in an RDF DB
15:32:16 <SteveH> e.g. piped in data on the command line
15:32:20 <pfps> I am totally confused as to how PROPOSED-A could possibly work.
15:32:22 <markus> ... one is to allow relative URIs even though you don't know how to resolve them (bit like hand-waving)
15:32:35 <AndyS> link to proposals email therad
15:32:38 <AndyS> ?
15:32:41 <davidwood> ack ericP
15:32:41 <Zakim> ericP, you wanted to ask what these relative URIs look like in an RDF DB
15:33:18 <markus> eric: is that approach possible? What do I put in for the URIs if I put such data in my DB?
15:33:48 <pfps> +1 to eric
15:33:53 <markus> ... there needs to be some specific ID to prevent clashes
15:33:53 <SteveH> q+ to ask what people do now?
15:34:16 <markus> sandro: I know I made that proposal but since I don't like it I don't wanna defend it
15:34:55 <SteveH> +1 to Pat re. option A and B
15:35:02 <pfps> options?  aren't we on "PROPOSED"
15:35:07 <markus> pat: my reaction is that options 1 and 2 are possible but should not be mandated
15:35:12 <AndyS> http://lists.w3.org/Archives/Public/public-rdf-wg/2013May/0199.html -- NB Proposals do not align to options
15:35:26 <markus> ... I'm in favor of allowing bnodes as graph names (option 3)
15:36:07 <sandro> s/option 3/option 4/
15:36:23 <pfps> as far as semantics is concerned there is no issue with relative IRIs, so long as they are resolved by the time that the graph is constructed, but this resolution is the issue
15:37:06 <davidwood> ack SteveH
15:37:07 <Zakim> SteveH, you wanted to ask what people do now?
15:37:49 <markus> markus: JSON-LD allows that (currently feature at risk)
15:38:14 <sandro> SteveH: getting data from the command line ... something other than fetching it ... we just use the standard base as if it was sucked from that location.
15:38:22 <AndyS> There is always a base - if nothing else, the system has a fallback one (for us, usually unique)
15:38:42 <yvesr> SteveH, could you give some examples of some of these issues?
15:38:42 <davidwood> We do exactly the same thing that SteveH does - and agree that it feels "a bit dirty", but it works.
15:38:45 <sandro> SteveH: Having blank nodes as graph identifiers would be a mistake
15:39:18 <davidwood> q?
15:39:24 <sandro> "* ivan rdflib actually implements bnodes as graph names..."
15:39:31 <AndyS> PROPOSED-C is the "no change"
15:39:59 <yvesr> SWI does as well afair
15:40:32 <PatH> +1
15:40:51 <PatH> q
15:41:03 <PatH> q+
15:41:06 <yvesr> i confirm it does indeed work in swi-prolog too
15:41:26 <AndyS> Why not just the text as it is?  RFC says there is always a base.
15:41:34 <SteveH> +! to AndyS
15:41:40 <davidwood> From the JSON-LD minutes: RESOLUTION: JSON-LD will continue to support blank node identifiers for properties and graph names. When converting data to RDF 1.1, the specification will not introduce any special checks to handle these specific cases. It is up to the implementations to figure out how to convert this data to something conformant to RDF 1.1.  http://json-ld.org/minutes/2013-02-26/#resolution-3
15:42:14 <sandro> q+
15:42:15 <ericP> A:-.5 B:+1 C:-.9
15:42:18 <AndyS> q+
15:42:27 <davidwood> ack PatH
15:42:29 <AndyS> RFC 3986 ==> (5.1.4) Default Base URI (application-dependent)
15:42:58 <markus> pat: I keep hearing that current systems don't implement bnodes and thus we can't do it.
15:42:59 <sandro> pat: What's the problem with blank nodes as graph identifiers?
15:43:01 <pfps> parsers break
15:43:12 <ericP> i've not seen a prob with bnodes as graph identifiers
15:43:13 <markus> ... could someone explains what breaks if we do allow that?
15:43:55 <markus> ???: nothing breaks. We had a store which allowed it but users disliked it
15:44:09 <sandro> SteveH: 3store, 2 gens away, allowed it.    Mostly users didnt like it because it wasn't obvious where the blank nodes came from.   They were sort of minted anonymously.   They came from implicit graphs that came from reasoning.   This produced a general sort of feeling of unhappiness.
15:44:17 <ericP> s/???: nothing/SteveH: nothing/
15:44:53 <markus> ... there's all kind of interesting questions around having variables being existential identifiers
15:45:02 <markus> ... it wasn't clean or elegant in any way
15:45:06 <sandro> ... people wanted to know if they were really existential variables, like blank nodes, since they're not in the graph.      We never tries to solve that.   Interesting semantics around having an identifier be an existential vairable.   Having this further in the system didn't really help, and it was a bit ugly.
15:45:10 <davidwood> The RDF WG resolved "Datasets can use blank nodes as graph names, not just IRIs." at https://www.w3.org/2013/meeting/rdf-wg/2013-05-15#resolution_2, but rescinded it the following week at https://www.w3.org/2013/meeting/rdf-wg/2013-05-22#resolution_2 to an objection on the mailing list.
15:45:15 <sandro> ... it annoyed people and didnt acieve something over UUIDs.
15:45:19 <markus> ... there wasn't a fundamental problem why it didn't work
15:45:23 <davidwood> q?
15:45:27 <davidwood> ack sandro
15:45:43 <markus> sandro: I take to opportunity to ask the same question
15:46:17 <markus> ... I wanted to ask steve what happens if you do it multiple times (same base used?)
15:46:26 <sandro> steve: We used the same static base URI for every time.
15:46:29 <markus> steveh: it is a statically defined URI
15:47:14 <markus> andys: UUID generation is implemented in every OS these days
15:47:33 <markus> sandro: but you can't get to it from JavaScript e.g. You have to reimplement it yourself
15:47:55 <davidwood> There are JS libraries for UUID generation.  It is not like you have to implement it yourself.
15:48:00 <markus> davidwood: the fact that something is possible one way doesn't mean that we should make it illgegal doing it in another way
15:48:11 <davidwood> s/davidwood/???/
15:48:15 <PatH> that was path
15:48:24 <davidwood> s/???/patH/
15:48:33 <PatH> that was path
15:48:34 <davidwood> ack AndyS
15:48:56 <ericP> i think the static URI parsing trig will keep appending to the resolved graph names: { a :Diff ; :from <from> ; :to <to> } <from> { ... } <to> { ... }
15:49:08 <markus> andys: I'm just looking at the RFC about how to establish the base.. the last step is to fall back to an application supplied base
15:49:49 <davidwood> q?
15:49:50 <ericP> the implementation burden for UUIDs seems much higher than that for bnodes as graph labels
15:50:30 <markus> markus: but isn't the problem that there is no (stable) URL?
15:50:36 <SteveH> ericP, I strongly disagree
15:50:58 <markus> davidwood: typically you get it from the web server. sandro is concerned about situations when there's no web server
15:51:19 <SteveH> this is a VERY old problem
15:51:29 <SteveH> and it's not tripped people up yet
15:51:43 <markus> sandro: in RDF we typically handle this by allowing either URIs or blank nodes.. this introduces a third type.. a late bound/resolved URI
15:51:56 <ericP> i don't think i've ever had to code something funny to handle bnodes as identifiers. i just have to relax the trig parser.
15:51:57 <davidwood> +1 to SteveH
15:52:07 <markus> ... I'm concerned about that. There were no standardized datasets before
15:52:10 <SteveH> in application != having an explicit BASE
15:52:21 <PatH> q+
15:52:27 <markus> andys: I see that as an argument to not standardize that now
15:52:34 <markus> sandro: not standardizing datasets?
15:52:56 <markus> andys: yes.. not allowing bnodes as graph names (we don't have experience)
15:52:56 <sandro> q+ to ask about n-quads
15:53:32 <markus> davidwood: there is implementation experience (n3 formulae in triple stores)
15:54:24 <markus> path: the case for disallowing bnodes is ridicously weak
15:54:26 <davidwood> ack PatH
15:54:42 <markus> ... the strongest point is that it can be implemented by other means and that people disliked it
15:54:55 <SteveH> I've not heard a strong, or well-reasoned argument in favour
15:54:59 <markus> ... on the other hand we have people like the JSON-LD guys who have really strong arguments
15:55:09 <SteveH> RDF has too much rope as it is
15:55:29 <markus> ... allowing bnodes doesn't mean that other mechanisms are disallowed such as skolemization to get rid of the bnodes
15:55:53 <davidwood> ack sandro
15:55:53 <Zakim> sandro, you wanted to ask about n-quads
15:56:26 <markus> sandro: Am I right that N-Quads couldn't serialize one of these relative datasets because the require absolute URIs?
15:56:54 <markus> andys: that's a non-question
15:57:16 <markus> sandro: I don't consider that as a non-question is the defacto test-language isn't able to serialize it
15:58:23 <markus> andys: you can have relative IRIs in the documents you are sending but if you are putting it into a store you convert it to absolute IRIs which you can serialize in N-Quads
15:58:37 <markus> davidwood: but your application has to make up the base
15:58:39 <ericP> q?
15:58:41 <ericP> q+
15:59:02 <davidwood> ack ericP
15:59:22 <Guus> q+ to make a meta remark
15:59:59 <sandro> davidwood, I think we're okay going long on meeting time.
16:00:08 <davidwood> sandro, ack
16:00:48 <PatH> I am not sure what the semantics of a thing with relative IRIs could be. Semantically, each implementation of it would be a distinct entity with its own semantics.
16:00:54 <markus> ericp: if the test I'm executing is in N-Quads I need to extend the test harness to ensure that the generated base URI is deterministic
16:00:59 <PatH> :-)
16:01:00 <markus> ... the other option is to allow bnodes
16:01:20 <markus> ... so either we change N-Quads or we need to add something entertaining to the test harness
16:01:20 <SteveH> we don't need to change anything re. relative URIs, it already works that way now!
16:01:27 <SteveH> people think this is new for some reason?
16:01:33 <davidwood> q?
16:01:40 <davidwood> ack Guus
16:01:40 <Zakim> Guus, you wanted to make a meta remark
16:02:08 <markus> guus: if we are not going to resolve this within the next two weeks we are in trouble
16:02:16 <markus> davidwood: we can also do nothing
16:02:27 <sandro> -1
16:02:30 <pfps> I thought that the core of the "relative" process is that there is something that turns these relative IRIs into absolute IRIs just before the real triples are created
16:02:31 <markus> guus: than we agree on proposal C
16:02:52 <markus> sandro: I don't agree, we need to choose the best option
16:03:18 <PatH> pfps, but my point theni s that the datathingie *with relative IRIs* itself has no semantics.
16:03:21 <markus> guus: I don't say we choose proposal C, but that's what we need to do if we can't reach consensus within two week
16:03:31 <sandro> STRAWPOLL:   (proposed-a (relative URIs)), proposed-b (blank node graph names), proposed-c: do nothing)
16:03:35 <ericP> A:-.5 B:+1 C:-.9
16:03:36 <SteveH> technically, A and C are equivalent, as far as I can see
16:03:38 <PatH> It is a kind of schema.
16:03:46 <markus> sandro: no, we could also go with proposal A then
16:03:57 <pfps> path: yes, but the datathingie is not an RDF graph, nor does it contain RDF triples, so it doesn't deserve any semantics
16:04:23 <SteveH> it's not ambiguous, it's in the RFC
16:04:23 <pfps> how can proposal A be a "may"?
16:04:38 <markus> A: -0; B: +1; C: -1
16:04:49 <davidwood> A; +1, B: −0 (because cygri has already objected), C: −0.5
16:04:55 <SteveH> A: -0 B: -1 C: +1
16:04:58 <sandro> PROPOSED-A: Close ISSUE-131 with rdf-concepts saying that RDF syntaxes and toolkits MAY support RDF (graph and/or dataset) serializations which use relative URIs and a base that is only determined by the receiver.
16:04:59 <yvesr> A:0 B:+0.9 C: -0.9
16:05:02 <PatH> pfps: Well, OK, but that does seem strange (to me). Datasets are defined so that they cannot possibly have any data...
16:05:05 <zwu2> A:-0; B: -1, C:+1
16:05:12 <davidwood> PROPOSED-A: Close ISSUE-131 with rdf-concepts saying that RDF syntaxes and toolkits MAY support RDF (graph and/or dataset) serializations which use relative URIs and a base that is only determined by the receiver.
16:05:12 <davidwood> 
16:05:12 <davidwood> PROPOSED-B: Close ISSUE-131 by allowing blank nodes as graph names.
16:05:12 <davidwood> 
16:05:12 <davidwood> PROPOSED-C: Close ISSUE-131 without adding any text on this subject to our specifications.
16:05:18 <sandro> A:-.5 B+1: C: -.9
16:05:19 <AndyS> 0.5 / -0.5 / 1  (A is not good use of editor time)
16:05:25 <TallTed> A: +0.5  B: +1  C: -0.5
16:05:25 <Zakim> -Arnaud
16:05:44 <PatH> A:-1 B:+1  C:-1
16:05:44 <pfps> A: -0.5 B: +0.9 C +0.5
16:05:47 <markus> davidwood: proposal B is exactly what Richard objected to
16:06:10 <markus> andys: why do we reopen this one and nothing else
16:06:21 <ivan> A: 0.5 B: -0 (same issue as david with FO) C: -0.5
16:06:21 <markus> davidwood: I'm not comfortable with this one
16:07:10 <sandro> eric: Cost of B to SPARQL?       SPARQL implementors would feel a pressure to add support for blank node graph names.
16:07:23 <SteveH> SELECT * WHERE { GRAPH _:abc { ?s ?p ?o } } doesn't do what you want in SPARQL
16:07:25 <davidwood> So, we have objections to B and C, but none on A (even though people don't like it very much).
16:07:30 <pfps> I'm not seeing a *COST* there
16:07:31 <markus> eric: maybe we should do a quick cost-analysis. What do SPARQL implementers need to change? What if LDP decides to use that as patch format?
16:07:33 <davidwood> q?
16:07:56 <davidwood> Correction, we have objections to all three
16:07:59 <ivan> q+
16:08:07 <AndyS> +1 to SteveH
16:08:09 <ivan> ack ivan
16:08:11 <davidwood> ack ivan
16:08:22 <markus> ivan: I would like to understand that objection
16:08:50 <markus> ... relative URIs do not appear in the abstract syntax - just in serializations
16:08:58 <sandro> yeah, I really don't like serializations that are no g-snaps.
16:09:07 <markus> ... which means that they have to invent a way to transform them to absolute URIs
16:09:21 <markus> path: it's not an objection in the sense of a formal objection
16:10:00 <markus> ivan: it's purely a matter of serialization and has nothing to do with concepts and semantics
16:10:06 <SteveH> this ambiguous situation already exists, and is commonly used, no amount of wishing will make it go away - we could ban it, but that would be crazy
16:10:12 <gavinc> gavinc has joined #rdf-wg
16:10:13 <ericP> how about if we just use rdf:reification (which permits bnodes as rdf:Statement names)?
16:10:33 <markus> path: what happens if we go with route A? What is the result? Is it a document?
16:11:02 <markus> ... there's one dataset at one end and a different at the other end because there are different bases
16:11:11 <TallTed> q+
16:11:13 <markus> davidwood: there's no dataset on the first end
16:11:15 <TallTed> Zakim, unmute me
16:11:15 <Zakim> TallTed should no longer be muted
16:11:24 <davidwood> s/davidwood/path/
16:11:26 <markus> path: well, but different systems will produce different datasets
16:11:31 <markus> ivan: that's true
16:11:45 <SteveH> PatH, this happens now   <s> <p> "o" . is a legit RDF serialisation
16:11:50 <AndyS> Same is true of Turtle today of graphs.  <s> <p> <o> .
16:11:57 <SteveH> right
16:12:16 <markus> eric: the issue here is that bnodes as identifiers are useful
16:12:26 <davidwood> ack TallTed
16:12:30 <markus> sandro: I think you are misunderstanding, we are talking about relative URIs
16:12:36 <AndyS> And they are different bnodes in every dataset parsed out of same doc at different times.
16:13:03 <markus> davidwood: sandro, if you are not a web server you are creating this thingy but you are not publishing to the world
16:13:11 <markus> sandro: no, I would like to publish it to the world
16:13:34 <markus> davidwood: but once you hand it over to the web server it can decide how to do that
16:14:04 <markus> sandro: if I'm part of the web server then yes. But if there are intermediaries etc. then it gets problematic
16:14:10 <SteveH> how is that different to what happens in RDF?
16:14:13 <markus> ... it may end up in different places with different graph names
16:14:24 <SteveH> s/graph name/subject/ - so what?
16:14:31 <markus> q+
16:14:32 <TallTed> if I (stupid web client) am pushing to web server, server can assign a final URI ... which I should receive from that server for further pushes (whether to same or different web server)
16:14:45 <AndyS> ?? multiple names (of resources) are the norm -- no UNA
16:15:15 <markus> sandro: the problem is that you can't create an anonymous dataset on the client
16:15:54 <markus> eric: until you get to the server you can't use the standard specs/tooling
16:15:54 <sandro> sandro: Only a webserve can create a standard dataset.    Anything else has to work as part of an application, in a not-standard way.
16:16:33 <SteveH> *sigh* we don't get to disallow relative URIs
16:16:41 <davidwood> ack markus
16:16:49 <sandro> no one is trying to disallow relative URIs.
16:17:20 <yvesr> markus, +1
16:17:25 <SteveH> a significant proportion of real users don't like bNodes in graphs - adding them as graph (non)identifiers isn't going to be popular
16:17:25 <ivan> zakim, mute me
16:17:25 <Zakim> Ivan should now be muted
16:17:26 <sandro> marcus: How is a blank-node graph name any different from a blank-node subject?
16:17:28 <PatH> +1
16:17:55 <markus> markus: my question is what the difference to bnodes for subjects is. if I can't assign a URI to a subject I use a bnode, why can't I just do the same for graphs?
16:18:08 <markus> s/marcus/markus/
16:18:16 <yvesr> SteveH, hmm - i wouldn't think so - all RDFa users or JSON-LD users are creating lots of bnodes without even realising it
16:18:44 <markus> sandro: Richard didn't make a formal objection he just said he might do
16:19:42 <ivan> q+
16:19:45 <markus> davidwood: let's please try to continue this discussion on the mailing list
16:19:53 <ivan> zakim, unmute me
16:19:53 <Zakim> Ivan should no longer be muted
16:20:21 <pfps> it would be nice to get more information from Deri on why they are threatening to object
16:20:37 <davidwood> ack ivan
16:20:42 <markus> eric: can I propose that the editors of concepts and semantics would remove all references to datasets from the documents so that we can go to LC?
16:20:52 <markus> davidwood: prob. simpler to mark as at risk
16:20:58 <SteveH> I would object too
16:21:09 <ericP> s/eric: can I/PatH: can I/
16:21:15 <SteveH> I wasn't here two weeks ago
16:21:20 <markus> ivan: Richard said he will raise a formal objection could the chairs talk to Richard directly
16:21:24 <SteveH> will explain in email
16:21:38 <markus> sandro: maybe we can ask anyone who would object to send a mail to the mailing list
16:22:16 <markus> ivan: I know that Richard is extremely busy but we need to get the discussion rolling
16:22:16 <AndyS> zhe?
16:22:37 <zwu2> I would object too, AndyS
16:22:46 <markus> eric: could we just mark it as feature at risk which would give SPARQL implementers time
16:22:56 <davidwood> Richard's latest comments: http://lists.w3.org/Archives/Public/public-rdf-wg/2013May/0201.html
16:23:00 <markus> ... which might satisfy people who are concerned
16:23:23 <markus> ivan: we can certainly do that but I'm not sure that it's wise given we have two potential formal objections within our group
16:23:53 <markus> ivan: Richard's response doesn't contain arguments against it
16:24:10 <markus> andys: well, he said why he doesn't like it
16:24:50 <markus> path: my basic point is that we should only disallow if there are very good reasons to disallow
16:25:03 <SteveH> yes, I beiieve quite strongly in don't change without a good reason
16:25:08 <markus> ... other people think that we should not change unless there are very good reasons to change
16:25:08 <sandro> pat: We should only disallow something when there is a strong reason for that.   Other people are taking a "don't change unless there's a strong reason" position, which is also reasonable.
16:25:15 <AndyS> Want to hear a response to -- (17:07:23) SteveH: SELECT * WHERE { GRAPH _:abc { ?s ?p ?o } } doesn't do what you want in SPARQL
16:25:17 <SteveH> we;ve spent millions of dollars implementing the specs we have
16:25:43 <SteveH> anyway, email later
16:25:51 <Zakim> -Ivan
16:25:51 <sandro> AndyS, I want to understand that.        What does it do?   What should it do?        Steve, can you explain in email?
16:25:52 <Zakim> -davidwood
16:25:54 <Zakim> -TallTed
16:25:56 <zwu2> thanks
16:25:57 <Zakim> -PatH
16:26:05 <Zakim> -zwu2
16:26:07 <Zakim> -pfps
16:26:15 <Zakim> -yvesr
16:26:26 <PatH> I dont find the millions of dollars argument at all persuasive. In a word: tough. Or, go tell that to Microsoft.
16:27:16 <AndyS> sandro - you should provide the explanation as part of your proposal - it's an implication of the design.
16:28:25 <Zakim> -Guus_Schreiber
16:39:20 <AndyS> zakim, who is on the phone?
16:39:20 <Zakim> On the phone I see Sandro, SteveH, ericP, markus, AndyS
16:40:52 <AndyS> GSP requires naming?
16:41:11 <sandro> sandro: Blank nodes are the worst possible approach, except for all the other approaches.
16:42:21 <SteveH> IMHO <http://graph.prefix/$uuid> is better in every way
16:42:41 <SteveH> in reality there a no clients that can't generate something equivalent to a UUID
16:42:44 <sandro> <sandro> pat: We should only disallow something when there is a strong reason for that.   Other people are taking a "don't change unless there's a strong reason" position, which is also reasonable.
16:44:31 <sandro> zakim, who is making noise?
16:44:44 <Zakim> sandro, listening for 12 seconds I heard sound from the following: Sandro (20%), AndyS (50%)
16:46:13 <SteveH> people are always free to add non-standard features to RDF...
16:46:35 <davidwood> You, too, SteveH ;)
16:47:39 <SteveH> there's lots of different entailment rules that could come out...
16:47:49 <SteveH> yet
16:48:01 <SteveH> owl people will want them
16:52:24 <Zakim> -SteveH
16:52:26 <Zakim> -markus
16:52:39 <AndyS> I could support some text that noted it MAY happen in a later WG (and other possible features).
16:52:46 <Zakim> -AndyS
16:52:46 <Zakim> -Sandro
16:52:47 <Zakim> -ericP
16:52:47 <Zakim> SW_RDFWG()11:00AM has ended
16:52:48 <Zakim> Attendees were pfps, Sandro, AndyS, davidwood, Ivan, TallTed, SteveH, Guus_Schreiber, ericP, zwu2, yvesr, markus, Arnaud, PatH
17:32:24 <gavinc> Please see recent email regrading proposed changes to Turtle to meet todays resolution. ( gkellogg, ericP, AndyS )
17:32:53 <AndyS> yes.
17:34:28 <AndyS> What do you want from me? (it will have to be tomorrow)
17:34:51 <gavinc> Nothing, if you can live those.
18:00:56 <manu1> manu1 has joined #rdf-wg
19:01:17 <Zakim> Zakim has left #rdf-wg
20:57:21 <gkellogg> gkellogg has joined #rdf-wg
22:18:56 <AndyS> AndyS has joined #rdf-wg