Session Start: Fri Aug 02 15:16:30 2002 [15:16] *** Now talking in #rdfcore [15:16] *** Topic is 'rdfcore teleconference - 2002-08-02' [15:16] * DaveB arrives [15:16] Zakim, who's here? [15:16] sorry, DaveB, I don't know what conference this is [15:16] On IRC I see gk, danbri, RRSAgent, Zakim, DaveB, em, logger_1 [15:16] Zakim, list conferences [15:16] I see no active conferences [15:16] too early? [15:17] I'm in [15:17] I dialed in, nobody was there. I'm trying my new speakphone so excuse any funny noise. [15:17] * DaveB listens to the silence [15:17] *** Guha_ has joined #rdfcore [15:17] zakim, list conferences [15:17] I see no active conferences [15:17] zakim, list conferences. [15:17] I see no active conferences [15:18] hmm... [15:18] anyone have the w3c bridge phone number handy? [15:18] sure [15:18] Phone: +1-617-761-6200 (Zakim)#7332 [15:18] thanks! [15:18] zakim, this is RDF_CORE [15:18] sorry, em, I do not see a conference named 'RDF_CORE' [15:19] zakim, this is SW [15:19] sorry, em, I do not see a conference named 'SW' [15:19] zakim, this is SW_RDFCore. [15:19] sorry, em, I do not see a conference named 'SW_RDFCore.' [15:19] Zakim, list conferences [15:19] I see no active conferences [15:19] agenda http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Aug/0007.html [15:19] zakim, this is SW_RDFCore [15:19] sorry, em, I do not see a conference named 'SW_RDFCore' [15:19] odd [15:20] *** jjc has joined #rdfcore [15:20] * Zakim is departing [15:20] *** Zakim has quit IRC (Leaving) [15:21] * danbri phone trouble [15:21] bridge: 7332? [15:21] yes [15:21] re-retrying. [15:21] *** gk is now known as gk-scribe [15:22] 2: Roll Call [15:22] *** Zakim has joined #rdfcore [15:22] zakim, this is SW [15:22] ok, em [15:22] zakim, who is here? [15:22] On the phone I see ??P6, Manola, EricM, ??P14, +1.650.965.aaaa, ??P15, DanBri [15:22] On IRC I see Zakim, jjc, Guha_, gk-scribe, danbri, RRSAgent, DaveB, em, logger_1 [15:22] -DanBri [15:22] em, any reason why I can't get connected back from Zakim? [15:23] did my answerphone pickup? [15:23] on the call.... [15:23] guha, daveb, gk... [15:23] +Mike_Dean [15:23] and jjc [15:23] Jeremy [15:24] *** DanC has joined #rdfcore [15:24] danbri: we got a US long distance operator voice [15:24] Regrets: Patrick [15:24] agenda - http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Aug/0007.html [15:24] 3: Review Agenda [15:24] [off] it worked yesterday, i've no idea what's up. [15:24] agenda +lbase [15:24] * Zakim notes agendum 1 added [15:24] * danbri trying again, sorry if it causes more noise [15:24] +DanBri [15:24] agenda +datatypers [15:24] * Zakim notes agendum 2 added [15:24] em: also include lbase [15:24] no me. [15:24] agenda +new doc [15:24] * Zakim notes agendum 3 added [15:24] Zakim, disconnect DanBri [15:24] DanBri is being disconnected [15:24] -DanBri [15:25] 4: Next telecon Aug 9 2002 [15:25] +DanC [15:25] regrets 9 Aug [15:25] Date OK [15:25] 5: Review minutes of 2002-07-26 teleconference [15:26] Jeremy did issue minutes in record time [15:26] jjc's minutes http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jul/0162.html [15:26] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jul/0163.html [15:26] -- [15:26] No objections [15:26] 6: Status of Action Items [15:26] where are the minutes? 0162 isn't it, is it? [15:26] *** sergey has joined #rdfcore [15:26] Open items: [15:27] minutes 26Jul: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jul/0163.html [15:27] + +1.850.434.aabb [15:27] ACTION: danbri to circulate LBase document before Wednesday [15:27] * RRSAgent records action 1 [15:27] New lbase circulated today [15:27] see http://www.w3.org/2002/06/lbase/ [15:28] ACTION: jang [15:28] update test cases in light of bugs [15:28] * RRSAgent records action 2 [15:28] Continues... [15:28] ACTION: eric [15:28] Look into why jang gets dropped on rdf-comments list and fix problem [15:28] * RRSAgent records action 3 [15:28] Fixed - do0ne. [15:28] (Skip bwm's actions) [15:28] * danbri tries zakim once more [15:28] DaveB actions ... all continue [15:29] ... New draft next Tuesday [15:29] have you guys seen the rdflib parser with nodeID support? [15:29] yup [15:29] +DanBri [15:29] eikeon said it was relatively straightforward [15:29] http://rdflib.net/2002/07/29/rdflib-0.9.4.tgz <- http://rdflib.net/latest/ [15:29] (wohoo-- i got thru to phone bridge) [15:29] All actions to review GK doc are closed [15:29] sergey: on call? [15:30] (I started code-reviewing rdflib... looks pretty cool) [15:30] in a second... [15:30] em: lots of discussion on e-mail, are we ready to go? [15:30] +??P18 [15:30] * DanC wonders who just joined [15:30] gk: issues are refs to striped syntax [15:31] zakim, +??P18 is rond. [15:31] sorry, em, I do not recognize a party named '+??P18' [15:31] * danbri tries speakerphone, let me know if causes problem [15:31] gk: that allow ordinary XML to be RDF compatible [15:31] gk: there has been discussion 2.3.3 & 2.3.4 (last two sections of rdfms-assertion [15:31] +??P19 [15:31] zakim, +??P19 is sergey? [15:31] sorry, em, I do not understand your question [15:31] zakim, +??P19 is sergey [15:31] sorry, em, I do not recognize a party named '+??P19' [15:32] danbri on call now [15:32] gk-scribe, when you say "needs more work", do you mean "cannot be released without more work"? [15:32] issue about assertions and mimetype [15:33] +??P20 [15:33] zakim, +??P20 is JosD. [15:33] sorry, em, I do not recognize a party named '+??P20' [15:33] jos joined [15:33] zakim, +??P20 is JosD [15:33] sorry, em, I do not recognize a party named '+??P20' [15:34] issue about higher level organizational component (em's message) [15:34] q+ [15:34] * Zakim sees DaveB on the speaker queue [15:34] so the editors haven't gotten together? [15:34] em: I don't want to hold group up if there was consensus [15:34] gk: this is a reasonable topic [15:35] * DanC q+ to address the striping bit... I could let it go [15:35] * Zakim sees DaveB, DanC on the speaker queue [15:35] *** JosD has joined #rdfcore [15:35] q? [15:35] * Zakim sees DaveB, DanC on the speaker queue [15:35] ack DaveB [15:35] * Zakim sees DanC on the speaker queue [15:35] jjc: I thought that considering the coherency was meant to happen now [15:35] q- [15:35] * Zakim sees DanC on the speaker queue [15:35] daveb: let's set time limit [15:36] em: 5 minutes more on this discussion [15:36] danc: propose publication without any of the contentious text [15:37] danc: stuff that can't be excised let's put note. [15:37] path, jos support publciation [15:37] q+ [15:37] * Zakim sees DanC, DaveB on the speaker queue [15:37] ack danc [15:37] ack DanC [15:37] DanC, you wanted to address the striping bit... I could let it go [15:37] * Zakim sees DaveB on the speaker queue [15:37] * Zakim sees DaveB on the speaker queue [15:38] the "ordinary XML" bit should go in the primer, if anywhere. [15:38] ack DaveB [15:38] * Zakim sees no one on the speaker queue [15:38] I agree, re primer. [15:38] gk: making rdf compatible with ordiunary xml is highly persuasive [15:38] (I could say so in audio track too, but we're short of time) [15:38] q+ [15:38] * Zakim sees Em on the speaker queue [15:38] db: I could support this doc without that text on striping, ordinary xml etc. [15:38] q- [15:38] * Zakim sees no one on the speaker queue [15:39] frank: some of that material will go in the primer [15:39] q+ [15:39] * Zakim sees Em on the speaker queue [15:39] ack Em [15:39] * Zakim sees no one on the speaker queue [15:40] frank: the objection is not to the point but making it now in this doc [15:40] em: i would like this just to be the data model, and other stuff in primer [15:40] em: i would like a note indicating that some stuff will move into other docs [15:41] * DanC q+ to say I don't see the need for another WD before last call [15:41] * Zakim sees DanC on the speaker queue [15:41] em: we would then need a second draft before a last call [15:41] +q [15:41] * Zakim wonders where q is [15:41] danc: i think editorial materail can move in the last call [15:42] q+ [15:42] * Zakim sees DanC, Jjc on the speaker queue [15:42] ack DanC [15:42] DanC, you wanted to say I don't see the need for another WD before last call [15:42] * Zakim sees Jjc on the speaker queue [15:42] ack danc [15:42] * Zakim sees Jjc on the speaker queue [15:42] ack jjc [15:42] * Zakim sees no one on the speaker queue [15:42] danbri does not want striping ref [15:43] connoolly manola deroo hayes vote for publish [15:43] manola would liek some of the abstract material in the intro [15:43] I PROPOSE: to authorize publication, after any edits the editor wants to make given advice he's seen so far. [15:43] go: danc, jos, pat, frank (suggest abstract should also go in intro ... and clearer that this in normative) [15:44] go: daveb, danbri, mike [15:44] gk acks changes to XML things in http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Aug/0026.html [15:44] danbri: "we can move things around later" [15:44] abstain: sergey, ron, eric [15:44] action eric get ball rolling on getting things out the door [15:45] 8) Datatypes [15:46] Guha - proposal - default local idiom and that's it. [15:46] Mike: want global idion [15:46] Frank: want global idiom [15:46] * DanC q+ to explore going to last call without datatypes [15:46] * Zakim sees DanC on the speaker queue [15:46] Patrick: (in absentia) wants global idiom, we think [15:47] q+ [15:47] * Zakim sees DanC, Em on the speaker queue [15:48] Guha: start with local idion, allow application-specific or other layers to add global idion [15:48] Mike: local idiom adds triple-bloat [15:48] q- [15:48] * Zakim sees DanC on the speaker queue [15:48] Guha: "local idiom" means some way os specifying the type of a particular literal -- not necessarily extra triples [15:49] danc, can i hold off on the discussion you abuot to ask for a sec? [15:49] ok [15:49] er... no, "that we cannot ship without" is what I want to discuss [15:50] ok [15:50] q+ [15:50] * Zakim sees DanC, Jjc on the speaker queue [15:50] Guha: proposes we recignize that we can't ship without some way of saying a liuteral is (say) an integer [15:50] DanC: can we ship everything but datatypes? [15:50] ack danc [15:50] DanC, you wanted to explore going to last call without datatypes [15:50] * Zakim sees Jjc on the speaker queue [15:51] q +pat [15:51] q+ pat [15:51] * Zakim sees Jjc, Pat on the speaker queue [15:51] q+ frank [15:51] * Zakim sees Jjc, Pat, Frank on the speaker queue [15:51] ack jjc [15:51] Guha: process overhead concern [15:51] * Zakim sees Pat, Frank on the speaker queue [15:51] JJC: what does graph syntax look like with local typing? [15:51] "more literals" looks like a little bit of functional terms, to me. Not a bad thing. [15:52] q+ guha [15:52] * Zakim sees Pat, Frank, Guha on the speaker queue [15:52] ack guha [15:52] * Zakim sees Pat, Frank on the speaker queue [15:52] Guha: in the graph, should be able to have (say) a numeric literal node [15:53] ... then, questions of how to represent in RDF/XML, N-triples, etc [15:53] q+ [15:53] * Zakim sees Pat, Frank, Gk on the speaker queue [15:53] ack pat [15:53] * Zakim sees Frank, Gk on the speaker queue [15:53] q+ gk-scribe [15:53] * Zakim sees Frank, Gk, Gk-scribe on the speaker queue [15:53] ack gk-scribe [15:53] * Zakim sees Frank, Gk on the speaker queue [15:54] sooo... we're talking about 10 ? [15:54] q+ [15:54] * Zakim sees Frank, Gk, Em on the speaker queue [15:54] q+ [15:54] * gk-scribe DanC, I think so [15:54] * Zakim sees Frank, Gk, Em, Jjc on the speaker queue [15:54] ack frank [15:54] * Zakim sees Gk, Em, Jjc on the speaker queue [15:54] q+ [15:54] * Zakim sees Gk, Em, Jjc, JosD on the speaker queue [15:55] Frank: this seems close to discussion of rdf:value in the Primer [15:55] q- [15:55] * Zakim sees Gk, Jjc, JosD on the speaker queue [15:56] q+ guha [15:56] * Zakim sees Gk, Jjc, JosD, Guha on the speaker queue [15:57] * DaveB can scribe if gk wants to talk [15:57] yes please DaveB [15:57] thanks [15:57] gk-scribe: support guhas idea [15:57] q, what is the breadth of datatypes that would be allowed in literal space? [15:57] how do we determine? [15:57] ack guha [15:57] Guha_: + answering frank. [15:57] * Zakim sees Gk, Jjc, JosD on the speaker queue [15:58] ... distinguishing issue 1) [15:58] ack gk [15:58] * Zakim sees Jjc, JosD on the speaker queue [15:58] * DanC whee! design on the telephone! [15:58] ... how do we repreent in the graph that value/target of (missed it) is a number or a string [15:58] Guha: distingish 2 issues: (1) how to represent a value is a number, not string... [15:58] ... bigger issue 2) [15:58] ... (bnodes, ...) [15:58] back to you gk [15:59] (2) what are the other sorts of things? [15:59] Start by aligning with XML schema [15:59] ack jjc [15:59] * Zakim sees JosD on the speaker queue [16:00] There are 2 parts to what I am saying: [16:00] JJC: has also suggested something like this -- still question of tidy vs untidy [16:00] q+ mike [16:00] * Zakim sees JosD, Mike on the speaker queue [16:00] ack JosD [16:00] * Zakim sees Mike on the speaker queue [16:00] We should be able to say SanJose--population-->800000, where 800000 is a number, not a string. There are a couple of ways of doing this, one with bnodes, one without. [16:00] We've heard people who don't want to restruct ourselves to local, but the issue of tisy/untidy still needs to be addressed [16:01] JosD: don't buy triple-bloat argument -- prefer's "interpretation properties" approach, and believs it is scalable [16:01] ... summary: favours local idiom [16:01] q+ PatH [16:01] * Zakim sees Mike, PatH on the speaker queue [16:01] ack mike [16:01] * Zakim sees PatH on the speaker queue [16:02] q+ [16:02] MikeD: concern with local-only; no way to indicate expectation of what a property value will be [16:02] Then, there is an issue of how to determine from an XML serialization that a particular reference to 800000 or 800,000 or "eight hundred thousand" refers to the number [16:02] * Zakim sees PatH, Jjc on the speaker queue [16:02] q+ [16:02] * Zakim sees PatH, Jjc, Em on the speaker queue [16:02] I propose that we separate these two issues and deal with the second latter. [16:02] * DanC q+ to ask if anybody has implementation experience with xsi:type="dt:integer" [16:02] * Zakim sees PatH, Jjc, Em, DanC on the speaker queue [16:02] MikeD: would be comfortable with global-only, not local-only [16:02] ack path [16:02] * Zakim sees Jjc, Em, DanC on the speaker queue [16:03] i was going to ask with danc is about to ask.... [16:03] q- [16:03] * Zakim sees Jjc, DanC on the speaker queue [16:03] PatH: agree triple-bloat argument is weak; others argue differently (e.g. API construction); [16:04] harder to map into XML with a bnode local idiom, rather than attaching datatype to literal [16:04] ack jjc [16:04] * Zakim sees DanC on the speaker queue [16:04] ack DanC [16:04] DanC, you wanted to ask if anybody has implementation experience with xsi:type="dt:integer" [16:04] * Zakim sees no one on the speaker queue [16:04] q+ [16:04] * Zakim sees Jjc on the speaker queue [16:04] DanC: anyone have implementation experience with xsi:type= kind of approach? Guha, Sergey both [16:05] q+ [16:05] * Zakim sees Jjc, Em on the speaker queue [16:05] Sergey: supports making primitive datatypes part of abstract syntax [16:05] q+ guha [16:05] * Zakim sees Jjc, Em, Guha on the speaker queue [16:05] ack jjc [16:05] * Zakim sees Em, Guha on the speaker queue [16:05] There are two ways of indicating (in the graph) that a particular occurance (in the XML) of "800000" or "eight hundred thousand" refers to the number [16:05] q+ [16:05] * Zakim sees Em, Guha, Danbri on the speaker queue [16:05] JJC: MikeD wants schema to indicate expected type -- is this consistent with local typing [16:06] extra-logical? I don't see that. [16:06] PatH: consistent, yes; enforced by MT, no [16:06] One is to use b-nodes and the other is to make a literal a 4-tuple, wherein a literal is a {string, bit-for-xml, language, data-type} [16:06] em: (ref DanC recent email) can define a local idiom that can be handed up to higher layer to define global datatyping? [16:06] I was going to ask whether a schema might declare an application-specific subpropertyof an xsd datatype property... [16:07] q- [16:07] * Zakim sees Em, Guha on the speaker queue [16:07] DanC: that looks unacceptably risky. [16:07] ack em [16:07] * Zakim sees Guha on the speaker queue [16:08] Guha: adding to Sergey; have done 3 big implementations (10Ks, 100Ks of files); with different ways of dealing with numbers. These implementations are not interoperable. [16:08] q+ [16:08] * Zakim sees Guha, Em on the speaker queue [16:08] ack guha [16:08] * Zakim sees Em on the speaker queue [16:09] Hard to take seriously a representation that can't deal with numbers. [16:09] ack em [16:09] * Zakim sees no one on the speaker queue [16:10] q+ [16:10] * Zakim sees Em on the speaker queue [16:10] q- [16:10] * Zakim sees no one on the speaker queue [16:10] sooo... we're talking about 10 ? [16:10] yes [16:10] em: can you write down the graph ? [16:10] Guha_: options are to add abone, or to add in litral [16:10] s/abone/a bnode/ [16:10] more specifics wrt Guha's suggestions: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Aug/0012.html [16:10] Guha: (a) introduce a bnode, (b) make the literal carry its type (a 4-tuple) [16:11] per guha's mozilla comment, here's the current Mozilla (and presumably NS 6.2) RDF literals implemenation: http://lxr.mozilla.org/mozilla/source/rdf/base/idl/nsIRDFLiteral.idl [16:11] q? [16:11] * Zakim sees no one on the speaker queue [16:11] q+ [16:11] * Zakim sees Em on the speaker queue [16:11] (I thought we discussed this at F2F1 and decided not to do it, can't remember why) [16:12] (f2f2 - cannes) [16:12] (global idiom) [16:12] I like the proposal that literals carry their type. [16:13] Guha: (about extensibility) would be nice, [but is necessary?] [16:13] danbri: me too, but want to see graph/model [16:13] Question: would tidy literals all be acceptable if they didn't all have to be strings? [16:13] * DanC nominates Sergey and Guha to draft a proposal [16:13] yes, let's get a proposal [16:13] another proposal - sigh! :) [16:14] q+ [16:14] * Zakim sees Em, DaveB on the speaker queue [16:14] em: asks for a proposal, asap. [16:14] well, I got new info in the last two weeks; in the proposal, let's please acknowledge the community input on xsi:type="" [16:14] q+ frank [16:14] * Zakim sees Em, DaveB, Frank on the speaker queue [16:14] yes, what's status of the current pre-pre-draft? who had it last? [16:15] I'll read/review, fwiw. [16:16] jeremy: I'll take an action item to do a test case (I have one in mind) [16:16] Who will work on this: Guha, Sergey, PatH, Mike, Jos; jjc will do test case [16:18] ACTION: Guha, lead preparation of proposal [16:18] * RRSAgent records action 4 [16:18] q? [16:18] * Zakim sees Em, DaveB, Frank on the speaker queue [16:18] ack em [16:18] * Zakim sees DaveB, Frank on the speaker queue [16:18] * DanC notes the time, moves to adjourn [16:18] When can we expect this proposal? [16:18] q+ mike [16:18] * Zakim sees DaveB, Frank, Mike on the speaker queue [16:18] DaveB: if the syntax changes, there are dangers here (last draft due tuesday) [16:18] ack daveb [16:18] * Zakim sees Frank, Mike on the speaker queue [16:19] * DaveB wary of last minute changes, for some value of last [16:19] (dave I don't think anything can get to 'last call' w/o datatypes) [16:19] Frank: hopes the mechanism would be extensible to other than XSD types [16:19] I have no requirement for extensibility. [16:19] ack frank [16:19] * Zakim sees Mike on the speaker queue [16:19] acnk mike [16:19] ack mike [16:19] * Zakim sees no one on the speaker queue [16:20] timecheck? [16:20] re plugging in other datatyping specs, see http://www.oasis-open.org/committees/relax-ng/xsd-20010907.html " Guidelines for using W3C XML Schema Datatypes with RELAX NG" [16:20] MikeD: triple-bloat argument -- has a number of databases in the 0.5million triple range -- taking these to 2 million would be a problem. [16:20] (aside) [16:20] Mike: do I need to get some HP salesreps to contact you for more disk, memory? [16:21] ?: The new approach won't introduce new triples [16:21] ----meeting closed----