13:01:03 <logger> logger has joined #rdfcore
13:01:03 <irc.w3.org> Users on #rdfcore: @logger 
13:33:54 <bwm> bwm has joined #rdfcore
13:52:37 <AaronSw> AaronSw has joined #rdfcore
13:52:50 <AaronSw> zakim, this will be rdf
13:52:50 <Zakim> Zakim has joined #rdfcore
13:52:53 <AaronSw> zakim, this will be rdf
13:52:54 <gk> gk has joined #rdfcore
13:52:54 <Zakim> ok, AaronSw
13:56:39 <DanCon> DanCon has joined #rdfcore
13:58:11 <Zakim> SW_RDFCore()10:00AM has now started
13:58:17 <Zakim> +??P0
13:58:51 <DanCon> brian, do you want an RRSAgent log?
13:58:59 <Zakim> +FrankM
13:59:18 <Zakim> +??P2
13:59:27 <Zakim> +??P3
13:59:34 <Zakim> +??P4
13:59:37 <gk> zakim, ??P4 is GK
13:59:39 <Zakim> +GK; got it
13:59:45 <bwm> zakim, ??p2 is bwm
13:59:46 <Zakim> +Bwm; got it
13:59:47 <DanCon> Zakim, agenda + 18Oct http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0209.html
13:59:48 <Zakim> agendum 1 added
13:59:58 <DanCon> DanCon has changed the topic to: RDFCore. chair: bwm. scribe: ???
14:00:00 <jang> jang has joined #rdfcore
14:00:00 <bwm> DanCon: yes please
14:00:11 <RRSAgent> RRSAgent has joined #rdfcore
14:00:12 <RRSAgent> * RRSAgent is logging
14:00:18 <DanCon> DanCon has changed the topic to: RDFCore. chair: bwm. scribe: jang
14:00:40 <JosD> JosD has joined #rdfcore
14:00:43 <jang> jang is now known as jang_scri
14:01:07 <bwm> zakim, who is on the phone
14:01:09 <Zakim> I don't understand 'who is on the phone', bwm
14:01:12 <bwm> zakim, who is on the phone?
14:01:13 <Zakim> On the phone I see ??P0, FrankM, Bwm, ??P3, GK
14:01:23 <Zakim> + +1.503.222.aaaa
14:01:30 <AaronSw> zakim, where is +1.503?
14:01:31 <Zakim> North American dialing code 1.503 is Oregon
14:01:32 <jang_scri> zakim, ??p0 is jang
14:01:33 <Zakim> +Jang; got it
14:01:45 <Zakim> +??P7
14:01:57 <gk> zakim, where is +44.1235
14:01:58 <Zakim> gk, I do not see a party named 'where'.  If you meant to ask a question you need to add '?'
14:02:04 <jang_scri> zakim, ??p7 is jos_d
14:02:06 <Zakim> +Jos_d; got it
14:02:10 <Zakim> +AaronSw
14:02:13 <gk> zakim, where is +44.1235?
14:02:14 <Zakim> country code 44 is United Kingdom
14:02:24 <AaronSw> zakim, 1.503 is PatH
14:02:25 <Zakim> sorry, AaronSw, I do not recognize a party named '1.503'
14:02:29 <AaronSw> zakim, +1.503 is PatH
14:02:30 <Zakim> +PatH; got it
14:02:39 <AaronSw> zakim, who's on the phone?
14:02:40 <Zakim> On the phone I see Jang, FrankM, Bwm, ??P3, GK, PatH, Jos_d, AaronSw
14:02:48 <DanCon> * DanCon has a little trouble dialing in...
14:03:05 <bwm> zakim, who is on the phone?
14:03:05 <DanCon> * DanCon saw regrets from danbri, but wonders whether to expect EricM
14:03:06 <Zakim> On the phone I see Jang, FrankM, Bwm, ??P3, GK, PatH, Jos_d, AaronSw
14:03:07 <jang_scri> AGENDA: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0209.html
14:03:13 <DanCon> agenda?
14:03:14 <Zakim> * Zakim sees 1 item remaining on the agenda:
14:03:15 <Zakim> * Zakim 1. 18Oct http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0209.html [from DanCon]
14:03:35 <AaronSw> zakim, ??p3 is SteveP
14:03:36 <Zakim> +SteveP; got it
14:04:03 <Zakim> +??P8
14:04:15 <jang_scri> * jang_scri is on a handset today, will have a hunch by 4pm :-)
14:04:17 <AaronSw> zakim, ??p8 is patricks
14:04:19 <Zakim> +Patricks; got it
14:04:36 <jang_scri> kicking off...
14:04:36 <Zakim> +??P10
14:04:38 <jang_scri> fine
14:04:42 <AaronSw> regrets: danbri, ..?
14:04:59 <AaronSw> zakim, ??p10 is jjc
14:05:00 <Zakim> +Jjc; got it
14:05:09 <jang_scri> scribe next week:
14:05:12 <jang_scri> jos!
14:05:14 <jang_scri> thanks jos
14:05:17 <jang_scri> roll call
14:05:19 <jang_scri> zakim, who is here?
14:05:19 <bwm> zakim, who is on the phone?
14:05:20 <Zakim> On the phone I see Jang, FrankM, Bwm, SteveP, GK, PatH, Jos_d, AaronSw, Patricks, Jjc
14:05:21 <Zakim> On the phone I see Jang, FrankM, Bwm, SteveP, GK, PatH, Jos_d, AaronSw, Patricks, Jjc
14:05:23 <Zakim> On IRC I see JosD, RRSAgent, jang_scri, DanCon, gk, Zakim, AaronSw, bwm, logger
14:05:47 <jang_scri> next teelcon next week
14:05:50 <jang_scri> minutes of last meeting:
14:05:55 <DanCon> * DanCon tries dialing in a 3rd time
14:05:59 <JosD> * JosD can not type as fast as Jan :-(
14:06:04 <jang_scri> http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0131.html
14:06:11 <jang_scri> completed actions:
14:06:12 <jang_scri> ok
14:06:15 <jang_scri> withdrawn :
14:06:16 <jang_scri> ok
14:06:22 <jang_scri> 8. book review
14:06:32 <Zakim> +DanC
14:06:38 <jang_scri> http://rdf.burningbird.net/
14:06:47 <jang_scri> "practical rdf"
14:06:52 <jang_scri> call for review, esp. wrt reification
14:06:54 <AaronSw> reification: the rdf big ugly, http://rdf.burningbird.net/chapters/chap4.htm
14:07:23 <jang_scri> 9: lbase document
14:07:33 <DanCon> who has looked at it?
14:07:36 <jang_scri> see and kobolds are _too_ weedy to begin with 
14:07:38 <jang_scri> ?
14:07:41 <jang_scri> see http://lists.w3.org/Archives/Public/www-archive/2002Oct/0046.html
14:07:51 <jang_scri> * jang_scri wonders wtf taht came from
14:08:15 <jang_scri> comments: bwm
14:08:21 <jang_scri> doc status, minor comment
14:08:30 <jang_scri> talk about "addressing two requirements"
14:08:36 <Zakim> -Jjc
14:08:51 <DanCon> * DanCon q+ note that the doc status is actually the responsibility of The Director, i.e. w3c staff, not the editor
14:08:52 <jang_scri> (wordsmithing discussion)
14:08:52 <Zakim> * Zakim DanCon, you typed too many words without commas; I suspect you forgot to start with 'to ...'
14:08:55 <DanCon> * DanCon q+ to note that the doc status is actually the responsibility of The Director, i.e. w3c staff, not the editor
14:08:56 <Zakim> * Zakim sees DanCon on the speaker queue
14:09:19 <Zakim> +??P9
14:09:27 <jang_scri> danc: w3 staff writes status
14:09:41 <AaronSw> zakim, ??p9 is jjc
14:09:42 <Zakim> +Jjc; got it
14:10:02 <gk> q+ to ask whether there should be some kind of WB review period if we are approving/requesting publication
14:10:03 <Zakim> * Zakim sees DanCon, Gk on the speaker queue
14:10:18 <jang_scri> jjc formal review?
14:10:22 <jang_scri> path: good idea, I think.
14:10:44 <jang_scri> bwm: call for reviewers.
14:10:51 <jang_scri> to publish next week
14:10:57 <jang_scri> path: would be happier like that
14:11:10 <jang_scri> reviwers: jjc, jos, jang.
14:11:14 <jang_scri> gk also
14:11:23 <gk> q-
14:11:24 <Zakim> * Zakim sees DanCon on the speaker queue
14:11:28 <jang_scri> ACTION: jos, jang, gk, jjc to review
14:11:29 <RRSAgent> * RRSAgent records action 1
14:11:32 <DanCon> * DanCon has a pile of TAG and WebOnt actions queued ahead of anything like reviewing lbase
14:11:33 <jang_scri> moving on
14:11:34 <AaronSw> ack danc
14:11:35 <Zakim> DanCon, you wanted to note that the doc status is actually the responsibility of The Director, i.e. w3c staff, not the editor
14:11:37 <Zakim> * Zakim sees no one on the speaker queue
14:11:38 <jang_scri> 10 abs syntax
14:11:55 <jang_scri> o Are datatypes restricted to XSD datatypes?
14:12:10 <AaronSw> zakim, who's talking?
14:12:11 <DanCon> Zakim, who's talking?
14:12:22 <Zakim> AaronSw, listening for 10 seconds I heard sound from the following: Bwm (32%), GK (95%), PatH (4%)
14:12:31 <jang_scri> zakim, mute gk
14:12:32 <Zakim> GK should now be muted
14:12:35 <Zakim> DanCon, listening for 11 seconds I heard sound from the following: Jang (4%), FrankM (21%), Bwm (29%), GK (52%), AaronSw (8%), DanC (67%)
14:13:09 <gk> * gk my phone's done something ... don't know what ... trying to hang up & redial
14:13:10 <jang_scri> jjc: trying to edit this, noticed difference in emphashs
14:13:17 <jang_scri> jjc wants to emphasise compatability
14:13:24 <jang_scri> patc, frankm wants to deemphasise it
14:13:34 <jang_scri> but noone seems to want to say, only xsd
14:13:40 <jang_scri> s/patc/pats
14:13:45 <DanCon> * DanCon doesn't hope to get group consensus on 'emphasis'; 'emphasis' is up to editorial discretion.
14:13:46 <Zakim> -GK
14:13:48 <jang_scri> jos: keeping it open is important
14:14:30 <jang_scri> DECISION: datatypes other than xsd are allowed
14:14:36 <DanCon> action test case?
14:14:38 <jang_scri> o Do typed literals have a language identifier?
14:14:45 <jang_scri> pats: yes
14:14:59 <jang_scri> for the practical reason that all strongly dted ontologies
14:15:04 <jang_scri> (eg, at nokia)
14:15:19 <Zakim> +??P4
14:15:23 <jang_scri> we want to differentiate different language literals for titles
14:15:24 <gk> zakim, ??P4 is GK
14:15:26 <Zakim> +GK; got it
14:15:38 <jang_scri> q+
14:15:39 <Zakim> * Zakim sees Jang on the speaker queue
14:15:49 <DanCon> * DanCon q+
14:15:50 <Zakim> * Zakim sees Jang, DanCon on the speaker queue
14:15:52 <jang_scri> bwm: is this only strings?
14:16:05 <jang_scri> pats: no, subclasses of strings (tokens) also want constraining by language
14:16:07 <AaronSw> what's a linguistic token?
14:16:20 <jang_scri> jjc: I think they shouldn't
14:16:26 <jang_scri> I think this comes down to emphasis
14:16:40 <jang_scri> pats' enumeration examples aren't within xsd datatypes, which explicitly doesn't include languages
14:16:46 <jang_scri> pats: i think they are
14:16:54 <jang_scri> we're talking about an rdf typed literal
14:17:03 <jang_scri> the rdf literal is language qualified
14:17:23 <jang_scri> jjc: the way of handling language that xsd assumes is that you do the correction on entry
14:17:30 <jang_scri> pats: that's locale, not language
14:17:42 <jang_scri> danc: if the range is xsd:string
14:17:49 <jang_scri> then "abc" = "abc", surely?
14:17:55 <gk> q+ to express a preference for no language for typed literals
14:17:57 <Zakim> * Zakim sees Jang, DanCon, Gk on the speaker queue
14:17:59 <jang_scri> regardless of lang tags
14:18:00 <jang_scri> q+
14:18:01 <Zakim> * Zakim sees Jang, DanCon, Gk on the speaker queue
14:18:39 <jang_scri> danc suggests xm literals
14:19:26 <jang_scri> danc: is carrying lang down into serialisation a c18n issue?
14:19:37 <jang_scri> danc:
14:19:58 <jang_scri> specific case, published rdf schema for rdfs has a french subproperty label
14:20:06 <jang_scri> what's that end up as in the abstract syntax?
14:20:18 <jang_scri> jjc: labels shouldn't be xsd strings as I've already pointed out
14:20:24 <jang_scri> a label is a human readable thing
14:20:36 <jang_scri> so the most natural way to express it is an xml literal
14:21:07 <gk> q-
14:21:09 <Zakim> * Zakim sees Jang, DanCon on the speaker queue
14:21:22 <jang_scri> q-
14:21:24 <Zakim> * Zakim sees DanCon on the speaker queue
14:21:51 <jang_scri> pats: for every single stringlike type, you need a geometric explosion of type x lang tag
14:21:56 <Zakim> * Zakim sees DanCon on the speaker queue
14:21:59 <AaronSw> ack dancon
14:22:00 <Zakim> * Zakim sees no one on the speaker queue
14:22:01 <DanCon> I sure hope for a test case for the <rdfs:label lang="fr">su...</rdf:label> case.
14:22:07 <jang_scri> danc: I want to ensure we have a test case for rdfs label case
14:22:15 <jang_scri> jjc: langstring gets to the heart of this
14:22:29 <jang_scri> jang can invent jang:langstring
14:22:38 <jang_scri> but it's not part fo xsd datatypes
14:23:01 <jang_scri> deciding that lang tag is ont part of all literals...
14:23:14 <jang_scri> is about supporting xsd as xsd and dropping some possibilities of supporting on-xsd datatypes
14:23:20 <gk> q+ to say I think that we should resist adding complexity to literals (RDF can handle this stuff anyway)
14:23:21 <Zakim> * Zakim sees Gk on the speaker queue
14:23:30 <jang_scri> jjc: I still think it's right to drop it, we should be concentrating on supporting xsd as xsd
14:23:31 <DanCon> amen, gk
14:23:35 <jang_scri> not closing the door on others too much
14:23:40 <jang_scri> but emphasise compat
14:23:42 <AaronSw> ack gk
14:23:42 <bwm> ack gk
14:23:43 <Zakim> Gk, you wanted to say I think that we should resist adding complexity to literals (RDF can handle this stuff anyway)
14:23:44 <Zakim> * Zakim sees no one on the speaker queue
14:23:45 <Zakim> * Zakim sees no one on the speaker queue
14:23:49 <jang_scri> gk: don't agree with alng as part of literals
14:23:57 <jang_scri> we should resist adding complexity to rdf literals
14:24:01 <AaronSw> +1 to that
14:24:11 <jang_scri> bwm: any more substantive comments?
14:24:21 <jang_scri> path: pats' need is clear and compelling
14:24:22 <JosD> +1 to that too
14:24:33 <jang_scri> other objections appear to be "i don't like that"
14:24:45 <jang_scri> pats:
14:24:53 <jang_scri> the way we do it now reflects the M+S methodology
14:24:59 <DanCon> it's not "I don't like that"; it's "that's observably more complex and costly"
14:25:10 <gk> * gk have we got a URL for PatrickS' use-case?
14:25:14 <gk> * gk have we got a URL for PatrickS' use-case?
14:25:23 <gk> * gk have we got a URL for PatrickS' use-case?
14:26:11 <jang_scri> (discussion of details of ways to deal with this)
14:26:40 <jang_scri> * jang_scri notes that ""-en is <> ""-fr. saying nothing in french is completely different :-)
14:27:04 <jang_scri> does lang participate in lexical-> value mapping?
14:27:12 <jang_scri> pats says no, but it's available.
14:27:17 <gk> q+ to note complexity in defining typed literal equality  "10"<xsd:inteer>-en vs "10"<xsd:integer>-fr
14:27:18 <Zakim> * Zakim sees Gk on the speaker queue
14:27:35 <jang_scri> jjc: xsd string doesn't regard datatype
14:27:44 <jang_scri> so the items have the same _value_ but are different lexically
14:27:51 <jang_scri> pats: waht about xsd in 10 and xsd int 010
14:28:00 <jang_scri> differ lexically, same value
14:28:22 <jang_scri> danc: lang people I've seen use rdf properties, not xml:lang
14:28:37 <jang_scri> you can't query for stuf in french!
14:28:48 <jang_scri> * jang_scri (notes that depends on your query language)
14:28:59 <AaronSw> I thought the whole point of datatypes was that jenny[age] would return 10, not "10". now it's supposed to return (10, "en")?
14:29:25 <jang_scri> bwm: straw poll..?
14:29:28 <jang_scri> question is:
14:29:31 <DanCon> bwm, pls do prefer and live-with
14:29:36 <jang_scri> do dt literals have a lang string?
14:29:37 <bwm> zakim, who is on the phone?
14:29:39 <Zakim> On the phone I see Jang, FrankM, Bwm, SteveP, PatH, Jos_d, AaronSw, Patricks, DanC, Jjc, GK
14:29:51 <jang_scri> path: "are they allowed to have a lang string in the abs syntax"?
14:30:09 <DanCon> "how many prefer can have lang"
14:30:16 <DanCon> "how many cannot live with can-have-lang"?
14:30:22 <jang_scri> proposal:
14:30:38 <jang_scri> dt literals CAN have a lang tag in the abs syntax
14:30:49 <jang_scri> prefer:
14:31:05 <jang_scri> path, pats, frank prefers
14:31:11 <jang_scri> steve prefers
14:31:21 <jang_scri> can't live with:
14:31:29 <jang_scri> nobody
14:31:39 <AaronSw> i'm not going to implement it.
14:31:40 <jang_scri> prefers NOT:
14:31:49 <jang_scri> danc jos jjc
14:31:55 <jang_scri> aaron
14:32:00 <jang_scri> gk
14:32:12 <DanCon> (it's not essential to get names in straw-polls; just rough numbers)
14:32:13 <jang_scri> can't live with cannot:
14:32:14 <jang_scri> pats
14:32:42 <jang_scri> jjc: as editor, sounds like put it in, we might get feedback
14:32:44 <DanCon> yes, one cannot-live-with trumps 4 prefers.
14:32:55 <AaronSw> is refusing to implement a cantlivewith?
14:33:21 <AaronSw> I guess not.
14:33:24 <jang_scri> DECISION:  dt literals CAN have a lang tag in the abs syntax
14:33:30 <DanCon> AaronSw, if *nobody* implements it, the spec can't go to PR. But if everybody else implements it, your sorta out in the cold.
14:33:49 <jang_scri> moving on...
14:33:56 <jang_scri> o Does the abstract syntax require a datatype URI to refer to a
14:33:57 <jang_scri> datatype and that the lexical form be from the lexical space of that
14:33:57 <jang_scri> datatype
14:33:58 <gk> q?
14:33:59 <Zakim> * Zakim sees Gk on the speaker queue
14:34:06 <AaronSw> I guess I'm already in the cold in some sense.
14:34:22 <jang_scri> jjc: xsd:foobar
14:34:34 <jang_scri> is an error
14:34:41 <jang_scri> at what point does that error occur...
14:34:47 <jang_scri> xsd:int "foobar" is an error too
14:34:52 <jang_scri> where does _that_ occur?
14:34:53 <jang_scri> options:
14:34:59 <jang_scri> syntactic
14:35:03 <jang_scri> or mtetic
14:35:07 <jang_scri> or alter on
14:35:16 <jang_scri> jjc: my interest is whether it stops at abs syntax
14:35:24 <jang_scri> danc: when I asked that question,
14:35:31 <jang_scri> I got the answer: well-formed abs syntax
14:35:34 <jang_scri> not a mt error
14:35:48 <jang_scri> I got, "all uri + string paifrs are ok"
14:36:06 <jang_scri> danc: it's acceptable to me, although odd
14:36:16 <jang_scri> jjc: in the latest wording I've talked about generating the value...
14:36:31 <jang_scri> so eventually we get from xsd:int "10" to the integer 10.
14:36:41 <jang_scri> it's full of ifs and buts...
14:36:58 <jang_scri> one option is to not talk about values or even datatypes in the abs syntax
14:37:11 <jang_scri> it's possible, but feels strange
14:37:33 <jang_scri> bwm: path - what does it do for you if the abs syntax can have not-well-formed dt literals?
14:37:37 <jang_scri> path: doesn't bother me
14:37:50 <jang_scri> bwm: sounds like the easy path is to say it's not an abs syntax error..?
14:38:00 <gk> * gk rdf vs rdf+[some datatype] entailments, allows for datatype extensibility?
14:38:03 <jang_scri> path: there's something clearly wrong with it
14:38:20 <jang_scri> path I can live with it either way, I can see arguments both was
14:38:22 <DanCon> it seems straightforward to say "the use of a datatype literal is a claim that the string is in the value space of the datatype denoted/indicated by the URIref, but this claim isn't guaranteed to be true by the RDF spec"
14:38:31 <jang_scri> jjc: doesn't requre all parsers to know all datatypes
14:38:35 <jang_scri> q+
14:38:36 <Zakim> * Zakim sees Gk, Jang on the speaker queue
14:38:42 <JosD> q+
14:38:44 <Zakim> * Zakim sees Gk, Jang, JosD on the speaker queue
14:38:47 <Zakim> -AaronSw
14:38:53 <jang_scri> ack gk
14:38:54 <Zakim> Gk, you wanted to note complexity in defining typed literal equality  "10"<xsd:inteer>-en vs "10"<xsd:integer>-fr
14:38:55 <Zakim> * Zakim sees Jang, JosD on the speaker queue
14:39:15 <jang_scri> pats: jjc, you can still talk about values in the abs syntax in the sense of...
14:39:34 <jang_scri> the way rdf:datatype is defines that the dt plus lexical form unambiguously denote a value
14:39:44 <jang_scri> that's defined. you don't know what the value is in the abs syntax...
14:39:51 <jang_scri> bwm I don't think that's right.
14:39:53 <DanCon> what PatS is saying doesn't compute with (xsd:int, "foo")
14:40:01 <Zakim> +AaronSw
14:40:35 <jang_scri> pats: that conforms to eh abs syntax because otherwise
14:40:44 <jang_scri> the abs syntax, parsers need to know about all dts
14:41:01 <jang_scri> adding dts means modifying abs syntax, mt, etc.
14:41:14 <jang_scri> jjc: practical applications will have a fallback syntax
14:41:24 <jang_scri> danc: pats' position is a coherent one to me
14:41:43 <gk> * gk What DanC said.
14:41:46 <bwm> zakim, who is on the queue
14:41:47 <Zakim> I don't understand 'who is on the queue', bwm
14:41:48 <bwm> zakim, who is on the queue?
14:41:49 <Zakim> I see Jang, JosD on the speaker queue
14:42:07 <jang_scri> pats: you can have triples that are vald in the abs syntax ...
14:42:15 <jang_scri> at a higher level you find a problem
14:42:22 <jang_scri> danc: what about int 10 and int 010
14:42:36 <jang_scri> jjc: cardinality constraints act at mt level
14:42:53 <jang_scri> the mt gives denotations for nodes in the abs syntax
14:43:13 <jang_scri> danc: jjc's position doesn't soud coherent to me.
14:43:19 <gk> q+ to comment on possible MT distinction between rdf- and rdf+[datatype] entailment
14:43:20 <bwm> ack jan
14:43:20 <Zakim> * Zakim sees Jang, JosD, Gk on the speaker queue
14:43:21 <Zakim> * Zakim sees JosD, Gk on the speaker queue
14:44:02 <bwm> ack JosD
14:44:03 <Zakim> * Zakim sees Gk on the speaker queue
14:44:12 <jang_scri> jang: xsd emphasis question again
14:44:14 <jang_scri> jos: agree
14:44:23 <DanCon> I could live with a datatype syntax that *only* handled integers. no dates, months, etc.
14:44:23 <jang_scri> xsd is well-known and well-defined
14:44:28 <bwm> ack Gk
14:44:29 <Zakim> Gk, you wanted to comment on possible MT distinction between rdf- and rdf+[datatype] entailment
14:44:31 <Zakim> * Zakim sees no one on the speaker queue
14:44:36 <jang_scri> [danc: yes]
14:44:57 <DanCon> brian, please don't separate the abstract syntax from the MT. My position on the questions are not separable.
14:45:02 <jang_scri> gk: I can see the possibility of distinguishing betwen rdf-entails an drdf+dt - entails
14:45:13 <bwm> DanCon: noted
14:45:14 <JosD> Oh Dan that's actually a very good poit!
14:45:42 <JosD> ...just numbers
14:45:44 <jang_scri> ... and that's monotonic
14:46:05 <jang_scri> ie, rdf-entailments don't hold, rdfrdt-entailments _do_ hold
14:46:06 <JosD> s/poit/point
14:46:13 <jang_scri> pats: just posted code to wg to demonstrate this
14:46:14 <gk> ack gk
14:46:16 <Zakim> * Zakim sees no one on the speaker queue
14:46:29 <jang_scri> danc: either int 10 and int 010 are distinguishable, or they're identical
14:46:41 <jang_scri> path: owl's already there, it's got equality
14:46:54 <jang_scri> danc: the point of literals is that it's evident from the syntax what they denote
14:47:01 <jang_scri> because they're tidy
14:47:16 <jang_scri> bwm: klook at next question...
14:47:30 <jang_scri> o is there a generic value entailment in the model theory:
14:47:31 <jang_scri>   <a> <b> <dt1>"foo" .
14:47:31 <jang_scri> entails
14:47:32 <jang_scri>   <c> <d> <dt2>"bar" .
14:47:32 <jang_scri> whenever dt1.value("foo") = dt2.value(bar)?
14:47:52 <jang_scri> entails <a> <b> <dt2>"bar"
14:47:55 <jang_scri> not c, d
14:48:13 <jang_scri> danc: eg, dt1 = decimal, foo = 10, dt2 = int, bat = 010
14:48:19 <jang_scri> in xsd, those denote the same value
14:48:28 <jang_scri> pats: my problem with this...
14:48:31 <jang_scri> as gk mentioned...
14:48:39 <gk> I think "in the model theory" isn't clear -- RDF-entailment, or ?-entailnent?
14:48:43 <jang_scri> is that distinguishing between rdf-ent and rdfdt-entailment
14:48:54 <jang_scri> is much clearer
14:49:07 <jang_scri> path: that sounds a sensible thing to say
14:49:20 <jang_scri> jjc: gk's proposal would clarify a number of these issues
14:49:34 <jang_scri> gk: proposal: to distinguish between rdf-entailment (without dt knowledge)
14:49:52 <jang_scri> and rdfdt-entailment, which is what you get when you've got knowledge of dt mappings
14:49:56 <DanCon> hmm... this smells a lot like untidy literals.
14:49:59 <Zakim> -AaronSw
14:50:04 <jang_scri> that is, knowledge of dts relevant to the entailment you're testing
14:50:11 <DanCon> maybe I can live with that for these new things.
14:50:15 <jang_scri> path: the mt would include this if we choose this
14:50:36 <Zakim> +AaronSw
14:50:54 <jang_scri> danc: that would be incompletely specified, eg danc:fhiwuhfeiw is unknown ?
14:51:11 <jang_scri> is there one new class of entailments, or 2^n new classes?
14:51:14 <jang_scri> path: 2^n
14:51:27 <jang_scri> where n = | { datatypes } |
14:51:42 <jang_scri> mt is parameterised in terms of the dts you give it
14:51:49 <jang_scri> danc: how do we get out of cr on this?
14:52:01 <jang_scri> jjc: we do it for a base conformance test with xsd schema types?
14:52:20 <jang_scri> path: the rdfdt-entailment takes a set of dts as a parameter...
14:52:29 <jang_scri> conformance level for CR we use xsd datatypes
14:52:41 <jang_scri> pats: that's quite good, the code I posted shows that that works
14:52:55 <jang_scri> path: one qualification: we want to watch xsd union dts
14:53:02 <jang_scri> danc: don't feel compelled to support those
14:53:09 <jang_scri> bwm: I hear agreement here.
14:53:32 <Zakim> * Zakim sees no one on the speaker queue
14:53:46 <jang_scri> pats: describing an app that does rdf-entailment, not rdfdt-entailment is important
14:54:01 <jang_scri> danc: must have test suite with various distinctions.
14:54:23 <DanCon> [he was asking 'Dan' but Dan is happy that Jan said yes]
14:54:46 <jang_scri> action: jang to add test suite stuff for this
14:54:47 <RRSAgent> * RRSAgent records action 2
14:54:55 <jang_scri> o Does the abstract syntax require a datatype URI to refer to a
14:54:56 <jang_scri> datatype and that the lexical form be from the lexical space of that
14:54:56 <jang_scri> datatype
14:54:58 <DanCon> * DanCon is prepared to stay for 20 extra minutes
14:55:01 <jang_scri> jjc: that's now resolved
14:55:11 <jang_scri> the answer is the abs syntax doesn't know about any specific dts
14:55:12 <gk> q+ to pose a test case
14:55:13 <Zakim> * Zakim sees Gk on the speaker queue
14:56:01 <jang_scri> bwm: can we give a dt to old-style literals?
14:56:08 <jang_scri> jjc: it would take longer
14:56:18 <jang_scri> q+
14:56:20 <Zakim> * Zakim sees Gk, Jang on the speaker queue
14:56:29 <gk> q+ to ask does "10"<xsd:integer>-en entail the same things as "10"<xsd:integer>-fr under rdfdt-entailment(xsd:integer)
14:56:30 <Zakim> * Zakim sees Gk, Jang on the speaker queue
14:56:40 <jang_scri> path: is the proposal to have a dummy type for untyped literals?
14:56:50 <jang_scri> jjc: not xsd:string, eg rdf:langstring
14:57:06 <jang_scri> danc: previous decision means xsd:string works, since they come with lang tags now
14:57:27 <Zakim> -AaronSw
14:57:37 <jang_scri> path: don't want to default to xsd:string
14:57:55 <jang_scri> q+ to say stuff about old-style literals 
14:57:56 <Zakim> * Zakim sees Gk, Jang on the speaker queue
14:58:47 <jang_scri> jjc: specific prblem is with xml:lang
14:58:59 <jang_scri> if a doc has half literals explicitly typed xsd:string
14:59:03 <jang_scri> no langs anywhere
14:59:15 <jang_scri> and another doc with root elt xml:lang tagged
14:59:31 <jang_scri> then there's entailment if literals are typed
14:59:33 <jang_scri> there's not typed
15:01:03 <jang_scri> [a brawl ensues over lang tags]
15:01:16 <gk> Does "foo"<xsd:string>-en entail the same things as "foo"<xsd:integer>-fr with knowledge of xsd:string?
15:01:17 <jang_scri> meeting formally closes... bwm
15:01:22 <jang_scri> danc: I'd like to extend the meeting
15:01:36 <jang_scri> extend until 10 past
15:01:40 <jang_scri> danc:
15:01:47 <jang_scri> the lang tags are orthogonal to the dt values, right?
15:01:56 <jang_scri> jjc: in the abs syntax, ...
15:02:05 <jang_scri> a literal is a triple (uri, string lang tag)
15:02:14 <jang_scri> danc: which maps to (value, lang tag)
15:02:35 <jang_scri> path: the record was about syntax, not denotation
15:02:52 <jang_scri> danc: then I'm not happy with "DECIDED" above
15:03:11 <jang_scri> bwm: the question should have been, "does the abs syntax have ..."
15:03:33 <jang_scri> bwm: it's up to entailment for dt to determine if lang tag has a meaning
15:03:42 <jang_scri> danc: can't live with that
15:04:12 <jang_scri> xsd:string maps strings to values.
15:04:15 <jang_scri> danc: yes!
15:04:35 <jang_scri> pats: that's why a dt xsd:string is compatible with, but is not the same as, xsd:string
15:04:49 <Zakim> * Zakim sees Gk, Jang on the speaker queue
15:04:51 <jang_scri> danc: strings <> string x lang tags
15:04:53 <Zakim> * Zakim sees Gk, Jang, Jjc on the speaker queue
15:04:59 <jang_scri> q- to take it to rdf-comments
15:05:00 <Zakim> * Zakim sees Gk, Jjc on the speaker queue
15:05:25 <jang_scri> pats puts his pov over lang tag not affecting literal->value mapping
15:05:30 <jang_scri> it's there for applications, that's all
15:05:42 <jang_scri> * jang_scri can't scribe this, one at a time
15:05:58 <gk> Well, does 10-en == 10-fr?
15:05:59 <jang_scri> danc: so xsd:int "10", fr -> (integer 10, "fr")
15:06:16 <jang_scri> path: lang tag is like the font.
15:06:20 <jang_scri> jjc:
15:06:45 <jang_scri> I think the fragile position on language has collapsed, but we had a good consensus on how to do the mt
15:06:54 <DanCon> if the datatype mapping can throw away language tags, I object.
15:07:01 <jang_scri> I'll take an action on writing up position thus far, and take it to next week for lang tags
15:07:26 <gk> DanCon, I tend to agree, which why I preferred no lang tags for typed literals
15:07:35 <jang_scri> danc - if the dt mapping _can't_ throw away lang tags for some things, it makes no sense, surely?
15:08:05 <jang_scri> bwm clarifies everyone's position, or tries to.
15:08:18 <jang_scri> as to where the lang tag lives in abs syntax, denotation, etc.
15:09:21 <jang_scri> jjc: for me, the whole goal of an rdf graph is to convey meaning that's realised through the mt
15:09:42 <jang_scri> to carry information that's not realised theouth the tm makes no seense, and seems contrary to entire design criteria of rdf mt
15:10:08 <jang_scri> path: if literals can be subjects then this becomes straightforward:
15:10:16 <jang_scri> "the literal itself is french, and it denotes 10"
15:10:40 <Zakim> -SteveP
15:11:03 <gk> I think if have a MT where a typed literal node denotes (10,en) rather than just 10 is messy (unless, maybe, the language tag applies to *all* values denoted in a graph?)
15:11:32 <DanCon> * DanCon plays "I told you so" card w.r.t. option C in last week's decision. B is soooo much simpler and really does meet the requirements just as well.
15:11:41 <jang_scri> jang: that is, the literal (xsd:integer, "10", fr) is french and denotes 10
15:12:04 <bwm> DanCon: right re simplicity
15:12:24 <jang_scri> path: DLogicians will tear us limb from limb re lits as subjects, it's a no go
15:12:25 <DanCon> yes, we tried. oh well.
15:12:43 <jang_scri> bwm: we're out of time, we can't avoid to debate this forever
15:12:58 <jang_scri> ACTION: jjc to write up current position, and to try to make a coherent proposal
15:12:59 <RRSAgent> * RRSAgent records action 3
15:14:17 <jang_scri> (jjc accepts action to get at least jjc, path on teh same page)
15:14:47 <Zakim> -PatH
15:14:48 <jang_scri> meeting closes.
15:14:53 <Zakim> -Jjc
15:14:54 <Zakim> -Jang
15:14:55 <Zakim> -Patricks
15:14:55 <Zakim> -Jos_d
15:14:56 <Zakim> -FrankM
15:14:57 <Zakim> -GK
15:14:58 <Zakim> -Bwm
15:15:00 <Zakim> -DanC
15:15:01 <Zakim> SW_RDFCore()10:00AM has ended
15:15:06 <JosD> JosD has quit
15:15:10 <DanCon> bwm, was my reversal too incoherent? did it make sense that the WG isn't in the critical path now?
15:16:09 <gk> Ummm... I *think* I know what you meant ... something like what we need is a coherent position based on last week's decision, which the WG should accept absent new information?
15:16:39 <DanCon> yes.
15:16:56 <jang_scri> last week's decision sounded like stake-in-the-ground: gives a basis for editor's work, but future decisions should be based in the light of it
15:17:01 <jang_scri> ONT that it's all done and dusted
15:17:04 <gk> For myself, I find it *really* hard to follow the discussions in a telecon:  having a written statement to pick holes in is much easier for me.
15:17:22 <DanCon> well, jang, there will be a decision to publish.
15:17:39 <jang_scri> I sympathise strongly with the "i want to finish" danc :-)
15:18:07 <DanCon> but if we wanted to see text from the editors before deciding the datatypes issue (which, ideally, I would have much preferred), we should have said so *before* deciding.
15:18:33 <jang_scri> it might have been preferable, but wuld have required two or three versions of the documents, it seems.
15:18:40 <jang_scri> positions looked irreconcilable.
15:19:03 <gk> yeah... mind you, we did have several texts -- too many, one might say.  Now where did I leave that silver bullet?
15:19:20 <jang_scri> I dunno,  but please shoot me with it :-)
15:19:41 <DanCon> yes, gk, I definitely felt like I was designing while playing video game or something today. Thanks for getting us a concepts draft to look at asynchronously.
15:19:44 <jang_scri> gotta go write this up. I'll make sure danc's objections to wording of decision are in there.
15:20:12 <DanCon> jang, I'd rather that the record just showed that the WG advised the editors in a certain way, not that we made any WG decisions.
15:20:39 <DanCon> cuz like I said, if the question that was clarified after-the-fact were put to me, I think I would have to object.
15:20:56 <jang_scri> ok, in which case I'll write it up like that and let people fight over the minutes.
15:21:02 <bwm> bwm has quit
15:22:06 <gk> This raises a meta-question about the proper granularity of WG decisions.  When there's no conflict, it seems easy, but when there's conflict how do we avoid descent into micro-design-by-committee?
15:22:08 <jang_scri> heh, better to get people to agree and then clarify the question afterwards :-)
15:22:16 <gk> :-)
15:22:44 <gk> Now, is it that the victors write history, or the historians get to declare the victors?
15:23:18 <jang_scri> I dunno, I think we're stuck in the bowels of the colliseum somewhere debating RDF while the goths armpage overhead
15:23:28 <jang_scri> we'll emerge and wonder where everypone's gone :-)
15:23:37 <jang_scri> my typing's awful today
15:23:51 <DanCon> re meta-question, I would have like to see the "do we *really* need datatypes in this version of RDF?" question escalated. I'd like the chair to get some of TimBL's time to see if he thinks we should push the WG to finish this design, when it's clear that the WG has diverse, if not irreconcileable, views on it.
15:23:52 <gk> I'm glad we can still chuckle about it :-)
15:24:30 <jang_scri> I think DTs are very important, but concerned also that position C from last week doesn't represent much of a compromise.
15:24:36 <jang_scri> have to drop a line to -comments about it
15:24:45 <DanCon> i.e. maybe the right answer is "publish what you've got without datatypes; go play for 6 months and come back with more implementation experience."
15:24:59 <gk> Re  "do we *really* need datatypes in this version of RDF?";  Guha madea compelling case a few weeks back that RDF without the capability to express basic ideas like integers would be pretty lame.  Not an answer, maybe, but a view difficult to ignore.
15:25:12 <DanCon> yup.
15:25:28 <jang_scri> yeah. like the phraseology too.
15:25:45 <jang_scri> rdf - dts = lame. rdf + dts = '1337 or whatever the hep cats call it
15:25:56 <DanCon> the problem with the "publish what you've got" issue is we'd still have to decide whether literals were tidy or not.
15:26:11 <gk> ... mind you, some of our conflicts now are precisely because of the implementation experience
15:26:25 <jang_scri> that's decided, isn't it?
15:26:26 <gk> Yes, we couldn't entirely dusk that one.
15:27:06 <jang_scri> right see you later.
15:27:12 <jang_scri> jang_scri has quit
15:27:52 <gk> OK, time for my weekend.  Bye.
15:28:03 <gk> gk has left #rdfcore