W3C RDF Core Working Group IRC Chat Logs for 2003-08-08

13:01:02 Users on #rdfcore: @logger

13:56:15 <jang> hiya

13:56:24 <jang> jang is now known as jang-scribe

13:57:34 <jang-scribe> agenda http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Aug/0090.html

13:58:06 <DaveB> hi from HP - dialling

13:58:36 <jang-scribe> zakim, who is here?

13:59:02 <jang-scribe> zakim, who is here?

13:59:02 <Zakim> sorry, jang-scribe, I don't know what conference this is

13:59:03 <Zakim> On IRC I see Zakim, jang-scribe, bwm, DaveB, danbri, logger

13:59:09 <jang-scribe> zakim, this is rdf

13:59:09 <Zakim> ok, jang-scribe

13:59:16 <jang-scribe> zakim, who is here?

13:59:16 <Zakim> On the phone I see ??P15, ??P18

13:59:17 <Zakim> On IRC I see Zakim, jang-scribe, bwm, DaveB, danbri, logger

13:59:24 <jang-scribe> zakim, ??p15 is ilrt

13:59:24 <Zakim> +ilrt; got it

13:59:28 * danbri on his way

13:59:28 <jang-scribe> zakim, ilrt has jang

13:59:28 <Zakim> +jang; got it

13:59:29 <Zakim> +FrankM

13:59:35 <jang-scribe> zakim, ??p18 is hp

13:59:35 <Zakim> +hp; got it

13:59:41 <jang-scribe> zakim, hp has daveb bwm

13:59:41 <Zakim> +daveb, bwm; got it

13:59:46 <jang-scribe> zakim, who is here?

13:59:46 <Zakim> On the phone I see ilrt, hp, FrankM

13:59:47 <Zakim> ilrt has jang

13:59:48 <Zakim> hp has daveb, bwm

13:59:49 <Zakim> On IRC I see Zakim, jang-scribe, bwm, DaveB, danbri, logger

14:00:27 <Zakim> +DanBri

14:00:43 <Zakim> +??P21

14:00:48 <DaveB> we got danbri's voicemail

14:01:02 <jang-scribe> danbri, what voice over IP are you using?

14:01:06 <jang-scribe> zakim, ??p21 is jjc

14:01:07 <Zakim> +jjc; got it

14:01:11 <danbri> crap

14:01:14 <danbri> zakim, drop danbri

14:01:14 <Zakim> DanBri is being disconnected

14:01:15 <Zakim> -DanBri

14:01:17 <danbri> vonage

14:02:01 <Zakim> +Mike_Dean

14:02:12 * danbri reboots voip box, sorry fror delay

14:02:36 <danbri> I'm wearing Staff Contact hat today btw... (Eric can't make this week's call)

14:02:43 <bwm> Zakim, who is on the phone?

14:02:43 <Zakim> On the phone I see ilrt, hp, FrankM, jjc, Mike_Dean

14:02:44 <Zakim> ilrt has jang

14:02:45 <Zakim> hp has daveb, bwm

14:03:24 * jang-scribe has typical Englishman weather aversion: too hot in the summer, too wet in the winter

14:03:48 <jang-scribe> danbri on the way

14:03:54 <Zakim> +DanBri

14:03:59 <jang-scribe> regrets: gk

14:04:17 <jang-scribe> aob

14:04:17 * danbri rewires telephone to landline

14:04:20 <jang-scribe> xmlsch-02

14:04:26 <jang-scribe> (bwm, since jjc's here)

14:04:58 <jang-scribe> bwm: regrets for next week, on holiday, seeks chair

14:05:05 <jang-scribe> danbri looks at ceiling

14:05:11 <jang-scribe> danbri to chair next week

14:05:27 <jang-scribe> scribe next week: for next week

14:05:31 <jang-scribe> minutes last telecon

14:05:42 <jang-scribe> approved

14:05:51 <jang-scribe> danbri with AOB: french translation review would be good

14:06:14 <jang-scribe> completed actions

14:06:28 <jang-scribe> frankm: intentional use of subclassof, subproperty of

14:06:39 <Zakim> +Pat_Hayes

14:06:50 <jang-scribe> the primer is fine if that just means "if" and not "only if"

14:07:14 <jang-scribe> bwm: frankm: that action was one of several along the lines, "make sure it's not inconsistent"

14:07:29 <jang-scribe> [scribe begs for someone else to paste URLs as we go, mouseless at the moment]

14:07:33 <jang-scribe> all done.

14:07:36 <jang-scribe> withdrawn actions:

14:07:59 <jang-scribe> approved as withdrawn

14:08:28 <jang-scribe> item 9

14:08:32 <jang-scribe> <br /> testcase

14:08:41 <jang-scribe> DaveB: martin duerst tc suggeted

14:08:52 <jang-scribe> might help users to see how canonicalisation works

14:09:01 <jang-scribe> additional chamnge, people would have to test

14:09:21 <jang-scribe> jang-scribe: don't see that it hurts to add it

14:09:27 <jang-scribe> DaveB: two options:

14:09:29 <Zakim> +??P17

14:09:33 <jang-scribe> either with or without html namespace

14:09:46 <jang-scribe> zakim, ??p17 is pats

14:09:46 <Zakim> +pats; got it

14:09:59 <jang-scribe> jjc: preference for adding it, but a weak one.

14:10:28 <jang-scribe> DaveB: we're strictly testing another spec, but it looks harmless

14:10:57 <jang-scribe> ACTION jang to add test case option 1. (without html namespace)

14:11:09 <jang-scribe> prop: daveb second jjc, no against, no abst

14:11:12 <jang-scribe> approved

14:11:18 <jang-scribe> item 10

14:11:33 <jang-scribe> davbe: two ways to put XMLLiteral in the graph

14:11:47 <jang-scribe> martin suggested cutting out the possibilty of using datatype= to put it in the graph

14:11:59 <jang-scribe> it would cause something to break if we add this; I'm less in favour of this

14:12:28 <jang-scribe> re: item 9: ACTION daveb to inform them we've adopted that TC

14:12:49 <jang-scribe> item 10 again...

14:12:49 <danbri> danbri has changed the topic to: rdfcore 2003-08-08 http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Aug/0090.html

14:13:00 <jang-scribe> jjc: don't like it.

14:13:18 <jang-scribe> the current documents do restrict the lexical forms to be exccan form

14:13:31 <jang-scribe> you CAN put in an ill-formed DT, you can do that with anything, it's just wrong

14:13:44 <jang-scribe> DaveB: does concepts mention ill-formed dts near xml lits?

14:14:03 <jang-scribe> jjc: nope. We could add a test case with rdf:dt markup and ill-formed XML literal

14:14:18 <jang-scribe> it's a non-entailment test, although they cause problems for semantics implementation

14:14:30 <jang-scribe> DaveB: pfps pointed out you need to do this anyway

14:14:34 <jang-scribe> bwm: pats, any thoughs?

14:14:39 <jang-scribe> pats: wasn't really happy with this

14:14:48 <jang-scribe> it's yet another way that xMLLiteral is special

14:15:04 <jang-scribe> it's impoprtant to say, here's syntactic sugar, but here are DT literals, this is how you do it

14:15:13 <jang-scribe> so that it's easier for a user to do it editing by hand

14:15:13 * Zakim jang-scribe, you typed too many words without commas; I suspect you forgot to start with 'to ...'

14:15:32 <jang-scribe> zakim, shut up

14:15:32 <Zakim> I don't understand 'shut up', jang-scribe

14:15:54 <jang-scribe> pats: could say, "we did this for a reason, here be dragons..."

14:15:59 <jang-scribe> DaveB: can put that in syntax perhaps

14:16:03 <jang-scribe> bwm: anyone to argue for it

14:16:07 <jang-scribe> noone speaks

14:16:17 <jang-scribe> ok, that's a decision that we're not going to so restrict the RDF datatypes

14:16:27 <jang-scribe> what about meliorating action?

14:16:42 <jang-scribe> DaveB: advice in the doc? say JMUST and SHOULD?

14:17:03 <jang-scribe> bwm: I'm worried about filling our documents with caveats and corner-cases

14:17:07 <jang-scribe> bwm: something for the faq?

14:17:18 <jang-scribe> bwm: path, can I use rdf-dt to write XMLliteral

14:17:30 <jang-scribe> answer: yes, but there's another way to do it, which you might prefer manually

14:17:49 <jang-scribe> pats: I'd rather dump canonicalised lex form and dump it as a typed literal

14:18:05 <jang-scribe> pats: faq sounds good, but use less flippant language.

14:18:33 <jang-scribe> ACTION pats to draft an FAQ for XMLLiteral via rdf:datatype or otherwise

14:19:12 <jang-scribe> jjc: recently updated the serialiser to be tunable to do either

14:19:16 <jang-scribe> moving on

14:19:21 <jang-scribe> item 11: i18n update

14:19:43 <jang-scribe> bwm attended another call

14:19:55 <jang-scribe> his belief: that martin was pushing the decision to formally object

14:20:11 <jang-scribe> but haven't checked yet, not sure about the status of that

14:20:13 <jang-scribe> moving on

14:20:15 <jang-scribe> item 12:

14:20:22 <jang-scribe> denotation of rdf:XMLLiteral

14:20:46 <jang-scribe> martin pointed out the test case

14:20:50 <jang-scribe> (someone grab that url?)

14:21:10 * jang-scribe has a cursor drifting all over the place, can't paste urls.

14:21:22 <jang-scribe> nobody says we shouldn't make a change

14:21:27 <jang-scribe> jjc: like path's proposal

14:21:37 <jang-scribe> path: really pfps' proposal

14:21:50 <jang-scribe> path: summary is: we don't choose something, we say simply

14:21:53 <jang-scribe> there are things call xml values

14:22:03 <jang-scribe> they aren't character strings, they're not xsd values

14:22:10 <jang-scribe> they're just things in a set 1:1 with can form

14:22:25 <jang-scribe> pats: the problem is very simple. we're not saying the things are related to xml

14:22:50 <jang-scribe> pats: we need to define a dt such that implementors know what to do with the values

14:22:55 <jjc> q+ to mention msg 104 note

14:22:55 * Zakim sees jjc on the speaker queue

14:23:04 <jang-scribe> q+

14:23:04 * Zakim sees jjc, jang-scribe on the speaker queue

14:23:26 <jang-scribe> pats: at the lest, I'd want to say the value space of XMLLiteral is a subset of the XML infoset

14:23:57 <jang-scribe> path: I'd be happy to say that, the infoset isn't particularly well-defined either

14:24:17 <jang-scribe> pats: on that, we have loads of specs that define mappings from infoset to explicit things

14:24:25 <jang-scribe> so it can't actually be that the infoset is so vague

14:24:27 <DaveB> q+ to say PatH's lexical space isn't quite right, it gives octet sequences

14:24:27 * Zakim sees jjc, jang-scribe, DaveB on the speaker queue

14:24:33 <jang-scribe> that these definitions are built on sand, surely?

14:25:03 <jang-scribe> path: are you SURE the infoset is 1:1 with lexical forms?

14:25:17 <bwm> q?

14:25:17 * Zakim sees jjc, jang-scribe, DaveB on the speaker queue

14:25:50 <jang-scribe> pats: we still end up with the situation that we can do this right

14:25:51 <jang-scribe> ack jjc

14:25:51 <Zakim> jjc, you wanted to mention msg 104 note

14:25:52 * Zakim sees jang-scribe, DaveB on the speaker queue

14:26:09 <jang-scribe> path's folloowup message... I suggest replacing "not specified" value space

14:26:20 <jang-scribe> with note: one possibility is to use nodeset...

14:26:25 <jang-scribe> (see message for details)

14:26:48 <jang-scribe> bwm: pats, how woulod you feel about that?

14:26:55 <jang-scribe> pats: just responded to that with the comment that...

14:27:13 <jang-scribe> ... if we define the value space as xpath nodeset, it's just as syntactic a value as the canonicalised octet sequence

14:27:26 <jang-scribe> it's almost a serialisation of the infoset

14:27:34 <jang-scribe> jjc, would you agree with that?

14:27:43 <jang-scribe> jjc: it doesn't include everything in the infoset, right

14:28:04 <jang-scribe> pats: not happy with the amendment, about 30% happy with the suggestion

14:28:22 <danbri> q+ to ask how big a decision this?

14:28:22 * Zakim sees jang-scribe, DaveB, danbri on the speaker queue

14:28:27 <jang-scribe> path: I want to keep out of these debates about the identity of these wierd xml things

14:28:28 <bwm> ack jjc

14:28:28 * Zakim sees jang-scribe, DaveB, danbri on the speaker queue

14:28:42 <jang-scribe> if the xml experts are confident of these things, that there isn't a black hole

14:28:44 <bwm> ack jang

14:28:44 * Zakim sees DaveB, danbri on the speaker queue

14:29:31 <jang-scribe> jang-scribe: applications will deal with SOME representation

14:29:34 <jang-scribe> ie, lexical form

14:29:35 <bwm> ack daveB

14:29:35 <Zakim> DaveB, you wanted to say PatH's lexical space isn't quite right, it gives octet sequences

14:29:37 * Zakim sees danbri on the speaker queue

14:29:42 <jang-scribe> so why bother with the value space other than to say it's 1:1

14:29:44 <jang-scribe> ?

14:30:02 <jang-scribe> DaveB: there are problems at the moment that you can't get unicode strings for everything (..?

14:30:12 <jang-scribe> other thing: normal form c of that string is worthwhile putting in ther

14:30:17 <jang-scribe> so lexical space might need tweaking

14:30:19 <bwm> ack danbri

14:30:19 <Zakim> danbri, you wanted to ask how big a decision this?

14:30:20 * Zakim sees no one on the speaker queue

14:30:37 <jang-scribe> danbri: how important is this? jan's opinion reassures me

14:30:49 <jang-scribe> if we pick one and write it down, does it matter?

14:31:05 <jang-scribe> the relationship between rdf and xml has been the big topic that's haunted us since we started this

14:31:14 <jang-scribe> I'm lacking a sense of how technically important this is

14:31:26 <jang-scribe> jjc: I'd like to answer that if there's time

14:31:39 <jang-scribe> bwm: I'm thinking, how can I write a test case to tell the difference?

14:31:44 <jang-scribe> the only way I can tell is,

14:32:02 <jang-scribe> if someone else invented a datatype, whether they can say if their test case is definitely different from XMLLiteral

14:32:21 <jang-scribe> bwm: any other cases?

14:32:41 <jang-scribe> what I've heard pats say, and to some extend danbri say, is that this sounds more political than technical

14:32:48 <jang-scribe> that these values are intended to be fragments of xml

14:32:55 <Zakim> +??P0

14:33:00 <jang-scribe> pats: that's a fair synopsis

14:33:06 <jang-scribe> zakim, ??p0 is jos

14:33:06 <Zakim> +jos; got it

14:33:17 <danbri> zakim, jos is JosD

14:33:17 <Zakim> +JosD; got it

14:33:55 <jang-scribe> path: we don't have to have a form of words that fails pats' test

14:34:06 <jang-scribe> we can say the value space is, say, pairs of...

14:34:24 <jang-scribe> pats: that's like saying xsd:integer _could_ be integers, or pairs, or...

14:34:34 <jang-scribe> path: that's what xsd DO say - they use type markers

14:35:15 <jang-scribe> pats: my concern is that saying that we have a canonical rep

14:35:58 <jang-scribe> it smacks of saying "here's how you must implement this"

14:36:15 <jang-scribe> we've got people using dom, etc to model "the same infoset"

14:37:01 <jang-scribe> jang-scribe: we just need to test if two things are the same or not

14:37:16 <jang-scribe> pats: martin made the point that people are talking not about octet sequences

14:37:21 <jang-scribe> they're talking about richer things

14:37:32 <jang-scribe> saying the l2v is identity loses that semantics

14:38:01 <jang-scribe> path: the 1:1 is cute because it's invertible

14:38:09 <jang-scribe> so you can go back via canonical lexical form

14:38:25 <jang-scribe> I just want to say, the things in a 1:1 relationship ARE infoset things

14:38:33 <jang-scribe> jjc: my worry about infoset as opposed to nodeset

14:38:44 <jang-scribe> is that if we mention infoset we have to go through infoset conformance step

14:38:51 <jang-scribe> identifying things that are used etc.

14:39:01 <jang-scribe> we wind up reding the xpath nodeset work

14:39:04 <DaveB> q+ to say we can't just pick one of (infoset, xpath nodeset (which version), DOM, ...)

14:39:04 * Zakim sees DaveB on the speaker queue

14:39:30 <jang-scribe> path: "that infoset corresponding to the xpath 1.0 nodeset corresponding to the can lexical form"

14:39:46 <jang-scribe> bwm: what's the difference observable to software between deciding it's infoset as opposed to nodeset

14:40:18 <jang-scribe> pats: is an xpath nodeset considered to define a sequence of markup and content?

14:40:27 <jang-scribe> jjc: xpath nodeset is an abstract structure?

14:40:47 <jang-scribe> jjc: i can't answer that question, neither can software

14:41:01 <jang-scribe> pats: feel painted into a corner here

14:41:13 <jang-scribe> bwm: not deliberately, I'm trying to focus on the real issues

14:41:29 <jang-scribe> eg, if the argument is religious but important to send the right signals to the xcml community, that's a valid position

14:41:42 <jang-scribe> pats: I think it's impoprtant insofar as relating to the rest of the xml world

14:42:05 <jang-scribe> I'm doing this on behalf of martin because he has such a strong concern about distinguishing between a sequence of characters vs characters plus markup

14:42:11 <jang-scribe> and this distinguishment is important

14:42:19 <jang-scribe> bwm: we're not saying this is a sequence of characters

14:42:32 <jang-scribe> bwm: proposal: that we say the value space of XMLliterals is a set of xml fragments

14:42:46 <jang-scribe> xml fragments are disjoint from the set of unicode character sequences

14:42:55 <jang-scribe> and in 1:1 correspondence with (whatever the lexical space is)

14:43:12 <jang-scribe> ie, sense of abstraction, but we say "these really are bits of xml"

14:43:27 <danbri> q+ to ask re 1:1 underspecified? (what's the precise term, injective, surjective etc?)

14:43:27 * Zakim sees DaveB, danbri on the speaker queue

14:43:27 <jang-scribe> JosD: I fully agree. I find that a good proposal

14:43:37 <jang-scribe> path, pats, jos like it

14:43:48 <jang-scribe> bwm: jjs can you work this up?

14:43:51 <jang-scribe> jjc: two comments

14:44:06 <jang-scribe> one is that you've lost one of path's text that it's disjoint from xsd datatypes

14:44:15 <jang-scribe> pats: I'd like to have someone justify the inclusion of this?

14:44:32 <jang-scribe> jjc: "primitive datatypes"

14:44:37 <jang-scribe> pats: that changes a lot

14:44:59 <jang-scribe> jjc: that does mean that we're aying an xml literal is not a string, even when they only have a single string node

14:45:14 <jang-scribe> pats: wasn't that a problem for martin?

14:45:26 <jang-scribe> pats: why do we have to mention the relationship to xml schema?

14:45:29 <jang-scribe> DaveB: people keep asking

14:45:37 <jang-scribe> path: to avoid future discussions about this

14:45:58 <jang-scribe> pats: it strikes me as careless, we don't know if it's true...

14:46:03 <jang-scribe> path: we do if we define it like that

14:46:12 <jang-scribe> pats: but we don't know if we onstrain ourselves in that manner

14:46:28 <jang-scribe> we may later want to do something formally to equate xsd types and xml literals

14:46:37 <jang-scribe> I'm not convinced that it's even correct

14:46:48 <jang-scribe> path: "correctness" is meaningless, we just define it.

14:47:10 <jang-scribe> bwm: I think martin has in mind that xml literla swithout literals are really just plain literals

14:47:20 <jang-scribe> path: but he HAS to be wrong about that in order for there to be a logic

14:47:30 <jang-scribe> he wants names whose meaning changes according to context

14:48:38 <jang-scribe> pats: saying that these things are all distinct for simplicity's sake closes the door on useful stuff

14:48:52 <jang-scribe> path: xsd didn't say exactly that for simplicity sake, but for logical coherence

14:49:10 <jang-scribe> JosD: empty markup is still markup.

14:49:29 <jang-scribe> secondly, there can always be matching inference rules to do what pats wants, so this decision doesn't close the door on those options

14:49:52 <jang-scribe> bwm: is only pats arguing against disjointness?

14:49:56 <jang-scribe> jang-scribe: martin will argue that

14:50:39 <jang-scribe> bwm: we should discuss this, decide what we think is best, then go ask martin

14:50:53 <jang-scribe> or go to him and say, we can't choose, which one of these is best?

14:51:01 <jang-scribe> jjc: I'd prefer to choose one.

14:51:26 <jang-scribe> path: when pfps made this suggestion, martin reacted positively to it

14:51:42 <jang-scribe> pats: martin just sent a list of preferences

14:51:58 <jang-scribe> [scribe can't paste urls, laptop ipple in a twist]

14:52:34 <jang-scribe> pats reads martin's recent email

14:52:53 <jang-scribe> path: we're not going to get away with being agnostic, we'll get shot, we've tried that before

14:53:07 <jang-scribe> if they have to be allowed to be the same we have to say they are and build it into the semantics

14:53:14 <jang-scribe> and that WILL get rather complicated

14:53:27 <jang-scribe> we'll need disjoint lexical spaces of xml strings with and without markup

14:53:36 <jang-scribe> because the datatype relationships will be different

14:54:03 <jang-scribe> pats: in xsd you have integer and nit

14:54:06 <jang-scribe> int even

14:54:20 <jang-scribe> path: in xsd, the valuespaces there are pairwise disjoint, they designed it that way

14:55:00 <jang-scribe> frankm: people are going to say that in practical terms, there IS an overlap between value spaces even if the xsd specs don't say that

14:55:16 <jang-scribe> path: the spec says that they are NOT the same

14:55:22 <jang-scribe> you can, of course, map them onto the same thing

14:55:47 <jang-scribe> frankm: but a processor can decide 1 and 1.0 ARE the same thing

14:56:04 <jang-scribe> that's the way we should go: don't say anything about mappings are currently undefined

14:56:25 <jang-scribe> say formally as the datatype specs are defined, these are disjoint, applications are free to do what they want

14:56:38 <jang-scribe> bwm: I'd like to ask jjc to write up a proposal and seek feedback from martin

14:56:48 * danbri zakim, q?

14:56:48 * Zakim sees DaveB, danbri on the speaker queue

14:56:49 <jang-scribe> apts: and seek feedback from the WG before martin?

14:57:06 <jang-scribe> bwm: ok, but I'd rather not wait for WG agreement before we approach martin

14:57:18 <jang-scribe> path: do it in parallel, ie, enrol him in the WG on nthis issue

14:57:22 <jang-scribe> ack daveb

14:57:23 <Zakim> DaveB, you wanted to say we can't just pick one of (infoset, xpath nodeset (which version), DOM, ...)

14:57:24 * Zakim sees danbri on the speaker queue

14:57:44 <jang-scribe> the xml community has no clear single definition of what "xml" means

14:57:54 <jang-scribe> and we should not in particular depend normatively on any one of those

14:57:55 <bwm> ack daveb

14:57:56 * Zakim sees danbri on the speaker queue

14:57:59 <jang-scribe> ack danbri

14:57:59 <Zakim> danbri, you wanted to ask re 1:1 underspecified? (what's the precise term, injective, surjective etc?)

14:58:01 * Zakim sees no one on the speaker queue

14:58:02 <bwm> ack danbri

14:58:02 * Zakim sees no one on the speaker queue

14:58:25 <jang-scribe> jjc: the text in path's doc was adequately precise

14:58:40 <jang-scribe> ACTION jjc to put a proposal together on XMLLiterla for WG and martin

14:59:38 * jang-scribe notes that danc would like to see test cases for this

14:59:47 <jang-scribe> item 13:

14:59:50 <jang-scribe> implementation report

15:00:06 <jang-scribe> test cases we have don't provide good coverage of overall semantics

15:00:18 <jang-scribe> that is, they illustrate corner cases we've argued about rather than general coverage

15:00:40 <jang-scribe> so is having implementations that implement test acses a guarantee that we have interoperable implementations?

15:00:48 <jang-scribe> JosD: owl test cases use a lot of rdfs

15:00:52 <jang-scribe> we might be able to point there

15:01:02 <jang-scribe> I don't think we should extend our test cases now

15:01:14 <jang-scribe> jjc: I was reviewing the QA test groups guidelines

15:01:30 <jang-scribe> their suggestions of systematic analysis [was not good]

15:01:57 <jang-scribe> DaveB: I poked the sesame people into producing these emails

15:02:10 <jang-scribe> you'll see similar comments about datatypes, etc.

15:02:31 <jang-scribe> bwm: we might then need a cr period to build up stronger test cases

15:02:35 <jang-scribe> AOB: xmlsch-02

15:02:42 <jang-scribe> [danbri: I'm pro-CR by the way]

15:03:14 <jang-scribe> bwm: we're getting implementation reports that we may need stronger interop tests

15:03:22 <jang-scribe> so this is a thought at the back of my mind at the mometn

15:03:35 <jang-scribe> jjc: [xmlsch-02]

15:03:52 <jang-scribe> it seems that in general implementors are doing other than what we've written

15:04:02 <jang-scribe> it behooves us to change what we've written

15:04:16 <jang-scribe> if implementors think that what we've done is wrong, we ought to reexamine it

15:04:29 <jang-scribe> however, xmlschema say that the lexical space, value space, l2v is what it is

15:04:47 <jang-scribe> it would seem strange for us to say, sod xmlschema, l2v INCLUDES ws processing

15:04:59 <jang-scribe> it's a possible decision, it's very ugly

15:05:07 <DaveB> as jjc says in 2003Aug/0013

15:05:15 <jang-scribe> pats: one other option is that the definition of dts is unbounded.

15:05:24 <jang-scribe> xsd savvy apps are going to have to be away of, eg, facets

15:05:43 <jang-scribe> so apps using xsd + ws processing are applying xsd facets as part of their processing

15:06:00 <jang-scribe> jjc: we need to have our cake and eat it

15:06:33 <jang-scribe> I'm erring towards either an implementation note, OR making something normative that says implementations MAY do ws processing but the formal model does not support it

15:06:50 <jang-scribe> JosD: what if we described these test cases in rdf/xm rather than ntriple?

15:06:58 <jang-scribe> jjc: canonicalisation wouldn't work on that

15:07:06 <jang-scribe> jos ok, that wouldn't help then

15:07:19 <jang-scribe> it would go if we used post-schema-validated infoset

15:07:38 <jang-scribe> bwm: my instinct is the same as jjc: find a set of words that give implentations

15:07:47 <jang-scribe> room to do their thing

15:08:12 <jang-scribe> that is, room to do thier ws processing, and withdraw those test cases

15:08:29 <jang-scribe> jjc: I'm hearing bwm say, keep the MAY and dro pthe SHOULD produce warning messgae

15:08:35 <jang-scribe> bwm: let's take it to email

15:08:41 <jang-scribe> DaveB: ask implementors?

15:09:17 <jang-scribe> pats: we have to have somewhere an explicit rec that implementors try to use valid lexical forms, rather than rely on ws processsing

15:09:21 <jang-scribe> bwm: moving on

15:09:26 <jang-scribe> primer: frank...

15:09:34 <jang-scribe> has responded to outstanding actions

15:10:07 <jang-scribe> in the case of danc-01 I need clarification...

15:10:44 <jang-scribe> frankm: I believe both of these to involve the relationship between xml literals and language tags

15:10:50 <jang-scribe> DaveB: danc-01 is graph equality

15:11:16 <jang-scribe> DaveB: danc-01 is to change "graph equality" to "graph equivalence"

15:11:22 <jang-scribe> frankm: in that case, this is DONE

15:12:03 <jang-scribe> now, em isn't here but that action is out of date

15:12:14 <jang-scribe> typed literals can no longer have a language tag

15:12:32 <jang-scribe> frankm: current state of primer is that it doens't talk about any literals having language tags

15:12:59 <jang-scribe> frankm: the problem is that people keep on bringing up xml literlas and language tags

15:13:18 <jang-scribe> to cut it short, what I'd like folks to do, is to look at the current editor's draft of the primer, section 4.5

15:13:23 <jang-scribe> about which I sent a message on monday

15:13:48 <jang-scribe> in which I no longer have the triple - that is, it's a mistake to show people what these XMLliteral values actually look like

15:14:24 <jang-scribe> ACTION bwm to review section 4.5 of current ed's draft primer

15:14:55 <DaveB> i.e. http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-primer-20030117/Overview.html#xmlliterals

15:14:56 <jang-scribe> also, frankm: I believe pfps-15 is done

15:15:04 <jang-scribe> bwm: please ping him to verify that

15:15:39 <jang-scribe> frankm: expect to be done by the start of next week

15:15:44 <jang-scribe> concepts:

15:16:01 <jang-scribe> reviews outstanding

15:16:14 <jang-scribe> concepts to figure 5 in the primer, should be figure 6

15:16:26 <jang-scribe> frankm: te figure itself should be changed

15:16:31 <danbri> q+ to confirm meeting extended til half-past?

15:16:31 * Zakim sees danbri on the speaker queue

15:16:44 <danbri> (or did I blank on that and it happend?)

15:17:13 <jang-scribe> syntax:

15:17:28 <danbri> (all: yes we'd love to extend meeting.)

15:17:35 <jang-scribe> jang-scribe: that action's done (by dave as it happens)

15:17:53 <jang-scribe> new stuff: including literal text, I've showed to the group

15:18:12 <jang-scribe> the lexical form definition is difficult, pfps points out we should have NFC in there somewhere. that's where we are

15:19:18 <jang-scribe> DaveB: was going to add a line "this unicode form must be in NFC"

15:19:32 <jang-scribe> jjc: I was hoping to catch up with pfps' commetn on nfc today.

15:19:48 <jang-scribe> DaveB: if you can catch up with my proposal from monday this week

15:19:55 <jang-scribe> nfc needs to be added to that

15:20:04 <jang-scribe> lexical form for xmlliteral IS unicode characters

15:20:34 <jang-scribe> bwm: action arising? none

15:20:36 <jang-scribe> schema:

15:20:48 <jang-scribe> schema has two LC comments with no WG response

15:20:56 <jang-scribe> danbri: the list thing is pretty much wrapped up

15:21:01 <jang-scribe> the main sticking point is...

15:21:06 <jang-scribe> pfps-11

15:21:17 <jang-scribe> peter doesn't like the spec saying anything that can't be expressed in logic

15:21:29 <jang-scribe> his proposed text guts the content of sectiuon 5.3.1

15:21:43 <jang-scribe> pfps just wants us to say "statement is an instance of class, give range, give domain"

15:21:50 <jang-scribe> not say what it's for or what it describes

15:21:58 <jang-scribe> my cowardly way foprward is to just say "whatever"

15:22:24 <jang-scribe> path: pfps can't have any rational objection to wording saying "this is a class. the intention is to describe..."

15:22:41 <jang-scribe> ie, wording that clearly indicates that while this isn't necessarily logically true, that's what it's for.

15:23:25 <jang-scribe> bwm: text in primer about reification, that's non-normative, argues in favour

15:23:36 <jang-scribe> of danbri just doing the cutting

15:23:53 <jang-scribe> path: the semantics doc gives a fairly long discussion (normative) of the indended use of reification vocab

15:25:17 <danbri>http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Aug/0014.html

15:25:35 <bwm> q?

15:25:35 * Zakim sees danbri on the speaker queue

15:28:25 * jang-scribe is flagging, sorry

15:28:37 <bwm> ack danbri

15:28:37 <Zakim> danbri, you wanted to confirm meeting extended til half-past?

15:28:38 * Zakim sees no one on the speaker queue

15:28:39 <jang-scribe> [discussion of reification vocab]

15:29:10 <bwm> propose accept peters text and cross ref primer

15:29:20 <bwm> proposer: danbri

15:29:27 <jang-scribe> jjc, path abstain

15:29:30 <jang-scribe> pats abst

15:29:38 <jang-scribe> jjc wants to talk with colleagues about this

15:29:57 <bwm> action pat propose alternative text

15:30:12 <jang-scribe> ACTION path to draft a sentence for alternative text in schema

15:30:27 <jang-scribe> semanitcs:

15:30:47 <jang-scribe> bwm: you have 5 objectios now. one is to do with xml literal value spae

15:30:56 <jang-scribe> pfps -04 and -50: clsure rules

15:31:01 <jang-scribe> is that going to terminate?

15:31:14 <jang-scribe> pfps is now objecting to the style. I'm uninclined to persue this

15:32:07 <jang-scribe> path: if you have an inconsistency, the closure rules are not and have never been strongly complete

15:32:22 <jang-scribe> it'd be crazy to try to cover that

15:33:36 <jang-scribe> jang-scribe: will address on monday

15:33:38 <jang-scribe> meeting closes

15:33:39 <Zakim> -JosD

15:33:41 * jang-scribe melts

15:33:42 <Zakim> -pats

15:33:46 <Zakim> -Pat_Hayes

15:33:47 <Zakim> -Mike_Dean

15:33:53 <Zakim> -FrankM

15:34:24 <Zakim> -hp

15:34:26 <Zakim> -ilrt

15:34:29 <Zakim> -jjc

15:34:33 <Zakim> -DanBri

15:34:34 <Zakim> SW_RDFCore()10:00AM has ended