W3C RDF Core Working Group IRC Chat Logs for 2001-08-31

13:00:04 Users on #rdfcore: logger_1 Jema AaronSw jangII bwm

13:00:06 <Jema> -aob ... // add an item of another business

13:00:08 <Jema> -action owner/... // record an action item

13:00:10 <Jema> -decision ... // record a decision

13:02:19 <AaronSw> well, so much for Jema

13:02:24 <AaronSw> ok, gotta run

13:55:23 <spetschu> gday

13:57:47 <Jema> Hi. I'm Jema, the Jena meeting assistant

14:01:33 <jangII> can you get to the phone Art?

14:01:43 <jangII> can you get to the phone Art?

14:01:47 <em> +ericm

14:02:16 <spetschu> +SteveP

14:02:23 <bwm> -open

14:02:24 * jangII tests... can you hear me?

14:02:26 <Jema> RDFCore WG Telecon 2001-08-31 is now open

14:02:26 <Jema> The agenda can be found at http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0214.html

14:02:26 <Jema> Agenda item 1: Allocate scribe

14:02:35 <bwm> -net

14:02:37 <jangII> jang to scribe

14:02:40 <bwm> -next

14:02:40 <Jema> Agenda item 2: Roll Call

14:02:41 <jangII> jangII is now known as jang-scribe

14:03:00 <jang-scribe> reg: danbri

14:03:03 <jang-scribe> eric here

14:03:10 <jang-scribe> art? absent

14:03:12 <jang-scribe> dave b here

14:03:17 <jang-scribe> (art coming once phone working)

14:03:23 <jang-scribe> frank b no

14:03:28 <jang-scribe> jeremy reg

14:03:32 <jang-scribe> danc reg

14:03:33 <ArtB> +ArtB

14:03:39 <jang-scribe> art here

14:03:42 <jang-scribe> ron d here

14:03:47 <jang-scribe> bill d abs

14:03:48 <jang-scribe> jos here

14:03:53 <jang-scribe> rael abs

14:03:55 <jang-scribe> jan here

14:03:56 * DanC chooses between 2 telcons...

14:03:57 <jang-scribe> martin here

14:04:03 <jang-scribe> yoshi abs

14:04:05 <jang-scribe> graham here

14:04:11 <jang-scribe> mich kop abs

14:04:13 <jang-scribe> kwon abs

14:04:17 <jang-scribe> ora abs

14:04:20 <jang-scribe> frank m here

14:04:25 <jang-scribe> satoshi abs

14:04:28 <jang-scribe> steve p here

14:04:32 <jang-scribe> pierre r abs

14:04:35 <jang-scribe> aaron here

14:04:36 <bwm> -next

14:04:36 <Jema> Agenda item 3: Review Agenda

14:04:36 <Jema> See: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0214.html

14:04:44 <jang-scribe> bill d here (just arrived)

14:04:58 <jang-scribe> mike dean regrets, guha abs, pat here sergei reg

14:05:12 <jang-scribe> no ao

14:05:12 * em wants to learn more about jema ... encourages bmw to discuss this in #rdfig

14:05:15 <jang-scribe> no aob, even

14:05:21 <bwm> -nexty

14:05:22 <bwm> -next

14:05:22 <Jema> Agenda item 4: Review Minutes of previous meeting

14:05:22 <Jema> See: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0149.html

14:05:22 <Jema> See: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0153.html

14:05:22 <Jema> See: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0162.html

14:05:23 <Jema> Agenda item 5: Review Completed Actions

14:05:39 <jang-scribe> minutes approved

14:05:39 <bwm> -next

14:05:40 <Jema> Agenda item 6: Review Open Actions

14:06:23 <DanC> +DanC

14:06:29 <jang-scribe> you in the telecon danc?

14:06:33 <DanC> yes

14:06:36 <jang-scribe> danc here

14:06:47 <jang-scribe> note : xml use cases (bill) closed

14:07:01 <jang-scribe> no objections to marking as closed (agenda item 5)

14:07:20 <DanC> one word: "please continue" or "done".

14:07:24 <jang-scribe> danbri cont

14:07:35 <jang-scribe> bill d xml use cases withdrawn

14:07:39 <jang-scribe> eric: cont

14:07:42 <jang-scribe> dave b: cont

14:07:46 <jang-scribe> jan g cont

14:08:05 <jang-scribe> brian: cont

14:08:09 <jang-scribe> eric newsletter cont

14:08:11 <jang-scribe> martin: cont

14:08:14 <jang-scribe> (going well)

14:08:22 <jang-scribe> bri: continued (plenary)

14:08:32 <jang-scribe> bri (groups interested in meeting?) cont

14:08:41 <jang-scribe> dave b: reifying done

14:08:54 <jang-scribe> -next

14:08:54 <Jema> Agenda item 7: REVIEW SYNTAX WORKING DRAFT V1.23

14:08:54 <Jema> See: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0165.html

14:08:56 <bwm> -next

14:08:56 <Jema> Agenda item 8: ISSUE rdfs-no-cycles-in-subClassOf

14:08:56 <Jema> See: http://www.w3.org/2000/03/rdf-tracking/#rdfs-no-cycles-in-subClassOf

14:08:56 <Jema> See: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0172.html

14:08:56 <Jema> See: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0192.html

14:09:05 <jang-scribe> agenda item 7

14:09:06 <DanC> jema, help

14:09:18 <jang-scribe> dave b refers to list summary

14:09:25 * DanC can't hear well

14:09:31 <bwm>http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0252.html

14:09:46 <jang-scribe> comments: typos all accepted

14:10:00 <AaronSw> -help

14:10:01 <Jema> Jema recognises the following commands:

14:10:01 <Jema> -hi // tests whether Jema is alive

14:10:01 <Jema> -help // prints this help message

14:10:01 <Jema> -open // begin the meeting

14:10:01 <Jema> -close // close the meeting

14:10:02 <Jema> -next // next agenda item

14:10:04 <Jema> -aob ... // add an item of another business

14:10:06 <Jema> -action owner/... // record an action item

14:10:08 <Jema> -decision ... // record a decision

14:10:13 <jang-scribe> aaron's rdf:RDF - aaron should propose a solution here

14:10:34 <jang-scribe> this depends on context

14:10:44 <jang-scribe> art: container removal - to ask for feedback

14:11:05 <jang-scribe> danc, brian: <ins> and <del> stuff - need advice

14:11:16 <jang-scribe> AP: dave to correct ID stuff

14:11:28 <DanC> re ins/del: the spec meets the guidelines in letter. It's not clear in my browser, meanwhile.

14:11:44 <DanC> re title: good point.

14:11:45 <AaronSw> what browser?

14:11:46 <jang-scribe> document title: to change, "refactoring RDF/XML syntax"

14:11:52 <DanC> browser: NS4 on linux

14:12:10 <jang-scribe> the document as a whole: dave B still unsure as to quality, but open to comments

14:12:16 <jang-scribe> views?

14:12:30 <jang-scribe> dan c - ship after title change

14:12:36 <jang-scribe> bill d - agree

14:12:42 <jang-scribe> no show-stoppers (modulo title)

14:13:09 <jang-scribe> bwm: the strikeout stuff doesn't make sense in duff browsers

14:13:16 <jang-scribe> need a warning at the least

14:13:43 <jang-scribe> AP: dave to ship as is (modulo changes discussed above)

14:14:10 <jang-scribe> bwm: concerned about getting an "is that all?" reaction to this first working draft

14:14:25 <jang-scribe> alternative is to do a bit more work before making a splash

14:14:28 <AaronSw> strikeout doesn't work in NS4/Win either

14:15:00 <jang-scribe> people seem happy with this

14:15:09 <jang-scribe> (the do we ship? issue)

14:15:28 <jang-scribe> ron: no objection to this going out

14:15:39 <jang-scribe> bwm: proposes that this is shipped with the corrections:

14:15:41 <jang-scribe> changing the title

14:15:46 <jang-scribe> issuing the warning re: strikeout

14:15:51 <DanC> is the group deciding on a title or delegating title choice to the editor?

14:15:58 <jang-scribe> (plus dave's corrections outlined in the email)

14:16:08 <jang-scribe> we trust editorial decisions here

14:16:11 <jang-scribe> no objections

14:16:14 <AaronSw> this being shipping, not doing more work, no?

14:16:41 <jang-scribe> correct aaron, we ship this was the idea. you still ok with this?

14:16:47 <jang-scribe> dave asks for help with publication

14:16:52 <jang-scribe> eric offers help

14:17:03 <jang-scribe> eric: how ready are you to do this?

14:17:11 <jang-scribe> dave b: ok for next week?

14:17:14 <bwm> -decision ship syntax WD with corrections to title, strike out warning and editorial discretion to daveB

14:17:14 <Jema> Jema notes decision number 1 ship syntax WD with corrections to title, strike out warning and editorial discretion to daveB

14:17:47 <jang-scribe> danc: if eric's not here next week, we need another staff contact

14:17:57 <jang-scribe> bwm: can danc stand in?

14:18:05 <jang-scribe> danc: can this get done today?

14:18:23 <jang-scribe> daveb: probably not possible. one working day to do this

14:18:44 <jang-scribe> danc: not available for this.

14:18:50 <jang-scribe> eric welcomes art back into the fold :-)

14:19:33 <jang-scribe> art agrees to do this (act as staff contact)

14:19:46 <bwm> -action art to get syntax doc thru w3c publishing process

14:19:46 <Jema> Jema could not parse the action command. Please separate the action holder from the action by a '/' character

14:19:57 <bwm> -next

14:19:58 <Jema> Agenda item 9: Review MODEL Theory WD

14:19:58 <Jema> See: http://www.w3.org/2001/08/rdf-mt/rdf-mt

14:20:24 <bwm> -action art/ to get syntax doc thru w3c publishing process

14:20:24 <Jema> Jema notes action number 1 ar to get syntax doc thru w3c publishing process

14:20:30 <jang-scribe> we note that monday is labo(u)r day

14:21:23 * jang-scribe notes that jema needs a "-prev"

14:21:32 <jang-scribe> action item 8

14:21:32 * DanC can barely hear Pat

14:21:55 <jang-scribe> pat's response from daml was "do this right"

14:22:18 * gk not my mouse squeaking! my phone's currently muted

14:22:25 <DanC> my position on subclass loops: A subClass B and B subClass B is allowed; it means A's and B's class extensions are the same, but it does *not* necessarily mean that A=B.

14:22:34 <jang-scribe> pat: political and technical opinion of DAML is that subclass loops is the only rational way

14:22:44 <jang-scribe> they strongly recommend we consider this point of view

14:23:10 <jang-scribe> pat: subClassOf does NOT mean "proper subclass of"

14:23:44 <jang-scribe> this would be a change to schema

14:23:50 <jang-scribe> danc seconds this

14:24:25 <jang-scribe> ron d: is DAML considering class generation (in the OO sense) here?

14:24:37 <jang-scribe> pat: that's the alternative I proposed (getting a different subClassOf for DAML)

14:24:50 <jang-scribe> response: we _could_ do that, but that cuts all ties between rdfs and daml

14:25:00 <jang-scribe> therefore destroys the "layering" everyone lauds

14:25:33 <jang-scribe> also: techie feeling was generally that this is a neat way to express class identity

14:25:41 <jang-scribe> it's desirable to _detect_ subclass loops

14:25:56 * DanC wonders where the static is coming from

14:25:58 <jang-scribe> so the code to do this is probably going to be written

14:26:19 <jang-scribe> in DAML+OIL subclass loops is not parse-time checkable

14:26:35 <jang-scribe> danc: subclass loops mean "coextensional" not "identical"

14:26:49 <jang-scribe> pat agrees with this interpretation

14:27:19 <jang-scribe> bwm: choice: go with what DAML+OIL needs (to permit building on top of rdfs)

14:27:32 <jang-scribe> ph: daml:class still required (datatyping stuff)

14:28:20 <jang-scribe> bwm: other choice is to maek rdfs:subClassOf a subproperty of daml:subClassOf

14:28:35 <jang-scribe> ph: it would be worse than that. In practice, people just wouldn't use rdfs:sobClassOf

14:28:57 <jang-scribe> bwm: and we lose the ability to turn these schemas into OO hierarchies directly?

14:29:12 <AaronSw2> Note that not all OO languages prohibit loops

14:29:25 <jang-scribe> ph, GK: the ability to map into OO language hierarchies is not an option

14:29:34 * gk I'm not sure that direct translation to OO hierarchies works either

14:30:06 <jang-scribe> danc: OO inheritance tends to be a bit rubbish anyway (lots of people get it wrong for the sake of implementation inheritance)

14:30:15 <DanC> not rubbish; just not "subset"

14:30:50 <DanC> i.e. java:extends is just a different thing from rdfs:subClassOf altogether

14:30:57 <jang-scribe> frank: in OO you create an object as an instance of a particular class (or classes); on the web, people may determine after the fact that...

14:31:02 <jang-scribe> it's a member of other classes

14:31:47 <jang-scribe> ron: the OO translation is useful

14:32:05 <jang-scribe> PH: but people can carry on doing it; if they hit a loop, they just get an exception (or compile-time error)

14:32:22 <jang-scribe> frank: the question is: what capabilities do we want our classes to have?

14:32:42 <jang-scribe> danc: use cases should dominate the design

14:33:56 <jang-scribe> danc: reiterates and expands on differences between OO and this proposed points of view

14:33:57 * gk I can't fully agree with DanC's assessment here -- having more features/values doesn't mean derived class is not in some senses representing a subclass

14:34:09 * jang-scribe agrees with GK here

14:34:54 <jang-scribe> gk: a derived class in an OO sense _can_ represent a subset of a base class

14:35:21 <jang-scribe> dave b: DAML has implementation experience that this _can_ be implemented

14:35:30 <jang-scribe> thus I'm less concerned that making this change will be a problem

14:35:50 <jang-scribe> ron: has sergei commented on this?

14:36:06 <jang-scribe> (historical note: his objection was based on java implementations at the f2f)

14:36:14 <jang-scribe> bwm: not that I'm aware of

14:36:21 <jang-scribe> ...but this is probably because he's been away

14:36:44 <jang-scribe> dave: the point is, you have to be doing checking as you generate OO classes, and DAML applications are doing the same thing

14:37:19 <jang-scribe> danc: having two pieces of rdfs vocab might be the right resolution here

14:37:50 <jang-scribe> danc: eg, we might want a subclass of rdfs:Class called "set" where identity depends on extensionality

14:37:59 <jang-scribe> danc: in other words, have both sorts of subclass

14:38:17 <jang-scribe> so a subclass loop is legal but entails identity in teh Set case? (PH)

14:38:34 <jang-scribe> ph: these two ideas are somewhat orthogonal

14:38:56 <jang-scribe> danc: I'd like to have the option. I think the best thing is to have two terms in the rdfs vocab

14:39:00 <jang-scribe> one a subproperty of the other

14:39:11 <jang-scribe> PH: the difference is precisely the identity case

14:39:18 <jang-scribe> <= means that you can define =

14:39:33 <jang-scribe> allowing the loop case gives expressibiliity

14:39:57 <jang-scribe> also, i must emphasise this, is that DAML feedback is that saying that two classes have the same extension is VERY useful

14:40:09 <jang-scribe> jos: what about the same issue with subproperty of?

14:40:23 <jang-scribe> ph: not asked explicitly, but I think this is taken for granted (the identity case)

14:41:13 <jang-scribe> jang: does DAML distinguish coextensionality and identity?

14:41:17 <jang-scribe> ph: not currently

14:41:33 <jang-scribe> they'd probalby not have any objection to update their MT to agree with ours (I think)

14:41:52 <jang-scribe> bwm: then it seems there is a proposal:

14:42:08 <jang-scribe> to change the def of rdfs:subClassOf to allow loops and to investigate in the future refinements of this

14:42:22 <jang-scribe> that include the oo subclass restrictions

14:42:39 <jang-scribe> ron: the other alternative is to introduce a new this that is a superproperty of subClassOf

14:42:52 <jang-scribe> the installed base is trivial, though, so now is the time to break it

14:43:11 <jang-scribe> danc: there seems to be more installed base that treats rdfs:sco in the DAML sense

14:43:13 * gk OK for me; I think implementations can still use tricks that assume < rather than <= IF they know the classes they define have this property

14:43:27 <jang-scribe> bwm: the proposal is as I suggested then:

14:43:35 <jang-scribe> we change rdfs:sco to allow cycles

14:43:47 <jang-scribe> and investigate future specifications of this

14:43:54 <jang-scribe> ron: is this final and binding?

14:43:56 <jang-scribe> bwm: yes

14:44:00 <jang-scribe> ph: all decisions are reversible

14:44:15 <jang-scribe> ph: but if we don't make this it'll give rise to a huge debate

14:44:34 <jang-scribe> ron: i think sergei's point was a good one at the f2f and I'd like to hear more from him before I agree with this

14:44:45 <jang-scribe> danc: I'm not in a big hurry

14:44:57 <jang-scribe> eg, making a provisional decision pending expert input from sergei

14:45:17 <jang-scribe> dave b: I'm happy but would like to hear from sergei

14:45:40 <jang-scribe> frank m: happy to wait for sergei, but I repeat that the issue is that

14:45:58 <jang-scribe> mapping into programming languages shouldn't be a show-stopper; that's not what we're defining our classes for

14:46:09 <jang-scribe> so I'm happy to wait but I don't think my opinion will change

14:46:12 <jang-scribe> PH: seconds that

14:46:25 <jang-scribe> programming languages may change

14:47:04 <DanC> do we have test cases? I think it's *VERY* valuable to have test cases in hand when making a decision.

14:47:05 <jang-scribe> bwm: our position is that we're considering this but we're still looking for feedback?

14:47:08 <jang-scribe> ron: yes

14:47:21 <jang-scribe> more specifically, we know a member of the WG has a stance on this and we'd like to hear from him

14:48:03 <jang-scribe> frank: even if we delay, we want a concrete mechanism to resolve the delay

14:48:15 <jang-scribe> so we need to specify exactly what we're waiting for

14:48:37 <jang-scribe> ph: the nature of the "urgency" - DAML are on the point of publishing and setting in stone what they're doing

14:48:51 <jang-scribe> it's clear that they are presuming that this change is going to be made

14:49:10 <jang-scribe> otherwise they'll have to go back and %s/rdfs/daml/g their publications

14:49:25 <jang-scribe> danc: this is important, but not urgent

14:49:31 * spetschu notes that coextentional java classes won't even compile... forget exceptions

14:49:32 <jang-scribe> ph: we must give DAML the right signals

14:49:37 <jang-scribe> bwm: daml's deadlines?

14:49:49 <jang-scribe> ph: deadline, no; jim hendler holding the whip

14:50:40 <jang-scribe> the question here is of the urgency.

14:50:51 <jang-scribe> danc: i don't have any personal sense of urgency; DAML does

14:51:14 <jang-scribe> ph: we don't risk "damage", but delay will mean that the debate will spill out into public

14:51:22 <jang-scribe> and deluge (eg) rdf-logic

14:51:27 * gk doesn't this all rather depend on the length of any delay???

14:51:53 <jang-scribe> and such a debate illustrating (or interpreted as illustrating) rancor is unfortunate

14:52:16 <jang-scribe> scribe notes: that's a poor transcription: in summary, we want to get the politics right

14:52:45 <jang-scribe> sergei's back soon (6 days) but may take some time to get back in the swing

14:53:20 <jang-scribe> em: generating java classes from rdfs has come up a lot on rdfig

14:54:17 <jang-scribe> rdfs:sco and java "extends" are not interchangable.

14:54:48 <jang-scribe> ph: this isn't going to hurt people much.

14:55:10 <jang-scribe> who bears the responsibility for the rare cases where something does break?

14:55:21 <jang-scribe> is it rdfs' fault or a java translator author problem?

14:55:34 <gk> I can think of ways of using Java classes to implement rdfs classes with loops. Maybe not as neat as using Java extends for some cases.

14:55:41 <jang-scribe> steve: the danger I think is that people are going to think, "this is strange, let's not use it at all"

14:55:46 <jang-scribe> em: agrees with this

14:56:02 <jang-scribe> frank m: that's part of the job

14:56:24 <jang-scribe> aaron: anyone dealing with rdfs "in the wild" is going to run into this and is going to have to deal with this

14:56:29 <jang-scribe> ie, fix their code

14:56:41 <jang-scribe> ron: I just want to hear more from the person (sergei) who raised the original objections

14:56:50 <jang-scribe> ph: we don't want to railroad this

14:57:00 <jang-scribe> bwm: (as chair)

14:57:05 <jang-scribe> sergei is back in roughly 6 days

14:57:25 <jang-scribe> ap must go on someone to chase him to get him to contribute the information/opinion ewe need

14:57:39 <jang-scribe> ron is motivated in this...

14:58:32 <jang-scribe> -action frank m / to prompt sergei to feedback to us on this

14:58:33 <Jema> Jema notes action number 2 frank m to prompt sergei to feedback to us on this

14:59:02 <jang-scribe> next meeting: same time, next week

14:59:08 <DanC> does jema note the correspondence between action items and agenda items in which they occur? if not, please consider this an RFE

14:59:08 <jang-scribe> we're unforuntately out of time

14:59:16 <jang-scribe> bwm: incluined to take MT next week

14:59:32 <DanC> I hope we can we decide next ween to publish the other 2 docs.

14:59:33 <jang-scribe> PH notes thanks to aaron + danc for editorial work on mt

15:00:24 * gk cut off, just on cue?

15:00:26 <bwm> -close

15:00:27 <Jema> The meeting is closed

15:00:30 <AaronSw2> bye bye all

15:00:55 <gk> Bye folks

15:02:15 <bwm> -close

15:02:16 <Jema> The meeting is closed

15:07:57 <bwm> -close

15:07:57 <Jema> The meeting is closed

15:08:28 <bwm> -close

15:08:29 <Jema> The meeting is closed