W3C RDF Core Working Group IRC Chat Logs for 2001-04-27

13:49:20 <mhorner> Good afternoon!

13:50:50 <bwm-gone> hi martin

13:50:58 <bwm-gone> bwm-gone is now known as bwm

13:51:16 <bwm> bwm is now known as bwm-db-jan

13:51:17 <mhorner> Ah good. Someone there...

13:51:37 <bwm-db-jan> dajobe (dave beckett) says hi

13:52:03 <bwm-db-jan> dave beckett, jan grant and bwm are sharing a phone

13:52:30 <mhorner> Martyn Horner is trying to find a phone that works..

13:57:28 * em waves to all

14:02:29 <danbri> hi. i'm in a business centre in the hotel in hong kong; will dial in via mobile shortly.

14:02:40 <AaronSw> hi everyone

14:02:48 <danbri> I expect (background noise) to mostly lurk tonight. But hi!

14:03:48 <mdean> mdean is now known as _mdean

14:05:21 <AaronSw> Aaron Swartz

14:05:22 <_mdean> Mike Dean

14:05:27 <em> Eric Miller

14:05:28 * danbri apologises for delay. joining you shortly...

14:05:35 <bwm-db-jan> brian mcbride, dave beckett, jan grant

14:05:36 <guha> R.V.Guha

14:05:37 <mhorner> Martyn Horner

14:05:54 <_mdean> Ora Lassila is on cell phone from Hong Kong, but not on IRC

14:07:11 <_mdean> Mike Dean will scribe to IRC, capture actions, etc.

14:07:12 <bwm-db-jan> logs at http://ilrt.org/discovery/chatlogs/rdfcore/

14:07:33 * danbri connects.

14:07:35 <_mdean> Frank XXX, HTML Writers Guild

14:07:39 <_mdean> Frank Manola

14:07:44 <_mdean> Pete ?, IBM

14:07:57 <_mdean> Michael ?, Briance

14:08:10 <AaronSw> That's Frank Boumphrey from the HTML Writer's Guild

14:08:13 <_mdean> sorry for typos in names -- will be fixed in minutes

14:08:18 <danbri> +Dan Brickley, W3C/RDF

14:08:28 <danbri> s/RDF/ILRT ;)

14:08:50 <em> Or Lasilla, Nokia and Michael Kopchenov, Brience

14:08:55 <em> s/Or/Ora

14:09:27 <_mdean> thanks

14:10:00 <em> Stephen Petschulat, IBM

14:10:26 <_mdean> Brian McBride chairing

14:10:28 <em> Frank Boumphrey, HWG

14:10:31 <_mdean> Brian reviewed the charter

14:10:48 <AaronSw> agenda: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Apr/0058.html

14:10:58 <_mdean> backward compibility with existing RDF is a priority

14:11:02 * danbri nods

14:11:08 * mhorner nodes

14:11:18 <_mdean> no issues raised with charter

14:11:29 <_mdean> Frank Manola: devil may be in the details

14:11:49 <_mdean> Brian suggested driving WG from the issues list + subsequent additions

14:12:24 <_mdean> ask members to "own" individual issues: analyse, write up suggested fix

14:12:46 <_mdean> consensus with this approach

14:13:08 <danbri> I like this approach...

14:13:11 <rael> Rael Dornfest

14:13:27 <_mdean> 3 issues with proposals to be discussed today

14:13:34 <_mdean> regrets from 4 people: Barstow, deRoo, Ianella, Graham Klyne

14:13:43 <bwm-db-jan> rael, can review logs at http://ilrt.org/discovery/chatlogs/rdfcore/

14:13:56 <_mdean> Dave: namespace prefix problem

14:14:45 <_mdean> confusion due to XML NS interpretation of unprefixed attributes (ID, about, etc.)

14:15:06 <AaronSw> inconsistency in implementation

14:15:41 <_mdean> 2 suggestions: don't allow unprefixed, allow but deprecate

14:15:59 <em> namespace prefix open issue: http://www.w3.org/2000/03/rdf-tracking/Overview.html#rdf-ns-prefix-confusion

14:15:59 <_mdean> suggestions -> alternative proposals

14:16:07 <_mdean> thanks

14:16:47 <_mdean> Martin Horner prefers deprecation

14:17:16 <_mdean> Ora asked for clarification: deprecate default namespace?

14:17:43 <frank> frank boumphrey pinging

14:17:44 <_mdean> clarification: just for attributes

14:17:55 <_mdean> danbri: because of XMLns specification

14:18:17 <_mdean> Brian: xmlns doesn't apply to attributes

14:19:09 <_mdean> Aaron: apply to both elements and attributes? per proposals

14:19:38 <_mdean> should change proposal to apply only to attributes

14:19:46 <_mdean> Ora: sounds like bug in NS

14:19:54 <danbri> re namespace myths: http://www.xml.com/pub/a/2000/03/08/namespaces/

14:20:14 <_mdean> Eric? we're caught in the middle -- we should note that this seems like a xmlns bug

14:21:00 <_mdean> Frank B: XHTML also addressed this issue -- told they just had to live with xmlns

14:21:08 <danbri> see 'myth 4: unprefixed attributes are in XML namespaces'

14:21:27 <_mdean> Brian: consensus on using rdf: namespace for attributes

14:21:38 <_mdean> remaining issue: forbid vs. deprecate

14:22:06 <_mdean> Frank Manola: any ambiguity re unqualified attributes?

14:22:24 <_mdean> Brian: unqualified are in "global" namespace

14:22:45 <_mdean> emerging consensus toward deprecation

14:23:20 <mhorner> Good one!

14:23:23 <_mdean> RESOLVED: strongly recommend namespace qualified attributes, allow but strongly deprecate unqualified attributes

14:23:25 <danbri> danbri: i prefer 'depracate', it's in the spirit of the charter

14:23:28 <_mdean> 42 issues to go :-)

14:23:48 <_mdean> next issue: containers and syntax

14:23:55 <_mdean> can someone enter the URI?

14:24:11 <AaronSw> discussing http://www-uk.hpl.hp.com/people/bwm/rdf/issues/containersyntax/

14:24:33 <_mdean> Bag/Seq/Alt: 2 issues:

14:24:37 <AaronSw> issue: http://www.w3.org/2000/03/rdf-tracking/Overview.html#rdf-containers-syntax-ambiguity

14:24:42 <AaronSw> http://www.w3.org/2000/03/rdf-tracking/Overview.html#rdf-containers-syntax-vs-schema

14:24:50 <_mdean> 1) typedNode matches 2 productions

14:25:05 <_mdean> typedNode somewhat richer than containers

14:25:22 <_mdean> anyone who participated in original working group want to comment?

14:25:30 <_mdean> 2) subclassing containers

14:26:01 <_mdean> hard for parser to recognize subclass (requires RDFSchema processing -- undesirable)

14:26:24 <_mdean> existing parsers don't do this -- unable to process subclasses of containers against container productions

14:26:45 <_mdean> Ora: original requirement: processing shouldn't require knowledge of schema

14:27:18 <_mdean> Guha: have to be able to produce graph from syntactic expression without retrieving schema

14:28:06 <_mdean> Ora: BNF isn't best way of expressing this type of syntax

14:28:18 <_mdean> Ora: whole container portion of M&S is somewhat broken

14:28:50 <_mdean> danbri: closed world assumption

14:28:58 <_mdean> lots of issues related to containers

14:29:13 <_mdean> danbri: to own container problem -- Ora to help

14:30:03 <em> _mdean: asks if this problem can be addressed by prohibiting subclassing of containers

14:30:13 <_mdean> Mike: can address this problem by prohibiting subclassing of container classes

14:30:36 <danbri> danbri: offered to help describe possible overlap between surface syntax container problems with deeper model-related issues. Will talk w/ Ora in H.K. and summarise

14:30:46 <_mdean> Ora: should parsers signal an error if li occurs outside a pre-defined container class

14:31:07 <_mdean> proposal addresses all of these issues

14:31:34 <_mdean> Ora: think of the model, not the syntax

14:31:43 <_mdean> danbri: this is a syntax bug we're addressing

14:31:50 <_mdean> Ora: we have to get the model right first

14:32:10 <_mdean> danbri: an issue is using the abbreviated syntax to hide RDF content in HTML documents

14:32:30 <AaronSw> (a historical reason)

14:32:33 <_mdean> danbri: we should record more of the history of these issues

14:32:46 <_mdean> ?: ordering of elements

14:33:03 <danbri> danbri: in doing so... run into XML's constraint that you can't have multiple XML attributes with the same name

14:33:11 <danbri> ...hence _1 _2 etc

14:33:48 <_mdean> Ora: radical move could be to deprecate container types

14:34:13 <_mdean> ?: remove containers and reification to get very simple core

14:34:37 * em raises his hand

14:34:38 <_mdean> same ?: would simplify lots of things

14:34:51 <_mdean> ? = Guha?

14:34:56 * AaronSw raises hand

14:35:04 <mhorner> What is the real-world use of containers?

14:35:04 <_mdean> Ora: some of these require syntactic support

14:35:35 <_mdean> Eric: likes Guha's suggestion about modularizing different constructs

14:36:04 <_mdean> Eric: but that's a separate issue, which should be deferred

14:36:23 * mhorner raises hand

14:36:25 <_mdean> Eric: issue at hand is containers -- how we modularize is open issue itself

14:36:58 <_mdean> danbri: 2 ways to modularize: documents, namespaces -- very different solutions -- suggests using same namespaces for backward compatibility

14:37:37 <_mdean> Eric: could also introduce a new construct to let people pick and choose which features they want

14:37:59 <_mdean> Aaron: need to be able to do numbered properties without understanding schema

14:38:30 <AaronSw> aaronsw: we could allow the _1, _2, _n on an arbitrary node

14:38:40 <_mdean> Martin Horner: have never found containers very useful -- what's the real world use

14:38:45 <_mdean> ?: ordering

14:39:19 <_mdean> Guha: don't necessarily need them -- can be built from graph -- existing constructs just provide syntactic support

14:39:54 <_mdean> Ora: changes to syntax should imply new namespace -- allows for non-backward compatible changes

14:40:13 <_mdean> Brian: charter strives not to make incompatible changes

14:40:33 <_mdean> danbri: new namespace is not necessarily required -- can clarify definitions of current namespace

14:40:45 <_mdean> Guha?: not a lot of production code using RDF 1.0

14:40:58 <_mdean> Guha?: now is the time to make any critical changes

14:41:21 <_mdean> danbri: doesn't require new namespace

14:41:52 <_mdean> Guha?: Netscape uses RDF, but doesn't use reification and hardly uses containers

14:42:18 <_mdean> Brian: don't seem to be achieving closure

14:42:30 <AaronSw> guha: we need to find who uses what

14:42:32 <danbri> danbri: mozilla uses containers (ie. rdf namespace, but doesn't stick to spec's semantics)

14:42:55 <danbri> I absolutely agree with Guha: we need a better understanding of what will break if we change stuff

14:43:10 <AaronSw> em: guha and else, please send the list pointers to rdf users

14:43:12 <_mdean> Eric?: ask everyone to publish what RDF features the applications they know are using

14:43:47 <_mdean> ACTION: Guha to collect real-world usage info

14:44:09 <_mdean> Brian: re-address this issue in a broader context

14:44:18 <_mdean> ACTION (dan bri and Ora): above?

14:44:34 <_mdean> new topic: F2F in June -- requires 8 weeks notice

14:44:40 <_mdean> looking for host

14:44:48 <_mdean> end of June

14:44:51 <AaronSw> above? = interaction between containers and other probllems

14:45:05 <_mdean> could overlap with IWWS at Stanford

14:45:19 <_mdean> suggest August 1 and 2

14:45:32 <mhorner> Vote for that

14:45:42 <_mdean> general consensus that this is a good idea

14:45:44 <frank> aug1 and 2 works for me

14:45:55 <_mdean> anyone willing to host in SF Bay area?

14:46:02 <rael> o'reilly in sebastopol!

14:46:07 <_mdean> how about Stanford?

14:46:17 <_mdean> expect 20-25 people

14:46:32 <_mdean> Ora: Nokia has office in Mountain View -- will check

14:46:47 <_mdean> HP?

14:47:13 <AaronSw> contact bwm and danbri if you're willing to host

14:47:15 <_mdean> em: contact brian and danbri if you're in that area and willing

14:47:32 <rael> O'Reilly has volunteered to host

14:47:39 <_mdean> consensus on August 1-2 as dates for F2F

14:47:57 <_mdean> last topic: list of "low-hanging fruit" issues

14:48:05 <_mdean> looking for volunteers to own these issues

14:48:35 <_mdean> Ora: volunteer for rdfms-editorial

14:48:59 <_mdean> DanBri? and Guha? to help

14:49:38 <_mdean> reification-requirements: is a parser required to emit reification constructs

14:50:14 <_mdean> Guha has same phone he had 3 years ago? very faint

14:50:26 <_mdean> Guha: it is a different phone

14:50:57 <mhorner> I'll volunteer for either of the last two (What's a resource? Glossary)

14:50:57 <_mdean> DanBri: some parsers don't output reification -- some of their authors feel guilt

14:51:21 <_mdean> Guha: issue of "conformant RDF parser" -- can make proposal

14:52:18 <_mdean> Guha to own reification-requirements, Ora and DanBri to help

14:52:21 * em raises hand

14:52:33 <danbri> Danbri: I am happy to help with this issue, but I don''t want to take a lead on it.

14:53:18 <_mdean> DanBri willing to lead lots of other issues :-)

14:53:43 <_mdean> Frank Manola: anything to do with reification is a potential rats' nest

14:54:14 <_mdean> Ora: requirement to reify everything can result in infinite expansion

14:54:31 <_mdean> DanBri: reification is relatively broken -- willing to elaborate at length later

14:54:46 <_mdean> Guha: willing to be coordinating point for reification issues

14:55:15 <_mdean> Brian: this specific issue is much smaller and simpler: e.g. just say a parser is not required

14:55:24 <_mdean> ... to support reification

14:55:50 <_mdean> Guha: if parsers not required, user's can't depend on feature

14:56:13 <_mdean> Brian: anyone willing to look at empty-property-elements

14:56:32 <_mdean> jos de roo: willing to take this on

14:56:48 <_mdean> how about not-id-and-resource?

14:57:15 <_mdean> Dave Beckett will take not-id-and-resource

14:57:38 <_mdean> defining model for resources -- foundation for upcoming work

14:57:50 <_mdean> Martin Horner to work on defining model for resources

14:58:15 <danbri> (I don't really understand what this issue amounts to yet)

14:58:17 <_mdean> Martin Horner volunteering for terminology as well

14:58:26 <_mdean> refer to Graeme Klyne's glossary

14:58:39 <_mdean> Brian: any other business?

14:58:56 <AaronSw> jos deroo takes empty emements

14:58:56 <AaronSw> dave beckett takes not id and resource

14:58:57 * DanC_lap waves from the peanut gallery

14:59:01 <_mdean> Eric: 10's of T shirts on the way :-)

14:59:11 <_mdean> inside jokes about vegemite :-)

14:59:13 <rael> vegemite?

14:59:22 <rael> marmite!

14:59:35 <_mdean> Brian: thanks everyone -- meeting adjourned

14:59:43 <AaronSw> meeting ajourned

15:00:01 <bwm-db-jan> next meeting probably in 2 weeks

15:01:10 <DanC_lap> "stronly deprecate"???? I wonder what that means. Does the test pass or not?

15:01:46 <danbri> Pass with a warning...?

15:02:00 * danbri goes looking for Ora

15:02:11 <danbri> bye all!

15:02:13 <em> take care all...

15:02:24 <em> \leave