Semantic Web Interest Group IRC Chat Logs for 2002-03-21

00:56:57 <thc> thc is now known as TomC

00:59:11 <TomC> TomC is now known as tomc

02:18:47 <bijan> Hmmm:

02:18:51 <bijan> """Any protocol built to REST can access and manipulate

02:18:51 <bijan> anything with identity, which of course includes all URI using the HTTP

02:18:54 <bijan> URI scheme."""

02:19:11 <bijan> (From: http://lists.w3.org/Archives/Public/www-tag/2002Mar/0087.html)

02:19:36 <danbri> for some sense of 'can'

02:19:46 <bijan> "Anything with identity" is one of those phrases that crops up in web discussions.

02:19:53 <bijan> Like "URIs can name anything"

02:19:59 <danbri> are there any other kinds of thing?

02:20:07 <danbri> etc... blurgh...

02:20:13 <bijan> No entity without identity :)

02:20:15 <danbri> you won't catch me saying either

02:20:24 <bijan> Good!

02:20:31 <bijan> I now forgive you everything else :)

02:20:41 <bijan> We need a Web Wittgenstein.

02:20:47 <bijan> Hmm. There's a party game...

02:20:56 <bijan> ...which only me and you would like to play :)

02:21:11 <bijan> Which web figure corresponds to what philosopher!

02:21:13 * danbri brandishes a burning poker threateningly

02:21:30 <bijan> for example, TBL could be Plato, which would make RoyF Aristotle

02:21:47 <danbri> Pat Hayes could be Derrida

02:21:47 * DanC_jam waves

02:21:49 * danbri ducks

02:21:53 <danbri> hi danc

02:21:58 <danbri> btw what does _jam mean?

02:21:59 <bijan> Hmm. More like frege, I'd say.

02:22:03 <bijan> Hey danc

02:22:14 <niq> Exercise: extend URI to support uncountable address spaces <gd&r>

02:22:14 <DanC_jam> I call my laptop jammer.

02:22:16 <bijan> DanC, hmm. How about Popper?

02:22:21 <bijan> no...

02:22:31 <danbri> yeah, I was recalling PatH's various "they've probably got a humanities degree" outbursts :)

02:22:45 * danbri would go for Ryle

02:23:08 <danbri> (from his writings; don't know much about him)

02:23:25 <bijan> The problem with picking ryle (like hume or quine) is that they have their philosophical style and then their immense literary grace.

02:23:38 <danbri> I can aspire, can't I?

02:23:45 <bijan> And *then* their personalities :)

02:23:47 * danbri sighs

02:24:16 <bijan> And *then* their relationships to schools of thought/other philosophers.

02:24:33 <bijan> So, TBL can be Quine if preweb hypertext folks are the Positivists.

02:24:37 <danbri> I can't find a copy of Ryle's 'knowing how / knowing that' piece in the Web anywhere. Mid 1940s I think, woudl that put it out of copyright?

02:24:49 * DanC_jam proposes a rule: no philosophizing except while waiting for the compiler. or cwm. ;-)

02:25:02 <bijan> But merely being out of copyright doesn't put things on the web.

02:25:11 <danbri> I would type it by hand, happily.

02:25:12 <bijan> Well then, danc, you must philosophize a *lot* :)

02:25:18 <danbri> into IRC while Cwm waits....

02:26:06 <bijan> Wow, the tag list is busy.

02:26:09 * DanC_jam is hanging out with the family; doing a little iMac research, trying to decide wither to spend our tax return on an iMac

02:26:14 <bijan> How do these folks get the time?

02:26:23 * bijan was just fondling an imac

02:26:31 <bijan> In the student 'puter store.

02:26:47 * danbri just sent http://rdfweb.org/people/danbri/2001/05/hk/batch10/ani.html to www-rdfcore-wg by accident because of MacOS X

02:27:10 <danbri> (I forgot cut'n'paste worked differently from X, where it barely works...)

02:27:12 <bijan> Ouch!

02:27:26 <danbri> there are worse slips one could make.

02:27:55 * danbri heads off for the night

02:27:58 <danbri> cu!

02:28:01 <bijan> Ta

02:29:10 * bijan is going to meet jhendler next week, next tues in fact.

02:29:17 <DanC_jam> no hacking tonight, danbri?

02:30:15 <bijan> danc, does cwm take a long time on the pathCross.n3 stuff?

02:30:19 <bijan> For some definition of "long"? ;)

02:30:58 <bijan> I really need to get a cwm setup. It's sorta scary that i find it easier to *clone* it than to run it.

02:32:34 <bijan> I just don't do so well with command line interfaces.

03:02:23 <lilo> [ServerNotice] Hi all. We're in the process of upgrading this server to the newest code, and it's been removed from the rotation. If you could disconnect and reconnect to irc.openprojects.net, it would help us along. Thanks.

03:03:59 <sandro_> sandro_ is now known as sandro

03:08:22 <lilo> [Global Notice] Hi all. There will be small netsplits for the next few minutes as we convert more servers to the new release software. We'll let you know if your server will be affected. Thanks.

03:33:53 <sbp>http://lists.w3.org/Archives/Public/www-rdf-interest/2002Mar/0128

03:33:53 <dc_rdfig> A: http://lists.w3.org/Archives/Public/www-rdf-interest/2002Mar/0128 from sbp

03:34:01 <sbp> A:|Mapping Notation3 to XML: An Alternate XML RDF Syntax

03:34:01 <dc_rdfig> titled item A

03:34:44 <sbp> A:I was writing an article about various ways of expressing data in XML, and inadvertantly came up with something that looked a lot like Notation3 in XML

03:34:44 <dc_rdfig> added comment A1

03:35:40 <sbp> A:This is the result: Notation3 as an XML language. It's not as bad as XML RDF, but not as good as Notation3 (for readability), IMO

03:35:41 <dc_rdfig> added comment A2

03:36:49 <sbp> A:It's not *exactly* N3, since it's missing formulae (n3:formula?), is x of, and some othe rlittle bits

03:36:49 <dc_rdfig> added comment A3

03:36:59 <sbp> A3:It's not *exactly* N3, since it's missing formulae (n3:formula?), is x of, and some other little bits

03:36:59 <dc_rdfig> replaced comment A3

03:37:11 <sbp> A:But it's close enough for a ten minute hack

03:37:11 <dc_rdfig> added comment A4

03:39:12 <sbp> A:cf. [SURF|http://swag.webns.net/SURF-spec] and [BSWL|http://purl.org/net/bswl]

03:39:12 <dc_rdfig> added comment A5

03:39:37 <mnot> hmm... "RDF doesn't understand content negotiation, or anything like it." - RoyF

03:40:01 <sbp> yeah: that's an interesting thread

03:45:49 <mnot> Personally, I think that mostly misses the point - RDF only uses URIs as identifiers, *except* when mechanisms like log:content are used (which can and should be fixed to be conneg-capable). it's not an RDF-wide problem, IMHO.

03:49:19 <MarkBzzz> Well, RDF doesn't use URIs, it uses URI refs. So sometimes it's identifying resources, and other times it's identifying some fragment of a representation

03:49:34 <sbp> I agree. I'm not really sure what content negotiation has to do with the thread, except that the necessary identification of "http://[...]#frag" is under question too

03:49:44 <MarkBzzz> Very big issue for SW going forward, IMO

03:49:45 <sbp> [to mnot]

03:50:04 <MarkBzzz> Other reasons in http://logicerror.com/fragmentProblems

03:50:45 <sbp> the "fragment of a representation" argument is silly, since fragments of representations are resources in their own right, and there is no particular constraint upon those that I am aware of

03:51:08 <MarkBzzz> If they're resources, why don't they have a URI?

03:51:29 <sbp> they can have URIs. <http://blargh.org/> = <http://blargh.org/#blargh> .

03:51:42 <sbp> but they have URI references instead. What's the big deal?

03:51:43 <MarkBzzz> Ok. Implement it. 8-)

03:52:04 <sbp> O.K., done. daml:equivlaentTo gives me all the machinery that I need

03:52:20 <sbp> (along with the implementations of DAML)

03:52:35 <MarkBzzz> REST defines all its behaviour at the granularity of the resource

03:53:23 <MarkBzzz> in that blargh.org example, that is not a statement about two resources though. it's not as easy. you can't fix the problem by using features of the problem. 8-)

03:54:02 <MarkBzzz> we need a massive s/#foo/\/foo/g or something 8-)

03:54:34 <sbp> "features of the problem"? that's bizzare. I just said that URI-refs identify fragments of resources which are identifyable "things" and therefore potential resources in their own right according to RFC 2396... what's the "problem", and what feature of that am I using?

03:55:20 <MarkBzzz> "potential" is right

03:55:32 <MarkBzzz> but how do we realize that?

03:56:05 <sbp> that seems to be more of a question of philosophy to me than an actually quantifiable problem

03:56:36 <sbp> since I have just declared one potential resource to be equivalent to another resource, it kinda removes the uncertainty, doesn't it?

03:56:44 <MarkBzzz> real problems listed at http://logicerror.com/fragmentProblems

03:56:52 <MarkBzzz> all stem from the fact that the web server doesn't get the fragment id

03:56:59 <sbp> I *raised* some of those problems!

03:57:14 <MarkBzzz> I know

03:57:23 <MarkBzzz> We've had this discussion before 8-O

03:57:27 <sbp> :-)

03:57:43 <sbp> I think it's odd that HTTP doesn't pass the fragment to the server...

03:57:49 <sbp> did it have to be that way?

03:58:16 <sbp> it certainly doesn't have to be that way for other protocols, AFAIK

03:58:26 <MarkBzzz> it absolutely has to be that way

03:58:32 <mnot> the whole point of the fragment is that it's evaluated in the client context, not the server

03:58:51 <MarkBzzz> otherwise you're no longer coupling between abstract concepts, you're coupling between bags-o-bits

03:59:02 <sbp> that may well be the point, but where is passing the fragment to the server forbidden by any RFC?

03:59:26 <sbp> sorry, MB, I didn't get your last point...

03:59:28 <MarkBzzz> RFC 2616. Request line is METHOD URI proto-version

03:59:32 * sbp is tired at the mo'

04:00:04 <MarkBzzz> the contract between client and server is one where the discussion is about the abstract concept of things with identity

04:00:22 <sbp> er... that's the HTTP specification. if I'm inventing another protocol, then why is the HTTP specification relevant?

04:00:33 <MarkBzzz> if you permit discussion about bits, you're digging deeper into the interface than that, and have dependancies on bits

04:00:44 <MarkBzzz> best current practice

04:01:04 <sbp> who says it's about bits? do network protocols have to deal with bits? are bits of representations naturally a series of bytes?

04:01:07 <sbp> in RDF, they're not

04:01:22 <MarkBzzz> they are, because URI refs identify bits

04:01:23 <sbp> and if you say that they are, then the RDF ID attribute is very broken

04:01:29 <sbp> prove it

04:01:48 <mnot> MarkB, I think the problem you're raising is that if RDF talks about things at a finer granularity then resource, REST can't talk about it.

04:01:50 <mnot> no?

04:01:53 <sbp> i.e. ground that statement in RFCs and implementations, because I see no proof of it

04:01:54 <MarkBzzz> you have to have a representation (bits) to evaluate a fragment

04:02:21 <MarkBzzz> mnot, I'm sure there's some form of solution here, but the status quo ain't it

04:02:26 <sbp> just because you have to have those bits in order to evaluate the semantics of a fragment does not mean that the fragment identifies some bits

04:02:43 <MarkBzzz> how not?

04:02:54 <sbp> example: RDF

04:03:06 <sbp> <#MyCar> a :Car .

04:03:23 <sbp> #MyCar is a car: you know that by parsing and processing the bits

04:03:49 <MarkBzzz> So if I say http://foo.com/#dog is a dog, and http://foo.com/#cat is a cat, what's the value of the Resource-Type header when I do a GET?

04:04:15 * mnot has to go, but will think more on it...

04:04:18 <sbp> you don't do a GET on the URI-ref, you do it on the URI

04:04:19 <MarkBzzz> later mnot

04:04:24 <sbp> c'ya mnot

04:04:55 <MarkBzzz> ok, so what's the Resource-Type header on http://foo.com?

04:05:26 <sbp> :BagOResources, perhaps?

04:06:47 <sbp> or perhaps a class: :PetsOfSomeGuy :-)

04:07:02 <sbp> which would make sense. it could be an RDF document about someone's pets

04:07:19 <MarkBzzz> and if I want to relocate #cat from one bag to another?

04:07:32 <MarkBzzz> how do I get an image representation of that cat?

04:07:32 <sbp> what do you mean by relocate?

04:07:54 <sbp> <http://blargh.org/#cat> foaf:depictedIn <http://blargh.org/cat.png> .

04:07:58 <MarkBzzz> move it from "house" bag to "kennel" bag, in the temporary redirect sense

04:08:15 <MarkBzzz> (poor example)

04:08:20 <sbp> <http://blargh.org/#cat> daml:equivalentTo <http://blargh.org/newPage#cat> .

04:08:48 <sbp> or, to use your bag terms...

04:08:58 <sbp> <http://blargh.org/house#cat> daml:equivalentTo <http://blargh.org/kennel#cat> .

04:09:48 <MarkBzzz> but how do you automate that? daml:equivalentTo is automatable with HTTP response code 302

04:10:12 <sbp> what do you mean by "automate"? the expression above is machine readable, is it not?

04:10:43 <MarkBzzz> How do I represent it in HTTP?

04:11:02 <sbp> how do you represent that SomeCat is a type of Cat in HTTP?

04:11:24 <MarkBzzz> GET/HEAD returns a Resource-Type header of "Cat"

04:11:31 <sbp> ooh, neat :-)

04:11:39 <MarkBzzz> hey, it's your header 8-)

04:11:43 <sbp> heh, heh

04:11:53 <MarkBzzz> you should right it up, I-D style

04:12:01 <MarkBzzz> s/right/write - getting late

04:12:01 <sbp> Hmm... I guess we could always add a DAML-Equivalent-To header

04:12:11 <sbp> write it up: actually, I was in the process of doing so!

04:12:21 <sbp> I have a draft available... want me to put it up?

04:12:28 <sbp> actually... it's a bit rough at the moment

04:12:29 <MarkBzzz> earlier this evening, I realized I could do an "Assert" header, and throw Ntriples in there

04:12:38 <MarkBzzz> yah, please do

04:13:13 <sbp> Assert: that's a good idea. TimBL already proposed "N3:", but that takes just a pred and obj pair, and uses the resource (or representation???) as the subject

04:13:14 <MarkBzzz> have to figure out a fragment identifier system for other headers though - reification and all that

04:13:20 * sbp fiddles around with FTP

04:13:30 <MarkBzzz> hadn't heard that

04:13:40 <MarkBzzz> I think NTriples are more appropriate

04:13:44 <MarkBzzz> for HTTP headers

04:14:03 <MarkBzzz> MarkBzzz is now known as MarkB

04:14:11 <MarkB> oops

04:14:19 <MarkB> though I am feeling that way 8-/

04:15:50 <sbp> here we go: http://infomesh.net/2002/draft-palmer-resrep-type-00.txt

04:15:53 <sbp> zzz: heh :-)

04:16:25 <sbp> I think Tim's "N3" thing actually did mean NTriples... let me find a reference

04:17:27 <sbp> aha: http://ilrt.org/discovery/chatlogs/rdfig/2001-10-25.html#T03-18-06

04:18:00 <MarkB> cool

04:19:13 <MarkB> still need to be able to identify other headers though. lots of good HTTP headers work that way; Connection, Vary, Man/Opt, etc..

04:19:39 <sbp> identify them when?

04:19:57 <MarkB> When we spec out the Assert header

04:20:04 <sbp> ah, right

04:20:48 <MarkB> maybe JonB's http/mime stuff would help there

04:22:38 <sbp> very good point. Also, didn't PaulP do some work on that? I could be mistaken

04:22:48 <MarkB> er, oops, i mean identify them in the HTTP message. Reification.

04:23:07 <MarkB> not that I know of. could be though.

04:23:19 <sbp> .google Paul Prescod Notation3 HTTP

04:23:20 <xena> Paul Prescod Notation3 HTTP: http://ilrt.org/discovery/chatlogs/rdfig/2001-07-02.txt

04:24:27 <MarkB> header example ....

04:24:34 <MarkB> Content-Type: text/html

04:25:34 <MarkB> Assert: <#Content-Type> <http://example.org/has-subject> <http://rest.example.org/representation>

04:25:49 <MarkB> (not sure I did that right - "has subject"??)

04:25:59 <MarkB> meant what I talked about here ...

04:26:06 <MarkB> .google restwiki subject-is-representation

04:26:07 <xena> restwiki subject-is-representation: http://internet.conveyor.com/RESTwiki/moin.cgi/HttpAndTheSemanticWeb

04:26:36 * sbp reads, whilst still trying chase down PP's thing

04:26:48 * MarkB runs for a drink

04:27:59 <sbp> Hmm... that's very interesting. I think that it's a property of the headers themselves, so that on might say:-

04:28:28 <sbp> http:Content-Type rdf:type :SubjectIsRepresentation .

04:28:40 <sbp> er... which is basically what you just suggested (oops)

04:28:54 <sbp> well, kinda

04:29:29 <sbp> the rdfs:domain of your has-subject property would be the :SubjectIsRepresentation class

04:41:29 <MarkB> well, on the wiki I was thinking that "SubjectIsRepresentation" would just identify another assertion

04:41:46 <MarkB> in the example above, I'm not sure what I was thinking. 8-)

04:43:23 <MarkB> HTTP has a handy feature that if there are multiple headers of the same name, that you can treat them as merged. so I don't have to worry about identifying headers by name

04:50:12 <sbp> good point

04:50:26 * sbp was just finishing up the ID

04:52:11 <MarkB> you should probably explain how Repr-Type relates to Content-Type, such as what's a processor to do when they're inconsistent

04:52:20 <sbp> good point

04:52:32 <MarkB> Of course, another way to do this would be to define a Content-Type-Base header

04:52:37 * sbp should probably add some serious acknowledgements too -)

04:52:52 <sbp> I was going to say that the Content-Type header is the default...

04:53:03 <sbp> what would Content-Type-Base be for?

04:53:49 <MarkB> Content-Type-Base: http://www.iana.org/assignments/media-types/

04:53:56 <MarkB> Content-Type: text/rtf

04:53:59 <MarkB> or

04:54:09 <MarkB> Content-Type-Base: http://www.markbaker.ca/media-types/

04:54:13 <MarkB> Content-Type: foo/bar

04:54:56 <MarkB> or with that same base;

04:55:02 <MarkB> Content-Type: text/foobar

04:55:18 <sbp> Hmm... that's a good idea

04:55:45 <MarkB> Yah, just thought of it

04:56:03 * sbp wonders if media types are all necessarily disjoint

04:56:08 <MarkB> though the idea to think of content-type as a relative URI was mentioned to me a while ago by HenrikFN

04:56:24 <sbp> actually, of course they aren't. application/xhtml+xml is a subset of application/xml

04:56:49 <sbp> yep, it's been floating around for some time. Aaron mentioned it to me about a year ago

04:57:09 <MarkB> problem with Content-Type-Base and IANA stuff is that not all media types get URIs, only those aren't registered with an RFC 8-O

04:57:12 <sbp> but the idea of defining a base header is new, AFAIK

04:57:35 <sbp> heh, so I noticed...

04:57:42 <MarkB> (with the new IANA structure - isi had it right)

04:58:48 <sbp> aha: http://lists.w3.org/Archives/Public/www-tag/2002Jan/0088

04:59:00 <sbp> I thought I'd read something about that from you before :-)

04:59:52 <MarkB> yup

05:20:13 * DanCon fails to fall asleep; continues work on "how many WGs haven't published in 3 months" data gathering action item...

05:28:18 <MarkB> I suppose you could call that "counting sheep" 8-)

05:28:24 <DanCon> ;-)

05:28:43 <DanCon> it's kinda depressing how nosiy all the various data are

05:28:50 <DanCon> noisy

05:28:59 <MarkB> publish data? really?

05:29:22 <DanCon> yeah... little stuff like whether the WG home page ends with MarkUp/Group or MarkUp/Group/

05:29:37 <MarkB> heh. opacity sucks.

05:29:51 <DanCon> opacity rules. sloppiness sucks. ;-)

05:30:25 <DanCon> there's a heisenquality rule about data somewhere, isn't there?

05:30:46 <MarkB> a global RSS feed of pubs would be nice

05:30:57 <MarkB> heisenquality? dunno

05:30:58 <DanCon> i.e. unless you have some consumer that breaks if the data is bad in a certain way, you can bet it will decay in that way.

05:31:22 <MarkB> heh

05:39:17 <MarkB> http://www.w3.org/Member/Mail/ has some ok metadata in it. the wg/cg/ig URIs should be easily extracted. now you just need a "rss" rel attribute, and feeds that the WGs maintain

05:39:46 <DanCon> yes, I have /Member/Mail/ scraped into RDF: http://www.w3.org/2000/04/mem-news/groups.rdf

05:40:00 <MarkB> cool

05:40:15 <DanCon> and we have /TR/ scraped: http://www.w3.org/2000/04/tr2.rdf

05:40:32 <DanCon> now the trick is to join them in interesting ways

05:41:16 <MarkB> so that's what you're doing now?

05:41:25 <DanCon> yup.

05:41:33 <sbp> tr2.rdf: not found

05:41:52 * DanCon apologies for typing a URI by hand...

05:42:09 <DanCon>http://www.w3.org/2000/04/mem-news/tr2.rdf

05:42:10 <dc_rdfig> B: http://www.w3.org/2000/04/mem-news/tr2.rdf from DanCon

05:42:12 <MarkB> np

05:42:18 <sbp> thanks

05:42:37 <DanCon> B:|an RDF version of the W3C tech reports index

05:42:39 <dc_rdfig> titled item B

05:42:51 <DanCon> B:by Dom H. and DanC

05:42:51 <dc_rdfig> added comment B1

05:43:25 <DanCon> um... log:content and str:scrape are really wicked tools...

05:43:57 <DanCon> I'm watching cwm grab the content of each tech report and find links to activity statements.

05:45:11 <DanCon> say... folks have talked about little agents that keep certain pages in-cache... cwm could do that really well now.

05:45:40 <DanCon> e.g. every 30 min, trace all links from my home page to a depth of three thru my wwwoffle cache.

05:46:14 <DanCon> with various exceptions etc.

05:46:15 <MarkB> I'm planning to integrate some cwm-like functionality into our HTTP router

05:46:25 <DanCon> scary!

05:46:34 <DanCon> umm... HTTP router?

05:46:47 <MarkB> like KnowNow

05:47:21 <DanCon> fun... it seems to work:

05:47:23 <DanCon> <http://www.w3.org/TR/2000/WD-DOM-Level-1-20000929/> :_activityStatement <http://www.w3.org/DOM/Activity> .

05:47:23 <DanCon>

05:47:23 <DanCon> <http://www.w3.org/TR/CCPP-ra/> :_activityStatement <http://www.w3.org/Mobile/Activity> .

05:47:52 <MarkB> inferring routes based on CC/PP profiles, Accept-*, etc..

05:48:11 <MarkB> heh, neat

05:57:37 <MarkB> MarkB is now known as MarkBzzzz

10:06:37 * danbri gets misty eyed --- was gonna do a phd on this topic: http://www.cogs.susx.ac.uk/users/terrs/extrema.html (simulated evolution of neural nets...)

11:00:38 <dajobe>http://www.ics.uci.edu/~fielding/url/test1.html

11:00:39 <dc_rdfig> C: http://www.ics.uci.edu/~fielding/url/test1.html from dajobe

11:00:48 <dajobe> C:|Examples of Resolving Relative URLs, Roy Fielding

11:00:48 <dc_rdfig> titled item C

11:16:42 <dajobe> C:darn, can't see base URI "mailto:foo@example.com" relative "blah"

11:16:42 <dc_rdfig> added comment C1

12:26:07 <lilo> [Global Notice] Hi all. Let's see if we can finish up the conversion to 1.0.31+maint2 in the next few minutes. As the remaining servers are affected, we'll notify people on them. Thanks.

12:30:54 <dajobe> funny: http://www.rfc-editor.org/policy.html on URLs in RFCS: "The use of URLs in references in RFCs is discouraged in general, because URLs are often not stable references."

12:36:58 * maxf waves

12:37:18 <dajobe> hi nadia

12:37:26 <dajobe> oops

12:37:26 <dajobe> and hi maxf

12:40:22 <maxf> hi dajobe and nadia

12:40:26 * maxf likes the topic

12:52:49 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

12:53:01 <lilo> [Global Notice] Hi all. The next two server conversions are going to be big ones. The first one is a rehub, but it'll affect just about everybody. I'm going to just try to get both of them one right after the other. Please bear with us.

13:01:08 * nadia realizes too late that someone said hi

13:01:54 Disconnected from irc.openprojects.net (Connection reset by peer)

13:02:05 Topic now SemanticWeb 24x7 hack'n'chat -- entering completer-finisher phase, please check Google before coding! -- http://rdfig.xmlhack.com/

13:02:05 Users on #rdfig: logger danbri oierw|sleep-scho grove gerald pldab__ JibberJim nadia maxf sandro dc_rdfig AaronSw em deltab DanC_jam jang libby dajobe deus_x DanCon sbp

13:02:05 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

13:09:01 <lilo> [Global Notice] Hi all. Taking stock now, we should be just about finished.

13:58:51 <lilo> [Global Notice] Thanks again to Sir_Ahzz for making lackey.openprojects.net available to us for so long. His efforts are greatly appreciated, and we're sorry to see that server go.

15:04:55 <dajobe> hey jonb

15:25:53 <jonb> hey dajobe

15:27:00 <dajobe>http://www.w3.org/TR/2002/WD-rdf-primer-20020319/

15:27:01 <dc_rdfig> D: http://www.w3.org/TR/2002/WD-rdf-primer-20020319/ from dajobe

15:27:53 <dajobe> D:|RDF Primer 2002-03-19

15:27:53 <dc_rdfig> titled item D

15:28:52 <DanCon> wow... the primer's published?

15:29:28 * DanCon wonders if we should brace for impact, of if folks will just yawn at it

15:30:00 <DanCon> ok, danbri, everybody else has published. RDFS's turn!

15:30:46 <danbri> You're right.

15:31:20 <danbri> I'm looking at some slight mismatch between Frank's text and RDFS as I'm tweaking it. Sooner RDFS goes out the door the better. Watch this space.

15:33:36 * dajobe watches :)

15:35:08 <jonb> is RDFS going to be aligned with OWL? e.g. cycles restriction removed?

15:35:34 <danbri> cycle restrictions removed, yes.

15:35:41 <danbri> 'aligned with owl' is a slippery notion.

15:36:06 * danbri ignores IRC for a bit

15:39:55 <jonb> dajobe: what is bizarre about rdf:ID being NMTOKEN is that I thought the whole point of rdf:_1, rdf:_2 was to be an ID

15:40:08 <dajobe> they aren't

15:40:15 <dajobe> those are properties

15:40:19 <jonb> that is to say you could just do rdf:1 rdf:2

15:40:24 <dajobe> and in the XML syntax are just elements/attribtues like any other

15:40:31 <dajobe> no rdf:1 has no special meaning

15:41:03 <dajobe> so I'm not really opposed to changing rdf:ID's definition, just didn't hear anyone asking for it!

15:41:16 <jonb> so <rdf:Description rdf:ID="1" is intended to be legal?

15:41:27 <dajobe> sure, why not

15:41:35 <dajobe> that is equiv to rdf:about="#1"

15:41:44 <dajobe> which resolves to the in scope base URI

15:41:50 <jonb> because <rdf:1> is not a legal XML QName

15:41:54 <dajobe> oh

15:42:01 <dajobe> well that would be an XML error

15:42:35 <dajobe> rdf/xml is based on legal infosets; so if it can't be expressed in xml, it can't be in rdf/xml

15:42:37 <jonb> yes so <rdfs:Property rdf:ID="1"> etc. would be totally screwed up

15:42:52 <dajobe> no

15:43:10 <dajobe> the "1" is equiv to the URI-reference "#1"

15:43:12 <jonb> not because that is illegal according to syntax but because attempt to use

15:43:27 <jonb> <foo:1> as a QName is illegal

15:43:37 <dajobe> (uri-ref , in-doc referecnce whatever)

15:43:56 <dajobe> what has rdf:1 got to do with rdf:ID="1" ?

15:43:58 <jonb> that is to say you cannot have a _predicate_ with QName foo:1

15:44:12 <jonb> think about it

15:44:16 <dajobe> so

15:44:37 <dajobe> you can create rdf models with predicates that can't be expressed in XML

15:44:39 <dajobe> in rdf/xml

15:44:42 <dajobe> we know this

15:44:46 * dajobe shrugs

15:44:52 <dajobe> so don't do that :)

15:44:52 <jonb> that is the reason why an xml:ID is Name

15:45:26 <jonb> if rdf:ID were Name rather than Nmtoken, there would be no issue

15:45:28 * dajobe must be missing something

15:45:49 <dajobe> rdf predicates in the model have nothing to do with qnames

15:46:05 <jonb> we are talking syntax here

15:46:08 <dajobe> OK

15:46:51 <jonb> if rdf:ID were of type xml:ID or xml:Name then all legal values of rdf:ID would be associated with a legal QName

15:47:45 <dajobe> ok

15:48:24 <dajobe> but again, the model doesn't restrict the uris used for predicates. It would be advisable to create ones that could be made into legal XML IDs, but not required. You just have trouble serialising them (you can't)

15:48:43 <jonb> all you would give up would be allowing a number '.' '-' as the first character in an ID

15:48:54 <dajobe> which we note, nobody uses

15:48:59 <dajobe> so maybe we are agreeing? :)

15:49:01 <jonb> right

15:49:13 <dajobe> the low barrier for change has probably been breached then :)

15:49:19 <dajobe> ahem, did I say low?

15:49:21 <jonb> :-)

15:49:40 <jonb> yes you do need some barrier

15:49:52 <dajobe> actually IIRC, tim did say he wished - in IDs were banned

15:50:08 <dajobe> see - to _ re/writing strategy for n3

15:50:22 <DanCon> jonb, what does rdf:ID have to do with property names? rdf:ID isn't an XML ID. XML IDs have to be declared with <!ATTLIST>. rdf:ID is just shorthand for rdf:about="#foo"

15:51:03 <dajobe> that's what we were discussing DanCon, but they have different legal values

15:51:10 <dajobe> rdf:ID has nmtoken, xml:id has Name

15:51:19 <DanCon> nmtoken? where?

15:51:34 <dajobe> sigh :)

15:51:54 <jonb> i am just saying that the lexical values allowed for both should probably be the same

15:52:08 <dajobe> and I was just saying sure, why not, but I'd like a decent reason to change

15:52:39 <dajobe> rdf:ID is defined http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/#IDsymbol

15:52:40 <DanCon> change from what? where is the syntax of rdf:ID constrained to be anything smaller than uric*?

15:52:46 <dajobe> which says "6.21] IDsymbol ::= (any legal XML name symbol)"

15:52:53 <dajobe> which links to ttp://www.w3.org/TR/REC-xml#NT-Nmtoken

15:52:56 <dajobe> ugh

15:53:01 <dajobe> links to http://www.w3.org/TR/REC-xml#NT-Nmtoken

15:53:11 <dajobe> whereas xml:ID goes to ttp://www.w3.org/TR/REC-xml#NT-Name

15:53:17 <DanCon> ugh... so our grammar still isn't formal?

15:53:23 <dajobe> that's the old grammar

15:54:04 <dajobe> in the relaxng schema in the revised doc, I use: IDsymbol = xsd:NMTOKEN

15:54:18 <dajobe> but we were discussing if this should be xsd:Name like for xml:ID

15:54:29 <dajobe> and it seems yes, there are good reasons

15:54:36 <DanCon> ah... ok; dajobe's saying: as of Feb '99, RDF spec said what goes in rdf:ID is what XML calls a Name; and sees no reason to change. I suppose I don't either.

15:55:30 <dajobe> there wasn't any clamouring but I did notice this

15:55:38 <dajobe> and in fact ARP uses the xml:ID defn

15:55:43 <jonb> err, RDF says NMTOKEN, I am suggesting that NAME is more appropriate

15:55:48 <DanCon> seems to merit a test case, regardless.

15:56:18 <jonb> <rdfs:Property rdf:ID="1" />

15:56:44 <dajobe> I wouldn't use schema stuff in a test case; might imply more things

15:56:49 <dajobe> but <foo:bar rdf:ID="1"/>

15:56:56 <dajobe> is that legal or not?

15:57:08 <dajobe> rdf M&S says Yes I think at present

15:57:24 <dajobe> whereas I guess ARP would reject that or give a warning, sinc eit uses the xml defn.

15:57:57 <dajobe> woo, ARP does!

15:57:59 <dajobe> Warning: {W108} Not an XML Name: '1'[Line = 4, Column = 31]

15:58:05 <jonb> reason to use Property is that <foo:1>value</foo:1>

16:01:20 <dajobe> hmm

16:01:26 <dajobe> I'll try to make test cases for the wg

16:05:38 <jonb> do you have test case with two of the same rdf:ID="foo" in the same document?

16:10:15 <DanCon> we should; I'm pretty sure we decided that's a no-no.

16:10:30 <DanCon> s/decided/re-confirmed/

16:12:06 <dajobe> we did and yes

16:12:17 <dajobe> it is in the xmlbase tests

16:13:07 <dajobe> no, not there

16:13:08 * dajobe looks

16:13:18 * jonb wonders about this idea on the TAG thread that one ought not use the base URI to compose rdf:about="#foo" into a full URI

16:13:32 <jonb> s/URI/URIref/

16:14:18 <dajobe> (error) test is http://www.w3.org/2000/10/rdf-tests/rdfcore/rdfms-difference-between-ID-and-about/error1.rdf

19:13:01 <lilo> [Global Notice] Hi all. Please be aware that a bug has cropped up in the new maintenance release. INVITES are not being propagated across servers. The intent is to fix the problem this weekend, and also to include a new channel mode to allow you to decide whether non-operators should be able to issue INVITES. We'll keep you posted on the upgrade schedule. Apologies for the inconvenience, and thank you for using OPN.

19:42:19 * danbri reappears briefly

19:45:56 <AaronSw> D:What's with the <div class="exampleOuter"> <div class="exampleInner"><pre> stuff? It makes the document look pretty weird.

19:45:56 <dc_rdfig> added comment D1

19:48:50 * DanCon twiddles thumbs waiting for cwm

19:50:06 <danbri> BLURB:CouplaSOAP Things

19:50:06 <dc_rdfig> E: CouplaSOAP Things from danbri

19:51:31 <danbri> E:[http://lists.w3.org/Archives/Public/xml-dist-app/2002Mar/0205.html|Q - does SOAP encoding grok xsd:anyURI?], [http://lists.w3.org/Archives/Public/xml-dist-app/2002Mar/0189.html|SOAP Encoding as standalone XML format].

19:51:32 <dc_rdfig> added comment E1

19:53:34 <danbri> E:More on this next week. RDF folk really should take a careful look at this stuff. One reply to my latter question, '[http://lists.w3.org/Archives/Public/xml-dist-app/2002Mar/0207.html|I would like to see an independent spec on the serialization of a graph of name/value entities from the w3c]', I'm not sure how to answer. "We already did one?"

19:53:35 <dc_rdfig> added comment E2

20:25:40 <Seth>http://lists.w3.org/Archives/Public/www-rdf-comments/2002JanMar/0232.html

20:25:40 <dc_rdfig> F: http://lists.w3.org/Archives/Public/www-rdf-comments/2002JanMar/0232.html from Seth

20:25:56 <Seth> F:| Contexts at last :)

20:25:57 <dc_rdfig> titled item F

20:27:02 <Seth> F: [http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Mar/0253.html|in response to Pat Hayes things that go bump]

20:27:03 <dc_rdfig> added comment F1

20:27:56 <Seth> F: F: [http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Mar/0267.html|agrees with Grahm's response]

20:27:57 <dc_rdfig> added comment F2

20:28:45 <Seth> F: [http://www.ninebynine.org/RDFNotes/UsingContextsWithRDF.html|see also]

20:28:45 <dc_rdfig> added comment F3

20:29:37 <Seth> F2: [http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Mar/0267.html|agrees with Grahm's response]

20:29:59 <Seth> F2: [http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Mar/0267.html|agrees with Grahm's response]

20:30:00 <dc_rdfig> replaced comment F2

20:32:59 <Seth> F: would give us <rdf:RDF context="http://foo/"> ...</rdf:RDF>

20:33:02 <dc_rdfig> added comment F4

20:33:27 <Seth> F: and the ability to have dark triples as first class resources

20:33:28 <dc_rdfig> added comment F5

20:43:22 <sbp> Hmm... dark triples

20:44:24 <sbp> I can imagine Semantic Web historians in decades time searching for these mysterious "dark triples" that hold the Semantic Web together, and make up for the unaccounted data hinted at by the current data :-)

20:58:09 <dajobe> F:do not underestimate the dark side of the triple

20:58:09 <dc_rdfig> added comment F6

20:59:03 <Galahad> Galahad is now known as xena

21:23:51 <sandro> I want a Content-Type for n3. Is there one? Should I just use an HTTP URI for it?

21:31:09 <AaronSw> There's an HTTP URI for it.

21:31:18 <AaronSw> see http://infomesh.net/2002/draft-palmer-resrep-type-00.txt

21:36:48 <sandro> That being http://www.w3.org/2000/10/swap/log#N3Document

21:37:19 <sbp> yep

21:39:19 <sandro> It looks like apache (Apache/1.3.23) passes that through to CGI properly. yay.

21:46:33 <sandro> ...alas lwp-request will not allow it on a "-c <content-type>" parameter.

21:47:36 <sbp> it isn't a valid content type...

21:48:54 * sbp suddenly realises that he left Sandro off of the acks list - aaaargh!

21:49:40 <sbp> the FLD proposals were quite seminal

21:50:37 <sandro> Yeah, but you never know how strict various software will be. I'm not saying lwp-request has a bug, just that going to full-URI content-type is going require changing lots more software than I hoped.

21:51:09 <sandro> ack lists? FLD proposals? <wooshing over /me's head>

21:51:33 <sbp> why would you want to change the definition of Content-Type? It's defined as being an IANA registered media-type in RFC 2616...

21:51:40 <AaronSw> acknolwedgements, formal language definition

21:52:07 <sbp> acks in draft-palmer-resrep-type-00.txt, now fixed

21:52:23 <sbp> the FLD thing was in the Blindfold langIdent page

21:53:26 <sandro> Ah yes, "Formal-Language-Definition". Gotcha.

21:54:04 <sandro> Why change Content-Type? Because it's silly having iana have a lock on it.

21:54:16 <sbp> hence the proposal for Repr-Type

21:54:29 <sandro> Why not just extent Content-Typr?

21:54:42 <sbp> because Content-Type is already deployed over many, many, user agents

21:55:02 <sbp> why break any existing implementations that syntactically validate the Content-Type value?

21:55:05 <sandro> Yeah, but this is a backwards-compatible change.

21:55:23 <sbp> what is? Repr-Type is fully backwards compatible

21:55:32 <sbp> adding URIs to the Content-Type field is not

21:56:11 <sandro> Yeah, I see your point. Anyway, not my argument.

21:56:14 <sbp> :-)

21:56:23 <sbp> oops...

21:56:26 <sbp> Gotta run

21:56:29 <sandro> ciao.

22:05:33 <sandro> Ohhhh, lwp-request wont even allow "+" in the content-type.

22:10:50 <sandro> (so you can't use RFC 3023 with it.)

22:12:44 * danbri_ wonders if em around

22:13:06 <danbri_> BLURB:RDFSchema editorializing

22:13:07 <dc_rdfig> G: RDFSchema editorializing from danbri_

22:13:23 <danbri_> G:Now we have the Primer, I'm planning to remove all the current examples from RDFS spec

22:13:25 <dc_rdfig> added comment G1

22:14:09 <danbri_> G:In fact my paper copy (the dodgy iuniverse.com paperback) has more sections crossed out than remaining.

22:14:10 <dc_rdfig> added comment G2

22:14:29 * dajobe adding more examples to rdf/xml doc

22:14:51 <danbri_> G:But we need at least one example schema. Not a mix of people, Vans, Search services etc., but something that can be built up thru the document.

22:14:52 <dc_rdfig> added comment G3

22:15:50 <danbri_> G:This [http://www.amazon.com/exec/obidos/ASIN/0595132308/qid%3D1016748754/ref%3Dsr%5F11%5F0%5F1/102-5505709-5460913|dodgy shovelware book] proved useful at last (as did a big red pen).

22:15:51 <dc_rdfig> added comment G4

22:16:09 <danbri_> dajobe, what examples are you using? In the 'Ora/homepage' tradition?

22:16:21 <dajobe> look for yourself: http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/

22:16:27 <dajobe> but based on example in previous draft

22:16:38 <danbri_> G:I need an example that shows RDFS as a foundation for Semantic Web vocabulary description, ie. shows how WebOnt slots nicely in.

22:16:41 <dc_rdfig> added comment G5

22:16:43 <dajobe> this is a placeholder for redoing in next one

22:17:03 * dajobe should link to the SVG versions

22:17:57 <danbri_> BTW how do you manage that document? RDFS (even XHTML'd) is unmaintainable in current form. I nearly started with a fresh doc and pasted stuff in, instead of cutting. I'm happy using XHTML (instead of specprod DTD(s)) but would love a tool that built the table of contents for me.

22:18:02 <danbri_> don't suppose you've got one?

22:18:34 <dajobe> no

22:18:40 <dajobe> but there are perl scripts involved in that doc

22:18:45 <dajobe> plus xml validators

22:19:20 <dajobe> was wondering about the xml techrep stuff you can get now

22:19:22 <danbri_> I'm also concerned about presentational stuff: there's new CSS is the Primer, and you have CSS styles in yours.

22:19:33 <dajobe> yeah, they pinched from me, I took it from others

22:19:38 <dajobe> XSD IIRC

22:19:39 <danbri_> often the way :)

22:19:46 * danbri_ will pinch from primer

22:19:56 <dajobe> but I did ask months back about 'official' CSS for good accessibility etc.

22:20:08 * dajobe is just making it up

22:20:59 <danbri_> the specprod effort was interesting, but I never felt the need for much beyond XHTML and a toc generator

22:21:30 <danbri_> OK, brutal edits time :)

22:21:54 <dajobe> don't forget to add datatypes :)

22:22:40 <danbri_> Grab Pat's doc and work from there?

22:22:48 <danbri_> It's gotta get way slimmer before filling out again.

22:23:35 <danbri_> I thought it was reasonably pure, compared to the confusion of Model+Syntax, but there's a fundamental confusion that shapes the whole RDFS spec.

22:24:10 <danbri_> Infected with XML's notion of validity and constraints, basically. The extensibility stuff isn't what WebOnt need, but the basic idea is sound.

22:25:08 <danbri_> How does this sound re examples? Strip out all current examples (or use in primer, test suite). Then have a single, initially simple example that runs throughout, based on the old MCF spec: 'chatwin -wrote-> songlines'.

22:26:06 <danbri_> show it elaborated on, with domain/range. Then another property, 'author' ascribed to 'songlines', would show how domain and range work, and how ontology languages could enhance the machine processability of RDF vocab descriptions...

22:27:05 <dajobe> I don't remember the mcf example

22:27:08 <danbri_> Something else I'm dying to fix: it was always ambiguous whether 'RDF Schema Specification' meant 'the practice of specifying RDF Schemas', or 'A Specification for RDF Schemas'.

22:27:17 <dajobe> but if you can construct one to go through the doc, great

22:27:21 <danbri_> MCF: I remember it vividly. Didn't understand RDF before I saw it.

22:27:21 <dc_rdfig> Label MCF not found.

22:27:39 <dajobe> I always thought schema was a bad word to use; <tired-record/>

22:27:46 <danbri_> I think we need that. Zero examples would be too dry, but the current ones are unanimously rather lame

22:27:51 <danbri_> "always"? Why?

22:28:00 * danbri_ has several views on this topic :)

22:28:01 <dajobe> to add?: superclass of rdf:_<n> stuff rdfs:member

22:28:06 <danbri_> Indeedy.

22:28:31 <dajobe> oh, and can we actually have a schema too?

22:28:34 <danbri_> rdfs:member? (rdf:li would be obvious name for it, but I suspect the interaction with syntax spec makes it bad).

22:28:43 <dajobe> ie. <rdfs:Class rdf:about="#Schema">

22:28:46 <danbri_> Hold yer horses! One thing at a time.

22:28:49 <dajobe> hehe

22:28:54 * dajobe plays the add-a-thing game

22:28:58 <dajobe> makes a change

22:29:02 <danbri_> Oh, see what you mean. I thought you meant an update to the doc at the namespace.

22:29:24 <danbri_> So something we have trouble with: there are _no_ good rdfs:comments for the classes and properties.

22:29:31 <dajobe> I bet you can't correct that sub/super thing the right way round?

22:29:54 <dajobe> the comments are bad?

22:30:03 <danbri_> Hey Jim, just in time...

22:30:12 <danbri_> Talking about RDFS redraft.

22:30:24 <danbri_> <title>

22:30:25 <danbri_> Semantic Web Vocabulary Description: RDFCore Schemas 1.0

22:30:25 <danbri_> </title>]

22:30:29 <danbri_> ...perhaps?

22:30:37 <danbri_> Pluralist ;-)

22:30:55 <dajobe> RDFCore?

22:31:12 <danbri_> I'm not sure we can hit all the deadlines, but after this afternoon, I'm happy it can be a useful spec.

22:31:14 <danbri_> RDF Core

22:31:36 <danbri_> 'Core' is so ambiguous in our usage. RDFS currently uses it as a synonym for M+S namespace/spec/etc.

22:31:56 <danbri_> Dave, you gonna be around for a bit longer? Someone to bounce stuff off would be useful...

22:32:06 <dajobe> sure, but editing rdf/xml in othe rwindow

22:32:09 <dajobe> wonder if beer will help

22:32:09 <danbri_> Ideal.

22:32:25 <danbri_> I've just turned down a beer temptation, but when this's done, I'm planning much beer.

22:32:30 <danbri_> :)

22:32:38 <danbri_> How does that title strike you?

22:32:52 <danbri_> We can't say 'ontology', and I wouldn't want to anyway...

22:32:54 <dajobe> sure

22:33:05 <jhendler> I like vocabulary title

22:33:08 <dajobe> although some people will read an implied "The" at the start

22:33:11 <dajobe> rathter than "An"

22:33:23 <danbri_> I'm more and more convinced (sorry Jim et al.) that the Semantic Web is not fundamentally about ontologies....

22:33:34 <danbri_> I'll defend that another time!

22:33:57 <danbri_> It's about merging descriptions of individuals, not making sweeping generalisations about categories. IMHO.

22:34:08 <danbri_> vocab title: excellent. it'll do for now then.

22:34:12 <jhendler> while I'm more and more convinced we cannot function without them -- but I suspect we really agree because what I think is necessary is RDFS plus a little, and what you think is necessary is OWL minus some...

22:34:27 * dajobe nods

22:34:37 * sandro has a vague image of a hulking brute standing in front of danbri grunting "No, you'll defend that NOW."

22:34:46 <jhendler> danbri - I think it is relating intstances to categories, and having a WEB of semantics - not semantics on the web.

22:35:00 * danbri_ nods, tries not to get drawn away from emacs

22:35:30 <jhendler> JimH is late for an appt later, looks forward to first time he and danbri and a beer can be in the same place to have this discussion correctly :->

22:35:41 <danbri_> The absolutely most valuable bit of WebOnt/DAML+OIL isn't the description logic stuff. It's dpo:UnambiguousProperty. Since it helps fix what we're talking about -- which objects we mean to describe

22:35:52 <jhendler> by the way, yesterday I titles one of my talk slides "The MOSAIC of the Semantic web" and it needed ontologies :->

22:35:56 <danbri_> jim, absolutely

22:37:00 <jhendler> unambig is useful, but it's also things like If I know you are a person, then I know that when you name a link on the web "My mom" it is also pointing to a person -- those restrictions on the types/classes really are important.... ciao for now

22:37:38 <danbri_> they're handy, f'sure

22:37:41 <danbri_> cu

22:37:47 <danbri_> hence rdfs:range

22:40:20 <danbri_> Anyone know of a construct that's useful but not in DAML+OIL? ie. something like symmetric, transitive etc? Or something that a rule language might be needed for?

22:40:39 <dajobe> friend of a friend?

22:40:44 <dajobe> knows?

22:41:03 <danbri_> rather not use foaf, seems a bit selfpromo (and a bit too fun/silly)

22:41:04 <dajobe> could give examples that would be familiar to people?

22:41:10 <Seth> negation of assertions

22:41:10 <danbri_> that'd be... nice...

22:41:18 <danbri_> no way!

22:42:35 <danbri_>http://www.w3.org/TR/NOTE-MCF-XML/

22:42:39 <dc_rdfig> H: http://www.w3.org/TR/NOTE-MCF-XML/ from danbri_

22:42:48 <danbri_> H:Meta Content Framework Using XML (has a nice example)

22:42:49 <dc_rdfig> added comment H1

22:44:39 <Seth> implies ?

22:44:54 <danbri_> no!

22:44:57 * dajobe gets flashbacks

22:45:00 <dajobe> its Boston, 1995

22:45:15 <dajobe> This new language called Java has just been invented...

22:46:01 <Seth> reflexive ?

22:46:17 * danbri_ had just got a new job at ILRT, his Phd ambitions were starting to fade

22:48:52 <danbri_> Do you think the phrase 'type system' is useful in RDFS spec? I fear it conjurs up expectations that are mostly frustrated

22:50:37 * danbri_ chops the heritage stuff (KIF, Warwick framework) -- belongs in primer probly

22:50:42 <danbri_> (can always add it back later)

22:51:08 <dajobe> mmm type system

22:51:12 <danbri_> geek

22:51:18 * dajobe was just thinking

22:51:23 <danbri_> @All vocabularies are expressed within a

22:51:26 <danbri_> single well defined model

22:51:29 <danbri_> that can go

22:51:42 <dajobe> yeah, could take out "type system"

22:51:47 <dajobe> or hint at other work

22:51:56 <dajobe> not single model: yes

22:52:47 <danbri_> [[

22:52:49 <danbri_> All RDF vocabularies share some basic common structure: they

22:52:49 <danbri_> describe types of resource and types of relationships between resources.

22:52:49 <danbri_> This commonality allows for a finer grained

22:52:49 <danbri_> mixing

22:52:49 <danbri_> ]]

22:52:52 <danbri_> is that any better?

22:53:08 * danbri_ removes 'machine-undestandable'

22:53:30 <dajobe> a bit better

22:53:37 <dajobe> if you could say it without using 'type', that would be good

22:53:44 <dajobe> but I think that may be as good as you get

22:53:57 <dajobe> s/types/kinds/?

22:54:04 <danbri_> category

22:54:09 <dajobe> uh oh

22:54:13 <dajobe> is that overloaded too much?

22:54:16 <dajobe> prob not as bad as type

22:56:09 <danbri_> I think 'constraints' terminology just has to go. I can't maintain sane prose that uses that word without causing immense confusion.

22:56:39 <Seth> [[they describe resource and relationships between resources]]

23:00:52 <danbri_> [[

23:00:52 <dajobe> yeah

23:00:53 <danbri_> <p>

23:00:53 <danbri_> The RDF Schema vocabulary provides basic mechanisms for describing RDF

23:00:53 <danbri_> vocabularies. It allows vocabulary designers to represent descriptions of classes and properties in the World Wide Web, for example by describing ways in which combinations of classes, properties, and values can be used together meaningfully.

23:00:55 <danbri_> </p>

23:00:57 <danbri_> ]]

23:01:07 <danbri_> long sentence, ho hum. still shorter than was.

23:01:09 * dajobe turns americanizes the doc. ised->ized that is

23:01:32 <danbri_> Yeah, I saw that. Bummer. Maybe we could write an cascading XSLT stylesheet that repairs it?

23:01:37 <dajobe> hehe

23:02:31 <dajobe> I've got an excuse; can't break the anchors

23:02:38 <dajobe> a href="#section-Serialising">

23:02:58 <danbri_> Do you have a nice CSS for editorial-todos? (I'm just using 'color:red')

23:03:14 <dajobe> haven't done that much

23:03:27 <dajobe> but I graham hand wrote http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/wd3.html for me

23:03:37 <dajobe> or in some app (it's tidy-ed)

23:04:38 <danbri_> Wondering about the Dog/Mammal stuff in 2.1.

23:04:45 <danbri_> Looks primery.

23:04:53 <dajobe> don't rip out all the examples

23:06:09 <danbri_> It needs examples, for sure. Currently I think they subtract from the overall presentation: they're casual, ill worked out, and academicy.

23:06:30 <dajobe> well take them all out and put them back in if you think they deserve it

23:06:55 <dajobe> else leave a placeholder

23:07:00 <danbri_> Yeah. I think the author/Person/Document example has a lot going for it.

23:07:23 <danbri_> Isn't DC, so we might want to think about that.

23:07:26 * danbri_ returns to cutting.

23:07:49 <danbri_> Classes and Properties as sets and elements: that's gotta go. It doesn't match the MT set theoretic stuff.

23:08:04 <dajobe> anything example wise to nick from http://www.w3.org/TR/2002/WD-webont-req-20020307/ ?

23:08:11 <dajobe> yes

23:08:20 <danbri_> good idea. wouldn't want to meet too many requirements though.

23:08:51 <dajobe> I think pathayes has some words on what you might use for classes and properties; he did some rdfs words IIRC

23:09:08 * dajobe cheered up by rdfprimer being out

23:09:19 <dajobe> we should tell dublin core

23:24:54 <danbri_> dave?

23:25:04 <dajobe> mmm

23:25:32 * danbri_ looked at rdfs:isDefinedBy: I'd peged the three 'what this might be used for' as Primer fodder. Not so sure. Any view?

23:25:51 <danbri_> [[

23:25:52 <danbri_> The most common anticipated usage is to identify an RDF

23:25:52 <danbri_> schema, given a name for one of the properties or classes

23:25:52 <danbri_> defined by that schema.

23:25:53 <danbri_> etc

23:25:55 <danbri_> ]]

23:26:15 <dajobe> I always thought the <term rdf:isDefinedIn="schema url"> was useful but a bit of a hack

23:26:28 <dajobe> um, that was dublincore meeting in washington dc, IIRC

23:26:28 <DanCon> hack: how so?

23:26:41 <dajobe> hence the rdfs:Class rdfs:Schema would be nice

23:26:48 <dajobe> DanCon: can't discover a schema

23:27:00 <dajobe> with a specific class, you could

23:27:15 <dajobe> but then "an" rdf schema is rather a vague concept

23:27:17 <danbri_> Do we say anywhere (syntax?) that it is convenient/polite/etc to choose namespace URIs that end in a character that allows the namespace bit to be mechanically figured out?

23:27:36 <dajobe> sort of

23:27:37 <danbri_> Yes, we don't make an explicit notion of an 'RDF Schema Document' anywhere

23:27:39 <danbri_> where?

23:27:47 <dajobe> hold on

23:28:14 <dajobe> I thought it was here: http://www.w3.org/TR/2001/WD-rdf-syntax-grammar-20011218/#section-Serialising

23:28:24 <dajobe> but it just points to jeremys paper

23:28:29 <danbri_> No worries for now.

23:29:31 <DanCon> I made a pretty specific proposal (includeing test case sketches) re chosing namespace names; the WG adopted it, as I recall.

23:29:40 <danbri_> Ah, that's probably what I was thinking of.

23:30:41 <DanCon> danbri_, for possible prose to borrow, have you looked at "making vocabularies" in http://www.w3.org/2000/10/swap/Primer.html ?

23:30:43 <danbri_> There's *very* little in current RDFS about deploying this vocab in the Web. Documents, namespace URIs etc. If we want to find lots of machine readable RDF Schemas at namespace URIs, I think we better say so more explicitly...

23:31:16 <dajobe> well, whats at a namsepace uri is a hot potato

23:31:23 <dajobe> see various mailing lists etc.

23:31:43 <DanCon> meanwhile, it works real good to put an RDF schema at the end of an RDF namespace uri.

23:31:50 <dajobe> hehe

23:31:51 <danbri_> Me too.

23:32:08 <danbri_> swap primer: nice work, may pinch some

23:32:47 <danbri_> DanC, what do you think about the word 'Constraints' in RDFS? It seems to me to have caused the most confusion (by mkaing people think of DTD validation, closed world etc)

23:33:09 <DanCon> have you given yourself any space to think about re-titling the RDFS spec to avoid the XML schema/RDF schema stuff? maybe "Specifying RDF Classes and Properties"

23:33:25 <dajobe> LOL

23:33:31 <danbri_> See above.

23:33:33 <DanCon> I'm not joking.

23:33:39 <dajobe> neither was I

23:33:46 <dajobe> just I've been banging on about that for years

23:33:46 <DanCon> constraints: chuck it.

23:33:48 <danbri_> 'Semantic Web Vocabulary Description: RDF Core Schemas 1.0' being current idea...

23:34:14 <danbri_> yeah 'chuck it' happening big time. I was blocked on this stuff until I decided to do a full hatchet job. It wasn't easily tweakable...

23:34:35 <DanCon> why bother with "Core Schema"? stop after "Semantic Web Vocabulary Description in RDF"

23:34:49 <DanCon> from the department of redundancy department ;-)

23:35:24 <danbri_> Cos we're chartered to complete the work on RDF Schema :-/

23:35:34 <danbri_> Yours sounds better tho

23:35:49 <danbri_> Resource Description Framework Description Framework

23:35:52 <DanCon> charter: yes, and CR phase feedback says the word "schema" in the title doesn't help.

23:36:15 <danbri_> indeed. OK. Do we remove it from the entire doc? We might need a noun phrase to replace it...

23:36:36 <danbri_> also 'rdfs:' becomes 'swvd:' ?

23:36:43 <dajobe> lol

23:36:47 <DanCon> entire doc: I dunno; I'd have to read it to be sure. probably not; e.g. not from the namespace name (or are we changing that?).

23:36:51 <dajobe> well, the namesapce was changing anywya?

23:36:52 <dajobe> oh

23:36:53 <danbri_> that's a whole load more.

23:37:06 <danbri_> changing: we could almost avoid that... dunno.

23:37:15 <dajobe> if xsd: can be historic (right?), rdfs: can too

23:37:23 <danbri_> Dave, are we sticking w/ the 1999 one for Syntax?

23:37:28 <dajobe> yes

23:38:02 <dajobe> is rdfs changing too much that it needs a new ns?

23:38:05 <danbri_> I think stick w/ RDFS

23:38:08 <DanCon> the WG decided to chuck all the rdfs:ConstraintProperty stuff, yes? that motivates (but doesn't completely mandate) a namespace name change.

23:38:11 <danbri_> we have a new rdfs:member property

23:38:15 <danbri_> True.

23:38:32 <danbri_> Though one might argue that we've simply changed our mind about the existing namespcae

23:39:30 <DanCon> I think the benefit of keeping the existing namespace name (e.g. sesame doesn't have to retool) outweighs the cost (i.e. "where did rdfs:member come from?").

23:39:49 <danbri_> Yeah, me too. Also scares peopel away if we keep jiggling.

23:39:52 <dajobe> but the stuff that dropped?

23:40:07 <DanCon> I think we dropped that stuff on the grounds that nobody wanted it.

23:40:14 <danbri_> Expose content-negotiated CVS history at the namespace?

23:40:24 <DanCon> the only question is: did we give the implementor community enough notice that we dropped it?

23:40:36 <danbri_> content-type: application/xml+version=cruftyold

23:41:11 <danbri_> I think leave namespace intact for next WD, with a candidate new NS RDF doc as an appendix, and flag it in 'status of this doc'.

23:41:21 <DanCon> re ns hot potatoe: here's where push comes to shove: putting an RDF schema at http://www.w3.org/2000/XMLSchema to support datatypes.

23:41:44 <danbri_> That one's cooled down hugely.

23:41:52 <danbri_> Everyone knows we have to be schema pluralists nowadays...

23:42:41 <DanCon> but how many XML Schema tools use Accept: headers? what media type do you use to get the RDF schema?

23:42:48 * danbri_ renames 'Constraints' as 'Domain and Range'

23:42:58 <dajobe> rdf syntax has 12 normative references! (mebbe could make primer, model non-norm)

23:43:35 <DanCon> where in the text does rdf syntax depend on primer?

23:43:39 <dajobe> exactly

23:43:46 <DanCon> how about model?

23:43:47 <dajobe> I just stick it in wrong sec

23:43:59 <danbri_> It doesn't depend on model? What is it a syntax for?

23:44:00 * DanCon and IanJ developed tools for checking that sort of thing in the HTML 4 specs

23:44:06 <danbri_> oooh

23:45:37 * DanCon recalls that Arnaud probably did most of the developing; I, as chair, just played customer, and Ian played user

23:48:10 * danbri_ hacks the constraints bit. Quick read-thru?:

23:48:13 <danbri_> [[

23:48:26 <danbri_> For example, while

23:48:27 <danbri_> an RDF schema can assert that an <code>author</code> property

23:48:27 <danbri_> is used to indicate resources that are members of the class

23:48:27 <danbri_> <code>Person</code>, it does not say whether or how an

23:48:27 <danbri_> application should act in processing that class information.

23:48:27 <danbri_> Different applications will use this information in different ways.

23:48:29 <danbri_> For example, a data checking tools might use this to help discover errors

23:48:31 <danbri_> in some dataset, an interactive editor might suggest appropriate values,

23:48:35 <danbri_> and a reasoning application might use it to infer additional information from

23:48:35 <danbri_> instance data.

23:48:37 <danbri_> ]]

23:49:05 <danbri_> RDF Schema provides a mechanism for describing this information,

23:49:05 <danbri_> but does not say whether or how an application should use it.

23:49:10 <danbri_> ...is the lead in text.

23:49:11 <dajobe> ok

23:49:46 <DanCon> yup; reads well.

23:50:34 <danbri_> this is so much fun

23:53:37 <danbri_> How bout this?:

23:53:39 <danbri_> [[

23:53:42 <danbri_> <p>

23:53:42 <danbri_> This specification introduces an RDF vocabulary for making

23:53:42 <danbri_> statements about the meaningful use of properties and

23:53:42 <danbri_> classes in RDF data. For example, an RDF schema might

23:53:42 <danbri_> describe limitations on the types of values that are appropriate

23:53:45 <danbri_> for some property, or on the classes to which it makes sense

23:53:45 <danbri_> to ascribe such properties.

23:53:48 <danbri_> </p>

23:53:52 <danbri_> ]]

23:54:17 <danbri_> that's slipperyer territory ('meaning'...). happy for now.

23:54:21 <DanCon> "making statements about" seems wordy... "describing"?

23:54:32 <DanCon> i.e. describing properties and classes in RDF data.

23:55:22 <dajobe> would have too describes in 2 sentences then

23:55:35 <dajobe> s/have too/have two/

23:55:40 <DanCon> declaring?

23:55:44 <danbri_> we got a lot of 'describing', but sure.

23:57:14 * DanCon can't think of anything really good, doesn't think what's there is critically bad

23:58:22 <danbri_> good enough for me

23:58:38 <danbri_> first cut will be rough; this is mostly about purging dead and daft bits

23:58:57 * dajobe 's link checking is failing cos of http://blogspace.com/rdf/mimetype which is timing out :)

23:59:06 * DanCon goes back to three-month-rule audit...