Semantic Web Interest Group IRC Chat Logs for 2001-09-20

NOTICE: #rdfig logs are being turned off 2004-12-03. Please switch to the new and shiny #swig channel for Semantic Web Interest Group chat. Change your client to #swig and enjoy the new experience. Or read the latest #swig logs to see what you've been missing :)

00:38:01 * AaronSw waves

01:47:00 <AaronSw> <DanC> what's the python-land convention for naming modules the closely correspond to classes? In modula-3, the module was called foo.m3 and the class was foo.T

01:47:07 <AaronSw> Ahh, that explains ircAsync.py

01:52:17 <AaronSw> <tim2> All CS students -- look, I have just forkedout for the 3 volume Knuth -- I am doing my best ;-)

01:52:17 <AaronSw> <tim2> It has a reference to my mother on p230 of vol 1 (!)

01:52:19 <AaronSw> lol!

01:52:44 <sbp> He he he

02:03:25 <AaronSw> re: urlopen and error messages... spent a long time fighting with that, finally got it to work with help from bijan and deltab

02:03:59 <AaronSw> code I used:

02:04:27 <AaronSw> try:

02:04:38 <AaronSw> urllib2.urlopen(...)

02:04:51 <AaronSw> except: urllib2.HTTPError, e

02:05:06 <AaronSw> rather, except urllib2.HTTPError, e:

02:05:28 <AaronSw> and then there's a bunch of method

02:05:32 <AaronSw> and then there's a bunch of methods on e

02:14:09 * AaronSw wonders about pyunit's applicability to Clean CWM...

02:21:54 * AaronSw wonders if tav and libby know each other...

02:22:54 <tav> we have greeted each other

02:24:22 <AaronSw> <tav`> <jang> I don't know quite where aaron got that from... you _can_ do random sigs in eudora on mac os x like in any other unix

02:24:29 <AaronSw> greeted: noticed that

02:24:40 <AaronSw> I didn't mean to imply Eudora couldn't do random sigs

02:24:48 <AaronSw> merely that jang has a cool web hack ;-)

04:59:02 <AaronSw>http://preview.openbits.org/

04:59:02 <dc_rdfig> A: http://preview.openbits.org/ from AaronSw

04:59:10 <AaronSw> A:|Bitzi Catalog available in RDF

04:59:11 <dc_rdfig> titled item A

04:59:37 <AaronSw> A:Bitzi is the "patron" of [MusicBrainz|http://www.musicbrainz.org] and they're doing some pretty cool stuff with file metadata.

04:59:37 <dc_rdfig> commented item A

05:00:16 <AaronSw> A:Everything is experimental at this point -- I assume feedback is invitied.

05:00:16 <dc_rdfig> commented item A

05:00:33 <AaronSw> A:Ouch, 404 on [the namespace|http://bitzi.com/xmlns/2001/09/10/experimental#]

05:00:34 <dc_rdfig> commented item A

05:00:51 <deltab> uh, it's a namespace... :-)

05:01:02 <AaronSw> A:Otherwise looks pretty good... pretty simple.

05:01:02 <dc_rdfig> commented item A

05:01:25 <AaronSw> namespace: yeah, but if you're going to go to the work of making it an easily accesible URI, you might as well put something there!

05:01:34 <AaronSw> It's always good to document your work.

05:28:34 <deltab> AaronSw: where do you see 404 at http://bitzi.com/xmlns/2001/09/10/experimental# ?

05:29:39 <AaronSw> Hmm, well they clearly meant to say 404, but apparently their server is screwed up... double :-(

05:29:56 * deltab nods

05:29:56 <AaronSw> A:And their 404 message isn't even returned with a 404 status :-(

05:29:56 <dc_rdfig> commented item A

05:31:03 <AaronSw> Boy, ArsDigita's shutting down their free services faster than the market is going down!

05:31:12 <AaronSw> :-( http://www.arsdigita.com/bboard/q-and-a.tcl?topic%5fid=1615

05:33:31 <AaronSw>http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Sep/0273.html

05:33:32 <dc_rdfig> B: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Sep/0273.html from AaronSw

05:33:38 <AaronSw> B:|URI terminology demystified

05:33:38 <dc_rdfig> titled item B

05:33:43 <AaronSw> B:by Dan Connolly

05:33:43 <dc_rdfig> commented item B

05:35:18 <AaronSw> Heh: "the hash mark (aka octothorpe; a fifty-cent word you can have for free ;-)"

05:38:14 <AaronSw> B:Dan [wants a term for absoluteURI-with-optional-fragment-identifier|http://lists.w3.org/Archives/Public/uri/2000Jan/0002], I interpret the absence of such a term to be a polite hint to stop using it in specs

05:38:14 <dc_rdfig> commented item B

05:39:44 <AaronSw> Heh: "Well, remember the asterisk (*) when I said the value of an href/resource attribute is a URI reference? I lied."

05:40:33 <AaronSw> LOL: "What a mess! Are you serious? For a technology so architecturally core to RDF and the Web, that's quite a kludge-tower!"

05:40:33 <dc_rdfig> Label LOL not found.

05:40:47 <AaronSw> B:This message made me laugh three times, so it must be good.

05:40:47 <dc_rdfig> commented item B

09:17:13 <debcon> debcon is now known as DanC_

09:17:43 <DanC_> well, I have to say this win4lin thing really does work.

09:18:07 <DanC_> I got the kinks out of this debian install (X windows, sound, ...) and the win4lin install went pretty smoothly.

09:18:27 <DanC_> hmm... does sound work under win4lin now?

09:19:15 <DanC_> indeed it does

09:20:46 <DanC_> hmm... the windows start-up sound works, but the media player complains about no signal.

09:24:28 <DanC_> hmm... now it's worknig.

09:34:08 <timbl-gone> timbl-gone is now known as t_zzzz

09:52:58 * dajobe wonders if somebody could update http://www.w3.org/2000/07/rdf.xsd to use XML schema REC

10:09:14 * dajobe struggles with updating above XSD

13:03:19 <dajobe>http://ilrt.org/discovery/2001/09/rdf-xml-schema/

13:03:19 <dc_rdfig> C: http://ilrt.org/discovery/2001/09/rdf-xml-schema/ from dajobe

13:03:26 <dajobe> C:|XML Schema for RDF/XML

13:03:26 <dc_rdfig> titled item C

13:03:40 <dajobe> C:current state - broken

13:03:40 <dc_rdfig> commented item C

13:34:40 <bwm> I'm having trouble accessing Pat's model theory for RDF at http://www.coginst.uwf.edu/~phayes/RDF%20MT-currentdraft.html

13:35:01 <bwm> Does anyone have a cached copy we can post to the archive?

13:36:32 <dajobe> hold on, I'll look

13:38:26 <dajobe> stil looking ... big cache ...

13:39:12 <bwm> thanks dajobe

13:39:14 <dajobe> got a version from 22:16 last night?

13:39:23 <bwm> That will be fine.

13:39:42 <bwm> Could you just mail it to me please?

13:39:50 <dajobe> yes

13:39:59 <bwm> Ta muckle.

13:40:58 <dajobe> done

13:41:20 <dajobe> ugh, might be from 18th, not 19th

13:42:30 <bwm> So long as it was post the announcement of the candidate WD, it will do.

13:43:20 <bwm> I take it you too cannot currently access the URI pat gave?

13:43:44 <dajobe> bwm: yes

13:43:49 <dajobe> bwm: I cannot

13:44:04 <bwm> k. ta.

13:44:33 * AaronSw waves

13:59:19 * tim wonders whether DanC worked until 5am or until 5am last night

14:16:10 <AaronSw>http://www.coginst.uwf.edu/~phayes/RDF%20MT-currentdraft.html

14:16:10 <dc_rdfig> D: http://www.coginst.uwf.edu/~phayes/RDF%20MT-currentdraft.html from AaronSw

14:16:19 <AaronSw> D:|RDF Model Theory Working Draft *to be*

14:16:19 <dc_rdfig> titled item D

14:16:19 <dajobe> D:|don't click here, site offline

14:16:20 <dc_rdfig> titled item D

14:16:24 <AaronSw> D:|RDF Model Theory Working Draft *to be*

14:16:24 <dc_rdfig> titled item D

14:16:34 <AaronSw> give me a sec...

14:16:47 <AaronSw> D:cached copy [in www-archive|http://lists.w3.org/Archives/Public/www-archive/2001Sep/att-0051/01-html_filename_x.html]

14:16:47 <dc_rdfig> commented item D

14:17:01 <AaronSw> D:because site is currently down

14:17:02 <dc_rdfig> commented item D

14:17:35 <dajobe> should have changed the status before archiving

14:18:00 <AaronSw> yeah, oh well

14:24:16 <dajobe> oops, forgot to grab picture from cache too

14:31:51 <sandro> I think bNodes should have URI-Reference names, too, so their encodings are not sensitive to document boundaries.

14:32:46 <sandro> I'm not sure if they should be distinguished from normal nodes entirely out of band, or by using some other part of the URI-Reference namespace.

14:37:27 <AaronSw> Hmm, sandro and Sean seem to agree on this matter.

14:42:45 <sandro> I get the feeling Sean and I agree on a lot, but I'm not really keeping track. :-)

14:42:55 <AaronSw> :-)

14:43:51 <AaronSw> Hmm, I'll still owe you a response to your identification thing... guess i'll do that now

14:47:05 <sandro> I don't remember where we left it.

14:47:21 <AaronSw> From 20001-09-02:

14:47:22 <AaronSw> [19:53:37] AaronSw

14:47:22 <AaronSw> Sandro, re: your recent email on "The RDF Identification Problem" my only point was that you made a misleading (and I think wrong statement) about HTTP URIs -- I don't disagree about the example in the RDF spec

14:47:23 <AaronSw> [19:53:53] AaronSw

14:47:23 <AaronSw> And I totally understand about the RDF identification problem

14:47:23 <AaronSw> [19:54:31] AaronSw

14:47:25 <AaronSw> However, if you insist that "RFC 2616 says that HTTP URIs denote "network data objects or services" (not people)." then I have a real problem.

14:50:38 <sandro> There are perhaps two separate issues: (1) what does 2616 say? and (2) what is agreed upon by a vast majority of people with an opinion on the subject. A possible third issue is what certain key people or documents once said or thought about the subject.

14:52:52 <AaronSw> Well, yes, I suppose that is interesting... but I'd sort of like to stay in compliance with the spec. ;-)

14:54:37 <sandro> I'd like to stay in compliance with both, if that's possible!

14:54:58 <sandro> And I think it is, generally.

14:55:01 <AaronSw> Indeed.

14:55:28 <AaronSw> So my question is whether you see this as conflicting with the spec (I see it as common practice, so I hope not).

14:55:47 <sandro> I don't think identifying a person with HTTP URI is in compliance with either the spec or popular opinion. Do you disagree?

14:56:47 <AaronSw> Yes.

14:57:04 <AaronSw> I think it is in compliance with both.

14:57:50 <sandro> I don't feel ready to argue about 2616 right now. Can we try popular opinion instead (first, at least)?

14:59:10 <sandro> A little experiment: type "aaronsw" into Google and press I Feel Lucky, and you get a web page with the address http://swartzfam.com/aaron/

14:59:10 <AaronSw> Fine with me.

14:59:49 <AaronSw> I do not argue that that web page is me.

15:00:40 <AaronSw> But, I do insist that: http://logicerror.com/myWeavingTheWeb identifies my book.

15:01:46 <sandro> So if I click on that and a get a 404, what does that mean?

15:02:23 <AaronSw> It means that my server cannot locate any informaiton about my book.

15:02:49 <AaronSw> Or at least not enough to send you an HTTP Response on the subject.

15:05:45 <AaronSw> I found such an interpretation was apparent from the care in terminology uses by 2616.

15:06:15 <AaronSw> From 2616:

15:06:16 <sandro> The string "http://logicerror.com/myWeavingTheWeb" is 38 characters long. How long is http://logicerror.com/myWeavingTheWeb?

15:06:39 <AaronSw> Umm, what do you mean by the latter?

15:06:51 <AaronSw> How long is the URI? How long is the thing the URI denotes?

15:07:25 <sandro> How many characters long is the thing called "http://logicerror.com/myWeavingTheWeb", in rough terms? About 100 characters? Or many thousands of characters?

15:08:19 <AaronSw> the thing called "http://logicerror.com/myWeavingTheWeb" is a physical book, and I do not generally measure physical things in character... how long are the contents of the book?

15:09:03 <sandro> Okay -- how long are the contents of the thing called "http://logicerror.com/myWeavingTheWeb" ?

15:09:26 <AaronSw> The response received from logicerror.com about that URI at 2001-09-20 10AM is ~100 characters.

15:09:44 <AaronSw> the contents of the thing called "http://logicerror.com/myWeavingTheWeb" are many many characters long

15:09:59 <sandro> My point is that it's completely reasonable, and perhaps what most people would do, to say the contents of the thing called "http://logicerror.com/myWeavingTheWeb" are the string "Aaron's Copy of Weaving the Web

15:09:59 <sandro> The URI: http://logicerror.com/myWeavingTheWeb identifies Aaron Swartz's copy of Weaving the Web. This copy of

15:09:59 <sandro> the book can be identified by the fact that it has this URI written in the front cover.

15:09:59 <sandro> Links to: Aaron Swartz | Weaving the Web

15:10:00 <sandro> Last Modified: 2001-02-04 14:49:49

15:10:00 <AaronSw> I do not feel in the mood to count them right now ;-)

15:10:02 <sandro> Powered by Blogspace, an Aaron Swartz project. Email the webmaster with problems."

15:10:32 <AaronSw> I think that's rather unreasonable -- you left out all of the HTML formatting/

15:11:08 <AaronSw> (please don't paste it, however)

15:11:43 <sandro> As you would have in counting the characters in the book..... The point is, there are some question one can ask about the thing called that which apply to both a book and a web page.

15:11:48 <AaronSw> People speak loosely about URIs, yes

15:12:50 <sandro> Not really, no. They speak about URIs as things you use in interacting with browsers, to get the browser to do something. Web Addresses. places you can make your browser "go" -- which really means things you can make your browser do in a fairly reproducable manner.

15:12:58 * azaroth butts in his 2p - http URIs, IMO, should only represent web pages. Otherwise how do you distinguish between the triples: http://logicerror.com/myWeavingTheWeb, linksTo, http://someotherurl/ and http://logicerror.com/myWeavingTheWeb, isOwnedBy, (foaf)AaronSw ?

15:13:29 <AaronSw> azaroth, you say the correct thing, which is:

15:14:03 <AaronSw> [ a :Response; :for <http://logicerror.com/myWeavingTheWeb> ; :at "date here" ] :linksTo <fooz> .

15:14:20 <AaronSw> sandro, if that were literally true then web pages could never change.

15:14:36 <AaronSw> People think of URIs as conceptual entities which they can use to do things.

15:14:44 <sandro> I said "fairly reproducable".

15:14:46 <AaronSw> One of those things is typing them into a browser and getting something back.

15:14:56 <AaronSw> But there are other things.

15:15:02 <sandro> The conceptual entities are bits of browser functionality, not things like books and people.

15:16:10 <AaronSw> But a changing, growing web page is not a piece of browser functionality!

15:16:23 <sandro> Unfortunately, I need to step out for about two hours.

15:16:58 * azaroth fails to grok, but it was an example rather than something specific. Z3950 URLs wouldn't represent the book in the library, just the record of it's existence. Otherwise when you destroy a book, the Z URI becomes non attributed ? (If I understand the conversation)

15:17:38 <AaronSw> Well, it depends on how Z3950 URLs are/would be defined -- likely they would be defined as identifiying records, I would think.

15:17:50 * lasDesk1 skims the conversation, decides that she's unsure of whether it's reasonable to say that the URI http://logicerror.com/myWeavingTheWeb refers to a book.

15:17:54 <lasDesk1> lasDesk1 is now known as las

15:17:56 <sandro> A "web page" is an experience people get when they "go to" a "web address".

15:18:15 <AaronSw> Hmm. An experience?

15:18:22 <sandro> or not "experience" but a virtual-thing they interact with on their computer.

15:18:41 <AaronSw> The HTTP spec is really the only one I know of that is careful about saying that HTTP URIs do not refer to bags of bits.

15:19:06 <azaroth> You could have a Z URI refer to anything. For example, I have a Z database of trading cards. IMO z3950://gondolin.hist.liv.ac.uk:210/search/l5r?... doesn't refer to my copy of the card, but just the record thereof.

15:19:10 <sandro> gotta run. sorry.

15:19:17 <AaronSw> bye, sandro

15:19:25 <AaronSw> we will continue later, i suppose

15:19:43 * las waves at sandro, feels she's wandered in just as an interesting conversation was breaking up....

15:20:31 <AaronSw> Appears so. :-(

15:20:55 * las hopes that she'll be around when next Aaron and Sandro decide to be so engaging....

15:21:03 <AaronSw> What do you think http://logicerror.com/myWeavingTheWeb refers to, las?

15:21:29 <azaroth> Why don't you just have something like: subjectOf(http://logicerror.com.../) ?

15:21:44 <azaroth> Which is unambiguous.

15:22:42 * azaroth speaks from outsider's standpoint obviously as I haven't had time yet to read over everything.

15:22:50 <AaronSw> Well so is, representationOf(...), etc.

15:23:01 <AaronSw> And it's more compatible with various specs.

15:23:21 <las> Aaron, I'm going to answer the question thinking out loud, so don't hold me to these positions until I decide :o)

15:23:21 <AaronSw> I tend to use [ a :Person ; :homepage <foo> ] quite a bit, however.

15:23:27 <AaronSw> :-)

15:23:45 <las> http://logicerror.com/myWeavingTheWeb is just a name; in some sense, it can refer to anything.

15:23:45 <dc_rdfig> E: http://logicerror.com/myWeavingTheWeb from las

15:23:59 <AaronSw> E:What is this URI?

15:23:59 <dc_rdfig> commented item E

15:24:05 <las> One way to resolve it is to say, it's a URI, and specifically an http: uri.

15:24:17 <AaronSw> Well, that's obviously true?

15:24:19 <las> So then it takes its meaning from the rules governing http.

15:24:22 <AaronSw> err s/?/./

15:24:33 <las> No, it's not obviously true.

15:24:58 <las> From a philosophical standpoint, it's reasonable to say that URI rules, and http rules in particular, have no bearing on the meaning of the symbol.

15:25:04 <AaronSw> Oh, are you speaking of the string "http://logicerror.com/myWeavingTheWeb" and not the URI...

15:25:16 <las> Perhaps.

15:25:58 <las> To say that (which I think I agree with -- that calling it a URI implies tying it to certain interpretation rules) implies buying into a particular framework which isn't necessary.

15:26:15 * las wonders whether *anyone* will ever be able to parse that sentence.

15:26:58 <las> So if we go with the URI interpretation, then http://logicerror.com/myWeavingTheWeb is a web page (i.e., it's what is returned by an http get on that machine, etc.)

15:27:08 <las> Which means it is *not* a book.

15:27:09 <AaronSw> ummm.... no

15:27:14 <AaronSw> Who says that?

15:27:16 <las> (At least not according to my browser)

15:27:24 <las> ummm....no to which?

15:27:46 <AaronSw> both, I suppose -- since when did your browser define the meaning of a URI?

15:27:58 <AaronSw> Does that mean no page can change after your browser visits it?

15:28:19 <las> My browser claims that http://logicerror.com/myWeavingTheWeb is a web page that makes some assertions (in English, at least), including assertions about the meaning of the URI.

15:29:05 <las> "since when did your browser define the meaning of a URI?" -- well, my understanding is that the way to find out what an http: URI means is to do a get on it, and my browser is a convenient way of doing that.

15:29:28 * azaroth thinks the distinction is the one between URL and URI?

15:29:32 <las> "Does that mean no page can change after your browser visits it?" -- No, because a later http get might get something else.

15:29:52 <las> URL/URI -- maybe, but then I need to understand what the URI rules are.

15:30:30 <las> ...as opposed to URL.

15:30:38 <azaroth> As far as I understand it, a URI is an unchanging URL. That is, once a URI has been assigned it will always refer to the same object. (please correct me :) )

15:30:58 <AaronSw> "the way to find out what an http: URI means is to do a get on it" - Where does this assumption come from? It is not in the HTTP spec.

15:31:11 <las> Yes, but even if that's so (URI = same object), that object can be time-varying. Or do you mean an unchanging object?

15:31:37 <las> E.g., "the cnn webpage" has different content each day. Does it have a single URI? URL?

15:31:43 * azaroth would like to read up on it again before commenting ;)

15:33:35 <AaronSw>http://www.ietf.org/internet-drafts/draft-mealling-uri-ig-00.txt

15:33:35 <dc_rdfig> F: http://www.ietf.org/internet-drafts/draft-mealling-uri-ig-00.txt from AaronSw

15:33:35 <las> Http get -- maybe I just made it up. It's what I'd understood, but perhaps there is some Offical Way to know what an http URI means? I'd understood from various comments that (1) the protocol (e.g., http) got to determine how to resolve the ur(l/i) and (2) that resolving an http url (i?) was accomplished by doing a get.

15:33:47 <AaronSw> F:|URI Schemes and URN Namespaces

15:33:47 <dc_rdfig> titled item F

15:33:57 <AaronSw> F:Clarifies differences between URNs and URIs.

15:33:57 <dc_rdfig> commented item F

15:34:08 <jang> there is a difference between the notion of denotation and dereferencing (from a web point of view)

15:34:09 <AaronSw> F:New version should be coming soon.

15:34:09 <dc_rdfig> commented item F

15:34:17 * las thanks Aaron for the URl, goes off ot look briefly.

15:34:36 <las> JanG, yes, there is a difference between dereferencng/resolving and denoting.

15:34:39 <jang> and there is a natural shorthand, that "the web page at http://foo.bar" is denoted by "http://foo.bar"

15:35:00 <jang> however, this shorthand confuses people

15:35:02 <AaronSw> "perhaps there is some Offical Way to know what an http URI means" - ask the person who created it, or judge from the contexts of its use, are the only methods, I believe

15:35:04 <las> So I think part of the question is, what if any relationship exists between the wto.

15:35:20 <jang> eg, who is the dc.author of http://foo.bar/bramslullaby.wav ?

15:35:28 <jang> is it me (who ripped it shamelessly)?

15:35:32 <jang> is it brahms?

15:35:46 <AaronSw> s/dc:author/dc:creator/

15:35:47 <las> It's actually not just a shorthand, in the sense that there's a functional/operational connection as well.

15:36:03 <AaronSw> Only in a certain context.

15:36:03 <jang> for many things, poems, plays, recordings, the difference between a work and an instantiation of that work (a rendering) is obious

15:36:20 <las> "who is the dc.author": It depends what the UR(L/I) refers to, right?

15:36:24 <AaronSw> If I type a URI into Google, I get a very different page than when I type it into my browser.

15:36:43 <jang> however, that distinction is blurred by the notational convenience of talking about a URI denoting a website AND dereferencing to it

15:36:49 <las> google/browser: Yes, but Google is not abiding by the http rules

15:36:56 <jang> particularly since you can dereference the same URL and get different documents

15:37:06 <jang> eg, ones in english, spanish, wav files, jpegs, etc

15:37:39 <AaronSw> HTTP rules? HTTP specifies no rules on the use of URIs...

15:37:49 <AaronSw> It specifies one way of accessing them, yes.

15:37:55 <las> But separating those two notions -- the referent of a URI and its denotation -- requires having different ways of expressing the two.

15:37:55 <AaronSw> but provides room for proxies, etc.

15:39:28 <las> http specifies get-ability and dereferencing an http url according to the http protocol is the same as doing a get on that url. buying into interpreting the string as a url means letting the protocol dictate what meaning to accord to the thing, at least in the meaning=referent sense. urls don't have (qua urls) denotations.

15:39:34 <jang> "the" referent of a URI isn't necessarily well-defined, either

15:39:41 <las> there's an open questoin as to whether uris have denotations.

15:39:49 * las hates the sticky keys on her keyboard....

15:40:13 <las> I didn't say "the referent of a uri", I said

15:40:25 <jang> well, RDF has a proposed MT (nearly) that uses the notion of a denotation for a URI labelling a node in an RDF graph

15:40:46 <jang> it's a shame RDF uses URIs I suppose :-/

17:18:10 * AaronSw is trying to set up webcam -- i look like something out of an AI trailer!

17:18:48 <AaronSw> his email is real.... but he is not.

17:22:14 * rwenning waves

17:22:19 <rwenning> /nick danbri

17:22:22 <rwenning> ugh

17:22:26 <rwenning> rwenning is now known as danbri

17:22:28 <AaronSw> ooh!

17:22:30 <AaronSw> hi danbri!

17:22:35 <danbri> heya

17:22:40 <AaronSw> whose computer didya steal?

17:23:05 <danbri> I'm borrowing Rigo's login from Bert's desk (having accidentally left my laptop at home this morning).

17:23:22 <AaronSw> How is France?

17:23:57 <danbri> I'm a big Sophia fan now. I think there must've been an MIT conspiracy to keep the existence of this place hushed up ;-)

17:24:11 <danbri> And it only cos me 30ukp to fly to Nice from Bristol!

17:24:16 <AaronSw> heh

17:24:31 <AaronSw> How close is the next Technical Plenary to Sophia?

17:24:46 <danbri> Close, I believe. Not sure of details yet.

17:25:11 <danbri> Seen Libby around?

17:25:20 <dajobe> libby gone home

17:25:30 <AaronSw> she left about an hour or so ago

17:25:39 <danbri> the slacker!

17:26:26 * azaroth waves.

17:27:10 <danbri> hi azaroth

17:32:53 <dajobe> totally unrelated: http://www.manchesteronline.co.uk/business/ebusiness/content.cfm?story=114236

17:34:27 <danbri> NI invented the SW?

17:35:00 <dajobe> more quotes on http://www.networkinference.com/news.asp?ID=33&menu=3

17:36:08 <jhendler> they're betting that people will actually turn on the reasoners if they use DAML :->

17:37:05 <danbri> I'd be more inclined to believe the hype if the search script on their website appeared to be more than an SQL-based .asp tool...

17:37:24 <jhendler> but they don't claim to have rights to anything but the particular methodology for fast reasoning,and the implementaion outside of the "for free" that Ian gives away - so don't see this as a stmubling blck to SW

17:38:14 <danbri> yep

17:40:48 <jhendler> besides, I'd be the last to complain if more small businesses join the W3C and support the SW efforts

17:41:21 <danbri> i'm not complaining, we need folk excited about this stuff; just also need to keep a lid on the hype sometimes...

17:41:52 <jhendler> hmm, a certain amount of hype is okay - just need to keep it managable :->

17:49:59 * danbri heads off; back tommorrow (on and off...)

18:19:12 <jhendler>http://www.itl.nist.gov/div895/ebook2001/index.html

18:19:12 <dc_rdfig> G: http://www.itl.nist.gov/div895/ebook2001/index.html from jhendler

18:19:24 <jhendler> G:|NIST Electronic Books Conference

18:19:24 <dc_rdfig> titled item G

18:20:10 <jhendler> G: will have a Sem Web presentation - Tues 3PM - part of a session on "Standards and Interoperability: Setting the Standard

18:20:10 <dc_rdfig> commented item G

18:20:16 <jhendler> G: "

18:20:17 <dc_rdfig> commented item G

18:38:53 <jhendler>http://CEUR-WS.org/Vol-52/

18:38:53 <dc_rdfig> H: http://CEUR-WS.org/Vol-52/ from jhendler

18:39:19 <jhendler> H:| On-line proceedings of workshop on "ontologies for agents"

18:39:19 <dc_rdfig> titled item H

18:39:28 <jhendler> H: workshop at AAAI-2002

18:39:28 <dc_rdfig> commented item H

18:39:40 <jhendler> H: some interesting papers/presentations

18:39:40 <dc_rdfig> commented item H

18:48:05 AaronSw, you still around and have that handy mac illustrating tool?

18:48:19 <AaronSw> yes to both

18:48:35 * em wonders if you can provide an alternate image to Pat's RDF Model Theory document

18:49:28 <AaronSw> I couldn't understand that picture...

18:50:25 hmmm...

18:52:06 * AaronSw takes a shot at it...

19:02:25 <AaronSw> Do you want it in color, em?

19:05:29 color yes, provided BW printing can still degrade appropriatlye and by understandable

19:05:38 <AaronSw> what color is a universe?

19:08:25 <AaronSw> light orange

19:12:08 <AaronSw> D:here's a [nice new OmniGraffle figure|http://blogspace.com/rdf/modeltheory.png] for the draft

19:12:08 <dc_rdfig> commented item D

19:16:07 thanks AaronSw, this looks much better :)

19:16:56 * em labels this "RDF meets Dr. Suess"

19:17:03 <AaronSw> Dr. Seuss?

19:17:13 <AaronSw> The color scheme?

19:17:14 "thing1 and thing2"

19:17:24 <AaronSw> ahh, yes.

19:18:41 <AaronSw> I think IEXT and IR need better names... harrumph and ballwump?

19:19:51 <AaronSw> Want to mail a linkn to the list?

19:20:26 yes, please email this

19:21:35 thanks again

19:21:38 <AaronSw> sure

19:23:19 <AaronSw> this graph making stuff is fun! :-0

19:23:35 :)

19:25:34 <AaronSw> I already got the Omni guys to support DOT so I could read the 02pd stuff, and they said that they might have RDF export in a future version

19:28:28 excellent!

19:29:03 * em sounds like even a more compelling reason to move to mac osx :)

19:31:23 <dajobe> can you make an svg version?

19:32:00 <AaronSw> sadly no SVG... PNG GIF JPG TIFF PDF EPS

19:32:12 <dajobe> hmph

19:32:35 <dajobe> it isn't a vector program?

19:32:46 <AaronSw> yeah...

19:32:58 <AaronSw> """ Jim mentioned OmniGraffle, so I decided to try it out. This is what apps are supposed to feel like. I can't compare it to any similar tools, but it looks and feels beautiful. The only drawback is the lack of SVG support that should be mandatory in any vector-based app; I assume this just didn't make the first beta.""" - wmf, http://wmf.editthispage.com/2000/11/17

19:33:44 * DanC considers opening the fragments discussion with AaronSw

19:34:19 * AaronSw wonders if DanC has further developed his "way out" of the problem

19:35:08 <DanC> way out?

19:35:24 <DanC> I forget what the problem was.

19:35:47 <DanC> why on earth would one want to get rid of <rdf:Description rdf:about="http://example/vocab#term"/> ?

19:36:07 <AaronSw> because http://example/vocab#term is not a resource in any defined spec

19:36:21 <AaronSw> it isn't compatible with extant web architecture, or the HTTP protocol

19:36:27 <DanC> not a resource: it is in the RDF model theory spec

19:36:45 <AaronSw> hmm, perhaps so, but that isn't published yet

19:36:48 <DanC> it's consistent with web architecture back to '89/'90; RFC2396 is an anomaly

19:37:39 <AaronSw> i need to follow up on that claim, but roy presented a good case why it was changed

19:39:42 <DanC> evidence of fragments being in web arch since day 1: [[As an example we have the URL: "/TheProject.html#news" ]] -- http://www.w3.org/Library/src/HTParse.html

19:40:38 <AaronSw> I don't argue that relative URIs and fragment IDs don't exist -- merely that 2396 and 2616 don't treat them as resources

19:40:47 <AaronSw> and that's a lot of code to repair if you want to change things

19:41:12 <DanC> no code changes are involved. they're already treated as resources.

19:41:30 <DanC> e.g. browsers turn links purple or not based on resources such as "section 1 of the HTML spec"

19:43:15 <AaronSw> yes, but my HTTP server cannot do access control on such resources

19:44:11 <DanC> true; nor can it do access control on resources such as the string "abc" (aka data:,abc)

19:44:49 <AaronSw> yes, but there's no expectation for it to

19:45:11 <DanC> nor is there any expectation for it to do access control on "section 1 of the HTML spec"

19:45:20 <AaronSw> I can query an HTTP proxy about a "data" URI, but not about fragment

19:46:01 <AaronSw> expectation: i would think there would be, since if it was a resource, then someone created it and gave control of it to their HTTP server -- i'd expect the HTTP server to be able to do all the normal things it does, like access control

19:46:47 <DanC> well, I don't know where you got that expectation. Some resources are smaller than the fling-it-over-the-network size.

19:48:03 <AaronSw> and what about the issues of fragment identifiers and conneg?

19:48:28 <DanC> hmm... doing some archaology, it seems "resource" is a relatively modern term; the original terms were "document" and "anchor". (cf http://dev.w3.org/cvsweb/libwww/Library/src/HTAnchor.html?rev=2.61&content-type=text/x-cvsweb-markup)

19:48:41 <DanC> what about fragment identifiers and conneg?

19:49:20 <DanC> "It would actually be most useful if we had persistent anchors so that an

19:49:20 <DanC> application could build up a higher knowledge about the Web topology."

19:49:37 <DanC> ^early bits of RDF design, no?

19:49:41 <AaronSw> from HTAnchor?

19:49:47 <DanC> yes

19:50:43 <AaronSw> rdf design: i'm not quite sure -- it's difficult to follow the different terminology

19:51:10 <AaronSw> How come some resources appear to be there in some MIME type responses, and illegal in other responses?

19:51:33 <DanC> ?

19:52:21 <AaronSw> like http://example.org/symphony#0011:0020 is a valid MP300 URI, but not when I get back an HTML version of the symphony.

19:53:22 <DanC> rdf design: the original web browser had datastrcutures just like cwm's Thing and Fragment; they were HTAnchor and HTDocument. they were connected with URI-labelled-links, just like an RDF graph. RDF can be seen as a persistence format for that structure.

19:54:12 <DanC> it seems perfectly reasonable for http://example.org/symphony#0011:0020 to correspond to some section of an HTML representation of the symphony.

19:54:41 <AaronSw> err, except that it's an illegal HTML ID, no?

19:55:39 <DanC> yes, colons aren't allowed in SGML/XML names. That seems like an orthogonal detail, not an architectural issue.

19:57:08 <DanC> i.e. tomorrow, we could revised the HTML spec to add a new niftyID attribute that doesn't have the no-colons restriction.

19:58:17 <AaronSw> so how are the semantics of these fragment identifiers defined? they're not defined by the mime type they're associated with?

19:58:41 <DanC> (the no-colons restriction is not very popular; witness ebXML folks using URIs for identifiers and complaining that XML Schema doesn't allow colons in IDs... never mind /'s! The new key/keyref stuff has no such silly lexical restrictions. good riddance to ID/IDREF)

19:58:54 <DanC> how are semantics defined: see the model theory spec.

19:59:36 <AaronSw> So these RDF identifiers are not related to those old things called URIs?

19:59:41 <DanC> ?

20:00:35 <DanC> here's an informal account of the semantics of fragments: foo#bar denotes whatever bar denotes in a 200 OK reply (or equivalent) from foo.

20:01:10 <AaronSw> right... the model theory doesn't say that... it doesn't say anything about URIs other than they make nice names

20:01:19 <DanC> exactly.

20:01:51 <AaronSw> So isn't that an incompatibility between RDF and the URI spec?

20:01:54 <DanC> no.

20:02:08 <AaronSw> The URI spec says they have semantics from a data retrieval, RDF says they can be anything.

20:02:55 <DanC> I think you're looking for more stuff than any spec needs to define. What are the semantics of a dollar? The meaning of a dollar is an emergent property from the economy as a whole, not something anybody can formally define. (there is a formal grounding in the U.S. tax system, but it's almost incidental)

20:03:27 <DanC> "X can be anything" is not incosistent with "X must be blue." It's different, but not inconsistent.

20:03:38 <AaronSw> Thousands of RDF users have been using properties with '#' thinking they meant something like Bag or Alternative collections... now you tell me they're simply a bunch of XML infoset items?

20:03:53 <DanC> ?

20:03:54 <AaronSw> s/properties/terms/

20:04:18 <AaronSw> I thought 'http://www.w3.org/1999/02/22-rdf-syntax-ns#Seq' had the semantics of a Sequence collection, but when I do an HTTP GET...

20:04:31 <AaronSw> I find that it really is an XML infoset!

20:04:52 <DanC> huh? 'Seq' in that document denotes the class of RDF sequences.

20:06:20 <AaronSw> how did you get that from a data retrieval on the URI?

20:06:53 <AaronSw> URI RFC:

20:06:56 <AaronSw> The semantics of a fragment identifier is a property of the data

20:06:56 <AaronSw> resulting from a retrieval action, regardless of the type of URI used

20:06:56 <AaronSw> in the reference. Therefore, the format and interpretation of

20:06:56 <AaronSw> fragment identifiers is dependent on the media type [RFC2046] of the

20:06:56 <AaronSw> retrieval result.

20:07:19 <AaronSw> I got back an XML document, which seems to say it's a bunch of infoset items, not some class of Bags...

20:10:50 <DanC> yes, you got back an XML document... one written in RDF (it says so in the 1st tag)... so you check the RDF spec, and see that 'Seq' denotes a class of things.

20:11:30 <AaronSw> Wow... that's an incredible interpretation of things!

20:11:41 <DanC> seems plain as day, no?

20:12:12 <AaronSw> No... it's the most incredible spec-crossing interpretation I've seen in a long time...

20:12:13 <DanC> i.e. I just followed my nose.

20:12:49 <DanC> what spec crossing? I just followed my nose from the URI spec to the HTTP spec to the MIME spec to the XML/namespaces spec to the RDF spec.

20:13:00 <AaronSw> heh... exactly :-)

20:13:41 <DanC> ok... so that's what you meant by spec-crossing: following ones nose. What's incredible about that?

20:14:51 <AaronSw> Well, I suppose you might find it if your nose already had its destination in mind, but I'd never have guessed that's what the specs meant from a blank slate.

20:15:53 <AaronSw> Does xpointer explain this?

20:16:41 <AaronSw> XPointer: "Given a resource and an XPointer, an XPointer processor yields the subresource identified by the XPointer, expressed as a location-set, or else an error as described in 3.4 Classes of XPointer Errors "

20:16:53 <AaronSw> I'm not sure how you got from a location-set to the RDF spec.

20:18:53 <DanC> "given a resource and an XPointer""??? ugh. that's incoherent.

20:19:40 <DanC> I'm not sure XPointer is relevant to the case of an RDF document.

20:19:58 <AaronSw> Well, the document in question is returned as text/xml.

20:20:42 <DanC> and...?

20:21:18 <AaronSw> And the text/xml MIME type definition cites XPointer as the definition for its fragment identifiers.

20:21:51 <AaronSw> in following your nose you apparently jumed across few buildings... ;-)

20:22:09 * dajobe sticks PatH MT in RCS to track changes...

20:24:31 <DanC> are you sure? I suggest you look again at the text/xml MIME type spec.

20:25:11 <DanC> XPointer is in Candidate Rec, with a big SOTD saying "are we sure we want to use this as the definition for fragids in all XML docs?"

20:25:31 <AaronSw> SOTD?

20:25:38 <DanC> SOTD=status of this document

20:26:14 <DanC> "As of today, no established specifications define identifiers for XML

20:26:15 <DanC> media types." -- ftp://ftp.isi.edu/in-notes/rfc3023.txt

20:26:33 <AaronSw> and then a big but...

20:27:22 <DanC> yes, the hypothesis of the XPointer CR is that it'll work for all XML docs. But I have my doubts.

20:28:24 <AaronSw> So your nose is stopped at rfc3023 where the trail goes dead?

20:28:38 <DanC> it's also possible that there are actually two (or more?) different ways of pointing in the Web. I think TimBL mentioned at the Feb Cambridge RDF IG ftf about *addr vs **addr or some such... I had the same thoughts in reply to a message from LarryM in uri@w3.org

20:29:11 <DanC> when I got to rfc3023, I mixed in a healthy dose of namespaces and self-describing documents.

20:30:07 <AaronSw> trippy

20:30:21 <AaronSw> **addr?

20:30:26 <DanC> different kinds of pointing: that suggestion also comes up whenever TimBL looks at the idea of looking at XLink as an RDF syntax.

20:30:54 <AaronSw> oh, i see -- dereferencing the pointer

20:30:57 <DanC> *addr vs **addr <-- harking to C pointer syntax.

20:31:27 <sandro>http://www.w3.org/2001/03/identification-problem/

20:31:27 <dc_rdfig> I: http://www.w3.org/2001/03/identification-problem/ from sandro

20:31:49 <AaronSw> I:|How We Identify Things (on the Semantic Web)

20:31:49 <dc_rdfig> titled item I

20:31:50 <sandro> I:Sandro makes another attempt to gather up the threads

20:31:50 <dc_rdfig> commented item I

20:32:14 <AaronSw> I:Seems to be the topic of the day...even links to the conversation this morning

20:32:14 <dc_rdfig> commented item I

20:32:17 <DanC> different pointing: one idea is that href(addr) is sorta like *addr and resource(addr) is more like **addr. I don't think the analogy is quite on target... but I'm open to the possibility that the RDF model theory doesn't apply to XLinks.

20:32:38 <AaronSw> Woohoo! I get to be W3C Director in 2051! My dreams have been realized.

20:32:59 <AaronSw> But shoot... nobody trusts me.

20:33:22 <sandro> The not-trusting part was purely hypothetical. :-)

20:33:40 <AaronSw> Oh, good.

20:34:47 <AaronSw> Hmm, here's an interesting analogy:

20:35:14 <AaronSw> I see the anonymous nodes problem as analagous to the pointing-to-a-paragraph-in-an-html-doc-that-doesn't-have-an-id problem.

20:37:11 * DanC doesn't get the 2051 joke

20:37:27 * DanC sees it now

20:37:59 * sandro will add any quick revisions anyone suggests

20:38:38 <DanC> nifty: [[ SELECT data

20:38:38 <DanC> FROM giant_table_of_all_data_on_the_internet]]

20:38:45 <DanC> -- http://www.w3.org/2001/03/identification-problem/opacity-spectrum.html

20:39:05 <AaronSw> sanro, the "existential variable" technique section seems to have a typo...

20:39:18 <AaronSw> uniqueProperty. In n3 one can write "[ foaf:mbox ]",

20:39:27 <AaronSw> forgot to quote <?

20:39:33 <sandro> Ah!

20:40:21 <sandro> Fixed.

20:41:27 <AaronSw> [ foaf:homepage "http://www.drum.org/~natasha/pets/taiko.html" ] seems to be a misues of foaf:homepage (assuming the usual binding)

20:41:39 <AaronSw> change the "s into <>s

20:42:23 <sandro> Oh, okay -- I'll change the foaf to foaf2 then.

20:42:39 <AaronSw> why no <>s?

20:42:41 <DanC> I'd suggest using contact:homePageAddress in stead

20:42:50 <AaronSw> oh -- i see the issue

20:43:06 <AaronSw> hmm, perhaps for my RDF implementation I should just use anonymous resources with log:uri properties...

20:43:12 <sandro> Because < > get us back to denotation question, yeah.

20:43:17 <AaronSw> right

20:43:26 <AaronSw> log:uri would certainly simplify node merging

20:43:39 <JHendler_> JHendler_ is now known as Jah-class

20:43:47 <sandro> Thanks DanC, fixed.

20:44:11 <DanC> I used to rail on RDF users that hid their URIs in strings... then I started thinking about email processing... it's very clear that "connolly@w3.org" != "danc@w3.org" but much less so that <mailto:connolly@w3.org> != <mailto:danc@w3.org>. esp from a logical perspective

20:44:25 <AaronSw> AaronSw has changed the topic to: semanticWebChat: Another edition of classtime in #rdfig - http://www.w3.org/RDF/Interest/#irc

20:45:18 <AaronSw> AaronSw has changed the topic to: semanticWebChat: At the hour, we'll have another edition of classtime in #rdfig - http://www.w3.org/RDF/Interest/#irc

20:45:25 <sandro> Yeah. As you can probably tell, I think we're best off minimizing use of URIs as symbols.

20:45:43 <DanC> minimizing? I certainly don't want to go that far!

20:45:49 * AaronSw sometimes thinks we should just knock everything over and start anew

20:46:15 <AaronSw> we need the WebArch 3000 project to do these kinds of things -- what would web arch look like in an ideal world?

20:46:33 * Jah-class being a "newbie" still doesn't understand (in a deep sense) what the fuss is about...

20:46:58 * Jah-class seriously, there's years of philosophical papers on symbol grounding and the like that never solved this,

20:47:21 <DanC> if we really *need* that, we're doomed. It's nice to think outside the constraints of the present system at times, but I don't hope to abandon the hugely valuable connected set of resources we've built over the last 10 (or 40) years.

20:47:22 * Jah-class but when I say "aaronsw" in an appropriate context, there is no ambiguity...

20:48:57 <sandro> My question is just whether http://foo is useful as a logical symbol. It's good as a literal, when we're talking about web addresses, which I'm sure we'll want to do a lot, but as a Symbol we get into having to know what people "mean" by it.

20:49:00 <AaronSw> Well we don't *need* much.

20:49:15 <DanC> Jim, the big fuss is that the Semantic Web is an interaction of classical/formal systems (monotonic, every name has exactly one denotation, ...) with a the real chaotic world....

20:49:33 <AaronSw> URIs just aren't experts in the field of "mean"ing

20:49:38 <DanC> ... to take your aaronsw-in-context example, the computing world has yet to figure out how to formalize that.

20:51:28 <Jah-class> understood, but no one else ever figured it out either -- consider the "how big is http://www.cs.umd.edu/~hendler" question...

20:51:56 <Jah-class> If I run a byte counter - it's some number of bytes, if I run an inferencer that extracts content, it's "dex:short"

20:52:26 <Jah-class> so basically, the problem is trying to be too clean in the language, rather than letting pragmatics dictate a lot of the solution

20:52:45 <DanC> that's another natural language question... if you ground the how-big term formally, it should become evident whether you're asking about bytes or inches... that's an easy one, I think...

20:53:01 <AaronSw> Ouch... 404 at http://www.ncsa.uiuc.edu/demoweb/url-primer.html

20:53:32 <DanC> ... the harder ones are: how many members does W3C have?

20:54:13 <DanC> more formally: how many members are in the consortium whose home page is "http://www.w3.org/"?

20:54:17 <Jah-class> DanC: the notion of "formally ground" is part of the problem - on the web where we can tell easily if two things are exactly the identical URI, and then we can have mapping rules to others, why don't we do that

20:54:39 <DanC> yes, let's do ground things. But I'm not sure even that is a solution...

20:54:59 <Jah-class> We're going to have to accept that on the web we may never have a single answer to that question - I'd settle for something that could come back w/several answers, but justify them

20:55:31 <DanC> re how many members: one approach is to make sure all questions are phrased in the for-all-time-and-space context. i.e. that's a bad question. you can only ask "how many members are there this week?"

20:56:49 <DanC> yes, in the informal, chaotic web, there's no one answer to lots of things. But we're specifying formal mechanisms for machines to talk to each other, no? in those formalisms/idealisms, there's one answer (in each context or whatever, perhaps) yes?

20:57:34 <DanC> i.e. when I open an account and deposit $10, there's exactly one correct answer to "what's the balance of my account right now?", yes?

20:57:35 <Jah-class> i.e. How many members are in the consortium will depend more on how you process the question, as opposed to the exact links (IMHO)

20:57:50 <sat> there is no one answer even in real world. for ex: how many stem cell lines are available for research?

20:57:53 <DanC> I think you're trying to have your cake and eat it too, JimH.

20:58:29 <Jah-class> DanC - that's a good example, but doesn't it have more to do with the social rules of how the bank identifies you and your account than the technical issue of how the bank does the URIs?

20:58:51 <Jah-class> p.s. yes - I do want to have my cake and eat it too - what fun is it otherwise!

20:58:51 <DanC> yes, sat, there are all sorts of heisenberg-ish reasons why lots of questions have no clear/formal answer. But the semantic web is a formal idealism (or several in parallel or something).

20:59:32 <Jah-class> anyway, I gotta get class going - AaronSW - here or elsewhere - do we want to derail this day long conversation?

20:59:48 <DanC> bank social rules: yes; but the premise of the semantic web is that we can formalize the more mundane social rules in the world and trust the computers to execute them faithfully, right?

21:00:00 <AaronSw> i'd be happy to have in in #swhack which is publically logged -- others who are itnerested can join

21:00:51 <Jah-class> DanC right - and lots of real issues - I was only asking why the URI "syntax" issue was so crucial

21:00:53 <DanC> the students are welcome here as far as I'm concerned.

21:02:28 <sat> yes, Danc, but with many limitations, right?

21:02:52 <JHendler> JHendler is now known as CMSC498x

21:02:53 <AaronSw> if everyone else is OK with it, we can have it here -- i don't mind

21:03:08 <CMSC498x> sorry - lost connection for a min - we're back...

21:03:21 <CMSC498x> give me a min for some administrivia, we'll be back soon...

21:03:22 <AaronSw> AaronSw has changed the topic to: semanticWebChat: Another edition of classtime in #rdfig, please welcome CMSC498x - http://www.w3.org/RDF/Interest/#irc

21:06:15 <CMSC498x> AaronSw - class is here - care to say a few words about MusicBrainz

21:06:23 <AaronSw> Hi there.

21:06:37 <AaronSw> OK, well I'll assume you don't know much about it uet.

21:06:39 <AaronSw> err yet.

21:06:41 * CMSC498x class says "hi there"

21:06:52 <AaronSw> :-)

21:06:57 <AaronSw> MusicBrainz a large database of music metadata, and even though it's only in beta testing right now, it already has almost 300,000 tracks in the database. MusicBrainz information is all user-contributed, providing what some have termed the "cornucopia of the commons".

21:07:44 <AaronSw> Unlike many situations, where each user decreases the value of the shared space (the so called "tragedy of the commons"), the easy duplication of electronic information creates a situation where each users makes the system more valuable.

21:08:27 <AaronSw> Many of you have probably used CD-playing software that gets the track names off the Internet (likely from CDDB).

21:08:36 <DanC> say... is MusicBrainz subject to the same risks that materialized in the cddb case? Or does the license prevent the service operator from taking it proprietary somehow?

21:08:39 <AaronSw> MusicBrainz provides a similar function, but also works for MP3 files.

21:08:57 <AaronSw> MusicBrainz is licenesed under the OpenContent license, which is like the GPL for content.

21:09:05 <AaronSw> So it stays freely available forever.

21:09:18 <AaronSw> We're also putting in lots of safegaurds to prevent a CDDB-style takeover

21:09:37 <AaronSw> The data will be distributed across a number of servers, for one thing.

21:09:43 <DanC> is the data widely replecated? i.e. if the current operators of the service got corrupted, could the service spring up elsewhere easily?

21:09:58 <AaronSw> yes -- the data is all available in database dumps and rdf export.

21:10:04 <AaronSw> the source code is freely available too.

21:10:18 <AaronSw> so you could pretty simply set up a MusicBrainz clone

21:10:23 <AaronSw> in fact, some people already have

21:10:45 <AaronSw> mp3.nl is one example of such a server

21:11:03 * DanC has been thinking a lot about distributed trust stuff lately... I'd sure like a decentralized network of assurances that everybody I get on a plane with intends to survive the trip.

21:11:04 <CMSC498x> Aaron - how about something more basic - what is it? What does it do?

21:11:26 <AaronSw> Well, it's a project.

21:11:41 <AaronSw> One of the things we produce is a server which you can query for tracks and albums

21:11:42 <DanC> it supplies a CDDB-like service to freeamp, yes?

21:11:43 <CMSC498x> a little less basic than that///

21:11:48 <AaronSw> :-)

21:11:54 <AaronSw> cddb-style: yes

21:12:01 <AaronSw> So all of you go download FreeAmp! :-)

21:12:08 <DanC> is CMSC498x familiar with CDDB?

21:12:43 <CMSC498x> most, not all - perhaps a short background (or chump)

21:12:53 <AaronSw> The basic idea is that when you insert a music CD and open up FreeAmp, it queries the MusicBrainz server (using RDF) for album name and track names, etc.

21:13:05 <DanC> the CDDB service goes like this: fred catalogs his CD collection, indexed by metrics computed from the data on the CD... uploads it to cddb.org; 10,000 other folks do likewise. All CDs on the planet are thus catalog'd.

21:13:29 <AaronSw> It's worked pretty well.

21:14:05 <AaronSw> Of course, this is just one application of MusicBrainz's RDF data -- but probably their major application.

21:14:08 <DanC> in the CDDB case, the operators went "ah! this is worth money!" and betrayed the trust of the community. (or something like that?)

21:14:29 <AaronSw> yeah, they took out patents on the system and sold usage of the data to programmers.

21:14:41 <CMSC498x> what is the status - does it work yet or not (musicbrainz)

21:14:54 <AaronSw> MusicBrainz works today -- it's in the latest version of FreeAmp.

21:15:05 * DanC would like to quote musicbrainz deployment stats on w3.org/RDF

21:15:10 <AaronSw> It uses real RDF (I helped design the RDF schema they used).

21:15:13 <mjc> mjc is now known as alsoin498x

21:15:28 <AaronSw> :-)

21:15:40 <CMSC498x> can you give us a little background on Freeamp and what it is based on

21:15:41 <AaronSw> deployment stats: what sort? freeamp downloads?

21:16:15 <CMSC498x> comcast got aaron?

21:16:39 <DanC> deployment stats: how many albums/tracks catalogged? how many artists? how many queries/day? how is cataloged spelled?

21:16:42 <AaronSw> FreeAmp is a GPLed Audio Music Player -- it plays CDs, MP3s, etc.

21:16:59 <AaronSw> See http://freeamp.org/

21:17:09 <AaronSw> danc, all on http://musicbrainz.org website

21:17:16 <CMSC498x>http://freeamp.org

21:17:16 <dc_rdfig> J: http://freeamp.org from CMSC498x

21:17:17 <alsoin498x> Had the music industry looked at your schema as some kind of standard?

21:17:19 <AaronSw> look on the left side panel

21:17:25 <CMSC498x> J:| freeamp web page

21:17:25 <dc_rdfig> titled item J

21:17:26 <AaronSw> Server Stats

21:17:27 <AaronSw> Artists:16864

21:17:30 <AaronSw> Albums:37598

21:17:31 <CMSC498x> J: uses RDF/Musicbrainz

21:17:32 <AaronSw> Tracks:288328

21:17:32 <dc_rdfig> commented item J

21:17:33 <AaronSw> TRM Ids:44542

21:17:35 <AaronSw> DiskIds:9083

21:17:37 <AaronSw> Mods:22529

21:17:41 <DanC> grumble... no link from http://www.musicbrainz.org/MM/ to http://www.musicbrainz.org/

21:18:10 <AaronSw> mjc, a couple of organizations have looked at the specification, including Real (makers of RealAudio) for the basis of some Digital Rights Management programs

21:18:19 <AaronSw> (i.e. systems where they make you pay for all your MP3s)

21:18:30 <CMSC498x> students say there are a lot of bugs in the data, approve of the "moderation" - but wonder who does it all, and how that would work as it gets much bigger

21:18:40 <AaronSw> DRM: MusicBrainz tries to be neutral in the area, but anyone can use the data, obviously.

21:18:40 <dc_rdfig> Label DRM not found.

21:18:58 <DanC> hmm... 37598 albums vs. 9083 diskids... does that mean each disk is cataloged 4 times?

21:19:16 <AaronSw> no... it just means that there are albums in the database that aren't associated with a physical CD.

21:19:21 <AaronSw> well, moderation is all volunteer -- you can do it thru the website.

21:19:36 <CMSC498x> ok, ok, but let's get back to the semantic web -- where does RDF/SW concepts come in?

21:19:40 <AaronSw> the main idea is that you see a typo in your FreeAmp player and fix it -- and click submit to web.

21:20:18 <DanC> RDF/SW ... I wonder if we could do an swBot/musicbrainz trial...

21:20:19 <AaronSw> RDF/SW: well, the data is all in RDF, for one thing. secondly, others can use the openness, extensibilitty of RDF to add to it, build on top of it

21:20:34 <AaronSw> all the tracks, artists, albums have URIs, so other people can use the information for their own purposes.

21:20:46 <CMSC498x> SA (Student asks) -- how do we avoid "sabotage" and people doing bad things either originally or "mismoderating"

21:21:12 <AaronSw> moderation works on a voting system, so a majority of users would have to "mismoderate" something

21:21:24 <DanC> avoid sabatoge: I expect that in this case, natural market forces tend toward good data. Who is motivated to pollute the thing?

21:21:36 <CMSC498x> SA- what about having to ground the tracks, artists, etc. to some authoratative web site (URI)?

21:21:42 <AaronSw> Also, we have some predefined scripts that go over the data and clean up common problems.

21:22:04 <AaronSw> grounding the tracks: well, we don't need to do that -- we generate URIs for each track/artist/album in the database

21:22:16 <AaronSw> obviously as 3rd party could link these to promotional websites, fan pages, etc.

21:22:29 <AaronSw> such data may be added to the muscibrainz db someday, too

21:22:31 <DanC> e.g. a URI for an album: http://mm.musicbrainz.org/album/911e3f30-192e-4c3d-aa25-2a89d4202a3e

21:23:01 <CMSC498x> SA - meant it the other way - if, for example,

21:23:37 <CMSC498x> SA - you linked the artist to the home page of the artist, would give you a definite reference instead of a "made up" URI. If you use your kind, how can we know we're talking about same artist, etc.

21:24:06 <DanC> "the" home page of the artist? 1/2 ;-)

21:24:07 <AaronSw> Well, linking to the homepage causes all sorts of confusion, as we were discussing earlier today. ;-)

21:24:13 <CMSC498x> (or even - I trust FreeDB to authenticate the issue)

21:25:01 <alsoin498x> Maybee the artist or record company should be responsible for setting up the authoriative URI?

21:25:01 <AaronSw> "the same artist" moderator merge duplicate artists, and before submitting data the server asks you whether your artist is the same as some similar names already in the db

21:25:01 <CMSC498x> well, how would we know two albums are by same artist without just a string match (in which case why use URIs)

21:25:01 <DanC> it seems to me that, again, natural market forces tend toward the musicbrainz URIs being as good, if not better, than any other URI for an album/artist/etc.

21:25:01 <AaronSw> there is no "authoritatve URI" -- that's the point of the Web!

21:25:27 <AaronSw> there's no way of knowing whether two URIs are equivalent or not for sure.

21:25:44 <AaronSw> that's just how the Semantic Web works -- we can count on people and machines to assist us, but there's no guarantee

21:26:02 <DanC> but the market forces around musicbrains tend toward each artist having one URI shared by the musicbrains community.

21:26:22 <CMSC498x> SA - what about music that has different names in different countries - how could we map those?

21:26:54 <AaronSw> Hmm... I'm not sure the MB database has provisions for that, but it wouldn't be too hard to do it in RDF...

21:27:04 <AaronSw> just add a frenchTitle property, etc.

21:27:28 <DanC> in one sense, the recordings are grounded in bits. the trackids are computed from bits on the CD. In the different-countries scenario, do you expect the recorded bits to be the same or different?

21:27:32 <CMSC498x> SA - wouldn't attribute (language = EN) be better?

21:27:56 <AaronSw> xml:lang? umm, this is an issue that has generated some discussion. ;-)

21:28:00 <DanC> ;-)

21:28:06 <AaronSw> i tend to prefer things that are clearly available in triples

21:29:26 <CMSC498x> seems wrong to make single entities for every combination of language/attribute -- is that what you are suggesting (oops - should be "SA")

21:30:03 <AaronSw> single entities?

21:30:06 <AaronSw> I'm thinking:

21:30:20 <AaronSw> <http://mm.musicbrainz.org/album/911e3f30-192e-4c3d-aa25-2a89d4202a3e> dc:title "Dummy" ;

21:30:26 <AaronSw> :frenchTitle "lalala" ;

21:30:32 <AaronSw> :arabicTitle "????" .

21:30:32 <DanC> how about the CMSC498x take this titles-in-different-countries scenario as a case study and build something?

21:30:33 <AaronSw> etc.

21:31:19 <CMSC498x> S asserts - but you also need Songtitle:French, Songtitle:arabic, and and and

21:31:55 <AaronSw> Well, yes. you need one property for every language.

21:31:59 <AaronSw> You could also model it as:

21:32:04 <DanC> re Songtitle:French: yes, there are probably some reusable structures worth designing. Sounds like a great student project ;-)

21:32:18 <AaronSw> <...> dc:title [ :en "fooz" ; :fr "blaz" ; :ar "wham" ] .

21:32:33 * AaronSw assigns the asserting student the problem for homework. ;-)

21:32:38 <CMSC498x> students like this better

21:32:55 <CMSC498x> they will all use the N3 w/anonymous mode in assignment one...

21:33:19 <DanC> or <album> [ is rdfs:subPropertyOf dc:title; :lang "en" ] "fooz".

21:33:27 <AaronSw> OK... I think there's a predefined namespace for the :en stuff... give me a second...

21:34:13 <AaronSw> Yes, some folks in the IETF are working on @prefix lang: <urn:ietf:params:languiage:> for this kinda stuff

21:34:35 <AaronSw> Coming soon from Graham Klyne

21:35:17 <DanC> it would be great for the students to put together something out of real world data; i.e. real cases where the same album/movie/whatever has different names in different parts of the world, and look at how folks would want to exploit the "sameness". The RDF Core WG is making *very* slow progress on this xml:lang stuff, and more real-world data would help a lot, I think.

21:35:34 <AaronSw>http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Sep/att-0226/02-draft-klyne-urn-ietf-lang-00a.html

21:35:34 <dc_rdfig> K: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Sep/att-0226/02-draft-klyne-urn-ietf-lang-00a.html from AaronSw

21:35:42 <AaronSw> K:|A URN sub-namespace for langauge tags

21:35:43 <dc_rdfig> titled item K

21:35:47 <CMSC498x> aaron - think you said before that metadata from musicbrainz is downloadable - where is that? (i mean, is thi the whole DB, or just the metadata definitions)

21:35:49 <AaronSw> K:From Graham Klyne

21:35:49 <dc_rdfig> commented item K

21:36:12 <CMSC498x> SA - URN??

21:36:12 <DanC> downloadable: if you want just a few samples, there are pointers in http://logicerror.com/musicbrainzRDFCore

21:36:27 <DanC> "URIs for Artists, Tracks, Albums, etc. you can HTTP GET? for an RDF description ..."

21:36:37 <AaronSw> URN: Uniform Resource Name -- a URI that starts with the characters 'urn:'

21:36:37 <dc_rdfig> Label URN not found.

21:36:38 <DanC> e.g. http://mm.musicbrainz.org/album/911e3f30-192e-4c3d-aa25-2a89d4202a3e

21:37:30 <CMSC498x> SA - what others uses for metadata are there than the display stuff?

21:37:49 <AaronSw> You can get MB IDs like that from the musicbrainz website... http://musicbrainz.org/showtrack.html?trackid=7701

21:38:19 <AaronSw> other uses for metadata: some folks working on a project called Espra (espra.net) are working to build a Napster-style system on top of Freenet

21:38:29 <AaronSw> so metadata helps them to do searches, tagging of songs, etc.

21:38:45 <AaronSw> MusicBrainz uses software from relatable to do hashes of MP3s, etc. to get unique IDs for songs

21:38:52 <AaronSw> these are called TRM IDs

21:39:08 <AaronSw> another example is music comapnies doing DRM as i described earlier

21:39:09 <CMSC498x> (jimH tries to show the students, but is burned because Netscape 6.1 can't display rdf! foo)

21:39:51 <AaronSw> (thanks to DanC for digging up those URIs)

21:40:19 <CMSC498x> SA - is this based on the audible music or the binary data

21:40:21 <AaronSw> of course any music-related projects you guys want to do can build on top of the MB terms and schemas

21:40:47 <AaronSw> umm, i haven't looked at the algorithm in depth, but I believe it's based off of the binary data

21:41:30 <CMSC498x> (prof H points out we're moving to the low level, away from SW -- will try to get the students thinking metadata again...)

21:41:31 <AaronSw> Aha:

21:41:35 <AaronSw> """TRM advanced acoustic fingerprinting technology is a leading solution for

21:41:35 <AaronSw> identifying digital music files. TRM recognizes songs based on the acoustical properties in the audio

21:41:35 <AaronSw> itself, and has been developed to achieve maximum accuracy in discriminating between different songs,

21:41:35 <AaronSw> as well as identifying each and every digitized copy of a recorded song, regardless of audio file

21:41:35 <AaronSw> format, bit rate or common signal distortions. TRM delivers unprecedented speed and can scale to meet

21:41:36 <AaronSw> the needs of any size network."""

21:41:56 <CMSC498x> (Students are impressed)

21:41:57 <AaronSw> So it's based upon the actual acoustical properties -- sorry

21:42:17 <AaronSw> their stuff was so good napster uses it to filter songs! :-(

21:42:52 <AaronSw> "TRM will help ensure that the millions of music files transferred through the new Napster system will be accurately monitored and it will enable the appropriate allocation of royalties to artists, music publishers and record companies,"

21:43:17 <CMSC498x> SA - seems like a lot of useful stuff (they say business model) could come from this -- example, given a set of tracks/artists I could say find me a single CD that has the most of these (and where on the web to buy it)

21:43:21 <alsoin498x> wow...that's pretty impressive

21:43:33 * DanC reads "appropriate allocation" with a chunk of salt

21:43:43 <AaronSw> heh!

21:43:54 <CMSC498x> SA - what is 0% of nothing?

21:44:14 <AaronSw> buy a CD: yes, the original sponsor of MB was EMusic which wanted to use the metadata to sell people CDs and DRMed music

21:44:41 <CMSC498x> SA - Past tense?

21:45:29 <AaronSw> EMusic still hosts the MusicBrainz servers but fired the lead developer (Rob Kaye) -- he now works at a startup called Bitzi which does metadata for all files (not just music)... they also announced that they'll be releasing dumps in RDF the other day.

21:45:41 <AaronSw> where they = bitzi

21:46:29 <AaronSw> Rob Kaye has very cool hair, BTW... :-)

21:47:33 <AaronSw> pictures: http://www.mayhem-chaos.net/hair.html

21:48:29 <AaronSw> any other questions?

21:48:36 <AaronSw> have I gone mute?

21:48:54 <CMSC498x> we were out on the web looking at pictures...

21:49:38 <CMSC498x> class wonders if there is a meta-data repository of these hair styles

21:49:52 <AaronSw> Hmm, you should suggest that to him! :-)

21:49:56 <AaronSw> Ooh, class project!

21:50:43 <AaronSw> Another cool thing is that MusicBrainz has an RDF representation of music lyrics for some songs

21:50:55 <CMSC498x> JimH - class will probably split into 2-3 groups each doing a SW appl - what else than MB should we look at?

21:50:57 <AaronSw> so you can watch the lyrics scroll by in sync wiht the music

21:51:23 <CMSC498x> students say - cool - how does that work?

21:51:38 <AaronSw> They mark times off in the song and associate a series of words with it.

21:52:15 <AaronSw> Dublin Core is a big SW group, I suppose... ODP has a lot of RDF data, but it isn't exactly valid

21:52:26 <AaronSw> and RSS is also pretty popular.

21:52:43 <CMSC498x> SA - does lyric stuff run into copyright dangers?

21:53:24 <CMSC498x> SA - is there a system for doing the time stamp stuff, or is it a hard thing to do (student envisions stop watch with a lap counter)

21:53:32 <AaronSw> Lyrics examples in: http://www.freeamp.org/pipermail/mm/2001-February/000015.html

21:53:38 <AaronSw> copyright dangers: we haven't yet...

21:54:11 <AaronSw> lap counter: heh!

21:54:33 <AaronSw> There's a special lyric (SyncText) editor at: http://mp3.nl/mp3nl/synctext_editor.html

21:55:36 <AaronSw> It's a WinAMP plugin, for you windows users.

21:55:47 <CMSC498x> Aaron - we got to go - this has been great - any last thoughts for these guys?

21:55:58 <CMSC498x> (where guy is generic term)

21:56:20 <AaronSw> Well, it's nice having you! Please go out and download Freeamp and the Winamp plugin and try this stuff out.

21:56:53 <CMSC498x> SA - before you go, What OS is freeamp for

21:57:04 <AaronSw> I think there's Linux and Windows versions.

21:57:22 <AaronSw> Yep.

21:57:34 <CMSC498x> ok, g'night - Shana Tova.

21:57:41 <AaronSw> g'nite

22:00:10 <Xirzon> Hi, can you explain how a chumpbot works? We're thinking about using it on #infoanarchy.

22:00:29 <AaronSw> Hi there.

22:00:37 <AaronSw> Cool, the chumpbot works like this:

22:00:39 <Xirzon> Hey Aaron.

22:00:42 <AaronSw>http://infoanarchy.org/

22:00:42 <dc_rdfig> L: http://infoanarchy.org/ from AaronSw

22:00:46 <AaronSw> L:|InfoAnarchy

22:00:46 <dc_rdfig> titled item L

22:01:03 <AaronSw> L:A nice news website about the information future

22:01:03 <dc_rdfig> commented item L

22:01:12 <AaronSw> see http://rdfig.xmlhack.com/ for the output

22:01:42 <Xirzon> So everything with L: at the beginning is web-posted and | at the beginning= link title?

22:02:02 <AaronSw> right...

22:02:13 <AaronSw> the L: changes for each post -- like the first post was A: the next one will be M:

22:02:14 <Xirzon> Is the log for iA still active?

22:02:34 <AaronSw> Yes, you can continue to comment on anything on today's page

22:03:05 <Xirzon> L:We discuss copyright, "cyber liberties", patents and similar issues. We also report news about new relevant applications.

22:03:05 <dc_rdfig> commented item L

22:03:13 <Xirzon> Hmm, nice.

22:03:54 <Xirzon> How do I know that InfoAnarchy is L from looking at the page? The access letter does not seem to be listed there.

22:04:14 <AaronSw> yeah... you have to count up from A, or do

22:04:18 <AaronSw> dc_rdfig:view

22:04:18 <dc_rdfig> H: On-line proceedings of workshop on "ontologies for agents" (http://CEUR-WS.org/Vol-52/)

22:04:19 <dc_rdfig> I: How We Identify Things (on the Semantic Web) (http://www.w3.org/2001/03/identification-problem/)

22:04:20 <dc_rdfig> J: freeamp web page (http://freeamp.org)

22:04:21 <dc_rdfig> K: A URN sub-namespace for langauge tags (http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Sep/att-0226/02-draft-klyne-urn-ietf-lang-00a.html)

22:04:22 <dc_rdfig> L: InfoAnarchy (http://infoanarchy.org/)

22:04:54 <Xirzon> Hmhm..

22:05:18 <sbp> If you just type the "K:" or whatever symbol in the channel, it comes up with all of the info posted about it... but it tends to bug the crap out of people too

22:05:41 <Xirzon> would it be possible to start a topic with something different as the central focus point?

22:05:50 <AaronSw> you mean other than a URL?

22:05:55 <Xirzon> Yes.

22:05:56 <AaronSw> BLURB:Who is Xirzon?

22:05:57 <dc_rdfig> M: Who is Xirzon? from AaronSw

22:06:07 <AaronSw> M:This mysterious man just entered our channel all of the sudden.

22:06:07 <dc_rdfig> commented item M

22:06:20 <Xirzon> M:And who runs http://www.infoanarchy.org (mostly), I might add.

22:06:21 <dc_rdfig> commented item M

22:06:32 <Xirzon> Hmm, nice.

22:06:40 <sbp> I:The "Variables + Bootstrapping" section seems like a rather unweildy alternative to minting new identifiers

22:06:41 <dc_rdfig> commented item I

22:06:54 <AaronSw> M:Visit [InfoAnarchy|http://www.infoanarchy.org]. It's *cool*. ;-)

22:06:54 <dc_rdfig> commented item M

22:07:31 <Xirzon> Hehe. OK, I think we might install that for our channel. I'll talk to tav about it.

22:07:35 <sbp> lol! [[[Tim Berners-Lee, born in 1955, served as W3C director from 1994 to 2051, when he retired and was replaced by Aaron Swartz.]]] - http://www.w3.org/2001/03/identification-problem/ I'd actually give odds of 3:4 on

22:07:57 <AaronSw> Cool, Xirzon.

22:08:24 <AaronSw> sbp... "I'd actually give odds of 3:4 on" on what?

22:08:40 <sbp> 75% chance of it actually happening :-)

22:08:52 <AaronSw> heh.

22:09:04 <AaronSw> date could be off, tho...

22:09:12 <sbp> 2052?

22:10:06 <AaronSw> :-)

22:10:29 * sbp likes Chas. Munat's lame excuse for missing the GL call: "thought it was Wednesday today..."

22:10:42 <AaronSw> Heh... I do that too.

22:10:46 <sbp> me too!

22:10:47 <AaronSw> But I don't miss telecons.

22:11:01 <sbp> I slept through one once... shh!

22:11:06 <dajobe> didn't someone else have a problem with Thursdays :(

22:11:26 <AaronSw> eh?

22:11:37 <sbp> I was up until 6-7AM backing my Hard Drive up in order to transfer it... it was a necessity

22:11:45 <dajobe> answer: arthur dent

22:11:56 <sbp> Arthur Phillip Dent?

22:30:00 * DanC tries to remember how Peirce denoted quantification in his diagrams...

22:34:58 <DanC> I thought I was going to google up some old stuff, but what I found looks new (and interestingly so): http://users.bestweb.net/~sowa/peirce/ontometa.htm

22:37:44 <DanC> aaron, still around? small world: Sowa cites TimBL in support of my claim earlier that things that aren't HTTP-access-controllable are resources:

22:37:46 <DanC> [[[

22:37:48 <DanC> Not all resources are network "retrievable"; e.g., human

22:37:49 <DanC> beings, corporations, and bound books in a library can also be

22:37:49 <DanC> considered resources.

22:37:49 <DanC> ]]]

22:38:06 <AaronSw> I find it funny that TimBL's view and mine are almost exact opposites.

22:38:18 <AaronSw> Err, I agree with that statement.

22:38:52 <AaronSw> I'd certainly hope that humans, books, and corporations are resources!

22:39:16 <DanC> er... but you're not happy saying that http://example/buildings#e32 denotes a building? how am I to understand your agreement to this but disagreement earlier?

22:40:02 <AaronSw> I'm happy to say that http://example/buildings denotes a building.

22:40:07 <AaronSw> But http://example/buildings#e32 isn't clear.

22:41:48 <DanC> [I'm netutral, but to take TimBL's position:] if http://example/buildings behaves in the normal way in my browser, are you happy to say that what I'm contacting via my browser *is* the building?

22:42:12 <AaronSw> Because I'm not sure http://example/buildings#e32 is a Resource.

22:42:24 <AaronSw> What you're contacting? You mean the web server?

22:42:35 <AaronSw> No, I don't think the web server is the building.

22:42:51 <DanC> I mean the web resource that the browser contacts when I make a request.

22:43:12 <AaronSw> Hmm, I've never heard this "contact" terminology before...

22:43:18 <DanC> how about access?

22:43:33 <AaronSw> I don't see how you can access a Resource.

22:43:42 <AaronSw> You can do an HTTP GET, is that what you mean?

22:43:49 <DanC> yes, HTTP GET

22:44:21 <AaronSw> The HTTP Response is not a building, if that's the question.

22:44:36 <DanC> no, but the HTTP response is from a resource, right?

22:44:42 <AaronSw> It's what the HTTP spec claims it is: some information about a Resource.

22:45:04 <AaronSw> I don't know what it means for something to come "from" a resource.

22:45:24 <DanC> ok, I'll go with that: the HTTP GET response is some information about a resource. Are you happy to say that resource is the building?

22:45:38 <AaronSw> Yes.

22:46:23 <DanC> now the HTTP GET response is likely to have a Content-Length: header field, right? And this is, per the HTTP spec, metadata about the resource... so you're happy with the claim that the building's content length is 1223 or whatever?

22:46:37 <AaronSw> whoa!

22:46:45 <AaronSw> Spec says no such thing.

22:46:53 * DanC checks...

22:47:01 <AaronSw> Content-Length is the length of the Entity, not the Resource.

22:47:19 <AaronSw> the Entity is the thing information about the Resource than an HTTP server sends back.

22:47:25 <AaronSw> Come on, I thought you knew these things...

22:47:53 <AaronSw> The Content-Length entity-header field indicates the size of the

22:47:53 <AaronSw> entity-body, in decimal number of OCTETs

22:48:01 <AaronSw> - http://rfc2616.x42.com/

22:48:12 <DanC> ok.. quite... content-length is a bad example...

22:49:15 <DanC> ok... I don't think I can force you into a difficult position by taking TimBL's position. As I say: I'm neutral.

22:49:49 <AaronSw> However, by taking my position you can force TimBL into a difficult postion. ;-)

22:49:57 <DanC> I don't see any particular problem using http://...dc.../title as both an RDF property and an HTTP resource.

22:50:09 <AaronSw> Me neither.

22:50:16 <AaronSw> An HTTP Resource could be an RDF property.

22:50:23 <deltab> AaronSw: for that RFC in particular, I prefer http://www.w3.org/Protocols/rfc2616/rfc2616 :-)

22:50:29 * AaronSw too...

22:50:31 <AaronSw> :-)

22:50:48 * DanC wonders what difficult positoin TimBL would be in, but has family obligations now

22:51:49 * AaronSw can't remember what page it is, but one of the DesignIssues makes some claims that are pretty clearly false

22:52:05 <AaronSw> Ahh, this one: http://www.w3.org/DesignIssues/Fragment.html

22:52:50 <AaronSw> Like this, oft quoted statement: "A client which understands the http: protocol can immediately conclude that the fragementid-less URI is a generic document."

22:53:41 <DanC> that's tautologically true if you define "generic document" to mean "something you can GET a representation of"

22:53:56 <AaronSw> well, true, but that's not what most people think of...

22:54:07 <deltab> btw, thanks for those pages, DanC

22:54:09 <AaronSw> You can GET a representation of a building, but most people wouldn't call it a document

22:54:34 <DanC> which pages, deltab? if they're useful to more than one of us, we should chump them.

22:55:09 <deltab> the hypertext RFC2616 at http://www.w3.org/Protocols/rfc2616/rfc2616

22:55:34 <AaronSw> any work done on that code recently?

22:55:47 <AaronSw> I thought it should be used on 2396, etc.

22:55:57 <AaronSw> code being http://dev.w3.org/cvsweb/2001/rfc2html/

22:56:10 <DanC> well, I did put the rfc2html code in dev.w3.org

22:56:25 <AaronSw> AaronSw has changed the topic to: semanticWebChat: A rose by any other URI... - http://www.w3.org/RDF/Interest/#irc

22:56:41 <DanC> I tweaked it for use with the IMAP spec a while ago; haven't messed with rfc2html since

22:56:59 <DanC> topic: chuckle. good one.

22:57:23 <deltab>http://www.w3.org/Protocols/rfc2616/rfc2616

22:57:23 <dc_rdfig> N: http://www.w3.org/Protocols/rfc2616/rfc2616 from deltab

22:57:31 <AaronSw> :-)

22:57:49 <deltab> N:|HTTP/1.1, in hypertext

22:57:49 <dc_rdfig> titled item N

22:57:57 <DanC> N:powered by [rfc2html|http://dev.w3.org/cvsweb/2001/rfc2html/]

22:57:57 <dc_rdfig> commented item N

The IRC chat here was automatically logged without editing and contains content written by the chat participants identified by their IRC nick. No other identity is recorded.