Semantic Web Interest Group IRC Chat Logs for 2002-02-15

00:30:02 * sbp checks out Seth's latest mentography

00:31:03 <Seth> what do you think about statings being classes rather than individuals ?

00:31:46 <sbp> statings? does seth:statings = sean:Statements . ? [and rdf:Statement to everyone else!]

00:32:53 <Seth> maybe .. give me an example

00:33:18 <sbp> AFAICT, the Mentography is: :c [ [ rdfs:range :A, :B ] ]; rdfs:domain :C . :C rdfs:subClassOf :A, :B .

00:34:11 <sbp> to me, a statement is a set of three terms as a binary relationship asserted in an RDF graph

00:36:29 <sbp> you've invented a predicate which says "when you use this predicate and hang an rdfs:range predicate off of (the subject?), then you're ORing them"... but I'm really confused as to how that's supposed to be represented in ordinary statements (serializable in XML RDF and Notation3), and what the semantics of that really are

00:36:37 <sbp> s/subject?/object?/

00:37:47 <Seth> woops were talking about which mentograph .. one was to mac about range\domain and the other about statings being classes .. which are you talking about?

00:38:01 <sbp> I am refering to http://robustai.net/mentography/rdfs_domain_range4.gif

00:38:31 <sbp> where's the "statings are classes" one? Mac?

00:39:07 <sbp> ah: http://robustai.net/mentography/reifyRDF_stating_not_statement.gif

00:39:21 <Seth> well one at a time ... so do you agree that range\domain meets Mac's challenge .. he did later in personal email .. with proviso

00:40:33 <sbp> I am not sure how you would serialize your range/domain mentograph

00:40:50 <sbp> I do not think that what he asked is expressible using only those predicates found in RDF Schema

00:40:59 <Seth> just read off the Ntriples

00:41:19 <sbp> where are the NTriples? Can you paste them in here for me, please?

00:41:35 <Seth> let me look

00:41:43 <sbp> thanks

00:42:37 <Seth> well forget about the meta stuff

00:42:56 <Seth> and forget about the <>

00:43:09 <Seth> and the rdfs:

00:43:15 <Seth> c range ClassQ

00:43:18 <Seth> woops

00:43:22 <Seth> c range ClassA

00:43:33 <Seth> c range ClassB

00:43:44 <Seth> c domain ClassC

00:43:55 <Seth> ClassC subclass ClassA

00:44:04 <Seth> ClassC subclass ClassB

00:44:26 <Seth> domain subProperty DisjunctiveProperties

00:44:39 <Seth> range subProperty DisjunctiveProperties

00:44:48 <sbp> from that, I can imply:-

00:44:56 <sbp> c DisjunctiveProperties ClassA

00:45:04 <sbp> c DisjunctiveProperties ClassB

00:45:07 <sbp> c DisjunctiveProperties ClassC

00:45:11 <sbp> which doesn't seem to follow

00:45:21 <sbp> I think you mean "rdf:type" in place of "subProperty"

00:45:45 <sbp> because I am also fairly certain that "DisjuntiveProperties" is a class

00:45:46 <Seth> your poobably right .. dont really know subProperty yet

00:46:01 <sbp> sub property is a specialization of a property

00:46:30 <sbp> """If some property P2 is a subPropertyOf another more general property P1, and if a resource A has a P2 property with a value B, this implies that the resource A also has a P1 property with value B.""" - http://www.w3.org/TR/2000/CR-rdf-schema-20000327/#s2.3.3

00:46:34 <Seth> thing is it is a class of properties .. so i though a subclass would be a subproperty .. but im probably off my rocer there

00:46:41 <sbp> :-)

00:46:51 <sbp> you mean:-

00:47:01 <sbp> DisjuntiveProperties subClassOf rdf:Property

00:47:10 <Seth> ok :)

00:48:11 <sbp> secondly, from your triples above, and the fairly widespread assumption that all triples are taken to be asserted conjuntively, c has the range of ClassA and ClassB, thanks to RDF Core's decision reversing the semantics. You already know about that, of course

00:48:12 <Seth> it would be fun to put that into cwm and see what happened

00:48:20 <sbp> I can do, if you like

00:49:39 <Seth> so there would be no way for me to overide the pervasive and-ness of a graph and say that some particular property was to be considered disjunctive

00:49:43 <sbp> IMHO, the question posed on www-rdf-logic is just someone complaining that the semantics of RDF Schema have changed, whilst not noticing that in the majority of use cases the decision is helpful

00:49:57 <sbp> no way to override AND: you've got it

00:50:00 <Seth> i agree

00:50:16 <Seth> so let me ask a question:

00:50:39 <sbp> the disjunctive override... Hmm...

00:51:21 <sbp> How about: { :c [ :disjunctiveOf rdfs:range ] :A, :B; rdfs:domain :A, :B }

00:51:34 <Seth> the basic lobical operators of (and,or,not,=>) are hard coded into cwm as log:implies etc .. they cant come in as rule definitions .. can they?

00:52:19 <sbp> rule definitions... I'm actually not quite clear on that myself. I know that one can nest rules, but I also know that CWM doesn't treal ":F a log:Falsehood" as it should

00:52:36 <sbp> s/treal/treat/

00:52:48 <Seth> well log:implies is hard coded .. isn't it?

00:53:00 <sbp> the semantics of the property are, indeed

00:53:31 <sbp> the domain and range of log:implies is n3:Formula, AFAIK

00:53:57 <Seth> in other words there is a piece of python code that basically does an (if .. then ...) that defines what log:implies means .. is that an accurate statement?

00:54:25 <sbp> yes: I believe your statement to be correct at this moment in time

00:54:48 <Seth> what other logical primitives are hard coded like that now?

00:55:42 <sbp> there are many properties and classes that have such "hard coding" in CWM, and Tim calls them "builtins". However, not all of them are logical primitives: most are for dealing with string equality etc.

00:55:56 <sbp> but you have: log:implies, log:Truth, log:Falsehood

00:56:17 <sbp> (=>, T, !(T))

00:56:22 <Seth> nothing for (and, or) ?

00:56:24 <sbp> or ¬ for not...

00:56:40 <sbp> and is quite pervasive in RDF, as you commented :-)

00:57:03 <sbp> as for OR, there's nothing that I know of, but there have been a handful of discussions about it

00:57:45 <Seth> so the ~anding~ of the triples is hard coded, but there is no OR or NOT primitive that is hard coded now?

00:58:52 <sbp> well, log:Falsehood is in there somewhere, but I'm not sure what effect it has

00:58:56 <sbp> if I may:-

00:58:56 <sbp> [[[

00:58:57 <sbp> 21:42:20 <sbp> e.g. to say "Bob is a Fish, or Bob is a Dog": { { :Bob a :Fish } a log:Falsehood . { :Bob a :Dog } a log:Falsehood } a log:Falsehood .

00:58:59 <sbp> [...]

00:59:10 <sbp> 21:42:41 <tim> OR? Operational Research?

00:59:11 <sbp> 21:42:52 <DanC> OR as in "this or that".

00:59:11 <sbp> 21:42:57 <sbp> yeah

00:59:11 <sbp> 21:43:08 <sbp> logical OR

00:59:11 <sbp> 21:43:31 <tim> ooops

00:59:12 <sbp> 21:43:49 <tim> Toolbox a :rusty;

00:59:15 <sbp> 21:44:13 <tim> doc:supercededBy :n3.

00:59:16 <sbp> [...]

00:59:25 <sbp> 21:45:47 <sbp> how would you express OR now? I thought of using an Alt bag (I thought was was what it was meant to do), e.g. [ a rdf:Alt; rdf:_1 { :Bob a :Fish }; rdf:_2 { :Bob a :Dog } ] .

00:59:25 <sbp> 21:46:59 <tim> I would say something like ( :F :G :H ) a :TrueDisjunction.

00:59:26 <sbp> 21:47:08 <tim> I would say something like ( :F :G :H ) a :OR.

00:59:27 <sbp> [...]

00:59:36 <sbp> 21:47:57 <SethR> where :F :G and :H are statements ?

00:59:37 <sbp> [...]

00:59:45 <sbp> 21:48:13 <tim> {[ a :OR; daml:first [a log:Falsehood]; daml:rest:x]} log:implies {:x a log:Truth} maybe

00:59:45 <sbp> 21:48:47 <tim> statement sets - yes i call them formulae at the moment - get sme into least trouble ;-)

00:59:49 <sbp> ]]] - http://ilrt.org/discovery/chatlogs/rdfig/2001-11-21.txt

01:00:15 <sbp> sorry for the huge paste, but it seems quite relevant

01:00:23 <Seth> cool i had forgotten that dialogue

01:01:02 <sbp> ah, CWM does have a built in to merge formulae: log:conjunction

01:01:16 <sbp> [[[

01:01:17 <sbp> log:conjunction a rdf:Property;

01:01:17 <sbp> rdfs:label "conjunction";

01:01:17 <sbp> rdfs:domain log:List;

01:01:17 <sbp> rdfs:range log:Formula;

01:01:17 <sbp> rdfs:comment """"A function to merge formulae: logical AND.

01:01:22 <sbp> ]]] - http://www.w3.org/2000/10/swap/log.n3

01:02:19 <Seth> so it is these built in formulae that give the RDF type languages fed into CWM their semantics

01:02:23 <Seth> ?

01:02:39 <sbp> not exactly: CWM understands more things besides its builtins

01:02:58 <sbp> these properties etc. are declared extra to the things that people hold to be truths in RDF (in general)

01:03:22 <sbp> e.g. that when you feed in a statement and then another statement, they are helf in the store conjuntively

01:03:29 <sbp> s/helf/held/

01:03:49 <Seth> well it understands rules built on those primitives .. but if you cant build a rule with the primitives given, then you cant's express that thing to a CWM process .. right?

01:04:02 <sbp> correct

01:04:25 <sbp> although that doesn't rule out such things being impossible in the future; possible near-future

01:04:36 <sbp> s/impossible/possible/

01:04:54 <Seth> i wonder why they didnt define these primitives first and agree upon them

01:05:11 <sbp> well, people seemed to be generally agreed upon the basics

01:05:26 <sbp> but then when you actually start *using* RDF, you need extra things

01:05:46 <sbp> so Tim (I guess) just started building them in, based upon his Toolbox stuff etc.

01:06:06 <Seth> have you been foolowing Peter Patel's stuff about the layers of the semantic web?

01:06:12 <sbp> which is good because you can do what you want - to some extent - in your own namespace...

01:06:17 <sbp> layers: only a little bit

01:06:26 <sbp> in WebONT, right?

01:06:30 <Seth> yes

01:06:40 <sbp> that's been going on for quite a while

01:07:03 <Seth> basically the paper .. want me to chump it?

01:07:11 <sbp> yes please

01:07:29 <Seth>http://www-db.research.bell-labs.com/user/pfps/webont/layering/layering.html

01:07:30 <dc_rdfig> A: http://www-db.research.bell-labs.com/user/pfps/webont/layering/layering.html from Seth

01:07:49 <Seth> A:|Layering the Semantic Web: Problems and Directions

01:07:49 <dc_rdfig> titled item A

01:08:01 <sbp> heh! "The Trouble with Triples" - http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0086

01:08:22 <Seth> A: Peter's alternate name is "The Trouble with Triples"

01:08:22 <dc_rdfig> commented item A

01:08:25 * sbp reads the document...

01:08:37 <sbp> that's the alternate name for that document?

01:08:50 <DanCon> no, for another one

01:09:26 <Seth> oh, i thought that was the same .. what is the trouble with tribles one?

01:10:36 <Seth>http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0076.html

01:10:37 <dc_rdfig> B: http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0076.html from Seth

01:11:04 <Seth> B:| The Trouble with Triples

01:11:04 <dc_rdfig> titled item B

01:13:19 <Seth> B: me thinks makes a compelling case for quads or greater tuples : (context, subject, predicate, object, ...)

01:13:19 <dc_rdfig> commented item B

01:15:20 <sbp> the former is a very interesting document

01:16:21 <Seth> to me its disturbing till i recoginzed that language processors have primitives and your really cannot trancend those primitives . which is what motivated my questions about CWM

01:16:23 <sbp> the latter is not really news

01:16:38 <sbp> yeah...

01:17:18 <Seth> thing is that RDF did not define any primitives at all really

01:17:46 <sbp> So what is the general climate in WebONT at the moment? Changes to RDF/OWL depature from the RDF world?

01:18:10 * sbp shall be reading through the list a bit more tonight

01:18:28 <Seth> ask DanC

01:20:03 <sbp> nah. He'll tell me to do my own research :-)

01:21:07 <Seth> so back to the statings question .. are statings a class .. or an instance?

01:22:31 <sbp> that's an interesting question

01:22:35 <Seth>http://lists.w3.org/Archives/Public/www-rdf-logic/2002Feb/0094.html

01:22:35 <dc_rdfig> C: http://lists.w3.org/Archives/Public/www-rdf-logic/2002Feb/0094.html from Seth

01:22:57 <Seth> C:|A stating of a triple is a class of things.

01:22:58 <dc_rdfig> titled item C

01:23:44 <sbp> "can one triple with the same members be different from another" is the question

01:24:16 <Seth> well i dont think that is the question

01:24:16 <deltab> Seth: 'is' or 'as'?

01:25:09 <Seth> isa .. as in rdf:type ... sumo:instance ... cyc:isa

01:26:09 <Seth> are two identical copies of the same book ... one on my desk and one on your desk the same thing?

01:26:47 <sbp> so by "stating", you actually mean a meta class

01:27:17 <sbp> stating != Statement

01:27:18 <Seth> i mean rdf:type rdfs:Class ... quite literally

01:27:30 <sbp> yep. I've got it now :-)

01:27:54 <sbp> interesting: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0355

01:28:41 <Seth> thing is M&S defined it as an instance and the class is rdf:Statement .. but maybe there is a better way .. might even help some of the probs with reification and PPS's kine of Russell's paradox issues . dont know but i have a gut feeling that it might

01:28:56 <sbp> they say that two statements with the same s p and o are different (properties applied to one do not entail the same p/o value on the other)

01:29:34 <Seth> right!

01:29:45 <Seth> just like instances of a class

01:31:15 <sbp> what they're saying doesn't seem counter intuitive. I can say that a car must have color, doors, and make properties: but that doesn't mean that two cards with the same color doors and make are actually the same car

01:31:31 <sbp> unless one of those properties were unambiguous :-)

01:32:18 <sbp> So I think that the question is whether rdf:subject, rdf:property, and rdf:object together make an unambiguous property set

01:32:55 <sbp> (and the answer: no)

01:33:20 <Seth> well obviously it dont .. not if this reified thing is gonna be a stating and not the abstract triple in the sky

01:33:41 <Seth> and that is the way the WG has been drifting

01:33:56 <sbp> so you agree with the descision? that's good

01:34:19 <Seth> oh yes i applaud it :)))

01:35:33 <Seth> how about you?

01:35:45 <sbp> Yep, I agree too

01:35:56 <sbp> We seem pretty much agreed on things tonight/day. We're doing well :-)

01:36:02 <Seth> how about AaronSW .. where does he fall in all this?

01:36:11 <sbp> I'm not sure - I'll have to ask him

01:36:40 * sbp reads

01:38:36 <sbp> er... I'm reading http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0391

01:39:58 <sbp> I love this debate. I've read so much about it that I don't think I care anymore

01:40:11 <Seth> me either

01:40:32 <sbp> FWIW, it's funny that Aaron brought up the old "make a QName and call it a Fragment" approach

01:40:56 <AaronSw> huh?

01:41:00 <sbp> it's beings up the problem of what happens when you use an HTTP URI to address that QName

01:41:32 <sbp> funny not as in "it sucks", but as in "quirky, nice that it was put forward"

01:42:50 <sbp> for example: <http://example.org/myFrag> :res <http://example.org/>; :frag "myFrag" . <http://example.org/> :res <http://example.org/myFrag>; :frag "blargh" .

01:43:16 <AaronSw> How is that a QName?

01:43:32 <sbp> it has a namespace and a local name

01:43:38 <sbp> looks like a QName of sorts to me

01:43:56 <sbp> there's nothing wrong with that. It just is

01:44:19 <AaronSw> Cool

01:44:45 <AaronSw> Cool! bwm is going along with the main thrust of my idea

01:44:57 <sbp> you could come up with a bit of syntax for it, even: _:x :resfrag "<http://example.org/>frag" .

01:46:57 <sbp> heh, I like Dan's reply

01:49:20 <sbp> oh, I really do love this whole conversation. In years of discussion, we haven't made one iota of progress

01:52:23 <sbp> A nice practical answer would be: "all the while the RDF namespace uses a URI-view, using URI-views is fine"

01:55:39 <AaronSw> Here's my real-world example:

01:55:40 <AaronSw> I want to create an HTTP proxy that will allow users to query an RDF database by simply typing URLs into their web browser. The browser will consult the proxy, which will consult the RDF database for everything with that subject. If the subject has a '#' in it, I won't be able to get that information (since it's not part of the HTTP protocol) and thus won't be able to respond properly to the user. Do you suggest we fix every web browser and HTTP implementation?

01:57:43 <sbp> heh, you've been working on that

01:58:14 <AaronSw> the project? yeah

02:00:22 <Seth> why wont you be able to get the frag id info?

02:00:36 <AaronSw> because the HTTP protocol explicitly does not send it

02:01:30 <Seth> oh

02:02:36 <Seth> have them type it in a form

02:03:31 <AaronSw> that would defeat the purpose, which is to make it feel like they're surfing the real web.

02:03:41 <AaronSw> i want the transition to be seamless, to increase adoption.

02:03:52 <Seth> so make your own client

02:04:42 <AaronSw> i don't want to do that. 1) it's hard to do right (see Netscape) and 2) there are tons of HTTP clients and people have odd preferences about them and 3) i'd like to be able to do this as a hosted service.

02:05:16 <Seth> could you make it a plug in for Mozilla?

02:05:59 <Seth> anyway i see your point, Aaron

02:06:14 <AaronSw> A lot of people don't like Mozilla.

02:06:30 <AaronSw> I don't use Mozilla as my daily browser.

02:06:34 * sbp is searching frantically for a quote about being able to describe one thing from multiple document on the Web independently, but can't find it! argh

02:06:48 <AaronSw> hm.

02:07:03 <AaronSw> (This isn't a made up goal, BTW, I want this for the Plex.)

02:07:08 <sbp> I read it just the other day, I'm sure. And I'm pretty certain that it was in DesignIssues

02:08:10 <Seth> thing is that it would be foolish to think that the whole w3c is gonna change all the uri references for just this one application

02:09:16 <sbp> what's wrong with xena? ".google" doesn't work...

02:09:25 <MikeM> fragmetns are client side thing.....

02:09:37 <AaronSw> exactly!

02:09:48 <AaronSw> .google test

02:09:55 <AaronSw> hm

02:13:10 <AaronSw> there's some design issue where timbl explains fragids as client-side things

02:13:23 <AaronSw> it's the one with the big crazy picture

02:14:12 <AaronSw> aha! Web Model

02:14:33 <Seth> anyway just putting in a URI without giving the context of the uri wont give you much of a useful retrieval .. me thinks

02:14:52 <AaronSw> I get great retrieval with URIs now-a-days.

02:14:53 <AaronSw> like http://www.w3.org/DesignIssues/Model

02:15:31 <AaronSw> that doens't look good

02:16:23 <sbp> A:"First, the triples are not connected. An RDF(S) document could be missing some of the triples." seems to run contrary to ""The design of the Web fundamentally differed from traditional hypertext systems in sacrificing link integrity for scalability. [...] no tool that looks at the Web as a whole can assume consistency. [...] A requirement, therefore, of a data model for the Semantic Web is that there should be no fundamental constraint relating what is said, what

02:16:23 <dc_rdfig> commented item A

02:17:04 <sbp> that's the closest quote that I can find :-)

02:17:48 <AaronSw> cut off: relating what is said, what

02:18:14 <sbp> ugh... thanks for pointing that out

02:18:26 <sbp> A: it is said about, and where it is said." in [http://www.w3.org/1999/04/WebData#tvpi|Web Architecture: Describing and Exchanging Data]

02:18:27 <dc_rdfig> commented item A

02:18:55 <AaronSw> I don't see the contradiction

02:19:04 <sbp> http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0095 is a good summary of the current WebONT position...

02:19:24 <MikeM> me thinks that RDF needed its own URI scheme: rdf:<url>:<rdfid>

02:19:29 <sbp> contradiction: I needed to quote a bit more material from the "Trouble With Triples" document

02:19:43 <AaronSw> mind if i quote you in my mail, MikeM?

02:19:49 <AaronSw> Trouble with Triples: ah!

02:19:53 * MikeM grins

02:20:03 <MikeM> sure! ;-)

02:20:08 <MikeM> I'll deny it in the morning though!

02:20:20 <sbp> heh

02:21:09 <sbp> A:cf. [http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0076|Trobule With Triples] above, which argues that "All triples are independent" is a problem in RDF

02:21:09 <dc_rdfig> commented item A

02:21:28 <MikeM> seriously, RDF's use of frags has caused so many problems.. If RDF needs something like it that badly then make it a new URI scheme that does it _right_

02:22:08 <sbp> s/rdf:<url>:<rdfid>/rdf:<rdfid>:<url>/

02:22:08 <AaronSw> Heh.

02:22:12 <sbp> and that still wouldn't work

02:23:40 <sbp> reminds me of some of Patrick's proposals...

02:24:13 <MikeM> possibly... but without all that "lets redefine the entire URI model while we're at it"

02:24:22 <sbp> heh. yeah

02:24:49 <monkinetic> monkinetic is now known as rm

02:24:49 <MikeM> I tried that argument for almost 3 years before I realized it wasn't really useful anyway....

02:25:57 <sbp> it seems that everyone answers the lesser important questions differently, but that there's general consensus on the main things

02:26:39 <sbp> for example, some people don't see a difference between HTTP URIs and URNs...

02:27:18 <MikeM> well, you still have to be careful. We're all really insulated and just because we say 'everyone' doesn't mean we've really actually asked _everyone_ who's using the stuff.... Come by an IETF meeting one day and see how the layer 3 and 4 guys are using URIs....

02:29:37 <sbp> the pivotal arguments revolve around the success of "http:" as a scheme, and as a protocol

02:30:56 <sbp> the success of that particular scheme is bizarre because it was intended to be that way, and now it's what most people think of when you mention "URL". OTOH, when people invent new schemes, they can't expect them to take off in the same way

02:31:17 <sbp> as RoyF always says, it's a social contract. The consensus is the most important aspect

02:31:48 <sbp> and you can't define a consensus: you can only drive a single bit, influence others as much as you can, and try to model and describe it

02:32:07 * sbp is in late-night rambling mode

02:33:04 <MikeM> right... but as a basic infrastructure element you can't let one communities concensus override another communities... that's why IP has been so succesful: everyone agrees that, at some point, you live with one standard. But you make sure that it can handle nearly anything under the sun..... I.e. the hourglass model.

02:34:37 * MikeM is rambling too....

02:34:38 <sbp> hence the need for standardization. But I still believe that if the masses decide to boycott a specification - even if they're wrong to do so - you have no choice but to go along with the crowd

02:34:51 <AaronSw> Dan described URIs as the hourglass model once.

02:34:54 <AaronSw> DanC

02:36:21 <MikeM> I guess my point is: which crowd are you listening too..... there are multiple groups using IP: sessions, command response, multicast, anycast, unicast, etc.... if one wants to say "lets change IP to make our lives easier" then who says that crowd wins over the other 'crowds'?

02:37:28 <AaronSw> What was Joe English referring to when he said 'Also, alas, one of the clearest descriptions of the "Semantic Web" that I have yet to see.'?

02:37:38 <sbp> ah, well then you have a fight on your hands :-) I meant that once you reach a critical points, there's no point in fighting it anymore. The "Point of No Return", I guess

02:38:45 <sbp> Aaron: is that a trick question, or are you really asking?

02:38:53 <AaronSw> I'm really asking.

02:38:57 <AaronSw> I don't read XML-DEV anymore.

02:39:30 * sbp neither

02:39:59 <sbp>http://bodoni.village.virginia.edu/conferences/xml2000/WEB/burkett_william.HTM

02:40:00 <dc_rdfig> D: http://bodoni.village.virginia.edu/conferences/xml2000/WEB/burkett_william.HTM from sbp

02:40:09 <sbp> D:|The Question of Semantics

02:40:10 <dc_rdfig> titled item D

02:40:21 <sbp> D:By William Burkett

02:40:21 <dc_rdfig> commented item D

02:41:17 <sbp> I'm surprised that a Google search for that quote yields nothing

02:41:28 <AaronSw> it's pretty recent

02:41:37 <sbp> perhaps you misquoted? I thought I'd come across it before

02:41:39 <AaronSw> thread is WSIO vs. Semantic Web

02:42:52 <rm> WSIO?

02:43:10 <AaronSw> Web Services In Orangutans?

02:43:31 <rm> oh cool - orangutans are cool

02:43:31 <sbp> http://lists.xml.org/archives/xml-dev/200202/msg00747.html

02:43:54 <sbp> [[[

02:43:55 <sbp> Also, alas, one of the clearest descriptions of the "Semantic Web"

02:43:55 <sbp> that I have yet to see. And I *still* don't understand it :-)

02:43:55 <sbp> ]]]

02:44:53 <sbp> in reply to http://lists.xml.org/archives/xml-dev/200202/msg00689.html

02:45:40 * AaronSw thinks semanticWeb-long!

02:45:53 <sbp> wow, that is a good description

02:46:10 <sbp> it's more like a semanticWeb-paragraph

02:46:11 <AaronSw> the semantic web is ... hooking up databases!

02:46:34 <rm> i don't see how web services and the semantic web are even in a "vs" relationship

02:47:07 <MikeM> same here....

02:47:25 <MikeM> at least to me its pretty clear exactly what the SW is....

02:47:33 <sbp> I wasn't too impressed with that article either

02:47:50 <sbp> the author takes a decidedly WS-oriented view

02:48:08 <sbp> it appears as if he feels threatened by the SW. At least, that it my cursorary interpretation of it

02:48:17 <sbp> pff: cursory

02:48:30 <AaronSw> what article?

02:48:44 <MikeM> yep....

02:48:50 <sbp> http://www.sys-con.com/webservices/articlenews.cfm?id=176

02:48:53 <rm> interesting: i see the SW as being about information humans use. WS are about computers talking to each other

02:48:58 <sbp> mentioned on RDF IG

02:49:03 <sbp> [the mailing list]

02:49:11 <AaronSw> oh, i don't read email any more

02:49:16 <sbp> heh, heh

02:49:35 <sbp> I agree, rm

02:49:43 <rm> thx.

02:49:43 <AaronSw> Heh, this JPG title image is hilarious all on its own: http://www.sys-con.com/webservices/archives/titleimages/w3c.jpg

02:49:58 <MikeM> at least to me the SW has always been about using the web as a knowledge store by using URIs as identifiers to talk about relationships between two other URIs.....

02:49:59 <sbp> well, the SW is information that humans use, but in a machine processable format. That's the important bit

02:50:17 <sbp> resource denoted by URIs...

02:50:22 <sbp> :-)

02:50:37 <rm> of course

02:50:47 <MikeM> but right now I"m finding that playing Ghost Recon on the 'net much more compelling. ;-)

02:50:52 <rm> but it's like comparing html to xml-rpc

02:51:27 <sbp> that'd be like comparing XML RDF to SOAP

02:51:45 <rm> LOL

02:51:57 * rm agrees

02:52:03 <sbp> :-)

02:52:19 <sbp> .google "Ghost Recon"

02:52:20 <xena> "Ghost Recon": http://www.ghostrecon.com

02:55:35 <MikeM> tom clancy military first person shooter....

02:56:18 <sbp> looks pretty good

02:57:01 <rm> too complicated

02:57:08 <rm> UNREAL UNREAL UNREAL

02:57:11 <rm> hehe

02:57:22 <rm> more twitch - less thinking :-D

02:58:50 <MikeM> hehe... I like using the weapons... i.e. I really doubt if I'll ever actually get to use a real OICW in my lifetime.... (a BMG .50 on the other hand.... one of those is sitting at the gun store down the street with my name on it)

02:59:32 <AaronSw> [MikeM, the crazy libertarian]

02:59:36 <AaronSw> ;-)

02:59:41 <MikeM> hehe....pretty much!

03:02:40 * MikeM thinks about getting completely out of this business and going into gun running.....

03:03:18 <MikeM> :'=_

03:03:24 <MikeM> ;-)

03:03:42 <sbp> [the former smiley was Mike shooting at his keyboard]

03:04:05 <MikeM> hehe.....that or its just to damn late to be typing.....

03:04:28 <MikeM> ok, time for bed....

03:04:34 <sbp> c'ya

03:04:40 <MikeM> ciao

03:08:03 * sbp works on EARL

03:20:35 <em> dc_rdfig:view

03:20:35 <dc_rdfig> A: Layering the Semantic Web: Problems and Directions (http://www-db.research.bell-labs.com/user/pfps/webont/layering/layering.html)

03:20:36 <dc_rdfig> B: The Trouble with Triples (http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0076.html)

03:20:37 <dc_rdfig> C: A stating of a triple is a class of things. (http://lists.w3.org/Archives/Public/www-rdf-logic/2002Feb/0094.html)

03:20:38 <dc_rdfig> D: The Question of Semantics (http://bodoni.village.virginia.edu/conferences/xml2000/WEB/burkett_william.HTM)

03:20:52 <em>http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0088

03:20:53 <dc_rdfig> E: http://lists.w3.org/Archives/Public/www-webont-wg/2002Feb/0088 from em

03:21:26 <em> E:A view from Danbri re layering

03:21:26 <dc_rdfig> commented item E

03:21:34 <em> E:bookmarking for future reference

03:21:34 <dc_rdfig> commented item E

03:21:39 <em> E:!A view from Danbri re layering

03:21:39 <dc_rdfig> commented item E

03:21:44 <em> E:|A view from Danbri re layering

03:21:45 <dc_rdfig> titled item E

03:22:12 * em looking for that next version of chump that provides editing!

04:31:17 <AaronSw> DanC is darting at shadows:

04:31:17 <AaronSw> RFC2396 does not say that a URI+fragid does not identify anything.

04:31:21 <AaronSw> I didn't say that it did.

04:31:40 <AaronSw> heh: "consensus by exhaustion"

04:53:25 <AaronSw> DanC, what's a failure mode in this context?

04:54:03 <AaronSw> .foldoc failure mode

04:54:04 <xena> error: unable to define failure mode

05:05:28 <AaronSw> Sometimes I feel like this URI-URIview discussion is DanC's way of teaching me debate.

05:05:34 <AaronSw> I'm definitely better at it this time around.

05:06:08 <sbp> heh, heh

05:06:59 <DanCon> failure mode: you said fragids don't work. what's a specific scenario in which they fail to work?

05:07:15 <AaronSw> Cool. I think I've got two good ones, then.

05:07:42 <AaronSw> By fail to work, do you mean a real-world story, or a model inconsistency?

05:07:48 <DanCon> real-world story

05:07:52 <DanCon> bits and bytes.

05:07:56 <DanCon> deployed software.

05:08:08 <AaronSw> Oh, well I have that story above a little bit.

05:08:24 <AaronSw> It's this HTTP proxy I want to make, but I can't, since all the web browsers won't give me the fragment ids

05:09:11 <DanCon> why would you expect them to?

05:09:24 <DanCon> that's not a failure. that's the system working as designed.

05:09:34 <AaronSw> Well, I wouldn't personally. Unfortunately, this HTTP proxy queries an RDF database which does expect fragids included.

05:10:29 <DanCon> in what way does this RDF database expect fragids included?

05:10:50 <AaronSw> the HTTP proxy does a query of the form: <uriUserTypedIn> ?x ?y ?

05:11:10 <AaronSw> there's no way for the user to get a # in that query, and thus they miss out on a ton of rdf

05:11:56 <DanCon> umm... EricP's algae server does that sort of stuff all the time; users type URIs in HTML forms and such. works great.

05:12:12 <DanCon> I'm not seeing how an HTTP proxy is relevant.

05:12:15 <AaronSw> yeah, but this is an http proxy -- i don't want users to type it into an html form

05:12:24 <AaronSw> i want to make it so their browsing experience is unchanged.

05:12:48 <AaronSw> [This is for the Plex project -- they'd be querying the Plex here.]

05:12:55 * AaronSw sends email

05:13:18 <DanCon> then what 'ton of RDF' are they missing out on? anything in the db of the form <pg> <prop> <val> where <pg> is an http-page isn't gonna have a # in the subject. Hence they won't miss anything.

05:13:46 <AaronSw> why isn't it going to have a # in the subject?

05:14:10 <DanCon> because addresses of HTTP-accessable things don't have #'s in them.

05:14:24 <AaronSw> erm. maybe i should explain again.

05:14:50 <AaronSw> they want to look up all things in the db

05:15:00 <AaronSw> not just http-accessable ones

05:15:17 <DanCon> 'all things' -- including, say, my nose?

05:15:28 <AaronSw> yep

05:15:30 <DanCon> ok

05:15:51 <AaronSw> [is your nose only identifiable via a #nose fragid on the Person format? ;)]

05:16:36 <DanCon> I dunno about only... but I'd be happy to say that http://www.w3.org/People/Connolly/personalKB#nose refers to my nose.

05:17:23 <AaronSw> yeah, unfortunately our users can't find that out, since the browser doesn't pass on the fragid

05:17:38 <AaronSw> nor can they do: http://get.theinfo.org/http://www.w3.org/People/Connolly/personalKB#nose for the same reason

05:17:50 <DanCon> can't find what out? they can GET http://www.w3.org/People/Connolly/personalKB and find it out just fine.

05:18:08 <DanCon> why can't they do get.theinfo?

05:18:39 <AaronSw> well, they'll get your whole personalKB, not the specific thing (yoru nose) they were interested in

05:18:48 <DanCon> yes; as designed.

05:19:11 <AaronSw> erm, the difference is instead of getting the file, they're getting back an rdf description

05:19:13 <DanCon> after they get the whole personalKB, they can look inside for info about #nose.

05:19:16 <AaronSw> which says nothing about your nose.

05:19:38 <AaronSw> because when people say stuff about your #nose in rdf, they don't repeat it about the file

05:19:45 <DanCon> why does it say nothing about my nose? I guess I don't understand what service get.theinfo provides.

05:20:09 <AaronSw> get.theinfo returns HTML-formatted RDF with the uri as the subject.

05:20:12 <AaronSw> just like my proxy

05:20:38 <AaronSw> if i were to assert something about your #nose, it'd never get found

05:20:47 <DanCon> oh; that's an interesting service. but... ouch! it's not proxying!

05:21:13 <AaronSw> it's a special case of proxying, like the crit.org mediator and stuff like that

05:21:23 <DanCon> no, mediation isn't proxying.

05:21:48 <AaronSw> if you say so, but i'd still like to use the "HTTP Proxy" function in my web browser for it.

05:22:09 <DanCon> that's contorted.

05:22:25 <AaronSw> why?

05:22:38 <AaronSw> was third voice contorted too then?

05:22:44 <AaronSw> i want to see crit annotations about every page i visit

05:22:53 <AaronSw> i want to see rdf annotations about every page i visit

05:23:21 <AaronSw> but even if we ignore that, and look at the 'just paste http://get.theinfo.org/ in front of the uri' case, the problem remains

05:23:32 <DanCon> didn't you say this service takes RDF and turns it into HTML? or did I misunderstand?

05:23:53 <DanCon> yes, I think the conversation will be simpler if we leave the proxy case aside.

05:23:58 <DanCon> now... what problem remains?

05:24:23 <AaronSw> so i want to know about http://www.w3.org/People/Connolly/personalKB#nose

05:24:35 <DanCon> ok.

05:24:38 <AaronSw> so i paste http://get.theinfo.org/ in front and get http://get.theinfo.org/http://www.w3.org/People/Connolly/personalKB#nose

05:24:56 <AaronSw> get.theinfo only gets /http://www.w3.org/People/Connolly/personalKB

05:25:04 <AaronSw> and thus doesn't return all the information it has about your #nose

05:25:05 <DanCon> yup. as designed. I'm with you.

05:25:31 <AaronSw> my user gets upset: Jane said she commented on Dan's nose and i want to see what she said!

05:26:18 <DanCon> then ask for http://get.theinfo.org/?about=http://get.theinfo.org/http://www.w3.org/People/Connolly/personalKB%XXnose

05:26:32 <AaronSw> but that's already a web resource

05:26:41 <AaronSw> %23, BTW as Mac OS X constantly reminds me

05:26:49 <AaronSw> now we're overloading that resource

05:26:57 <DanCon> overloading?

05:27:14 <AaronSw> well we're using it to mean the %23 resource and the resource+fragment

05:27:29 <AaronSw> and Jane just wants the latest nose-news, she doesn't understand all this %23 gunk -- usability goes down.

05:28:14 <DanCon> nowwhere in web architecture does it say you can take a URI+frag, whack some stuff on the front, and get something interesting as a result.

05:28:41 <AaronSw> hm, perhaps but it's very useful: web.archive.org, get.theinfo.org, crit, malkovic

05:28:55 <DanCon> jane need not understand %23; she just pastes http:...#nose into the form.

05:29:18 <AaronSw> form? usability goes down again

05:29:31 <DanCon> goes down vs what?

05:29:44 <AaronSw> vs. what she'd be able to do if rdf didn't have fragids

05:29:50 <DanCon> the usability of address-box-surgery is much lower than forms, I'd say.

05:31:12 <DanCon> goes down vs what? I ask again; pls answer directly, rather than by contrast. Are you talking about typing into address fields?

05:31:13 <AaronSw> well, it's not in the proxy case

05:31:37 <AaronSw> a vs. what question seems to be asking for a contrast!

05:32:06 <DanCon> nope

05:32:55 <AaronSw> i feel we're not connecting somewhere

05:33:12 <AaronSw> perhaps if i explain my point underlying the story it'll make more sense

05:33:17 <DanCon> ok

05:33:48 <AaronSw> [excuse me while i find my mail to quote from]

05:34:32 <AaronSw> you say that absolute-uri-refs are first class objects

05:34:45 <AaronSw> I expect first-class objects to be things I can ask an HTTP server/proxy

05:34:45 <AaronSw> about. (HTTP being one of many possible protocols to get information about

05:34:45 <AaronSw> Resources.)

05:34:46 <AaronSw> I can't do this with URI-references.

05:34:46 <AaronSw> Therefore, I don't think URI-references are first-class objects in the

05:34:46 <AaronSw> currently-deployed world.

05:35:25 <DanCon> you have odd expectations.

05:35:35 <AaronSw> really?

05:35:54 <DanCon> a first-class object is something that I can refer to in the global scope. something I can talk about with RDF.

05:36:03 <AaronSw> that's circular

05:36:27 <DanCon> there aren't many definitions of 'object' that aren't circular. Hm...

05:36:31 <AaronSw> in rdf we talk about first-class objects. first-class objects are things i can talk about in rdf.

05:36:41 <AaronSw> my point is that we can't talk about them on the web

05:36:43 <AaronSw> .time cst

05:36:43 <xena> Feb. 14, 2002 11:36 pm US/Central

05:37:38 <DanCon> we can talk about them on the web and we do.

05:37:57 <AaronSw> my HTTP server can't talk about them (specifically), my HTTP client can't ask about them

05:38:18 <jah-mac> .time est

05:38:18 <xena> Feb. 15, 2002 12:38 am US/Eastern

05:38:26 <DanCon> it can't GET a representation of them, but it can ask about them. e.g. via algae.

05:38:53 <AaronSw> that's rdf again. i'm looking for web examples

05:39:09 <AaronSw> the web seems to be the things I can GET representations of

05:39:18 <DanCon> ok, stick http://www.w3.org/TR/#xml-infoset in your browser.

05:39:32 * AaronSw command-clicks that URI

05:39:32 <AaronSw> I get a 404.

05:39:38 <AaronSw> at http://www.w3.org/TR/%23xml-infoset

05:39:56 <DanCon> ? who said #23xml-infoset?

05:40:05 <AaronSw> Mac OS X seems to assume that's what you meant.

05:40:10 <DanCon> bug.

05:40:15 <AaronSw> it's really strange.

05:40:35 <AaronSw> but, ignoring back, i get this big page unrelated to the XML infoset.

05:40:38 <DanCon> the web is more than things you can GET representations of. It's also things that those representations talk about.

05:40:55 <AaronSw> i disagree.

05:40:57 <DanCon> unrelated? XML infoset is one of the things that page talks about.

05:41:08 <AaronSw> oops i meant most of the stuff on it was unrelated

05:41:29 <AaronSw> if my page talks about my cat, but doesn't give it a URI, i don't think it's on the Web yet.

05:41:57 <DanCon> well, then we disagree.

05:42:02 <AaronSw> HTTP can't redirect fragments, it can't conneg fragments, it can't give 404s for fragments, etc. HTTP is the web for most people.

05:42:12 <DanCon> most people are confused.

05:42:44 <DanCon> but most people would agree that your cat is on the web, I bet.

05:43:25 <AaronSw> even if my page just says "And that is trans-nucleophusics...where was I, my cat stepped on me again."?

05:43:42 <DanCon> yup.

05:44:08 <AaronSw> perhaps in the same sense that when my mom saw me in the background of a tv shot, she said "look! aaron's on tv!"

05:44:18 <DanCon> yup

05:44:35 <AaronSw> but not in the more meaningful sense of, "there was a tv show with/about aaron"

05:44:49 <AaronSw> [more meaningful to my mom]

05:45:30 <AaronSw> practical problem: if foo#bar foo is a multigigabyte file, then how do I find out about bar?

05:45:51 <AaronSw> w/o overwhelming my 56k modem

05:46:16 <DanCon> you either GET foo or look in nearby knowledge bases or whatever.

05:46:23 <DanCon> web architecture is not devoid of practical problems.

05:47:08 <DanCon> nor usability issues/problems.

05:47:13 <AaronSw> true.

05:47:37 <AaronSw> well, we disagree about what the web is.

05:47:46 <DanCon> I think we have made some progress; it's pretty clear what we disagree about.

05:47:49 <AaronSw> yeah.

05:48:39 <DanCon> call it truce? Or jabber on? I can think of an argument I'm willing to try.

05:48:51 <AaronSw> i've got one more argument too

05:49:03 * DanCon tries to shoot first...

05:49:17 <DanCon> I've got a bookshelf full of books. A big space of information.

05:49:46 <DanCon> if you were in the room with me, we could use phrases like 'fourth one from the left on the top shelf' to refer to stuff on the shelf.

05:50:35 <DanCon> seems to me the space of information includes the things talked about in the book; i.e. I can refer you to any of the things discssed in the book ala 'the guy pictured on page 45 of that big blue book'

05:51:20 <DanCon> by analogy, the web is all the stuff people that use the internet can share... all the things that we can (fairly) easily access using a common naming syntax.

05:52:24 <AaronSw> your analogy clearly shows the flaw.

05:52:36 <AaronSw> it only works because books only have one form.

05:52:50 <DanCon> in the bookshelf example, the shared information space between me and somebody else in the room is not limited to talking about the books themselves. The fact that both of us can read the books means we also share the contents of the books.

05:53:09 <AaronSw> in the web, we are not limited to static books who only have one physical form

05:53:33 <AaronSw> on the web we talk about real resources in the real world, and ask HTTP servers, companies, and people to tell us more about them.

05:54:10 <DanCon> no, it would work if the books looked slightly different to you than they do to me; say... if we were in two different rooms, with books organized similarly, but your copies are in english and mine are in spanish.

05:54:33 <AaronSw> what if i'm blind and i have audiotapes? my tapes have no page 65.

05:54:58 <DanCon> the analogy relies on a common syntax for addressing.

05:55:18 <AaronSw> exactly!

05:55:33 <AaronSw> we have that with the books (they've got titles and authors)

05:55:39 <AaronSw> but we don't have that with their content

05:55:44 <DanCon> exactly what? how does that show that our conversations are limited to the books themselves, rather than the things talked about in the books?

05:55:46 <AaronSw> the books are the resources, their content is the entity.

05:56:27 <AaronSw> it only works within a small range of transformations of the entity (i.e. those which don't ruin your adressing mechanism). content-negotiation, HTTP, and certainly other forms of URI resolution aren't limited to that.

05:56:36 <DanCon> yes, they are.

05:56:48 <AaronSw> huh?

05:57:00 <AaronSw> conneg works between html files and audio files just fine

05:57:06 <DanCon> not really

05:57:32 <AaronSw> reading the book, vs. asking my friend john also works, but john's got no page 65.

05:57:46 <AaronSw> reading the large-print version, since my eyesight is bad also throws me off

05:58:12 <DanCon> if the page numbers don't survive large-print transformation, then your addressing syntax failed.

05:58:40 <AaronSw> yes it has. and the author never authorized that adressing syntax, it was a byproduct of the medium.

05:58:48 <DanCon> (believe me, page numbers *do* work in audio tapes and large-print editions. sighted people use them so much that the blind people are forced to support them.)

05:59:06 <AaronSw> (not in my experience)

06:00:04 <DanCon> re conneg across HTML and audio: that's a web architecture question that I've wanted to get nailed down for quite some time...

06:00:25 <DanCon> BLURB: how far is conneg supposed to stretch?

06:00:26 <dc_rdfig> F: how far is conneg supposed to stretch? from DanCon

06:00:32 <DanCon> F:from HTML to audio?

06:00:32 <dc_rdfig> commented item F

06:00:35 <DanCon> F:I don't think so.

06:00:35 <dc_rdfig> commented item F

06:00:37 <DanCon> logger_1, pointer?

06:00:38 <DanCon> See http://ilrt.org/discovery/chatlogs/rdfig/2002-02-15#T06-00-37

06:00:52 <DanCon> F:see [discussion|http://ilrt.org/discovery/chatlogs/rdfig/2002-02-15#T06-00-37]

06:00:53 <dc_rdfig> commented item F

06:01:06 <AaronSw> what if i put up an xml file and and someone uses xpointer on it?

06:01:15 <AaronSw> pretty much all conneg is useless at that point.

06:01:29 <DanCon> probably so.

06:01:31 <AaronSw> urirefs are device/medium-dependent in that way

06:01:45 <DanCon> to some extent, yes.

06:02:36 <AaronSw> the author doesn't get to choose them (other than choosing the medium) nor does she get to redirect them, or any of those useful things they can do.

06:02:41 <DanCon> F:I wonder if exploring this would be a good use of the TAG's time.

06:02:41 <dc_rdfig> commented item F

06:02:58 <AaronSw> it's really quite annoying in my experience. like when you want to expand a fragment of a page into a full page of its own.

06:03:07 <DanCon> author doesn't choose fragids? I think I misunderstood.

06:03:35 <AaronSw> not if they put up an xml document

06:03:50 <AaronSw> xpointer gives it a zillion possible fragids anyone can use

06:04:03 <DanCon> but they were all implicitly licensed by the author.

06:04:12 <AaronSw> in what way?

06:04:13 <DanCon> it's questionable whether XPointer is a good idea.

06:04:51 <DanCon> the author chose to use XML, and in doing so, licensed all the stuff that the specs say about XML.

06:05:21 <AaronSw> so do i have a responsibility not to create a new paragraph now?

06:05:27 <AaronSw> otherwise i'd break all those fragids

06:05:48 <DanCon> yes.

06:06:38 <DanCon> there's always a tradeoff around that sort of breakage, but yes, the author has a certain amount of freedom and control, and hence bears some responsibility.

06:06:58 <AaronSw> that's really annoying. i can't add paragraphs to all my html documents? urgh.

06:07:04 <AaronSw> i think my way makes a lot more sense

06:07:07 <DanCon> but so does the fool who linked to the 17th paragraph of a document marked "subject to change"

06:07:23 <AaronSw> fragment ids only make sense in the context of a frozen set of bits

06:07:26 <AaronSw> - my view

06:07:34 <AaronSw> or a relatively frozen

06:08:02 <AaronSw> my way makes sense ('cuz it's the REST way), your way brings up all these issues

06:08:06 <DanCon> yes, relatively frozen. quasi-static. slowly-changing. That's what web architecture is all about.

06:08:08 <AaronSw> F:I think so, for one.

06:08:08 <dc_rdfig> commented item F

06:08:43 <DanCon> I think you're playing ostrich-in-the-sand if you pretend these issues aren't there.

06:08:53 <DanCon> nothing in REST says that these issues aren't there.

06:09:25 <AaronSw> the issues aren't there because REST doesn't give you the license to assume that fragids won't break.

06:09:42 <AaronSw> REST says fragids make sense in context, which is how the Web today works

06:09:50 <AaronSw> RDF is dragging them out of that context.

06:10:11 <AaronSw> i mean, think of xanadu for a second. fine-grained links are attached to a particular vesion of a document.

06:10:21 <DanCon> the web today relies on fragids not breaking; not quite as much as it relies on 200 vs 404, but almost as much.

06:10:38 <DanCon> RDF isn't changing anything.

06:11:59 <AaronSw> in the current web fragids have this special relationship to uris, rdf ignores that.

06:12:38 <DanCon> if you put up a page about christopher columbus, advertise it widely as such, and then write over it with stuff about daffy duck, you do damage to the web. You're not going to kill the whole thing, but you're going to do a little damage. It might be damage that's offset by some larger good. I think adding a paragraph to an XML document does likewise. Or deleting an anchor from an HTML document.

06:12:47 <DanCon> RDF doesn't ingore it.

06:13:35 <AaronSw> what's it do with it then?

06:13:53 <DanCon> nothing special. just uses URIs as designed.

06:14:19 <AaronSw> but they're not uris!

06:14:36 <AaronSw> this is my final argument: RFC2396 says very clearly "The semantics of a fragment identifier is a property of the data resulting from a retrieval action, regardless of the type of URI used in the reference."

06:14:51 <AaronSw> it also says that the uri resulting from one is the part before the #

06:15:44 <DanCon> if you're not wiling to read 'URI' as in RFC1630, then you're being pretty antisocial.

06:15:58 <AaronSw> huh?

06:16:03 <AaronSw> oh, when you said it.

06:16:25 <AaronSw> i think you're antisocial. it's the pedantic web, after all.

06:16:38 <AaronSw> you're antisocial: err, being antisocial by confusing the terminology.

06:17:30 <DanCon> Shall I correct every grammatical and typographical error you make, even in this informal setting? Is that your idea of social?

06:17:53 <AaronSw> only when my typos and grammar errors mislead people

06:18:04 <AaronSw> your usage of URI is actively misleading, IMO

06:18:22 <DanCon> in this context?

06:18:39 <AaronSw> and when combined with other W3C actions it begins to look like part of a plan to change the definition of URI.

06:18:50 <DanCon> it's just habit, OK?

06:18:59 <AaronSw> ok, i didn't mean any offense.

06:19:09 <AaronSw> it's just confusing when it's the distinction itself we're discussing.

06:19:23 <DanCon> ok, if it's that much trouble, I'll rephrase:

06:19:29 <DanCon> RDF uses URI references as designed.

06:19:35 <AaronSw> anyway, you dodged the question: "The semantics of a fragment identifier is a property of the data resulting from a retrieval action, regardless of the type of URI used in the reference."

06:19:47 <AaronSw> - RFC 2396

06:19:53 <DanCon> what's the question?

06:19:54 <AaronSw> that does not sound like how rdf uses them to me

06:20:01 <AaronSw> hwo do you reconcile the two?

06:20:01 <DanCon> no?

06:20:08 <AaronSw> rdf doesn't retrieve

06:20:11 <DanCon> what's to reconcile?

06:20:37 <DanCon> er... data formats usually don't retrieve.

06:20:37 <AaronSw> CWM doesn't retrieve every URI-with-fragment in its knowledge base. How can it assume those semantics?

06:20:53 <AaronSw> [hmm, that's sort of a bad example]

06:21:17 <DanCon> you don't visit every place that I mention in conversation. Yet you understand what I mean when I say that I grew up in Prairie Village, a suburb of Kansas City.

06:21:43 <AaronSw> that's because the semantics of "Prarie Village" are independent of a retrieval action

06:22:11 <AaronSw> istm the only way that you know that my foo#bar and your foo#bar have the same semantics is to check hashes.

06:22:23 <DanCon> no, they're not. If I told you that Prairie Village has a population of 1,000,000, and you did a GET on it, i.e. came to visit, you'd discover that I lied.

06:22:51 <AaronSw> how does that change the meaning of prarie village?

06:23:17 <DanCon> change?

06:23:43 <AaronSw> we were discussing the meaning of the term, not whether you lied about it or not.

06:24:32 <DanCon> there are lots of ways to learn what pvks#population is; one is for me to tell you in conversation. (ala what you might get from an RDF file). another is to go visit first hand (ala GETting pvks).

06:25:04 <AaronSw> that would be true if pvks#pop had global meaning, but from that quote it seems it doesn't.

06:25:11 <AaronSw> "The semantics of a fragment identifier is a property of the data resulting from a retrieval action, regardless of the type of URI used in the reference."

06:25:21 <AaronSw> a retrieval action is allowed to return different things, no?

06:25:39 <AaronSw> global meaning -> global scope / consistent global meaning

06:25:58 <DanCon> return different: in the same way that the contents of web pages is allowed to change. slowly/gradually.

06:26:12 <DanCon> if all web pages returned totally random content, the web would break down.

06:26:17 <AaronSw> wha?

06:26:33 <AaronSw> so what about: us pop clock, current time, and rng

06:26:39 <AaronSw> those aren't allowed on the web?

06:27:06 <AaronSw> how about the random story generator? the random insult generator?

06:27:08 <DanCon> yes, if I told you /People/Connolly/ was about me, so you bookmarked it, but 2 days later it was about daffy duck, you'd be annoyed. If that happened to all your bookmarks and links all the time, you'd consider the web worthless.

06:27:25 <AaronSw> oh! i'm not saying the _resource_ changed, the entity changed.

06:27:39 <AaronSw> it's still a random story, it's just a different story every time you hit reload

06:27:51 <DanCon> if all web pages acted like the random story generator, the web would be worthless, no?

06:27:54 <AaronSw> the slashdot story generator does this, and people love to link to it.

06:28:09 <AaronSw> worthless: not at all!

06:28:32 <AaronSw> well, i mean, i'd like to have some static pages, but that's what Vary is about.

06:28:34 <DanCon> seriously? you'd be happy if *all* web pages returned random stuff each time you did a GET?

06:28:51 <AaronSw> you're missing the point.

06:28:57 <DanCon> no, you're missing the point.

06:29:02 <AaronSw> the web is about the resource staying the same and the entity changing.

06:29:31 <DanCon> and changing slowly, as a rule.

06:29:36 <AaronSw> not as a rule

06:29:47 <AaronSw> very much not as a rule.

06:29:56 <AaronSw> i mean, what's popular on the web? pages that change a lot

06:29:58 <DanCon> how about: changing predictably?

06:30:05 <AaronSw> stock quotes, weblogs, news sites

06:30:24 <DanCon> i.e. not changing randomly

06:30:29 <AaronSw> predictably: i think by that you mean the same thing as i mean by the Resource staying the same.

06:30:35 <DanCon> right.

06:31:19 <AaronSw> ok, so we agree that entities change

06:31:30 <AaronSw> now do we agree that the semantics of fragids change too?

06:31:49 <DanCon> I don't think so.

06:32:07 <DanCon> no more or less than the semantics of the rest of URI-references change

06:32:27 <AaronSw> i mean, if we're talking about pvps#pop and I say it's a prime number unless we agree on the instant in time for which i looked at the population.

06:32:39 <AaronSw> primenumber ^ it's not really useful ^ unless

06:33:24 <DanCon> but pvps#pop means 'the population of Prairie Village Kansas', not 12. it's late-binding, just like pvps

06:33:41 <AaronSw> not according to rfc2396!

06:33:52 <DanCon> s/12/120,000/. or whatever. it's less than a million

06:34:04 <DanCon> yes, even according to rfc2396

06:34:25 <AaronSw> "The semantics of a fragment identifier is a property of the data resulting from a retrieval action" does not say late-binding to me

06:34:31 <AaronSw> it goes on to delegate things to the mime type

06:34:44 <AaronSw> the mime type usually describes bits

06:34:47 <DanCon> yup; pvps#pop means 'whatever you get back'

06:34:58 <AaronSw> that sounds rather meaningless

06:35:11 <DanCon> no more or less meaningless than what pvps means

06:35:25 <AaronSw> exactly! we can't make any more assertions about #pop than we can about pvps

06:35:37 <AaronSw> (in fact, we can make rather less...)

06:36:09 <DanCon> er... have you switched to my side of the fence? it sounds like you're saying URI-references in general work just like URIs-without-fragments.

06:36:26 <AaronSw> hm, just when i thought you'd switched to mine. ;)

06:36:32 <AaronSw> i'm saying they don't work.

06:37:29 <DanCon> pvps : an entity you get back from a GET there :: pvps#pop : the actual population as of any GET

06:37:37 <AaronSw> if the semantics are defined by retrieval, than you can either: 1) not say anything interesting about them ("this is something i need to retrieve") or 2) say something about a bag of bits once identified by the fragment

06:37:58 <AaronSw> pvps#pop : the actual population as of any GET -- but within the context of a GET

06:38:08 <AaronSw> so we'd need to talk about the GETs, not the uri-ref

06:38:15 <DanCon> huh?

06:38:48 <AaronSw> what useful things can we say about it?

06:38:55 <DanCon> look, the phrase "the population of PVKS" is perfectly sensible, right? it has a certain time-invariant meaning. and it can be evaluated at any given point.

06:39:06 <AaronSw> yes, but the fragment is not that phrase.

06:39:19 <AaronSw> that phrase is analogous to a normal URI

06:39:21 <DanCon> it's not? I made it up. I say it is.

06:39:25 <AaronSw> fragments have this wacky propery.

06:39:40 <AaronSw> and no matter how much you say it is, you can't change that w/o going thru ietf process.

06:40:08 <AaronSw> "the format and interpretation of fragment identifiers is dependent on the media type [RFC2046] of the retrieval result."

06:40:20 <DanCon> change what? this is the way URIs and fragments work.

06:40:37 <AaronSw> you can't say what pvps#pop means

06:41:00 <DanCon> yes, I can. it has to agree with what you'd observe when you get there, but that doesn't stop me from saying things about it.

06:41:09 <AaronSw> you can't change its meaning from the RFC2396-defined value.

06:41:28 <DanCon> quite; I can't change its meaning. That doesn't mean I can't refer to that same meaning from RDF.

06:41:45 <AaronSw> ok, what's a useful rdf statement about it then?

06:41:49 <AaronSw> give me an example

06:42:08 <DanCon> e.g. pvps#pop < kansasCity#pop

06:42:20 <AaronSw> nope, can't say that.

06:42:22 <DanCon> which follows from suburb(pvps, kansasCity)

06:42:31 <DanCon> can't say that? why not?

06:42:37 <AaronSw> because "The semantics of a fragment identifier is a property of the data resulting from a retrieval action"

06:42:53 <AaronSw> nothing you retrieved at pvps#pop gave you the right to say that.

06:43:08 <DanCon> yup; and if you GET pvps and you GET kansasCity, you'll find that the populations obey the inequality.

06:43:25 <DanCon> I have the right regardless.

06:43:29 <AaronSw> maybe for you, but when I GET them they don't, because somebody bombed kansasCity

06:43:34 <DanCon> anyone can say anything about anytyhing.

06:43:51 <DanCon> yes, people lie, things change, such is life.

06:44:38 <AaronSw> it makes sense to say the value i got from pvps#pop when i downloaded it today is < the value i got from kanasasCity#pop

06:45:14 <DanCon> and it makes sense to say that inequality will always be the case (even though you might be lying)

06:45:34 <AaronSw> it's not a matter of lying

06:45:50 <AaronSw> it's simply something that no one will know.

06:46:39 <DanCon> er... I'm not 100% certain that China is still connected to the earth. But you and I both know what I mean when I write 'China', no?

06:46:57 <AaronSw> yes, but what do you mean when you write kanasasCity#pop?

06:47:07 <DanCon> I mean the population of kansas city.

06:47:26 <AaronSw> RFC2396 disagrees with you, it would seem.

06:47:34 <DanCon> nope.

06:47:57 <AaronSw> no?

06:48:01 <DanCon> the population of kansas city is what you'll get when you GET kansasCity, look up the mime type, and index by #pop

06:48:29 <AaronSw> ok, but it's still the number 100, not the pop of kansasCity

06:48:55 <DanCon> 100 is like the entity.

06:49:01 <AaronSw> maybe in a few hours it'll be the number 101 (or the ascii characters 1, 0 and 1)

06:49:10 <AaronSw> it's still not a resource :)

06:49:19 <AaronSw> 100: yeah, but it's the resource too

06:49:21 <DanCon> yes, it is. it's something with identity

06:49:35 <AaronSw> hm, i need a better term.

06:49:55 <AaronSw> it is a resource, but it's not a conept.

06:49:57 <AaronSw> err concept

06:50:12 <DanCon> 'the population of kansas city' is not a concept?

06:50:40 <AaronSw> it is, but it's not identified by that uriref

06:50:47 <DanCon> yes, it is.

06:51:07 <AaronSw> no, what's identified by that uriref "is a property of the data resulting from a retrieval action"

06:51:23 <AaronSw> the data is 100, not "this is something that really means the population of kansas city"

06:51:29 <DanCon> yup; the population of kansas city is a property of the data you'll GET from kansasCity

06:51:44 <AaronSw> that's so backwards.

06:51:49 <DanCon> nope.

06:52:12 <AaronSw> ok, so what if you get numbers#oneafter99, is that the population of kansas city too?

06:52:38 <DanCon> I dunno; I didn't mint numbers. you tell me.

06:52:52 <AaronSw> [Heh, the Beatles wrote 'Kansas City' and 'One After 909']

06:53:00 <AaronSw> [Well, performed at least]

06:53:12 <AaronSw> all i can tell you is that if you get/present you get back 100

06:53:19 <AaronSw> a property of 100 is that it's the population of kansas city

06:53:30 <AaronSw> so by your logic that identifies the population of kansas city also

06:53:34 <DanCon> what I don't get is: all this resource/entity stuff make sense to you, but you're unwilling to apply it to foo#bar.

06:53:45 <AaronSw> because the rfc says not to!

06:53:50 <DanCon> no, it doesn't.

06:53:56 <AaronSw> yes it does!

06:54:04 <DanCon> no, it doesn't. [whee!]

06:54:20 <AaronSw> maybe you should read the paragraph in its entirety

06:54:25 <AaronSw> the meaning is reallly quite clear

06:54:36 <DanCon> yup; and it agrees with RDF usage.

06:54:46 <AaronSw> nope, it disagrees

06:54:47 <AaronSw> i'd go ask some other folks but they're all asleep and out of town.

06:54:58 <DanCon> you can ask all the folks you like.

06:55:19 <AaronSw> when we meet in person i'll read it to you, i can't imagine you saying this wiht a straight face.

06:55:57 <AaronSw> either you're lying, or you're basing this on the meaning of something other than that spec.

06:56:11 <AaronSw> (like your experience with the WG)

06:56:32 <AaronSw> (or how you wish the web would be)

06:56:42 <AaronSw> - that's how it seems to me

06:57:00 <DanCon> I'm not wishing here. I'm perfectly happy to use exactly the words from the spec. witness: the population of kansas city is a property of the data you'll GET from kansasCity

06:57:17 <AaronSw> it's also a property of numbers#oneafter99

06:57:31 <AaronSw> and enron#stockPrice

06:57:42 <AaronSw> are these resources all equivalent?

06:57:53 <DanCon> you tell me; I didn't make them up.

06:58:06 <AaronSw> that's not the issue according to 2396!

06:58:14 <AaronSw> you seem to be arguing two contradictory things

06:58:26 <AaronSw> first you say it's a property of the GET, then you say it depends on who made them up

06:58:44 <DanCon> 2396 talks about 'the data you'll GET', right? the data you'll GET is up to the guy who makes up the URI-proper, yes?

06:58:54 <AaronSw> um, ok

06:59:25 <AaronSw> but unlike URIs, and Resources, 2396 ignores my intention in making up these URIs

06:59:34 <AaronSw> it turns off that feature

06:59:39 <DanCon> where?

06:59:51 <AaronSw> when it says "The semantics of a fragment identifier is a property of the data resulting from a retrieval action"

07:00:37 <DanCon> that explicitly refers to your intention in making up the URIs; we agreed that 'the data resulting from a retrieval action' is a function of the URI-proper-issuer's intent, no?

07:00:56 <AaronSw> With Resources it says "The resource is the conceptual mapping to an entity or set of entities, not necessarily the entity which corresponds to that mapping at any particular instance in time." (i.e. feature is on)

07:01:01 <AaronSw> refers to?

07:01:17 <AaronSw> I can write down "100" a zillion times with the same intent, but the resulting number is the same.

07:01:28 <AaronSw> uri-refs get intent deleted

07:01:32 <DanCon> where?

07:01:42 <AaronSw> the semantics do not derive from intent, but from a resolution operation

07:01:53 <DanCon> but the resolution operation is a function of intent.

07:01:55 <AaronSw> i'll quote it again if i have to: "The semantics of a fragment identifier is a property of the data resulting from a retrieval action"

07:02:21 <AaronSw> <DanCon> but the resolution operation is a function of intent.

07:02:24 <AaronSw> only partially

07:02:46 <DanCon> when you wrote "um, ok" at 00:59:22CT, it was very soon after I wrote 'the data you'll GET is up to the guy who makes up the URI-proper, yes?'

07:03:15 <DanCon> I undestood the relationship to be causal.

07:03:26 <AaronSw> It is, that doesn't make it an equivalence relation!

07:03:47 <AaronSw> I write 100, intending to write the current basketball score.

07:03:48 <DanCon> of course not.

07:03:57 <AaronSw> I got to choose that 100.

07:04:05 <AaronSw> But upon grabbing that 100, you did not also grab my intent.

07:04:16 <AaronSw> my intent is no longer connected to my output.

07:04:25 <DanCon> just like on grabbing an entity, I don't grab a resource.

07:04:31 <AaronSw> the hammer can be used for good or evil, regardless of my intent

07:04:37 <AaronSw> in its creation

07:05:02 <AaronSw> but with grabbing an entity, it's very clearly specified that you're saying: "hey what can you tell me about x" not, "give me x"

07:05:39 <DanCon> I think it's more than "what can you tell me about X"; it's "give me a representation of the state of X"

07:05:55 <DanCon> somewhat like "evaluate X for me"

07:06:10 <AaronSw> i don't see the relevance.

07:07:11 <DanCon> it seems to me that whatever it is that you do with URI-propers to get entities is the same thing we do with kansasCity#pop to get 100

07:07:39 <AaronSw> It might seem that way to you, but RFC2396 disagrees (feel like a broken record...)

07:07:43 <DanCon> i.e. we look it up in the global-shared-memory, which changes predictably.

07:07:54 <DanCon> no, RFC2396 doesn't disagree.

07:08:08 <DanCon> RFC2396 says exactly what I'm saying.

07:08:15 <AaronSw> er, i misread. yes we do the same thing to get them.

07:10:15 <DanCon> sleepy time, no?

07:10:38 <AaronSw> not for me :/

07:11:09 <DanCon> you have miles to go?

07:11:23 <AaronSw> no, work to do...

07:11:41 <AaronSw> anyway, when you wake up, try rereading that paragraph :)

07:12:48 <DanCon> er... 'miles to go' was an allusion... was it clear?

07:12:56 <AaronSw> maybe not.

07:13:17 <DanCon> .google and I have promises to keep and miles to go before I sleep and miles to go before i sleep

07:13:18 <xena> no results found.

07:13:30 <DanCon> it's a poem by Robert Frost

07:13:33 <AaronSw> ah

07:14:10 <AaronSw> DOWNLOAD_MONITOR:

07:14:10 <AaronSw> SHOW_TOO_MUCH_DETAIL_v2: yes

07:14:14 <AaronSw> err http://www.thundercloud.net/acpressions/Promises-to-keep.htm

07:14:30 <AaronSw> in that case i have many, many miles to go, but i don't think i need to travel them all tonight :)

07:15:53 <AaronSw> --

07:15:55 <AaronSw> Robert Frost was asked about the line, "I have promises to keep and miles to go before I sleep."

07:15:56 <AaronSw> "Were you writing about death, the big sleep?"

07:15:56 <AaronSw> "It's my poem." he replied, "A poet I can take credit for any interpretation, whether intended it or not."

07:15:57 <AaronSw> --

07:16:01 <AaronSw> - http://www.uuwestport.com/serm1126.htm

07:16:08 <AaronSw> Seems surprisingly relevant. :-)

07:17:40 <AaronSw> Heh, UU too.

10:19:24 <jang> DanCon: awake?

10:19:45 <jang> But how do you reconcile point 2. with text like

10:19:51 <dajobe> nah, he and Aaron were arguing late into the night :)

10:19:57 <jang> A statement and its corresponding reified statement

10:19:58 <jang> ? That's pretty clear that they're in 1-1 correspondence,

10:19:58 <jang> no?

10:20:20 * jang is wearing his underwear

10:20:32 <jang> that's pretty clear they're in 1-1 correspondence, no?

10:20:34 <jang> yech.

10:20:44 <jang> English isn't that unambiguous

11:08:08 <dajobe> BLURB:Web Services with RDF - it can be done

11:08:08 <dc_rdfig> G: Web Services with RDF - it can be done from dajobe

11:08:30 <dajobe> G:[Information Architects say|http://www.ia.com/ia/news/press_release_10032001.htm]: "Web Services built by IA SmartCode are SOAP-compliant and take full advantage of XML, RDF, and J2EE"

11:08:31 <dc_rdfig> commented item G

11:25:17 <dajobe> dajobe has changed the topic to: The Trouble with Triples - http://rdfig.xmlhack.com/

12:20:54 <dajobe> logger_1: hello

12:21:11 <dajobe> logger_1, grep 1 .

12:22:01 <logger_1> I'm logging. I found 166496 answers for '.' (showing 0...0)

12:22:02 <logger_1> 0) 2002-02-15 12:21:11 <dajobe> logger_1, grep 1 .

12:22:06 <dajobe> erk!

12:46:33 * chaals waves

12:46:42 * JibberJim waves back

12:47:52 <JibberJim> Those dublin core people call it ERL...

12:51:01 <chaals> well, those wai people called it quite a lot of things...

12:57:58 * JibberJim still isn't sure what the R is...

13:46:04 <JibberJim> Would it be bad form if I had xmlns:earl="someuri" and xmlns:earl2="someuri" (someuri being the same?)

13:46:59 <dajobe> bad style but not illegal

13:47:11 <JibberJim> would any tools break with it?

13:47:39 <JibberJim> (I'm combining different sources, and normalising that isn't proving too easy)

13:47:44 <dajobe> I'd say probably not

13:47:57 <dajobe> what XML is this?

13:48:06 <dajobe> or, what application of XML?

13:48:06 <JibberJim> EARL RDF

13:48:13 <dajobe> should be ok

13:48:28 <dajobe> Try http://www.w3.org/RDF/Validator/ to check stuff

13:48:37 <JibberJim> Yeah, it passed through that.

13:48:41 <dajobe> :)

13:58:39 <JibberJim> Erm, having the same namespace prefix pointing to different namespaces isn't invalid either?

14:06:10 <chaals> should be?

14:07:33 <JibberJim> erm.

14:07:40 <JibberJim> the validator didn't complain...

14:08:09 <JibberJim>http://jibbering.com/earldb/v.1?url=http://www.showcaster.com/

14:08:09 <dc_rdfig> H: http://jibbering.com/earldb/v.1?url=http://www.showcaster.com/ from JibberJim

14:08:15 <JibberJim> Growl

14:08:18 <JibberJim> Can I delete it?

14:08:19 <dajobe> you can't have the same prefix pointing, that's illegal XML

14:08:22 <dajobe> delete; no

14:08:46 * JibberJim tries to think of a comment for it then.

14:08:54 <dajobe> i.e <foo:bar xmlns:a="url1" xmlns:a="url2" ..> - duplicate attrs I mean

14:09:02 <chaals> H|quick slip of the keyboard...

14:10:12 <JibberJim> <foo:bar xmlns:foo="url1"/><foo:bar xmlns:foo="url2"> like that.

14:15:07 <JibberJim> H:|quick slip of the keyboard...

14:15:07 <dc_rdfig> titled item H

14:18:46 <AaronSw> .seen zooko

14:18:47 <xena> zooko seen in #mnet saying: [ Didn't youjust test it ? ] ~ 6 hr(s) 8 min(s) 54 sec(s) ago

14:18:47 <_xena_> zooko seen in #mnet saying: [ Didn't youjust test it ? ] ~ 6 hr(s) 8 min(s) 54 sec(s) ago

14:19:57 <JibberJim> Xena's been multiplying!

14:21:48 <dajobe> JibberJim: innermost xmlns overrides, again bad-ish style

14:22:09 <dajobe> OK for computers, confuses people who read it

14:50:28 <monkinetic> monkinetic is now known as rm

15:09:50 <chaals> jim, is this so you can mix different earl flavours in processing stuff?

15:10:17 <chaals> (if it is for computers not people it shouldn't matter, although you need to be strict about parsing and remind others to do so...

15:10:24 <chaals> ...which IMHO is a good thing really)

15:10:32 <JibberJim> Well, it's so that I don't need to do too much processing in the database.

15:11:23 <JibberJim> If two people submit reports about the same url, either using different EARL namespaces, or with different prefixes to the same, I can still return a nice valid understandable document.

15:11:32 * jonb waves

15:13:02 <chaals> OK. You should be able to use a squishing thing to merge two different reports, although this is where we need something like sean proposed for how to deal with apparent conflicts.

15:13:08 <JibberJim> My MSXML parser doesn't seem to have nice interfaces for working with namespaces.

15:13:28 <JibberJim> I'm ignoring the conflicts, that's up to the user tool.

15:13:51 <chaals> (If you can handle it without doing that step it sdoesn't matter of course, but lots of real parsers will need to know whether or not something got fixed.

15:14:32 <JibberJim> Oh yeah, but I (and we discussed this in one of the meetings) feel that's really down to the tool. Obviously a future version of the database will allow such querying.

15:30:23 <chaals> sure.

15:35:03 <JibberJim> I'm trying to fix it up anyway, so namespaces are only defined in the root node, and there aren't any duplicates.

15:42:56 <jonb> ah ns normalization ... yes a problem

15:43:17 * JibberJim is finding it a big problem...

15:43:41 * JibberJim is thinking about parsing it to triples and reconstructing... but that seems expensive.

15:43:58 <jonb> doesn't solve the prob

15:44:34 <jonb> whic can be that attribute values can have qnames

15:45:01 <jonb> e.g xpath, xmlschema

15:55:02 <chaals> reparsing and reconstructing: this is the sort of thing I was talking about having a proxy service to do - query a database and get a bunch of stuff, or query an intermediate service that can reparse and provide it (with logic applied if you like)

15:57:33 <JibberJim> I could normalise it when it is submitted, so all EARL 1.0 uses earl prefix earl0.95 user earl95 or something like that - it would mean I wouldn't be coping with other namespaces but the "normal" EARL ones would be ok.

16:54:51 <wendyout> wendyout is now known as wendy

17:01:11 * jonb is getting disconnected frequently

17:04:46 <jonb> catching up on the logs. i think that for RDF pov, an URIref and URI should be treated the same way

17:05:11 <jonb> however, the problem of fragid syntax being defined by media type exists

17:05:27 <jonb> a media type independent fragment id syntax would solve that

17:07:59 <jonb>http://www.rddl.org/fragment-syntax

17:07:59 <dc_rdfig> I: http://www.rddl.org/fragment-syntax from jonb

17:08:23 <jonb> I: media type independent fragment id syntax

17:08:23 <dc_rdfig> commented item I

17:08:43 <jonb> I:|Media-type independent fragment-id syntax

17:08:43 <dc_rdfig> titled item I

17:08:50 <DanCon> I:haven't read it yet. I hope it's just names.

17:08:50 <dc_rdfig> commented item I

17:09:05 <jonb> I: an I-D that I haven't sumitted yet

17:09:06 <dc_rdfig> commented item I

17:09:22 <jonb> not just names, it handles XPointer

17:09:46 <DanCon> how does XPointer work on image/gif?

17:09:47 <jonb> I: uses XPointer's xmlns= syntax

17:09:48 <dc_rdfig> commented item I

17:10:31 <jonb> it's just a syntax, so haven't said anything about what the media type of the entity is

17:10:51 <jonb> that is to say the URIref identifies a resource in the RDF sense

17:11:11 <jonb> or would identify if this became accepted

17:12:02 <jonb> An XPointer _could_ be useful for any media type however that's a different story

17:12:50 <jonb> you can develop an XML representation of anything based on EBNF or whatever syntactic rules something is defined to have

17:13:14 <jonb> so image/gif might have an internal structure represented as:

17:14:03 <jonb> <image><row><pixel>...</pixel> ...</row><row>...</row> ...</image>

17:15:39 <jonb> and a useful XPath might be: image/row[234]/pixel[35]/red

17:16:00 <jonb> get's back to the old 'groves' concept

17:19:24 <jonb> in any case I am concerned that a URIref has no 'meaning' outside a network operation at the present time, at least no 'official' meaning

17:20:09 <jonb> but we mostly all agree that it would be very useful to make assertions about URIrefs (or whatever they denote) outside of an HTTP GET operation

17:37:42 * jang has a great deal of sympathy for DanCon's DTing requirements

17:38:03 <jang> it's unfortunate that "KISS" wasn't explicitly on our desiderata. Not too late to add it, I hope.

17:39:12 <DanCon> <image><row><pixel> ... ok, I see. yeah, I think that's an interesting approach. (we used to throw it around the the context of the AudioVideo activity, i.e. SMIL)

17:59:08 <chaals> alternatively / as well, it would be able to pick up comment stuff in gif, which might well be real content...

18:04:17 <chaals> as in the stuff inserted into jpegs by RDFPic

19:13:25 <jonb> i happen to think that xpointer is just about right technically

19:13:53 <jonb> because it gives simple names, numeric paths and more extended paths

19:45:55 <jonb> regarding 'xpaths' for non-xml data: it can be done for any grammar that can be expressed as productions and is LALR(1)

19:48:06 <jonb> which I think is the requirement for being able to build an AST

20:30:20 <jonb> MikeM: do you see an absolute need to tie the fragid syntax to media type?

21:17:54 <DanCon> hmm... playing around: http://www.w3.org/2001/tag/pywiki?FrontPage

21:35:29 <sbp> ooh, a TAG Wiki

21:46:51 <sbp> in your face URIs: ugh indeed. Perhaps a chump-esque [URI|title] syntax would be better

21:47:00 <AaronSw> MoinMoin does that.

21:48:31 <DanCon> I think pywiki is the least featureful wiki around. The only features are (a) written in python and (b) RCS-backed

21:48:46 <DanCon> I figure it's less painful to add features than to take them away

21:48:54 <DanCon> or pywiki is easier to install. or something. ;-)

21:49:56 * AaronSw thinks DanC has been spending too much time in alternate logics ;)

21:50:04 <DanCon> I haven't really been inspired, but I sorta got started http://www.w3.org/2001/tag/pywiki?WebArchitecture

21:52:07 <DanCon> I think there should be a bit more hello-world level stuff in a web architecture document.

21:54:43 <MIkeM> wiki? as in "wiki wiki wiki" (does anyone remember Newcleus?)

21:59:06 <DanCon> as in WikiWikiWeb

21:59:22 <DanCon> no, not the little robot from that scifi show

21:59:41 * danja jam on it!

21:59:58 <AaronSw> wow, pat hayes isn't understanding me at all

22:00:02 <MIkeM> j-jj-jam on it..... cause I'm Cosmo D from outer space!

22:00:16 * MIkeM has the CD in his CD player...

22:01:21 * MIkeM thinks danja is the only one who knows what I'm blathering on about....

22:01:43 <danja> ;-)

22:02:20 <danja> I like wikis of all description...

22:02:25 <MIkeM> hehe... starting at Dan's wiki he posted earlier it took me 12 pages to actually get to something the explained what the hell it was... ;-)

22:02:28 <danja> g'night.

22:03:04 <AaronSw> MIkeM, but now that you know, you know you can fix that right?

22:04:18 <AaronSw> pf, pywiki doesn't appreciate my attention to detail

22:04:44 <DanCon> ?

22:04:44 <AaronSw> hm, you can sorta get rid of in-your-face-urls with the [n] thing

22:04:56 <DanCon> yeah... PoorMansHypertext

22:05:28 <AaronSw> where the wiki actually hosted? tux?

22:13:57 * DanCon tries to think of any risks associated with disclosing which host the service runs on... or any good reason why Aaron would want to know

22:15:23 <DanCon> it runs on cgi.w3.org aka lists.w3.org aka www19.w3.org

22:15:29 <AaronSw> cool

22:15:48 <DanCon> hmm... how to implement peer review.

22:16:31 <DanCon> ok... I'm pretty happy with http://www.w3.org/2001/tag/pywiki?UsingTheWeb

22:19:57 <michael_> michael_ is now known as MikeM

22:20:11 * DanCon has family obligations at 5pCT; tell me to scoot if I'm still around by then.

22:21:33 <DanCon> "pf, pywiki doesn't appreciate my attention to detail" <- I'm still puzzled by that.

22:26:48 <DanCon> ok... I think there's enough beef in this thing to invite the rest of the TAG to give it a whirl.

22:27:39 <DanCon> hmm... I have a TAG action item to produce some slides about web architecture. I wonder if I can use this wiki to do that.

22:42:43 <AaronSw> "pf, pywiki doesn't appreciate my attention to detail" - the original wiki said "Thank you for your appreciation to detail" as has every wiki clone i've used

22:42:47 <AaronSw> pywiki is the first that doesn't

22:43:42 <DanCon> I suppose I can change that. That sort of consciousness about quality is essential, I agree.

22:47:28 <DanCon> there.

22:47:34 <AaronSw> cool

22:48:43 <DanCon> I don't see any edits by you

22:48:57 <AaronSw> oh?

22:49:50 <sbp> "All of these are different Resources." - [1]. Actually, they are variants (different representations) of the same resources, AFAICT from all of the REST material and RFC 2616. [1] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0412

22:50:06 <AaronSw> variants are resources too, i think

22:50:17 <sbp> true

22:50:30 <sbp> ah... ambiguity in your text...

22:51:26 <sbp> FWIW, it's quite an accessibility problem

23:03:22 <MikeM> hey dan, didn't you say something about needing to do something at 5:00 CT?

23:05:26 <DanCon> yup; thanks. turns out to be 5:30 or 5:45

23:05:35 <DanCon> (and I just got part of it done)

23:07:21 * sbp tries to find an HTTP-ext test suite

23:10:58 <DanCon> crap; crosswinds.net yank free email service

23:11:02 <DanCon> yanked

23:11:12 <DanCon> so my efaxes are hosed

23:12:41 <sbp> ugh

23:16:30 <DanCon> sigh... "It?s easy to keep track of your account. Use this page to check all of your inbound and outbound faxes when you upgrade to eFax Plus."

23:16:51 <DanCon> efax has pretty much eliminated free services too.

23:16:59 <DanCon> I guess I had no reason to believe it would last.

23:22:45 <MikeM> can someone go to http://www.rocketforge.org/jetspeed/ and tell me what you get? thanks!

23:24:03 * MikeM thinks he has DNS issues but can't tell because he's to close to the server...

23:24:32 <sbp> >>> print urllib2.urlopen('http://www.rocketforge.org/jetspeed/').info()

23:24:32 <sbp> [...]

23:24:32 <sbp> URLError: <urlopen error (7, 'getaddrinfo failed')>

23:24:57 <MikeM> hmm... figures..... maybe granitecanyon wasn't such a good choice....

23:25:39 <MikeM> thanks!

23:26:55 <sbp> np

23:27:14 * DanCon bids good day to all

23:39:36 <MikeM> hey sean, can you try it again? (the nameservers started responding)

23:43:29 <sbp> >>> print urllib2.urlopen('http://www.rocketforge.org/jetspeed/').info()

23:43:29 <sbp> Date: Fri, 15 Feb 2002 19:48:49 GMT

23:43:31 <sbp> [...]

23:43:33 <sbp> works fine

23:43:58 <MikeM> cool!...thanks!

23:44:01 <sbp> "If you see this item, you've successfully installed and configured your Jetspeed portal." :-)

23:44:32 <MikeM> hhe.....yep!

23:45:12 <MikeM> I'm having a bitch of a time getting the RSS feeds working.....