00:00:49 <dajobe> dajobe has quit
00:29:14 <sbp> sbp has quit
00:37:32 <sbp> sbp has joined #rdfig
00:38:29 <danja> how why chart is passworded
00:39:35 <danja> LEAD sounds a spot-on approach though
00:40:43 <danja> dead image link in http://www.w3.org/2001/04/roadmap/ as well.
00:42:02 <danja> Sorry, Unauthorized. 
00:42:27 <danja> dan_c - weaving the road ahead???
00:43:42 <danja> g'night. 
00:43:44 <danja> danja has quit
01:01:49 <hazmat> hazmat has quit
01:13:48 <sbp> sbp has quit
01:28:13 <sbp> sbp has joined #rdfig
01:59:59 <sandro> I'd like another view to the same data, with one page per circle, focussed on that circle and showing only nearby circles.   I keep meaning to put it together.   
02:00:55 <AaronSw> Ooh, like the info visualization thing I chumped!
02:01:04 <AaronSw> Did you see that movie, Sandro? It's quite cool
02:01:08 <AaronSw> I want one of those for RDF...
02:01:15 <AaronSw> Perhaps I should email Ka-Ping...
02:04:24 <sbp> sbp has quit
02:08:43 <sbp> sbp has joined #rdfig
02:13:36 <AaronSw> Stefan?!
02:13:49 <AaronSw> What did you do to get http://www.ontoprise.de/com/start_products.htm to think you trademarked "Semantic Web"?
02:31:50 <sbp> sbp has quit
02:47:56 <sbp> sbp has joined #rdfig
02:48:00 <tav> tav has joined #rdfig
02:48:28 <tim-gone2> tim-gone2 has quit
02:52:25 <sbp> sbp has quit
02:54:08 <tav`> tav` has joined #rdfig
03:09:04 <sbp> sbp has joined #rdfig
03:11:33 <sbp> sbp has quit
03:22:48 <GabeW> GabeW has joined #rdfig
03:34:17 <sbp> sbp has joined #rdfig
03:37:20 <sbp> sbp has quit
03:55:30 <DanC> DanC has quit
04:02:28 <GabeW> GabeW has quit
04:22:34 <ElectricElf> [GlobalNotice] Okeydokey :) We'll be disconnecting services in about 60 seconds. We're moving them to a beefier server with a better connection. Isn't it nice we can do that now? :) Anyways, them to be back in a few minutes.
05:43:57 <DanC> DanC has joined #rdfig
06:17:57 <ircAsync> ircAsync has joined #rdfig
06:27:47 <ircAsync> ircAsync has quit
06:47:48 <tav> tav is now known as tav\movie
07:49:43 <tav\movie> tav\movie is now known as tav
08:31:01 <bijan> bijan has quit
09:08:29 <oierw`> oierw` has quit
09:08:43 <PONG> PONG has joined #rdfig
09:16:05 <PONG> PONG has quit
09:19:11 <PONG> PONG has joined #rdfig
09:39:17 <larsbot> larsbot has quit
10:14:00 <larsbot> larsbot has joined #rdfig
10:24:35 <dajobe> dajobe has joined #rdfig
10:30:45 <grove> grove has quit
11:02:18 <grove> grove has joined #rdfig
11:04:58 <libby> libby has joined #rdfig
11:12:50 <irc.openprojects.net> Disconnected from irc.openprojects.net (ERROR :Closing Link: logger_1[~rdfig-log@137.222.34.230] (Ping timeout: 181 seconds))
11:13:03 <logger> logger has joined #rdfig
11:13:03 <carter.openprojects.net> topic is: RDF/Semantic Web Chat: Capabilities or ACLs? - http://www.w3.org/RDF/Interest/#irc
11:13:03 <carter.openprojects.net> Users on #rdfig: logger libby grove dajobe larsbot PONG DanC tav` tav dmiles cmsc498x danbri tim-gone zooko_family_time jang AaronSw sandro kifbot em ericP chihchun_ deltab dc_rdfig 
11:13:03 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome
11:46:47 <grove> grove has quit
11:49:29 <grove> grove has joined #rdfig
11:51:45 <zooko_family_time> zooko_family_time is now known as zooko
11:51:49 <zooko> zooko is now known as zooko_breakfast
12:39:03 <zooko_breakfast> zooko_breakfast is now known as zooko
12:43:44 <grove_> grove_ has joined #rdfig
12:44:01 <danbri_> danbri_ has joined #rdfig
12:48:16 <grove> grove has quit
12:48:16 <danbri> danbri has quit
12:48:16 <em> em has quit
12:48:16 <ericP> ericP has quit
12:49:07 <ericP> ericP has joined #rdfig
12:49:07 <em> em has joined #rdfig
13:46:23 <ericP> danbri?
13:46:29 <danbri_> eric?
13:46:51 <ericP> where did you find `dot` docs?
13:47:05 <ericP> i have a man page, but it doesn't describe the parms in a dot file.
13:47:24 <danbri_> for graphviz? Not sure... I probably took an example one from the kind of output I was after, and tweaked it.
13:47:44 <danbri_> www.graphviz.org perhaps? some content migrated over from their old AT+T site
13:47:49 <ericP> i have a graph that i've fixed (imported into dotty and resaved) and want to make one of the nodes have a longer label.
13:48:11 <ericP> this seems to make it abandon the fixatives
13:48:19 <ericP> checking www.graphviz.org...
13:48:23 <danbri_> beyond my expertise, i'm afraid
13:48:46 <dajobe> hey ericP, are you using aboutEach?
13:48:57 <danbri_> hey, but while you're here. We're discussing aboutEach in the RDF Core WG. The suggestion is that we might drop it, and have WebOnt etc do that functionality at a higher level.
13:49:33 <danbri_> Your code, if I remember right, goes beyond M+S and preserves aboutEach generalisations behind the RDF API, in database tables etc.
13:49:53 <ericP> * ericP thinks about uses...
13:49:58 <danbri_> M+S says this is pure syntactic sugar, parsers resolve it and that's that
13:50:12 <ericP> i am, in the ACLs code, but just as a shorthand.
13:50:44 <danbri_> my story for dropping it is in: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0532.html
13:50:58 <dajobe> and a nice compelling story it is too
13:51:00 <danbri_> dave has a related version in terms of parsers, implementation etc.
13:51:04 <danbri_> thank you :)
13:51:08 <ericP> the reason for doing the late binding on aboutEach was 'cause it seemed like it would be a good excercise of the inferencing code.
13:51:37 <danbri_> eric, as someone who's implemented it, you get a louder say... 
13:52:09 <ericP> how's this? I AM, IN THE ACLS CODE, BUT JUST AS A SHORTHAND
13:52:14 <danbri_> interesting that you use it for rules; you've built a skeletal webont-ish system by parsing rules from the aboutEach structure, in effect
13:52:27 <ericP> or maybe I AM, IN THE ACLS CODE, BUT JUST AS A SHORTHAND
13:52:27 <danbri_> Hey, why doesn't IRC have a blink tag?
13:52:32 <danbri_> Oh, koool
13:52:41 <dajobe> * dajobe rolls his eyes
13:53:03 <danbri_> So, more interestingly, tell us if you'd support RDF Core dropping this feature of the language (for webont or rules to pick up later)
13:53:23 <ericP> i think it would make sense to strip it out of the core but haven't explored the logic effect on sequences and alts.
13:53:39 <danbri_> it makes them less magical and wierd
13:54:32 <ericP> i guess there's no way to retroactively add sequences. if you remove them, you are stuck in the car/cdr world.
13:54:52 <danbri_> MIT, you mean?
13:55:01 <ericP> yeah, that one.
13:55:03 <ericP> oops, have to run to chiropractor.
13:55:07 <rreck> rreck has joined #rdfig
13:55:17 <danbri_> ok, see you...
13:56:17 <AaronSw> * AaronSw disconnects
14:03:59 <larsbot> has anyone applied the Unix pipe way of thinking to RDF processing?
14:04:16 <larsbot> are there any tools that do this kind of thing with RDF?
14:04:49 <danbri_> cwm often gets used that way...
14:05:18 <larsbot> cwm is python, right?
14:05:29 <danbri_> yup
14:05:43 <larsbot> can you set up processing chains with it?
14:06:34 <larsbot> for example, by writing XML configuration files?
14:08:14 <danbri_> not sure. mostly commandline flags i think.
14:08:27 <larsbot> oh, ok
14:08:33 <larsbot> * larsbot is distracted, return soon
14:09:26 <zooko> zooko has left #rdfig
14:34:30 <rreck> rreck has quit
14:40:32 <chihchun_> chihchun_ has quit
14:40:32 <irc.openprojects.net> Disconnected from irc.openprojects.net (ERROR :Closing Link: logger[~rdfig-log@137.222.34.230] (Ping timeout: 181 seconds))
14:41:02 <logger> logger has joined #rdfig
14:41:02 <benford.openprojects.net> topic is: RDF/Semantic Web Chat: Capabilities or ACLs? - http://www.w3.org/RDF/Interest/#irc
14:41:02 <benford.openprojects.net> Users on #rdfig: logger em ericP danbri_ grove_ libby dajobe larsbot PONG DanC tav` tav dmiles cmsc498x tim-gone jang dc_rdfig deltab kifbot sandro AaronSw 
14:41:02 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome
14:41:23 <chihchun_> chihchun_ has joined #rdfig
14:42:50 <libby> libby has quit
14:43:38 <larsbot> larsbot has quit
14:47:29 <chihchun_> chihchun_ has quit
14:48:23 <em> em has quit
14:48:23 <ericP> ericP has quit
14:48:39 <chihchun_> chihchun_ has joined #rdfig
14:51:13 <larsbot> larsbot has joined #rdfig
14:52:22 <ericP> ericP has joined #rdfig
14:52:22 <em> em has joined #rdfig
14:52:22 <niven.openprojects.net> niven.openprojects.net has changed the topic to: RDF/Semantic Web Chat: Capabilities or ACLs? - http://www.w3.org/RDF/Interest/#irc
14:55:00 <tim-gone> tim-gone has quit
14:55:24 <tim-gone> tim-gone has joined #rdfig
14:55:55 <kifbot> kifbot has quit
14:56:27 <chihchun_> chihchun_ has quit
14:56:27 <danbri_> danbri_ has quit
14:56:27 <dmiles> dmiles has quit
14:56:27 <em> em has quit
14:56:27 <ericP> ericP has quit
14:56:27 <larsbot> larsbot has quit
14:56:27 <dajobe> dajobe has quit
14:56:27 <PONG> PONG has quit
14:56:27 <tav`> tav` has quit
14:56:27 <tav> tav has quit
14:56:32 <cmsc498x> cmsc498x has quit
14:56:43 <kifbot> kifbot has joined #rdfig
14:58:09 <em> em has joined #rdfig
14:58:09 <ericP> ericP has joined #rdfig
14:58:09 <larsbot> larsbot has joined #rdfig
14:58:09 <chihchun_> chihchun_ has joined #rdfig
14:58:09 <danbri_> danbri_ has joined #rdfig
14:58:09 <dajobe> dajobe has joined #rdfig
14:58:09 <PONG> PONG has joined #rdfig
14:58:09 <tav`> tav` has joined #rdfig
14:58:09 <tav> tav has joined #rdfig
14:58:09 <dmiles> dmiles has joined #rdfig
14:58:17 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome
14:58:20 <larsbot> danbri, you there?
15:02:34 <cmsc498x> cmsc498x has joined #rdfig
15:08:07 <danbri_> yes, though I'm in a meeting
15:14:35 <lasDesk> lasDesk has joined #rdfig
15:19:32 <lasDesk> lasDesk has quit
15:21:04 <cmsc498x> cmsc498x is now known as Iron_SpermWhale
15:23:15 <larsbot> ok, we and talk later
15:26:29 <libby> libby has joined #rdfig
15:28:53 <lasDesk> lasDesk has joined #rdfig
16:17:51 <grove_> http://www.advogato.org/article/376.html
16:17:51 <dc_rdfig> A: http://www.advogato.org/article/376.html from grove_
16:18:18 <grove_> oops, wrong channel. ;[
16:24:18 <tav> which channel shoulda that been for?
16:24:30 <tim-gone2> tim-gone2 has joined #rdfig
16:25:00 <larsbot> internal work channel
16:30:12 <DanC> tim, gottaminute re social impact of making RDF assertions?
16:31:09 <DanC> there's a proposal that the RDF Core WG shouldn't bother saying anything about it. http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0423.html
16:39:21 <tim-gone2> social impact... hmm
16:39:25 <tim-gone2> * tim-gone2 reads
16:39:35 <tim-gone2> tim-gone2 is now known as timbl
16:44:14 <timbl> Ok, so what I am supposed to do? I feel strongly about this.
16:44:31 <timbl>  I guess I should mail the cpomments list -- or join in?
16:45:13 <timbl> The semantics of a RDF document are that the author asserts each of hte statements.
16:45:39 <timbl> The semantics of a statement are defined by the predicate.
16:45:47 <timbl> The semantics of a statement are defined by the rdf:Property.
16:46:13 <timbl> The semantics of a rdf:Property in http: space or any other URI space which has a concept of
16:46:49 <timbl> delegated ownership are owned by the woner of hte URI, and any definition by that person should be considered definitive.
16:47:25 <timbl> A web server is a way of making informatoi about hte intended meaning of a Property public.
16:47:35 <timbl> for URIs in http: space.
16:50:51 <DanC> * DanC catches up...
16:52:14 <timbl> This applies of course to statements asserted at the top level of a document, not nested within some other formula F unless the semantics of F imply it specifically.
16:52:27 <reagle-tu> reagle-tu has joined #rdfig
16:52:52 <DanC> http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0423.html
16:52:53 <dc_rdfig> B: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0423.html from DanC
16:53:02 <DanC> B:a proposal to drop Issue rdfms-assertion
16:53:03 <dc_rdfig> commented item B
16:53:23 <DanC> no, tim, you don't need to participate directly in the WG discussion.
16:53:35 <r3agle-tu> r3agle-tu has joined #rdfig
16:53:57 <DanC> you just need to give me some ammunition as to why this belongs in RDF 1.0... especially when we published it as a REC without lots of related things (a MIME type, e.g.)
16:54:30 <DanC> best of all, give me some words to put in a spec.
16:54:32 <r3agle-tu> Is there proposed text as to what "it" is that should belong?
16:55:00 <DanC> no, that's the problem, reagle. I don't know how to say what TimBL (and I?) want.
16:55:16 <DanC> words like "The semantics of a statement are defined by the rdf:Property." won't work.
16:55:51 <r3agle-tu> Nothing quickly comes to mind with respect to something RDF itself should say...
16:56:49 <r3agle-tu> "An RDF statement is a statement which is intended to be as informative (if not moreso) and  normative as natural language expressions." is the best I can come up with
16:57:01 <DanC> hang on... there is a proposal... in a draft spec for a  MIME type for RDF.
16:57:43 <DanC> B:see also [aaron's MIME type draft|http://blogspace.com/rdf/mimetype] based on [DanC's draft|http://www.w3.org/2001/03mr/rdf_mt]
16:57:43 <dc_rdfig> commented item B
16:57:49 <DanC> phpht. I can't get to blogspace.com
16:58:21 <DanC> * DanC goes via tux.w3.org...
16:58:25 <DanC> got it: [[[ Because RDF is a format for semantically-meaningful information, it is
16:58:25 <DanC>    important to note that transmission of RDF under this media type,
16:58:25 <DanC>    whether via HTTP, SMTP or some similar protocol, means that the sender
16:58:25 <DanC>    asserts the content of the RDF document. ]]]
16:58:48 <DanC> it continues... [[[  If this is not desirable,
16:58:49 <DanC>    such as when a system is forwarding RDF written by someone else,
16:58:49 <DanC>    another applicable media type, such as application/xml or
16:58:49 <DanC>    application/octet-stream should be used. Also note that RDF provides
16:58:49 <DanC>    reification so that RDF statements can be sent and discussed without
16:58:49 <DanC>    actually being asserted themselves.
16:58:51 <DanC>  ]]]
16:59:01 <tim-gone> * tim-gone distracted brb
16:59:10 <tim-gone> tim-gone is now known as timmmit
16:59:37 <DanC> oops, tim, mailing the comments list is very much in order.
17:00:26 <r3agle-tu> * r3agle-tu is concerned about using the MIME/type as the semantic bootstrap...
17:00:35 <danbri_> * danbri_ catches up
17:00:36 <DanC> but keep in mind that stuff like  "The semantics of a statement are defined by the rdf:Property." won't be well-received. All sorts of stuff contribute to the semantics of statements.
17:01:05 <danbri_> * danbri_ nods
17:01:16 <DanC> well, r3agle-tu, the protocol context is important. If you send RDF in an attachment with content-disposition: refuted, you shouldn't be held accountable for it.
17:01:34 <timmmit> no - disaggree.  The Predicate rules.
17:01:43 <timmmit> This is important.
17:01:52 <timmmit> You can't define a noun which redefines my verb.
17:02:48 <timmmit> ... :mydoc "when used as a subject to the verb `sucks', has the significance `is really cool'.  ;-)
17:02:49 <r3agle-tu> If need be (without agreement/consensus), I would not be concerned if RDF punt (beyond saying what I quoted earlier "RDF is intended to be informative and normative").
17:02:53 <DanC> tim, "The predicate rules" is just nonsense. if I say { :Bob rdf:type :ChocoloateLover }, other statements about :ChocolateLover constrain the semantics just as much as other statements about rdf:type.
17:03:13 <r3agle-tu> Otherwise, the applications themselves need to be clear.
17:03:28 <timmmit> The semantics of type delegateto the semantci s of the class.
17:03:40 <r3agle-tu> the xmldsig-p3p-profile defines its bootstrap. http://www.w3.org/TR/xmldsig-p3p-profile/#sec-Semantics
17:04:12 <DanC> tim, you may have your own special meaning of "semantics". But it will do you little good to use it in communication with the RDF Core WG, which has adopted the traditional math/logic terminology.
17:04:17 <timmmit> No one except the owners of rdf:type can change the semnatics of rdf$type which basically say ... go see the unary prediate CX where C is the object class.
17:04:24 <danbri_> B:see also [my view|http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0558.html] on how we can partially postpone dealing with this
17:04:25 <dc_rdfig> commented item B
17:04:26 <r3agle-tu> and p3p itself does (in a way) via its extension mechanism.. http://www.w3.org/TR/2000/WD-P3P-20001018/#extension
17:04:55 <r3agle-tu> * r3agle-tu wonders what is the meaning of assert then...
17:05:10 <danbri_> does http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0558.html work for any of you?
17:06:27 <r3agle-tu> yes, and I like what you say, "propositional context" akin to my "capable of informative and normative"
17:06:37 <r3agle-tu> expression
17:07:02 <reagle-tu> reagle-tu has quit
17:07:29 <r3agle-tu> r3agle-tu is now known as reagle-tu
17:09:15 <DanC> danbri, what do you mean by "RDF core specs"? does that include the spec for the MIME type of RDF?
17:09:28 <DanC> or you not think the RDF Core WG owes the world a spec for a MIME type for RDF?
17:11:04 <DanC> * DanC thinks I've lost timbl's attention; wonders how to approach this, logistically.
17:13:01 <timmmit> We should not put this off to another time
17:14:04 <danbri_> "RDF core specs" was a bit vague. A MIME type for RDF would be a fine thing to give the world. Specifying what it means to receive a chunk of RDF/XML of that mimetype over HTTP, in the face of HTTP extension headers etc etc., feels riskier. We can't do protocol design in RDF Core.
17:14:07 <timmmit> * timmmit at been looking at bugzilla w/simon
17:14:42 <DanC> It would be nice if you could use the terminology of semantics that the WG is using, TimBL. I wonder how familiar you are with it. I wonder if Massimo is available for this; he taught it to me.
17:15:22 <danbri_> There's a whole side to 'semantics', closer to the T-and-S concerns you raised, that the MT/math approach explicitly distances itself from
17:15:35 <danbri_> mainly focussed around reference
17:15:35 <DanC> quite.
17:16:36 <timmmit> The MT can only define how you can manipulate information - it can't tell you what it MEANS without english.
17:16:46 <DanC> I don't believe statements -- bits of syntax -- have t-and-s:semantics outside the context of protocol events. i.e. accountability is grounded in actions, not syntax.
17:17:03 <timmmit> I could argue that one.
17:17:09 <timmmit> A document has a meaning.
17:17:10 <timmmit> Period.
17:17:11 <danbri_> * danbri_ agrees w/ timmit re MT
17:17:25 <danbri_> or parts of documents have meanings
17:17:26 <DanC> no, a document takes on meaning because of historical events.
17:17:58 <timmmit> The semantics of HTTP 200  response is itself a document which aasserts that the enclose bits represent in the given language the semantics of the requested URI within the qualifications given in all the various headers.
17:18:26 <danbri_> RDF defines an ingredient that people can use to make meaningful XML documents; shove it in the <head> it means one thing, shove it in <blockquote> means another. RDF Core's work is finished when it gives an account of the components of meaning for the RDF bit of the document.
17:18:34 <DanC> * DanC can't parse: "the semantics of a response *is* a document"
17:19:41 <timmmit> DanC: Yes, a document takes on meaning becase of historical events, specifically agreements between parties and otehr discussions which lead to an assumption that things will bne interpreted in particular ways, but in this particular case, the publcation ofhte spec is the primary and main historical event which defines everything directly or indirectly.  Hence the utility of standards.
17:20:25 <DanC> yes, but your position requies that all documents act like standards, tim.
17:20:44 <timmmit> Sorry, s/b "The  HTTP 200  response is itself a document which asserts that the enclosed bits represent, in the given language, the semantics of the requested URI subject to the qualifications given in all the various headers.
17:20:46 <timmmit> "
17:21:00 <lasDesk> lasDesk has quit
17:21:08 <timmmit> All documents act lik standards?
17:21:32 <timmmit> The RDF spec is a document which will be taken to be commonly understood and agreed by all those parsing RDF documents.
17:21:35 <DanC> like standards: for example, if a document asserts that W3C has 517 members, what do you do with it when W3C has 520 members?
17:22:06 <timmmit> My invoice to you for three milky way bars does not have the same wide ranging significance but they are both documents.
17:22:31 <lasDesk> lasDesk has joined #rdfig
17:22:39 <DanC> another parse/sort error/quesion: by "HTTP 200 response" do you mean the event itself or the syntactic contents of it? I regard an HTTP response as an event/action, grounded in physical/social history, not a sequence of bytes.
17:22:47 <lasDesk> lasDesk has quit
17:22:49 <timmmit> The RDF spec does not assert that the WG has 517 members.
17:24:54 <DanC> i.e. by "document" do you mean "event" or "character sequence"?
17:25:10 <timmmit> Dicument/action:  Theer are two models of the world, one (protocols) which exposes the messaging and is more basic, and the other (web) which is quasi-static and does not expose the protocol.   You can do everything with the first but often we find it is a simplification, when only HTTP GETs are used, to use a model in which the log:semantics are kept but the IP packets are left out of the model. 
17:25:43 <timmmit> You can't say that the document model is invalid.  It works and is consistent under certain (often reasonable) assumptions.
17:26:04 <timmmit> But i agree that it is interesting to track it down to the actions in the end.
17:26:24 <DanC> it is this quasi-static model that requires all documents to act like standards; i.e. to say things that are always-everywhere true.
17:26:39 <timmmit> However, when people talk about the semantics of a document, they describe it in terms of its being part of a protocol where necessary. 
17:26:40 <danbri_> http://www.arts.ed.ac.uk/philosophy/study_html/vade_mecum/sections/section2/2-1.htm
17:26:41 <dc_rdfig> C: http://www.arts.ed.ac.uk/philosophy/study_html/vade_mecum/sections/section2/2-1.htm from danbri_
17:26:55 <danbri_> C:|Propositions, arguments and truth
17:26:55 <dc_rdfig> titled item C
17:27:21 <timmmit> Quasi-static model requires consistency. Yes. 
17:27:29 <DanC> please don't generalize about how people talk about the semantics of a document, TimBL. you really do have a very provincial view on that account.
17:28:06 <timmmit> Sorry, I should have said when people talk about what something you get in an envelope means...
17:28:47 <danbri_> Could we re-couch the issue as "Do RDFCore believe RDF/XML documents (graphs?) are the kind of thing that can be true or false"?
17:29:03 <timmmit> Provicial view.   Naive, limited to affairs of the privinces?  Not metropolitain?  
17:29:24 <DanC> provincial: I meant "not widely shared". perhaps I misused the word.
17:29:39 <timmmit> Which bit - about the predicate rules?
17:30:05 <timmmit> I have a feeling that if people in the community nurdle on it they might realize that that is actually what is happening.
17:30:45 <DanC> well, in the conventional terminology, "the semantics of a URI/symbol" just doesn't parse. math/logic folk speak of the semantics of a language, not a symbol. They speak of an interpretation, which binds a symbol to its denotation.
17:30:58 <timmmit> In XML vocabular specs, for example, people define the meaning of tags.  And in other specs they define the meaning of attributes.  And the two specs can fight, as the attribute can change the meaning of the tag.
17:31:31 <danbri_> And particularly evil attributes can change the meaning of elements other than the one they're attached to.
17:31:43 <danbri_> I randomly assert, wishing there was an XML Semantics spec to prove me wrong.
17:32:21 <DanC> yes, and the math/logic folks developed all this model theory stuff so that any paper that does something like those evil attribute specifications gets a big BZZZT and never gets published.
17:32:39 <danbri_> RDF/XML is a little island of sanity amongst the tag-soup. But the RDF spec can't say anything about the meaning of the XML surrounding it; or alongside it in some DOM/Infoset. Not our job.
17:32:44 <timmmit> What is the word that the math/logic folks us eto express the relation between (what I call a formula) the instance of a langauge and that which a human brain is intended to derive from it?
17:33:18 <timmmit> danbri - xml around it - agreed. No semantics defined by rdf tags embedded in soup unless soup spec says so.
17:33:54 <DanC> good question, tim... lemme think...
17:34:08 <timmmit> I think I will stick to "meaning" in this discussion, wheer we are tying things down to english, and use "semantics" for thinsg you can define with more mathematics.
17:34:32 <danbri_> Well, they'd still have their own content, but their relationship to the enclosing document and protocol context would be unclear. It could be negation, query, imperative etc. The RDF itself still has meaning, just that meaning is detached from any usable context.
17:35:05 <DanC> they speak of I(F) = true or I(F)=false, where I is an interrpetaion, a mapping from syntactic expressions to their denotation. perhaps timbl:semantics = math-log:denotation.
17:35:33 <DanC> F being a formula
17:35:36 <timmmit> Danbri, Yes, I agree also that the RDF still has meaning associated with it.  Inherently. 
17:35:37 <DanC> or a symbol
17:36:03 <danbri_> So DanC/timbl, do you believe an RDF/XML document is the sort of thing that can be true or false? Or that it encodes something (...) that can be true or false?
17:36:23 <danbri_> ...do you believe that of an island of RDF/XML markup? of an arbitrary XML document?
17:36:28 <kifbot> kifbot has quit
17:36:29 <danbri_> my answers: yes, yes, no.
17:36:52 <DanC> good question, danbri. each interpretation assigns either true or false to each RDF document. The quasi-static view of the world assumes there's just one interpretation, or something.
17:36:53 <danbri_> (I mean, it encodes something that can be t/f; as do fragments; but generic XML doesn't)
17:37:16 <timmmit> Danbri, I agree yes, yes, no so long as you realize it can be true or false but may be neither
17:37:25 <timmmit> (paradox, etc)
17:37:32 <kifbot> kifbot has joined #rdfig
17:39:02 <timmmit>     s/you/one/
17:39:31 <DanC> I have a real worry that TimBL is headed toward making the quas-static view into The Web Architectecture, perhaps unwittingly.
17:40:23 <timmmit> Actually, the quasi-static view is a very important first approximation to web architecture.
17:40:56 <Iron_SpermWhale> Iron_SpermWhale has quit
17:40:56 <chihchun_> chihchun_ has quit
17:40:56 <dmiles> dmiles has quit
17:40:56 <danbri_> danbri_ has quit
17:40:56 <libby> libby has quit
17:40:56 <larsbot> larsbot has quit
17:40:56 <PONG> PONG has quit
17:40:56 <tav`> tav` has quit
17:40:56 <tav> tav has quit
17:40:56 <dajobe> dajobe has quit
17:40:57 <timbl> timbl has quit
17:40:57 <ericP> ericP has quit
17:40:57 <em> em has quit
17:40:57 <timmmit> It has simple axxiomatic structure, and for people it allows things like navigational metaphors etc.
17:40:58 <DanC> yes, but let's be/remain clear that it's an approximation.
17:41:28 <timmmit> The important thing is that it is an approximation which we can derive from the full(er) theory.
17:41:41 <DanC> hmm...
17:41:58 <timbl> timbl has joined #rdfig
17:41:58 <libby> libby has joined #rdfig
17:41:58 <Iron_SpermWhale> Iron_SpermWhale has joined #rdfig
17:41:58 <em> em has joined #rdfig
17:41:58 <ericP> ericP has joined #rdfig
17:41:58 <larsbot> larsbot has joined #rdfig
17:41:58 <chihchun_> chihchun_ has joined #rdfig
17:41:58 <danbri_> danbri_ has joined #rdfig
17:41:58 <dajobe> dajobe has joined #rdfig
17:41:58 <PONG> PONG has joined #rdfig
17:41:58 <tav`> tav` has joined #rdfig
17:41:58 <tav> tav has joined #rdfig
17:41:58 <dmiles> dmiles has joined #rdfig
17:42:16 <DanC> it's quite like the way the classical physics view is related to quantum physics, no? quite subtly.
17:42:33 <timmmit> Well, think of special relativity.
17:42:41 <DanC> I'm not sure I can.
17:42:48 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome
17:43:09 <timmmit> Not subtle. Just proovable equivalent to normal netonian cardesian stuff when velcities re much less that light
17:43:18 <danbri_> (one side reason for this is that an RDF/XML document can contain context-specific expressions such as 'rdf:about=""', which need resolving...)
17:43:23 <timmmit> Ok, think of your http stuff in Larch.
17:43:55 <timmmit> You can derive log:semantics from a whole lot of models including URI schemes, HTTP protocols, mime types and parsing XML documents.
17:44:01 <DanC> not subtle: but it is! it's equivalent for engineering purposes; i.e. the same to several orders of magnitude. But *not* identical.
17:44:27 <timmmit> But once you ahve done that, cwm can work with log:semantics as a black box, for the duration of a run,during which no document is got more than once.
17:44:59 <DanC> that's a HUGE design limitation for cwm; one that should be documented LOUDLY!
17:45:10 <DanC> it means you can't use cwm for long-running stuff like caching proxy servers.
17:45:16 <timmmit> But it is a deisgn limitation in every program.
17:45:35 <DanC> hell no! dns caching servers are very aware of time context of information.
17:45:36 <timmmit> Well, conventional programs which run and terminate.
17:45:38 <timmmit> Like make
17:45:45 <timmmit> Like CVS
17:45:48 <timmmit> Like cc
17:45:53 <timmmit> Like sed awk grep 
17:46:09 <DanC> yes, programs that don't do distributed computing.
17:46:38 <DanC> * DanC is still frustrated at undocumented assumptions in libwww that persist today in web clients that treat DNS records like they last forever.
17:46:56 <libby> libby has quit
17:47:33 <timmmit> re  DanC is still frustrated at undocumented assumptions in libwww that persist today in web clients that treat DNS records like they last forever.
17:47:52 <DanC> I want you to be *very careful* that you don't give the message that Web Architecture only works in non-distributed computations!
17:47:53 <timmmit> People make assumptions. ("make will only have 10 users ...")
17:48:03 <timmmit> Thee assumptions later have to be retracted.
17:48:47 <lilo> [GlobalNotice] Hi all.  We're aware that our routing has problems.  We're awaiting a patch we hope will arrive in the next 8 hours to resolve the link counting problem.  Thanks in advance for your patience. *Sighrin*
17:49:01 <reagle-tu> reagle-tu has quit
17:49:51 <timmmit> Suppose you start with a simple model, in which log:semantics is quasi-static, and then you turn on the underlying meachanisms live. That is you, go and replace the DNS assumption with the real DNS protocol.  The intial assumptions made a useful map, and then when you later ellaborate the dteails underneath, eitehr you justify the whole thing or yopu find a snag.
17:50:01 <DanC> at face value, if you take http://www.w3.org/DesignIssues/Axioms.html#unique and conjoin it with the RDF model theory, you get something that's patently absurd; you get a pretty strong argument *against* dealing with rdfms-assertion in this lifetime ;-)
17:50:19 <timmmit> * timmmit is enjoying this but hungry
17:50:26 <DanC> * DanC is hungry too.
17:50:52 <larsbot> larsbot has quit
17:50:55 <ericP> ericP has quit
17:50:55 <timbl> timbl has quit
17:50:55 <em> em has quit
17:50:55 <chihchun_> chihchun_ has quit
17:50:55 <Iron_SpermWhale> Iron_SpermWhale has quit
17:50:55 <dmiles> dmiles has quit
17:50:55 <danbri_> danbri_ has quit
17:50:55 <dajobe> dajobe has quit
17:50:55 <PONG> PONG has quit
17:50:55 <tav`> tav` has quit
17:50:55 <tav> tav has quit
17:51:19 <DanC> if you conjoin them, you get something equvalent to saying that newtonian mechanics is an exact model of the real world.
17:51:22 <timbl> timbl has joined #rdfig
17:51:22 <Iron_SpermWhale> Iron_SpermWhale has joined #rdfig
17:51:22 <em> em has joined #rdfig
17:51:22 <ericP> ericP has joined #rdfig
17:51:22 <chihchun_> chihchun_ has joined #rdfig
17:51:22 <danbri_> danbri_ has joined #rdfig
17:51:22 <dajobe> dajobe has joined #rdfig
17:51:22 <PONG> PONG has joined #rdfig
17:51:22 <tav`> tav` has joined #rdfig
17:51:22 <tav> tav has joined #rdfig
17:51:22 <dmiles> dmiles has joined #rdfig
17:51:29 <DanC> s/equvalent/analagous/
17:51:32 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome
17:51:40 <timmmit> * timmmit is going to find something to eat
17:52:11 <timmmit> In teh axiom, "meaning" or a URI can be "the current front page of the Times". 
17:52:24 <timmmit> You can't join axiom:meaning with rdfs:semantics
17:52:47 <timmmit> The "meaning'  bit in specs is where a lot of reference to protcol goes in in real life.
17:53:23 <lasDesk> lasDesk has joined #rdfig
17:58:20 <chihchun_> chihchun_ has quit
17:58:20 <Iron_SpermWhale> Iron_SpermWhale has quit
17:58:20 <danbri_> danbri_ has quit
17:58:20 <dmiles> dmiles has quit
17:58:20 <dajobe> dajobe has quit
17:58:20 <PONG> PONG has quit
17:58:20 <tav`> tav` has quit
17:58:20 <tav> tav has quit
17:58:20 <ericP> ericP has quit
17:58:20 <timbl> timbl has quit
17:58:20 <em> em has quit
17:58:59 <DanC> hhmm... now, where are we?
17:59:59 <bijan> bijan has joined #rdfig
18:00:08 <DanC> The RDF model theory, syntax, and test cases specs don't connect RDF syntax with real life in any way at all. I don't think RDF M&S did either. I think the RDF Schema spec tried to, but not in ways that I'm very happy with.
18:00:17 <dmiles> dmiles has joined #rdfig
18:00:17 <danbri_> danbri_ has joined #rdfig
18:00:17 <chihchun_> chihchun_ has joined #rdfig
18:00:17 <Iron_SpermWhale> Iron_SpermWhale has joined #rdfig
18:00:17 <tav> tav has joined #rdfig
18:00:17 <tav`> tav` has joined #rdfig
18:00:17 <PONG> PONG has joined #rdfig
18:00:17 <dajobe> dajobe has joined #rdfig
18:00:17 <ericP> ericP has joined #rdfig
18:00:17 <em> em has joined #rdfig
18:00:17 <timbl> timbl has joined #rdfig
18:00:30 <DanC> * DanC is seeing a lot of netsplits
18:00:41 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome
18:03:06 <DanC> Tim, back to WG tactics, today's meeting didn't reach resolution on #rdfms-assertion; we agreed to take a look at an RDF mime type spec. Do you think that's a reasonable place to treat this issue?
18:03:39 <DanC> the whole issue of how RDF documents and terms are grounded in the web is quite another one; i.e. whether daml:imports is redundant.
18:04:03 <timmmit> I think that it is better to set a precedent by saying what  an RDF document means than to put it offf to the mime type spec.
18:04:20 <timmmit> (redundant?)
18:05:06 <DanC> redundant: if I use rdfs:subClassOf, do I need to say "this ontology imports RDFS" explicitly, or have I already done so?
18:06:12 <DanC> "what an RDF document means" -- back to square one. harumph.
18:06:28 <bijan> bijan has quit
18:07:31 <timmmit> If you use rdfs:subClassOf, you are using it in the way that the RDFS authors intended or making a mistake.  Now *some* information, rather crude infromation, about the term is in the schema.  You may or may not want to  invoke those triples explictly as part of the message yuo are sending.
18:08:25 <chihchun_> chihchun_ has quit
18:08:55 <timmmit> So I don't think "imports" is redundant.
18:09:06 <DanC> ok.
18:09:19 <timmmit> But anyone sould be justified in using the schema in processing your document.
18:09:22 <DanC> I'm not sure how to formalize "... intended or making a mistake".
18:09:43 <timmmit> Specs don't talk about "or making a mistake".
18:09:51 <timmmit> Or it would have to be every otehr line.
18:09:57 <timmmit> That is understood.
18:10:07 <DanC> yes, but I don't know how to say "as intended" either./
18:10:28 <DanC> a lot of this dicussion... e.g. "justified" has to do with actions and their consequences, as much as truth/falsehood.
18:11:35 <timmmit> timmmit has quit
18:11:40 <DanC> hmm... another issue... hopefully related: there's an issue about boolean properties...
18:11:43 <timmmit> timmmit has joined #rdfig
18:11:49 <DanC> hmm... another issue... hopefully related: there's an issue about boolean properties...
18:11:50 <timmmit> * timmmit #$%^&*(#$%^&*(#$%^&
18:12:15 <timmmit> * timmmit got split this time.
18:12:24 <DanC> I suggested that a way to say "joe is not a chocolate lover" is: { :Joe a [ ont:complimentOf :ChocolateLover]} ...
18:12:49 <timmmit> * timmmit wishes therre was/hew knew a way to make mIRC keep the old text when it loses connection.
18:12:52 <DanC> PatH pointed out that one of the lemmas in the RDF model theory says that all RDF documents/formulas are satisfyable; there's no way to express contradition in RDF.
18:13:15 <timmmit> Eh?
18:13:49 <DanC> "Herbrand Lemma. Any RDF graph has a satisfying interpretation." -- http://www.w3.org/TR/2001/WD-rdf-mt-20010925/
18:14:03 <timmmit> satisfying what?
18:14:18 <DanC> satisfying the RDF graph/document/forumal.
18:14:35 <timmmit> under what definition of "satisfaction"?
18:14:48 <DanC> the one in the model theory spec ;-)
18:15:05 <DanC> i.e. the usual one for existential/conjunctive fragment of FOL
18:15:44 <chihchun_> chihchun_ has joined #rdfig
18:16:16 <timbl-mit> timbl-mit has joined #rdfig
18:16:16 <timmmit> timmmit has quit
18:16:49 <DanC> I thought surely ont:complimentOf could be layered on top of this RDF model theory, but I guess not...
18:16:57 <timbl-mit> I expect in a rather provincail way that "conjunctive" probably excludes "compliment".
18:17:36 <DanC> somehow, yeah.
18:18:08 <DanC> it was starting to look like RDF 1.0 and the ontology layer could share a model theory... that everything would dovetail nicely... but now I'm starting to wonder.
18:18:45 <timbl-mit> * timbl-mit is secretly glad that other folks are doing the worrying!
18:19:22 <timbl-mit> Now, what do you think it will take to get agreement on the points earlier which I will now number
18:19:35 <timbl-mit> 1. maning of document = conjunction of meaning of statements
18:19:48 <timbl-mit> 2. meaning of statement defined by spec of precicate
18:19:55 <DanC> hmm... cwm doesn't really bother with any sort of notion like "false" or "not"; if you say { Joe a :ChocolateLover, [ ont:complimentOf :ChocolateLover ]}, it doesn't do anything special.
18:20:09 <timbl-mit> 3. meaning of   x rdf$type c  defined by c.
18:20:29 <timbl-mit> You haven't give cwm the complimentOf axioms.
18:20:55 <DanC> is there any way to state complimentOf axioms that will cause cwm to do anything interesting?
18:21:03 <timbl-mit> One early test I did actully threw up an error when someone tried to delare that they were both emale and female.
18:21:51 <DanC> threw up an error? it just concluded something, no?
18:22:03 <DanC> * DanC would mind dropping the cwm/not thread and going to the numbered list...
18:22:37 <timbl-mit> this log:forAll :x,:y,:z. {:x a :y; [ont:complementOf :y]}log:implies { :something a :interesting}. 
18:22:47 <timbl-mit> * timbl-mit drops it
18:22:52 <DanC> 1 is pretty much in the model theory spec: "The denotation of a ground RDF graph in I ..."
18:23:02 <timbl-mit> check.
18:23:15 <DanC> if E is a ground RDF graph then I(E) = false if I(E') = false for some asserted triple E'
18:23:15 <DanC>            in E, otherwise I(E) =true.
18:24:00 <DanC> NOTE WELL: per the model theory, the "meaning of a document", i.e. the denotation of an RDF graph, is one bit: true/false.
18:24:09 <DanC> (one bit per interpretation, that is)
18:24:31 <timbl-mit> exactly.
18:24:43 <timbl-mit> Or loosely, the set of interpretations fro which that bit is 1.
18:25:06 <DanC> quite.
18:25:29 <timbl-mit> Or in terms of language, the instance of anoetrh languge which has the same set of interpretations.
18:25:30 <DanC> by the way... that's possible world semantics, in a nutshell....
18:25:53 <DanC> rather than interpreting forumlas as true/false, you interpret them as the set of interpretations that satisfy them.
18:26:01 <lilo> [GlobalNotice] Hi all.  Things seem to be a bit more stable, but let us know if you are having any unusual problems accessing the servers at this point.  Thanks.
18:26:19 <timbl-mit> Remember we are grounding this not in mT but in the language which is w3C specspeak.
18:26:47 <DanC> ok... check for #1...
18:27:02 <timbl-mit> The other two are obvious.
18:27:04 <timbl-mit> ;-)
18:27:29 <timbl-mit> The other two I declare to be true as director of the consortium.  0.95 ;-)
18:27:31 <DanC> well, for #1, I was worred you wanted "the meaning of a document" to be, e.g. a protocol specification, or a logic, or a poem.
18:27:55 <DanC> #2 is stated thusly in the MT spec: [[ if E is an asserted triple s p o . then I(E) = true if <I(s),I(o)> is in IEXT(I(p)), otherwise
18:27:55 <DanC>            I(E)= false. ]]
18:28:30 <timbl-mit> * timbl-mit might suggest at least that the spec scans and rhymes...
18:28:34 <DanC> you can read  IEXT(I(p)) as "spec of precicate" maybe?
18:29:33 <DanC> I(p) is what p denotes, and IEXT(P) is the "extension" of that denotation; i.e. the set of pairs in the relation.
18:29:35 <timbl-mit> Well, do you feel it is asserted in such a way as you could use it to demonstate that the author indirectly asserts the conjunction of the spec and the schema?
18:29:53 <DanC> no.
18:30:17 <timbl-mit> Then perhaps we need to get a bit more specific.  But I am not aking you to do this in math.
18:30:25 <DanC> remember: we compute a "meaning" of a document/statement on a per-interpretation basis.
18:30:33 <sandro> timbl-mit, dircproxy retransmits the last <n> lines of output per channel when you reconnect.
18:31:57 <DanC> hmm... in a way, I'm glad you're not asking me to do it in math, TimBL, but how the heck do we do it, then? i.e. how do we set up test cases or whatever?
18:32:15 <timbl-mit> Same as for any spec.
18:32:37 <DanC> no, not same as for any spec.
18:32:43 <timbl-mit> This is thr grounding in english. have to use people. Expensive, slow, unpredictable
18:33:00 <timbl-mit> but wonderful.
18:33:36 <timbl-mit> Same as any spec in that we have to use english. Different in that we are actually going to define a meaning rather than ignore the issue.
18:33:43 <larsbot> larsbot has joined #rdfig
18:33:59 <DanC> for most specs, there's a mathematical concept that we're trying to convey. noteable counter-examples include html:p; are you suggesting that sort of hand-waving?
18:34:13 <timbl-mit> Point two said "go read the property spec -- -the meaning is what it tells you"  That is english for people.
18:35:01 <DanC> but "the property spec" assumes the quasi-static view; we have to be careful to be clear that we've stopped speaking in crisp mathematical terms...
18:35:12 <timbl-mit> Well,  I am expecting things like "the ucc:cost preciate indicates that eth object is the cost in dollars of the subject".
18:35:21 <timbl-mit> People will define things like that.
18:35:34 <timbl-mit> they won't give the protocol.
18:35:42 <timbl-mit> But they will have a protocol in mind.
18:36:09 <DanC> but they protocol they have in mind isn't mathematical. it involves risk/reward, courtrooms, etc.
18:36:24 <timbl-mit> We can do better in that with a suitable language we can constrain the sequence of messages as well as the message. But in the end we use english to tie the $$$ to reality.
18:37:02 <DanC> what we really mean by "go see the property spec" is "gather evidence from the world about what folks think p means; act accordingly. There's a strong social convention that doing a GET at P should be respected."
18:39:01 <timbl-mit> No, we mean go see the spec. Period.
18:39:08 <DanC> I don't.
18:39:22 <DanC> I don't mean that there is certainly exactly one such spec.
18:39:34 <timbl-mit> We don't mean "ask whoever you like what they thing ucc:cost should mean"
18:39:41 <sandro> This seems like a good subject for the Joint Committee or WebOnt.    There are some hard ontology problems in connecting formal logic to reality.
18:40:02 <timbl-mit> I think it is called philosophy.  It is what we are all basically about.
18:40:22 <danbri_> Some people have massive and expensively created ontologies that may be ACL'd, or may be partially disclosed with additional information available in exchange for money or form-filling. A plain GET isn't always enough.
18:41:22 <danbri_> I spent enough time in a Philosophy department to not wish a Philosophy WG on anyone. WebOnt WG is the wrong place to do philosophy (except as b/g reading).
18:41:44 <DanC> clearly these are centuries-old philosophy questions. Today's agenda is: what should the RDF core specs say in answer, if anything?
18:41:54 <danbri_> Not disparaging philosophers in the slightest. Just discussions that have lasted centuries don't belong in our WGs.
18:41:57 <timbl-mit> We are defining a grounded set of specs.  We are not asking people to go find some strong social convictions.  We are telling them a set of convictions which, if they use them, will deliver certain results and have certain properties. We will by making it a Rec demonstarte a strong social conviction in the stack.  This will enable the distribuedt system.
18:43:02 <DanC> all that said, TimBL, I can't say, with a straight face, "go see the property spec. Period." I happen to know several properties whose spec has changed over time.
18:43:11 <timbl-mit> I actually think that is we answer with points 1-3 then we don't end up getting enmored in philosophical discussion and we do say something real and useful.
18:43:19 <danbri_> my answer: RDF graphs (and RDF/XML infoset fragments) have propositional content, determined as a function of (core specs, URI spec, other enumerated factors). Protocols, container formats, compound documents may use this feature of RDF to create documents and protocols that have bindings to social reality.
18:43:44 <timbl-mit> waffle.
18:43:50 <timbl-mit> * timbl-mit excuses himself
18:44:02 <DanC> his is waffly but yours isn't, tim?
18:44:08 <timbl-mit> Yup.
18:44:23 <timbl-mit> Mine will resolve a clash of definitions between two specs. His won't.
18:44:28 <danbri_> not waffle; a reflection of waffle in half-a dozen specs. The meaning of an RDF/XML doc is fixed in part by the waffle in the URI spec.
18:44:29 <DanC> well, you haven't given me what I need to defend your position, tim.
18:44:34 <timbl-mit> Mine will get you your money back in court.
18:44:56 <DanC> yours will also result in contraditions every time W3C gets another member, tim.
18:45:24 <timbl-mit> No, because the specs are not talking about number of members.
18:45:37 <DanC> I won't be party to a decision to bake the quasi-static view of the world into the RDF core specs.
18:46:10 <danbri_> saying that the 'meaning' of a doc is just the and-ing of the statements in it doesn't get us where we need to go; it doesn't connect those triply things to the world. There's another step missing, where we say these 3-part thingies _mean_ stuff about the world.
18:46:33 <DanC> the w3c:member spec isn't talking about the number of members?
18:46:42 <timbl-mit> If you can come up with something better, in other words a model which incldues the time and causality and protocols, and us eit to write something in the spec which in the quasistatic approximation means what I said, then you can put that in the spec.
18:47:01 <timbl-mit> We don't have a w3c:member spec.
18:47:04 <danbri_> * danbri_ won't hold his breath
18:47:13 <DanC> the obligation is not on me to come up with something better, tim. "First, do no harm" applies as much to standards work as medicine.
18:47:54 <DanC> no w3c:member spec? what do you call http://www.w3.org/2001/04/roadmap/org ?
18:48:19 <timbl-mit> Well, harm is done by releasing a language spec and not doing as good a job as you can at explaining what something written in that language means.  (Look at xmlns) 
18:48:36 <timbl-mit> http://www.w3.org/2001/04/roadmap/org wasn't on the TR page when I last looked.
18:48:36 <dc_rdfig> D: http://www.w3.org/2001/04/roadmap/org from timbl-mit
18:49:01 <DanC> hmm... so the RDF spec is only about properties defined in W3C recommendations?
18:49:20 <timbl-mit> So if you don't point someone to the predicate when they are looking for the meaning of a triple then you are doing harm.
18:49:25 <DanC> you speak of xmlns as if it were feasible to write down what you meant in such a way that the community would agree.
18:49:46 <timbl-mit> * timbl-mit suggets drop history and excuses himself for bring it up
18:49:47 <AaronSw> * AaronSw reconnects
18:50:19 <DanC> * DanC agrees to leave history in the past
18:50:54 <timbl-mit> I mean that the roadmap spec is also RDF, and the RDF spec applies to it, but that I am not going to worry about its consistency as much as I am about something (like RDFcore) which is heading for the TR page. 
18:52:17 <DanC> so you agree that your suggestion for the RDF spec will result in inconsistencies when the quasi-static view turns out to be observably inaccurate?
18:53:30 <timbl-mit> Chose: ":x a  w3c:member." means that the author of the document representing :x indended to communicate that at the time of writing or at least at some nearby time, thing then represented by :x was a member of the thing at the time called in common parlance the World Wide Web Consortium".
18:53:41 <timbl-mit> or: means x is  a member of W3C.
18:54:31 <dajobe> dajobe has quit
18:54:55 <timbl-mit> Actually, what will happen when you move to a time-aware model is that the emanings of the messages will be leveraged and still part of the model.  "+" will still have the same properties.  just eth model will employ them within a more complex framework.
18:55:22 <timbl-mit> In most cases.
18:55:46 <timbl-mit> So a formula "z = x + y"  will remain in that form in the system.
18:56:06 <timbl-mit> Thje conclusions about it will be qualifies by the metadat about the message in which the formula was transmitted.
18:57:26 <danbri_> There are a bunch of things jumbled up in this RDFCore open issue. (i) Q: Is RDF designed for writing assertions about the world, where the social/legal meaning of the triples constituting those assertions is determined largely by the content of RDF/XML docs in the Web? A: Yup (ii) A: what factors fix the meaning of some RDF/XML? A: RDF core spec/s, URI spec, + the schemas used (iii) Q: Who is making the assertions encoded in some RDF/XML 
18:57:26 <danbri_> le as a mime-typed HTTP-transmitted XHTML-wrapped chunk of markup? A: not our problem.
18:58:07 <timbl-mit> That seems to make sense.
18:58:44 <danbri_> If we move to a time-aware model, we'll need to radically rewrite the DAML+OIL formalisation. That's a bunch of work. Right now it doesn't ack time, so can't make certain distinctions I've needed in my apps. But we didn't budget for a massive rewrite, did we?
18:59:19 <DanC> it doesn't ack time, but it doesn't conflict with it either; i.e. it won't have to be re-written when we ack time.
18:59:35 <DanC> DAML+OIL is like + in that way.
18:59:47 <timbl-mit> Voila. And so are my points 1-3.
18:59:52 <danbri_> * danbri_ -> food
19:00:16 <DanC> Tim, please. Your 1-3 points are mathematical absolutes?
19:00:50 <DanC> i.e. there is exactly one spec for http://www.w3.org/2001/04/roadmap/org#member?
19:00:56 <timbl-mit> No, but they are things which are stated without mention of the time/version issues, but which will still be useful when one works with thos eissues.
19:01:47 <DanC> I don't agree.
19:02:15 <timbl-mit> If a spec is a document which purports to be a spec for something and is written by the authroity for that thing, then I don't think it has a spec.  but the spec must be a superset of the schema which I think might exists
19:02:33 <timbl-mit> * timbl-mit checks - yup
19:02:34 <DanC> again: look at the phrase "the spec of the property". That phrase will become useless in a non-static view, no?
19:02:53 <timbl-mit> No.
19:02:58 <DanC> ;lREAgjf;av
19:03:16 <timbl-mit> In a real-world situation, lawyers argue the diferences.
19:03:20 <DanC> given two specs for w3c:member, you would find the phrase "the spec of the property" useful how?
19:03:37 <timbl-mit> This is how.'
19:03:38 <lilo> [GlobalNotice] Hi all.  Just a note.  Please help us test the new channel forwarding features.  To do so, join #forwardfrom.  You should be sent to #forwardto.  If your client crashes, if you don't get the proper window presence or focus, or can't type into the channel, let us know.  support@openprojects.net.  Thanks.
19:03:50 <danbri_> modify it to 'the first published specification of the property'; and add to RDFS/WebOnt some metastuff for declaring a schema 'published'.
19:05:04 <timbl-mit> The lawyers would look at the misunderstanding.  They would see that the message could have been written on the understanding of spec v1 and interpreted on the understanding of spec v2.  they would argue as to what the first would mean, and see whether it is "reasonable" to hold the person to the other spec.
19:05:24 <timbl-mit> The HTML group do this all the time. I tend not to.  They change the spec so as to not cuase very mcuh damage.
19:06:00 <timbl-mit> i don't give any guarantee that a system which uses http://www.w3.org/2001/04/roadmap/org#member will work next year in the same way.  (Indeed, mine didn't!)
19:06:27 <DanC> hmm... seems we need a new RDF core spec: "RDF for lawyers" ala the P3P statement of principles.
19:07:06 <DanC> I can apprecaite the intent of statements 1-3, but they just don't fit into any of the specs we're working on.
19:07:07 <timbl-mit> There is a quasi-static model, and then there are (or not as the case may be) things which allow you to use it a non-zero tiume after it was published and be careful 100 years afetr it was published. 
19:08:04 <timbl-mit> In practice, the  stuff works quite well for things like SVG, and for RDF I think it will work much better, oin that new namespaces will be used and rules written to relate them, so there will be much mroe persistence and the quasistatic model will hold over long times.
19:08:37 <timbl-mit> Don't fit?  The spec has no hole or has holes which shape?
19:08:59 <timbl-mit> New short chapter?
19:09:26 <DanC> the specs that the RDF core WG is working on are completely black-and-white. Mathematics and test cases: model theory, XML syntax, test cases to be exact.
19:09:40 <DanC> there's a primer in the works. I can imagine it going there.
19:10:03 <DanC> I had thought of the primer as having a "not normative" label on the top, though.
19:10:04 <timbl-mit> Ah. So they don't specify what an RDF document means.  We should look elsewhere for an RDF spec?  :-)
19:10:55 <timbl-mit> Unforunately, we have mathematicians writing in xml/(xhtml_mathml) now and not able to publish it because of a lack of this crucial part of the XML NS spec.
19:10:58 <DanC> tim, points 1-3 don't specify what an RDF document means either. They don't specify anything. They suggest/describe/hand-wave, but they don't specify.
19:11:47 <timbl-mit> Specify schmesify.   They ground in english, which all TRs do.
19:12:03 <DanC> no, tim; there's a difference between math and other stuff.
19:12:26 <timbl-mit> Where is the math explaining what svg:circle means?
19:13:06 <sandro> * sandro is looking at http://www.w3.org/2000/01/sw/swad-chart.rdf with an eye towards making the circle-centric page view.   It's daunting.  
19:13:35 <DanC> are you serious? a circle is a set of points equidistant from another point. The SVG WG argues in fine detail about which bits get to be 1 and which bits get to be 0 in detailed test cases; the answers are justified by analytic formulas.
19:14:01 <timbl-mit> The SVG spec actually just says "The 'circle' element defines a circle based on a center point and a radius.
19:14:01 <timbl-mit> <!ENTITY % circleExt "" >
19:14:02 <timbl-mit> <!ELEMENT circle (%descTitleMetadata;,(animate|set|animateMotion|animateColor|animateTransform
19:14:02 <timbl-mit>                 %geExt;%circleExt;)*) >
19:14:02 <timbl-mit> <!ATTLIST circle
19:14:02 <timbl-mit>   %stdAttrs;
19:14:04 <timbl-mit>   %testAttrs;
19:14:06 <timbl-mit>   %langSpaceAttrs;
19:14:08 <timbl-mit>   externalResourcesRequired %Boolean; #IMPLIED
19:14:10 <timbl-mit>   class %ClassList; #IMPLIED
19:14:12 <timbl-mit>   style %StyleSheet; #IMPLIED
19:14:14 <timbl-mit>   %PresentationAttributes-Color;
19:14:16 <timbl-mit>   %PresentationAttributes-
19:14:18 <timbl-mit> ooooops!
19:14:28 <timbl-mit> ---------------------- sorry
19:14:31 <timbl-mit>  The SVG spec actually just says "The 'circle' element defines a circle based on a center point and a radius".
19:14:46 <timbl-mit> It doesn't define what a circle is.
19:14:56 <timbl-mit> It uses the english word "circle" and that is enough.
19:15:13 <DanC> I'm not talking about how things are specified; I'm talking about what is specified.
19:15:20 <timbl-mit> Our specs are not grounded in math when it comes to what they mean.
19:15:28 <DanC> the concepts in the SVG spec are, in the end, mathematical.
19:15:37 <timbl-mit> You can specify constraints and manipulations and translations in math.
19:15:41 <DanC> yes, the SVG concept of circle is grounded in math.
19:16:38 <DanC> the community agrees what a circle is, and in no cases is a court of law expected to tell us otherwise.
19:17:06 <timbl-mit> just as the SVG spec didn't need to ground "circle" as { (x,y). (x-ax)^2+(y-zy)^2=(r^2)} so the RDF spec doesn't have to ground "specification of".
19:17:38 <DanC> but there *is no mathematical concept* behind "specification of"
19:17:39 <timbl-mit> The community agrees what the spec of rdfs:subClassOf is too.
19:17:50 <timbl-mit> Evn though it not a mathematical concept.
19:18:01 <DanC> rdfs:subClassOf is grounded in math. it's a black-and-white concept with meaning outside of social context.
19:19:36 <danbri_> 1-3 don't ground RDF/XML for me, because they don't guide the Web community in determining when some lump of colloquial RDF/XML is true or not. They're pretty vague from a deployment/testing point of view. 
19:20:24 <timbl-mit> rdfs:subClassOf is grounded in english.  "This property specifies a subset/superset relation between classes". http://www.w3.org/TR/rdf-schema/#s2.3.2
19:20:36 <timbl-mit> Ther is some more english following it.
19:20:36 <DanC> nope.
19:21:07 <timbl-mit> I see no formulae in the spec.
19:21:16 <danbri_> also in the new MT spec, more precision
19:21:20 <DanC> again: I'm not talking about how somehting is specified; I'm talking about whether the concept itself has meaning outside of social context, courts of law, etc.
19:22:06 <timbl-mit> Look at type in same spec:  "This indicates that a resource is a member of a class, and thus has all the characteristics that are to be expected of a member of that class".
19:22:09 <DanC> if you tell me { :x a :C1. :C1 rdfs:subClassOf :C2 } then that entails that { :x a :C2} and no court of law has anything to say otherwise.
19:22:41 <DanC> yes, tim, that's the sort of language we're trying to get rid of.
19:22:57 <danbri_> * danbri_ nods
19:23:04 <timbl-mit> It would be an easy thing to argue I agree as the meaning of tyhe english and the concelt of subClass is very clear ( even though the spec was a  little affly about proper subclasses!)
19:23:29 <timbl-mit> That definition of type talsk about "the characteristics that are to be expected of a member of that class".
19:24:05 <timbl-mit> Not, if it said "the characteristics specified as expected of a member of that class" then we would be even closer.
19:24:22 <DanC> that text is gone, tim.
19:24:46 <timbl-mit> http://www.w3.org/TR/rdf-schema/#s2.3.1  has it.
19:24:47 <dc_rdfig> E: http://www.w3.org/TR/rdf-schema/#s2.3.1 from timbl-mit
19:25:25 <DanC> gone meaning: I don't agree to it, and I don't expect it to be in a W3C Recommendation when the RDF Core WG is done.
19:25:30 <timbl-mit> E: Definition of rdf:type in the schema spec as is on the TR page
19:25:31 <dc_rdfig> commented item E
19:26:07 <timbl-mit> In that case you have a logic but no langauge.  You have manipulation rules but  no  meaning.
19:26:15 <DanC> quite!
19:26:18 <timbl-mit> You need both.
19:26:58 <timbl-mit> Normally, often meaning goes in the spec and the math in the normative appendix.
19:27:12 <DanC> yes, let's get back to tactics...
19:27:16 <timbl-mit> In some cases, (like daml properties) the axioms comprise the ENTIRE meaning of something.
19:27:37 <timbl-mit> But not for rdf:type.
19:27:45 <DanC> no? hmm...
19:28:00 <DanC> ok, no, rdf:type isn't completely specified.
19:28:17 <timbl-mit> No, rdf:type indicates that the things specified in the specification of the class apply to the object.
19:28:37 <timbl-mit> ... including things said in english.
19:28:57 <timbl-mit> But the nice things is that only classes and properties need to be grounded in english.
19:28:59 <DanC> hmm...
19:29:22 <timbl-mit> If it doesn't include the things said in english then you can't use it in society.
19:30:00 <DanC> "indicated"... what specification of rdf:type we have developed in the RDF Core WG is given in the model theory and test cases; we don't use fuzzy stuff like "indicates..." and such.
19:30:19 <timbl-mit> For me it is important to make an authoritative system in whcih people have the possibility to define things and let those definitions stand whatever otthers define.
19:30:29 <timbl-mit> We must hook this up to ecommerce.
19:31:00 <DanC> ok, but to hook it up to ecommerce, you'd agree that we've left the realm of math/c.s. and gone into stuff like the HTML P spec and the WAI guidelines, yes?
19:31:12 <timbl-mit> It's no godo if you order something and then I deliver it and then you don't pay because you don't claim your order made any commitment.
19:31:30 <DanC> commitment: that's how we started this discussion. agreed. now, back to WG tactics...
19:31:32 <DanC> please.
19:31:34 <timbl-mit> We have left the realm of math and gone into a technical/financial specification.
19:32:00 <timbl-mit> WG tactics.  Stick in a <1 page chapter on what an RDF document means. Please?
19:32:07 <DanC> where?
19:32:16 <timbl-mit> I wish sooooo much we had done this with HTTP!
19:32:20 <danbri_> Never mind where, what's it gonna say?
19:32:30 <DanC> and... yeah... what danbri said: what's it gonna say?
19:32:43 <timbl-mit> Points 1, 2 and 3, basically, though point 3 has to go into the schema spec.
19:33:13 <timbl-mit> Stretch the words a bit if you have to.
19:33:39 <DanC> points 1 and 3 are captured in the RDF MT currently, I think. It's point 2 that I see no home for.
19:33:56 <DanC> or rather...
19:34:05 <danbri_> And 2. ("meaning of statement defined by spec of precicate") is the foundational one.
19:34:24 <DanC> or rather: I see a long discussion, inside the WG, in the web community, maybe in the TAG, etc. to actually get it adopted.
19:34:30 <danbri_> And bogus, imho. It ignores the reference aspect of meaning that's going to be critical in a lot of legalistic scenarios...
19:34:56 <timbl-mit> Could you elaborate that so that I can dispute it?
19:34:59 <timbl-mit> :)
19:35:05 <danbri_> "I didn't sell you that gold-embossed leather bound book, I sold you that HTML page about the book" 
19:35:31 <DanC> "what if the schema changes?"
19:35:52 <DanC> [insert URI-fud here, ala "what if the domain name gets yanked due to trademark foo"]
19:36:01 <danbri_> Meaning is a horrible word, but it can be dismembered. There's a whole side of it to do with inference rules, deduction etc. We've got that covered. And there's the side to do with denotation/reference; which entities in the world some referring expression picks out.
19:36:09 <timbl-mit> That (book vs page) is a mistake which RDF won't make and kids normally get tired of by the age of 10.
19:36:21 <DanC> HAHAHAHAHAHAH
19:36:35 <DanC> RDF users make that mistake ROUTINELY! <dc:creator>Dan Connolly</dc:creator>
19:36:57 <DanC> users: including schema authors.
19:36:59 <timbl-mit> Yes ... it isn't RDF making the mistake.  It was DC I think not being specific.
19:37:12 <timbl-mit> USers will always of course err, being people.
19:37:44 <danbri_> Even looking to our _simplest_ referring expressions, URI names, its clear that what some RDF description is talking about is tangled up with URIs. That what some RDF _means_ is determined in part by the meaning of the URI name.
19:38:21 <timbl-mit> Schema  authors ... awful .. saw one the other day.... adding rdfs:range clauses like it was some like of union or something. Don't talk about it.
19:38:22 <danbri_> For URIs, we can at least make the simplifying assumption that the name->world binding is static. That doesn't work for more ambitious referring expressions, eg. 'the person whose foaf:mbox is webmaster@w3.org'. 
19:39:13 <danbri_> So the legal/social/t-and-s aspect of the meaning of RDF which we need to nail down is less w.r.t. inference etc (fixed: see MT spec), and more the murk of naming. We can't "say what RDF docs mean" without at least a nod at this component of meaning.
19:39:14 <DanC> * DanC doesn't see why the referring-expression issue is any simpler for URIs than for more complex expressions.
19:39:20 <danbri_> And it'll be atough bit of spec to write.
19:39:57 <timbl-mit> But backing up from the real-world time-varying problems to the question as to whether I have to specify the meaning of p(s,o) for every triple (p,s,o) separately... I propose we stop that at once and say that it depends on p. Which it does. 
19:40:08 <danbri_> Because we can assume that uri:werwreewrew always denotes the same individual in all interpretations; but same can't be made for 'the person whose...' expressions (unless we populate the WebOnt language with tricks to allow this).
19:40:17 <DanC> btw.. re (fixed: see MT spec) I'm suddenly not so sure; if the RDF model theory can't be used with ont:complimentOf, i.e. if every RDF schema has to have its own model theory, then I wonder if we've gotten anywhere.
19:41:04 <timbl-mit> We have gotten a long way. 
19:41:08 <dajobe> dajobe has joined #rdfig
19:41:37 <danbri_> Yes, I'd hoped we could have an MT that subsequent specs could just build on (by (mumble) writing more axioms... or something...).
19:41:48 <timbl-mit> You have not got a model theory which allows you to alk about prime numbers even though I can define the URI meaning "the class of prime number" just like that. 
19:42:31 <danbri_> Yes, it depends on p, but also on s and o. And on other stuff in the graph it s and/or o are blank nodes (so-called anon resources)
19:42:33 <DanC> long way: really? even if the xyz: schema can say "it's not the case that { :s xyz:foo :o. :s xyz:bar :o} entails { s xyz:foo :o}; we don't use that part of the RDF model theory" 
19:42:45 <danbri_> * danbri_ got to go
19:43:10 <timbl-mit> on s and o too? No.  p rules. Absoluetly.  You cannot define s such that iy changes the meaning of the statement s p o. 
19:43:20 <timbl-mit> I have to go too.
19:43:36 <DanC> pls send points 1-3 to www-rdf-comments, timbl
19:43:48 <DanC> and reference issue rdfms-assertion
19:44:29 <danbri_> and write down what you mean by 'meaning' ;)
19:45:48 <DanC> logger, pointer to now?
19:45:58 <logger> I'm logging.  I found 1 answer for 'pointer to now'
19:45:59 <logger> 0) 2001-11-16 19:45:48 <DanC> logger, pointer to now?
19:46:10 <dajobe> hmm, logger could do that
19:46:11 <DanC> logger, where is the log of this message?
19:46:13 <logger> I'm logging.  I found 1 answer for 'where is the log of this message'
19:46:14 <logger> 0) 2001-11-16 19:46:11 <DanC> logger, where is the log of this message?
19:46:19 <timbl> timbl has quit
19:46:41 <timbl> timbl has joined #rdfig
19:46:47 <sandro> * sandro laughs pretty hard
19:46:47 <DanC> logger, what timezone do you think in, I wonder?
19:46:50 <logger> I'm logging.  I found 1 answer for 'what timezone do you think in, I wonder'
19:46:51 <logger> 0) 2001-11-16 19:46:47 <DanC> logger, what timezone do you think in, I wonder?
19:46:52 <dajobe> UTC
19:47:29 <DanC> good. better would be if it said so (ala 19:46:47Z), but UTC is the right answer.
19:47:52 <dajobe> It says Z in the RDF logs
19:48:00 <DanC> ah... good.
19:49:07 <dajobe> what command should give the URI of 'now' or 'this log'?
19:49:25 <DanC> sandro, still working on bubble-specific swad-chart? I think it should be straightforward.
19:49:32 <DanC> dajobe... hmm... logger, pointer
19:49:43 <dajobe> okay
19:49:50 <DanC> or bookmark
19:49:56 <DanC> or: where am I? ;-)
19:50:03 <dajobe> all of aboev
19:50:10 <DanC> ok.
19:52:23 <sandro> DanC, I'm swapping between things -- haven't really done anything on that.
19:52:59 <sandro> DanC, in concept it's straightforward - but the file is scarey.  I guess a lot of it is that I don't know what most of the terms mean.
19:53:30 <sandro> Or which namespaces are important, or what they mean.    I can figure it out, but it looks unpleasant.
20:03:09 <sbp> sbp has joined #rdfig
20:05:52 <DanC> would you like to learn, sandro? or would you prefer that I just do it?
20:06:09 <DanC> and if you'd like me to do it, sandro, are there a few bubbles in particular you'd like me to start with?
20:09:58 <AaronSw> A:|SourceForge Drifting
20:09:59 <dc_rdfig> titled item A
20:10:10 <DanC> old bits, no AaronSw?
20:10:15 <AaronSw> Indeed.
20:10:28 <AaronSw> I was about to say as much.
20:10:41 <sbp> sbp has quit
20:10:50 <DanC> ah... you just titled it; I though you chumped it.
20:11:18 <AaronSw> A:I blogged this [on #swhack|http://blogspace.com/swhack/weblog/2001/11/14/#1005706690.985625]
20:11:19 <dc_rdfig> commented item A
20:12:18 <DanC> B:|a proposal to drop Issue rdfms-assertion
20:12:18 <dc_rdfig> titled item B
20:12:46 <bijan> bijan has joined #rdfig
20:12:48 <DanC> B:discussed extensively among TimBL, DanC, DanBri today.
20:12:48 <dc_rdfig> commented item B
20:13:00 <AaronSw> I suppose I should read the logs on this before I send in my proposal...
20:13:15 <AaronSw> Logger's logs need to be colored...
20:13:49 <DanC> B:resulting in TimBL's "points 1-3"; 1 and 3 are pretty much in the RDF model theory and are generally agreed; DanBri and DanC feel insufficiently armed to defend point #2 in the community.
20:13:50 <dc_rdfig> commented item B
20:15:39 <dajobe> colo(u)red?
20:17:17 <DanC> B:timbl:1. maning of document = conjunction of meaning of statements
20:17:17 <dc_rdfig> commented item B
20:17:20 <sandro> I'm torn, DanC.   How would you approach doing it?
20:17:28 <DanC> B:timbl:2. meaning of statement defined by spec of precicate
20:17:29 <dc_rdfig> commented item B
20:17:42 <DanC> B:timbl:3. meaning of x rdf$type c defined by c.
20:17:42 <dc_rdfig> commented item B
20:17:54 <DanC> sandro, there are rules for computing the transitive closure of the graph; I'd just tweak those.
20:18:09 <DanC> to wit: { :s is g:digraph of [];
20:18:10 <DanC>     g:hasNode [ [ a g:EdgeProperty ] :n2]
20:18:10 <DanC>  } log:implies { :s g:hasNode :n2. }.
20:18:24 <DanC> and: { :s is g:digraph of [];
20:18:24 <DanC>     g:hasNode [ is [ a g:EdgeProperty ] of :n1]
20:18:24 <DanC>  } log:implies { :s g:hasNode :n1. }.
20:20:20 <DanC> sandro, got a bubble to suggest I zero in on?
20:23:09 <sandro> How about "meeting secretary agent"
20:23:30 <DanC> is that one still in there?
20:23:33 <sandro> In "how" I was thinking about like: python, cwm rules, XSLT, ....?
20:23:44 <sandro> Maybe not -- that could be an old version I looked at.
20:24:22 <DanC> I quoted it out, I think. pick again?
20:25:00 <sandro> Wow -- it's even more scary than befre.  :-)
20:25:07 <sandro> How about DAML FY2
20:25:43 <AaronSw> I have to go with TimBL timbl-2.
20:25:48 <AaronSw> err on timbl-2
20:26:27 <AaronSw> he may use the wrong syntax, but his semantics are clear ;)
20:26:57 <DanC> ok, DAML FY2.
20:33:58 <AaronSw> * AaronSw is glad to see sandro has found dircproxy as enjoyable as I have.
20:34:26 <DanC> * DanC finishes drafting "hourglass" rules; starts test run...
20:34:38 <AaronSw> hourglass rules?
20:35:20 <AaronSw> Heh: <danbri> I spent enough time in a Philosophy department to not wish a Philosophy WG on anyone. WebOnt WG is the wrong place to do philosophy (except as b/g reading).
20:35:26 <DanC> in howwhy terms, an { :s :hourGlassAround :o } means that :s is a graph of :o and everything it depends on and everything that depends on it.
20:36:00 <DanC> i.e. ancestors and descendants, but not descendants of ancestors
20:36:05 <AaronSw> Hmm.
20:36:14 <AaronSw> :s must be a very large bother.
20:36:37 <AaronSw> I find it hard to imagine hourglass graphs will not suck in a large portio of the world.
20:37:01 <bijan> Hey, is RDFS a proper subset of DAML+OIL?
20:37:36 <AaronSw> I don't think so, DAML+OIL redefines terms like rdfs:Class w/ daml:Class.
20:38:43 <larsbot> does anyone know why?
20:39:08 <AaronSw> Some Model Theory weirdness, as I recall.
20:39:20 <AaronSw> It was a very technical reason.
20:39:24 <dajobe> because of subclass loops
20:39:31 <larsbot> aha
20:39:33 <dajobe> which we now allow
20:39:39 <larsbot> but DAML doesn't?
20:39:48 <AaronSw> No, but they thought we wouldn't.
20:39:49 <bijan> Just noticed subclass loops (in DAML, not in RDFS)
20:39:50 <dajobe> it does, and rdf class now does
20:40:00 <dajobe> rather rdfs:Class
20:40:06 <larsbot> so they could now be unified?
20:40:13 <AaronSw> Doesn't it have to do something with their fear of rdfs:Class rdf:type rdfs:Class ?
20:40:16 <dajobe> well, class and property
20:40:52 <mea_cu|pa> mea_cu|pa has joined #rdfig
20:41:07 <dajobe> I think Pat Hayes satisfied them on that point; where them=Peter P-S
20:41:15 <AaronSw> yes.
20:41:33 <larsbot> so there isn't any reason to keep them apart any more?
20:42:00 <dajobe> I think there are other reasons, concerns they have
20:42:24 <larsbot> ok
20:42:27 <dajobe> there is an email somewhere
20:42:46 <mea_cu|pa> mea_cu|pa has quit
20:49:48 <AaronSw> logger, pointer
20:49:48 <AaronSw> I'm logging. I don't understand 'pointer', AaronSw.  Try /msg logger help
20:51:47 <danbri_> yes, it was in large part w.r.t. subclass loops (which the RDFS draft-in-progres linked from RDF Core WG homepage no longer forbids).
20:52:41 <danbri_> fears of Class type Class... I believe they've been reassured by the RDF MT spec, which works around possible gotchas there
20:54:38 <larsbot> what are the gotchas in C t C?
20:54:47 <larsbot> it seems pretty safe to me
20:54:56 <larsbot> (subclass loops, on the other hand...)
20:55:33 <dajobe> AaronSw: waiting for a good time to restart logger
20:55:59 <AaronSw> The best time to restart logger is when you kick someone :)
20:56:06 <dajobe> I fixed that too
20:56:07 <AaronSw> Unless that bug was fixed...
20:56:16 <AaronSw> Great! I'll have to upgrade.
20:56:27 <dajobe> get cvs 1.72
20:56:34 <danbri_> it sounded too close to 'the set of all sets is a set', had people muttering about Russell's paradox...
20:57:02 <danbri_> * danbri_ forgets the detail
20:58:26 <DanC> well, not so much russel's paradox as well-foundedness...
20:58:41 <DanC> in (normal, well-founded) set theory, X is never an element of X.
20:58:55 <bijan> ZF, that is. :)
20:59:02 <bijan> Er. ZF style.
20:59:13 <DanC> the mendelson IEXT indirection allows rdfs:Class rdf:type rdfs:Class.
21:00:33 <danbri_> "he use of an explicit extension mapping to allow
21:00:33 <danbri_> self-application without violating the axiom of foundation was suggested by Chris Menzel",  http://www.w3.org/TR/2001/WD-rdf-mt-20010925/
21:01:14 <DanC> Menzel.
21:01:30 <bijan> er..wait. isn't rdfs:Class rdf:type rdfs:Class?
21:01:42 <DanC> yup.
21:02:10 <bijan> Ok. Now I'm confused about the linkage of CtC and acyclic inherentence trees :)
21:02:12 <dmiles> * dmiles wakes up for the interesting conversation
21:03:16 <danbri_> bijan, the two are linked in that they were both aspects of RDF/S that the DAML+OIL committee expressed concerns about
21:03:18 <dmiles> so what is the issue with a class being a class ?
21:03:19 <DanC> * DanC is unpleasantly surprised at the outcome of the hourglass test.
21:03:35 <bijan> Ah! Thanks dan.
21:04:34 <bijan> * bijan is liking the Model Theory. Kudos to all involved.
21:04:51 <AaronSw> I still think it needs one of those click-thru age-checks at the beginning.
21:04:51 <danbri_> see also http://www.daml.org/2001/07/RDFS-DAML+OIL-coordination.html
21:04:53 <larsbot> when you say Class type Class, you don't mean things like
21:05:02 <larsbot> a type b type c type d, but like a type a?
21:05:26 <timbl-mit> * timbl-mit wonders if the hourglass has has a URI
21:05:46 <bijan> So...what's practice if you have a schema that can be mostly expressed by RDFS but you need a little more (e.g., cardinality constraints)?
21:06:08 <bijan> ooo, thanks for the coordination link, dan.
21:06:10 <bijan> Very helpful.
21:06:38 <larsbot> ah, the coordination document explained it
21:06:46 <larsbot> so a type b type c is disallowed
21:08:01 <danbri_> If you want a little more, you have a variety of options. Annotate the schema with additional stuff from DAML+OIL directly. Or note the constraints in prose (fancy cardinality constraints, eg. 'martians have between 3 and 14 parents' need this ;-). Or put an rdfs:seeAlso from the schema to a "more stuff about this vocab" document that can have such stuff added into later.
21:08:16 <xena> xena has joined #rdfig
21:08:44 <AaronSw> AaronSw is now known as DanC_
21:08:45 <DanC> hourglass: local disk so far.
21:08:56 <DanC_> .email person@example.org Please get the peanut butter
21:08:57 <xena> email successfully sent.
21:09:01 <DanC_> DanC_ is now known as AaronSw
21:09:21 <DanC> what does xena use for From:?
21:09:35 <DanC> * DanC can think of a few IETF security considerations violated by xena
21:09:36 <AaronSw> Umm, <nick>@espnow.com as I recall
21:09:41 <AaronSw> Heh heh heh.
21:09:52 <deltab> yeah
21:10:10 <deltab> bounce successfully delivered to DanC_@espnow.com
21:10:11 <AaronSw> What good is a bot if it doesn't violate any IETF security considerations?
21:10:18 <dajobe> ahem
21:10:21 <AaronSw> Heh.
21:10:52 <DanC> hm... 
21:10:54 <AaronSw> dajobe, don't worry, I'm sure logger violates something...
21:10:54 <DanC> { :s :hourGlassAround :n1. :n1 [ a :EdgeProperty ] :n2 }
21:10:54 <DanC>   log:implies { :n1 :how :n2 }.
21:11:10 <DanC> { :n1 :how [ [ a :EdgeProperty ] :n2 ] } log:implies { :n1 :how :n2 }.
21:11:29 <danbri_> So long as it isn't traced to the RDF IG channel, I don't care. But if some doofus sends the president death threats, I don't want the FBI kicking _my_ door down...
21:11:56 <danbri_> Can you make sure the messages come wrapped in a bunch of 'provenance unknown' headers please? (maybe they do already...)
21:13:26 <AaronSw> AaronSw is now known as danbri
21:13:33 <danbri_> .email danbri-ietf-security-loopholes@rdfweb.org Note to self: install whitelist-based mail filtering...
21:13:34 <xena> email successfully sent.
21:14:23 <danbri> .email president@whitehouse.com PLEASE SEND THE FBI OVER RIGHT AWAY :)
21:14:25 <xena> email successfully sent.
21:14:27 <danbri> danbri is now known as AaronSw
21:14:38 <danbri_> danbri_ is now known as danbri
21:14:45 <DanC> aha! :EdgeProperty missing g: prefix.
21:15:03 <DanC> * DanC chalks up 30 minutes wasted by lack of RDFS validation tools...
21:15:08 <timbl-mit> Doesn't that sort of thing get you down?
21:15:22 <dmiles> larsbot: what if there was a secondcontext of bottem class?
21:15:25 <bijan> * bijan working with some SWI-Prolog code for that sort of thing.
21:15:31 <timbl-mit> * timbl-mit could have validated it but it would have taken >30mins :-)
21:15:51 <bijan> I'm trying to get the schemaexplorer guys to clean up and release their code.
21:16:00 <DanC> bijan, meet tim. tim, meet bijan. you guys should talk. TimBL's working on schema validation.
21:16:07 <larsbot> dmiles: I don't understand; please explain
21:16:36 <dmiles> larsbot: i mean that everything has a Clas and therefore part of all its classes superper classes...
21:16:53 <dmiles> (me will be better grammatically in a minute)
21:16:59 <DanC> hmm... should rdf2dot take a parameter that tells which graph to extract? my data is going to have lots of graphs in it.
21:17:05 <bijan> Well, the code I'm dorking with is that behind: http://wonkituck.wi-inf.uni-essen.de/rdfs.html
21:17:14 <dmiles> larsbot: are you familiar with cyc ?
21:17:19 <larsbot> not really
21:17:29 <larsbot> I know what it is, but no more
21:17:40 <dmiles> i was asking so i could give you an example of what i mean
21:17:59 <larsbot> in CycL? you can give it a try, anyway
21:18:03 <dmiles> ah, well basicly everything has a class assertion somewhere
21:18:06 <timbl> I have schema validation enough to ctach spelling errors.
21:18:58 <dmiles> larsbot: i have had an issue where someone invokes a query (isa ?X ?Y)
21:19:08 <DanC> tim, is your schema validation stuff smart enough to know that :foo is  a property even though I only told it, explicitly, that :foo rdfs:domain :
21:19:10 <DanC> C?
21:19:25 <dmiles> maybe to list all entities and there bottem most classes
21:19:47 <larsbot> dmiles: looks fine to me -- what's the issue?
21:20:06 <dmiles> so.. i have adoptted a (bottemisa ?X ?Y)
21:20:41 <dmiles> this removves the combinatoriic issue that classes being classes have
21:20:43 <bijan> Dan, what happens if you run that through the schema explorer?
21:20:54 <danbri> doesn't cyc have something like instanceOf and allInstanceOf for that distinction?
21:21:08 <DanC> dmiles, pls consider another arg: (bottemisa ?X ?Y ?C) as in: "the most specific class of ?X given in knowledge-base ?C is ?Y"
21:21:17 <timbl> DanC, no it isn't yet.  For that I need to be able to merge the schema with axioms which needs the log:conjunction built-in. 
21:21:32 <larsbot> dmiles: then you're way ahead of me; I've yet to comprehend the problem with classes being classes
21:21:53 <timbl> :-)
21:22:07 <JHendler_> JHendler_ has joined #rdfig
21:22:09 <timbl>   <defaults.n3>     :ERROR_IN_SCHEMA_ACCESS_FOR gv:attributeName,
21:22:09 <timbl>               gv:hasNode,
21:22:09 <timbl>               gv:label;
21:22:09 <timbl>        :ERROR_NO_SCHEMA_OR_NOT_DECLARED log:outputString,
21:22:09 <timbl>               str:concatenation .
21:22:09 <timbl>   gv:attributeName     :SCHEMA_ACCESS_ERROR "" .
21:22:12 <timbl>   gv:hasNode     :SCHEMA_ACCESS_ERROR "" .
21:22:13 <timbl>   gv:label     :SCHEMA_ACCESS_ERROR "" .
21:22:30 <larsbot> dmiles: ok, so it evokes visions of the set of all sets, but...
21:22:38 <dmiles> larsbot: actually it hasnt been a problem yet
21:22:57 <larsbot> uh, ok.  but some people worry about it, right?
21:23:13 <dmiles> it could be if someone expected an existencial to only include things in 4 D space
21:23:36 <DanC> ok, hourglass test works better now, but I'm running into the modelling issue I pointed out a while ago:
21:23:54 <danbri> for grumbles about the flat structure of RDFS I think incl. the class-type-class thing, see http://citeseer.nj.nec.com/450519.html
21:23:55 <danbri> http://citeseer.nj.nec.com/450519.html
21:23:56 <dc_rdfig> F: http://citeseer.nj.nec.com/450519.html from danbri
21:23:59 <dmiles> or maybe an interaction with other quanitifiers.. but this is all based on the what forall type constraints are
21:24:13 <DanC> it's not enough to say "rm:needs is an edgeproperty." I have to be able to say whether it is or isn't on a graph-by-graph basis.
21:24:15 <danbri> F:|Metamodeling Architecture of Web Ontology Languages, by Jeff Z. Pan, Ian Horrocks
21:24:15 <dc_rdfig> titled item F
21:24:21 <larsbot> dmiles: I'm clueless, I'm absolutely clueless
21:24:49 <DanC> F:presents an unnecessarily constrained view of the world, IMO.
21:24:49 <dc_rdfig> commented item F
21:25:32 <dmiles> oh i mean to say if someone says (thereExistsAtLeast 4 ?Things (isa ?Things Entity))
21:25:51 <larsbot> ok, but what's that got to do with classes being classes?
21:26:15 <larsbot> what's the problem?
21:26:18 <dmiles> that could meran there are at least 4 classes .. but we are talking about instances..
21:26:43 <dmiles> it just it could maker an assertion like that cause an inconsistancy i guess
21:26:43 <danbri> * danbri trying to download the PDF to find the quote
21:26:48 <timbl> timbl has quit
21:26:51 <timbl> timbl has joined #rdfig
21:27:02 <larsbot> how could it? if resources can be both instances and classes,
21:27:04 <larsbot> where is the problem?
21:28:02 <logger> logger has joined #rdfig
21:28:02 <asimov.openprojects.net> topic is: RDF/Semantic Web Chat: Capabilities or ACLs? - http://www.w3.org/RDF/Interest/#irc
21:28:02 <asimov.openprojects.net> Users on #rdfig: logger timbl JHendler_ xena bijan dajobe larsbot timbl-mit chihchun_ em ericP PONG tav` tav Iron_SpermWhale danbri dmiles lasDesk kifbot grove_ DanC jang AaronSw sandro deltab dc_rdfig 
21:28:03 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome
21:28:04 <DanC> hmm... another problem with the DAMLFY2 hourglass diagram: :after is an edgeproperty (for layout purposes), but it shouldn't contribute :how arcs.
21:28:10 <larsbot> dmiles: if I'm being stupid now, just tell me
21:28:22 <dmiles> well it just casues overhead since an infernece engine cannot be optimized to only walk over the sencond argument of 'isa' assertions
21:28:45 <AaronSw> * AaronSw receives fanmail from "Computer Weekly" readers ;)
21:28:53 <dmiles> me4aning it removes oppertunity for optimization.. but really no problems
21:29:12 <larsbot> ah, ok, then I'm with you
21:29:16 <dmiles> but the inconsisatncy i was refering to happens becasue of pre-existing optimization
21:29:35 <dmiles> that people think ar eintrinsic to an iference mechanism
21:29:43 <larsbot> right, so it's not a mathematical necessity, it just seems that wa
21:29:44 <larsbot> y
21:30:08 <larsbot> what really confuses me, however, is this subclass thing
21:30:39 <larsbot> what does it mean if I say { :a :subclassof :b, :b :subclassof :a } ?
21:30:41 <dmiles> explain
21:30:41 <sbp> sbp has joined #rdfig
21:30:46 <larsbot> just did :)
21:30:55 <danbri> some complaints about rdfs:Class in (huge image) http://citeseer.nj.nec.com/cachedpage/450519/5 -- basically that RDF jumbles up instance and schema levels, which should allegedly be kept more carefully  separated...
21:31:06 <dmiles> is that a query or an assertion?
21:31:19 <larsbot> it's a query
21:31:31 <larsbot> what do those two RDF statements mean, taken together?
21:31:42 <larsbot> if subclass loops are allowed this must have some meaning
21:31:52 <danbri> that the two classes have a common extension
21:32:03 <larsbot> meaning that they are the same class?
21:32:08 <danbri> they may mean different things, but that anything that's a member of the one, is also a member of the other.
21:32:21 <dmiles> well i have an inference engine that detects that exact issue and return (forall ?X (isa ?X Class))
21:32:24 <larsbot> how can they then mean different things?
21:32:52 <dmiles> meaning it retunrs er ?a
21:32:58 <dmiles> ?A = ?b
21:33:09 <sbp> all the members of the chess club may be members of the swimming club...
21:33:26 <danbri> pick any two categories that have the same members, but differing definitions. The definitions could even be equivalent (but one in French, the other in English...).
21:33:32 <larsbot> club membership and class-instance are different
21:33:41 <sbp> not if you freeze it in time
21:33:56 <larsbot> imho, yes, they are semantically different
21:34:08 <sbp> (:Bob :Susan :Gary) a :2001ChessClub .
21:34:12 <larsbot> danbri: then they are the same class, but the tools don't know it
21:34:36 <dmiles> i agree with that freezing.. as in they are two differnt contexts of the same problems
21:34:37 <larsbot> sbp: fine, but bob is not an instance of ChessClub
21:34:51 <bijan> I suspect that you can't really divorce subclass loops from the form of inheritence.
21:34:57 <DanC> * DanC had to scratch his head the first time he saw one of the cyc guys say "just because they have the same members doesn't mean they're the same class"
21:35:07 <bijan> Er...
21:35:19 <larsbot> * larsbot has dandruff all over his desk from scratching
21:35:31 <dmiles> hehe
21:35:44 <sbp> heh
21:35:59 <sbp> no, :Bob a :ChessClub2001Member 
21:36:09 <bijan> Well, ok, I guess if you give "class" an intensional reading.
21:36:14 <sbp> cf. the iterating through lists stuff that DanC sent me
21:36:16 <bijan> i.e., as opposed to "set".
21:36:22 <larsbot> sbp: what's the 'a' here? rdf:type
21:36:24 <danbri> I could define a class whose defintion was 'People in [this house] at Fri Nov 16 16:34:47 EST 2001 who should be offline', and another 'People in [this house] at Fri Nov 16 16:34:47 EST 2001 who are logged into IRC'. Both classes have exactly the same membership (ie. me) but they mean differnt things.
21:36:25 <sbp> yes
21:36:27 <larsbot> bijan: that sounds right
21:36:37 <larsbot> sbp: in my world, that is crazy
21:36:48 <sbp> why would it be crazy?
21:36:51 <larsbot> sbp: ooooooops, I take that back
21:36:51 <danbri> I burned a timestamp into the class defintion in this case to avoid complications about their membership sometimes varying...
21:36:56 <dmiles> bijan: yeah a Class definition could be construed as a query that yeilds sets
21:37:13 <larsbot> sbp: you changed it. first the class was 2001ChessClub, then it became ChessClubMember
21:37:29 <DanC> danbri, talking about definitions doesn't really help, I don't think. What helps is: in RDFS, class identity isn't determined by membership. I can have a class whose members are {1,2,3} that is blue; that class is different from another class whose members are {1,2,3} which is red.
21:37:29 <sbp> no, I did not change it. Look at the statements carefully
21:38:01 <larsbot> I did, and I don't understand the semantics of your ( )
21:38:14 <sbp> () is a DAML list
21:38:17 <danbri> I don't get the red/blue thing. But I think I agree. I could probably have said 'meaning' or 'intension' instead of 'definition'.
21:38:35 <larsbot> in any case ChessClubMember as a class is also suspect, methinks
21:39:14 <sbp> Grpmgprmgh. O.K., Bob, Susan, and Gray are the only chess club members of 2001. They are also the only swimming club members
21:39:14 <larsbot> DanC: we got to this from subclass loops, which imply that all the classes in the loops have identical membership *by definition*
21:39:21 <sbp> s/Gray/Gary/
21:39:30 <danbri> I'd really like a better understanding of how containers (as a grouping construct) relate to classes. For modellers at least, the costs/benefits of each. Not to mention the interactions (subclassing Bags etc...).
21:39:43 <dmiles> a set could be a the result of a query against an insensional definiation of a class.. or a subset of some other pre-exesiting asserted set
21:39:53 <larsbot> that's fine
21:40:07 <danbri> careful: *by definition*? by which definition? of semantics of subClassOf, not the definitions of the classes to hand, right?
21:40:16 <larsbot> right
21:40:24 <JHendler_> I suspect we need to take turns - this time I really need you to do it, I'm still trying to write the "homework assignment" and get that out
21:40:30 <larsbot> but if subClassOf is part of the definition...
21:40:40 <bijan> (catching up) DanC: So classes have identity beyond their members?
21:40:52 <danbri> * danbri thinks JHendler_ is leaking out of of private /msg chat...
21:40:53 <DanC> yup, bijan. read the MT carefully.
21:40:55 <bijan> Or, for that matter, their definition?
21:40:57 <danbri> hi Jim :)
21:41:06 <DanC> properties have identity beyond their <subject,object> pairs too.
21:41:22 <larsbot> that makes very good sense
21:41:22 <bijan> Oh that's fine.
21:41:47 <bijan> I don't know if we're satisfying lars though :)
21:41:53 <DanC> yes, I agree it's fine; but it's different from set-theoretic relations.
21:42:00 <danbri> Yes. Thankfully. Since we're deploying it in a changing environment, it gives some stable grasp on reality...
21:42:06 <bijan> Yes.
21:42:11 <larsbot> DanC: yes
21:42:21 <bijan> Er.. To DanC. ??? to danbri :)
21:42:29 <dmiles> is subClassOf means that the instensional definition is subsummber by anothers classes definition right?
21:42:29 <DanC> wheee!
21:42:39 <danbri> doh. Thought I'd got a sanity check from bijan!
21:42:44 <larsbot> bijan: no, danc
21:42:57 <bijan> Huh?
21:42:57 <DanC> no, dmiles, subClassOf means that the extensions are subsets.
21:43:02 <larsbot> sorry, misunderstood
21:43:03 <DanC> er... oops.
21:43:05 <dmiles> does subClassOf mean that the instensional definition of the subclass is subsumed by anothers classes definition right?
21:43:15 <dmiles> sorry rewording
21:43:17 <bijan> Hmm. I thought subClassOf meant...subClassOf.
21:43:44 <danbri> I mean, we don't want want two say on tuesday that class1 and class2 are the selfsame thing, but then have some new state of the world on weds mean they 'become' different entities. That'd drive us crazy!
21:43:48 <DanC> subClassOf is defined formally by a couple rules; not much fuzziness to it.
21:43:51 <danbri> s/two/too/
21:43:59 <larsbot> bijan: no I am not satisfied. there seem to be two problems. 1) you use class in a different way from me (fine) 2) two classes can be in a subclass loop, and yet be different
21:44:09 <larsbot> danbri: sure, but if they are in a subclass loop that can never happen
21:44:44 <bijan> lars: Er... no I don't think *I* use class differently.
21:44:51 <DanC> a subclass loop defintely means the membership is the same. but no, it doesn't mean the classes are identical.
21:45:01 <larsbot> bijan: you don't know that, do you? :-)
21:45:07 <danbri> take my example above. Today, both classes have one member (me). Tommorrow, they'll possibly have different extensions. A subclass loop would ensure that they'd have to have the same different extension tommorrow, that's all.
21:45:17 <bijan> And 2) if classes have identity beyond membership, definiton, etc. then why not?
21:45:17 <dmiles> DanC: subClassOf axiomitized ?
21:45:35 <larsbot> danbri: exactly! so if they *always* have the same membership, why are they not the same?
21:45:47 <danbri> Think of the social properties of classes. They're created by people/orgnaisations, so have different properties such as s:createdByOrgWhoseHomePageIs...
21:45:53 <DanC> because, say, one is red and the other is blue, larsbot.
21:45:55 <dmiles> DanC: i mean is the definition of subClassOf axiomiztized somewhere?
21:46:05 <larsbot> DanC: red and blue can't be in a subclass loop
21:46:11 <DanC> yes, dmiles, in the MT. I think; lemme look it up...
21:46:23 <DanC> yes, a red class and a blue class can be in a subclass loop.
21:46:25 <bijan> Because identity of rdfs:classes is distinct from membership, even counterfactual membership.
21:46:31 <sbp> * sbp finds 4.2.1 in http://www.daml.org/2001/03/axiomatic-semantics.html but isn't sure that helps
21:46:40 <larsbot> DanC: by accident, perhaps, but not in a meaningful way
21:46:43 <bijan> class is more like type than like set.
21:47:05 <DanC> yes, in a meaningful way, larsbot. I'm not talking about a class whose members are blue, but a blue class.
21:47:10 <danbri> because they're not just sets. they pick out a set of individuals based on their meaning. The nature of the world means that the defintions of some classes will ensure they sometimes pick out the same set of individuals as another class.
21:47:20 <bijan> dan: no.
21:47:36 <DanC> * DanC suggests danbri doesn't get into "meaning" and "nature of the world".
21:47:37 <timbl-mit> * timbl-mit is susicious every time someone introduces time to make a point in the logic.
21:47:42 <timbl-mit> http://www.sfu.ca/counterfactual/
21:47:42 <dc_rdfig> G: http://www.sfu.ca/counterfactual/ from timbl-mit
21:47:44 <larsbot> DanC: I'm getting confused here. is the class the class of all things that are blue, or is it some class that just happens to be called blue, but means something completely different?
21:47:49 <bijan> it's not an intentionsal vs. extentional issue, if I understand thigns correctly.
21:47:50 <dajobe> (bijan: dan isn't a unique identifier here, can you use nicks?)
21:47:56 <bijan> ah!
21:47:59 <bijan> aiee!
21:48:03 <bijan> Yes. I keep doing that.
21:48:06 <bijan> danbri in the prior line.
21:48:29 <DanC> larsbot, I'm saying that classes have properties other than membership. for example, color. perhaps I picked a bad example.
21:48:38 <dmiles> hrrm I see that subClassOF exists in axioms that use it but is never the consequnet of any axioms
21:48:41 <larsbot> DanC: I think you did
21:48:59 <danbri> meaning/nature of the world.... Classes _mean_ something. We were making a big fuss about this earlier, right? Some things are members of each class, some aren't. And which things fall in a class is typically going to vary according to the state of the world.
21:49:01 <DanC> think of a class as a set of members plus a pile of other properties. properties OF THE CLASS, not of the members. 
21:49:09 <danbri> DanC: yup
21:49:30 <dmiles> * dmiles appends that it is still transitive
21:49:30 <DanC> yes, danbri, classes mean something, but it's not necessary to go into philosophy to make the point here.
21:49:40 <timbl-mit> You have a choice, I guess - to define that classs have no charactersitics except those which dtermine their membership, or otherwise.  If you make that determination then you get some nice simple axioms.
21:49:50 <larsbot> what I don't understand is how if A and B are in a subclass loop it does not follow that whatever definitions they have must be the same. how can the consequence only be that their membership must be equal for all time? if that is guaranteed, what difference can there be in their definitions?
21:50:07 <danbri> timbl, time is just the most commonly encountered way in which some of these issues become painful for implementors. nothing magic about time.
21:50:18 <danbri> * danbri was trying _not_ to go into philosophy. ouch.
21:50:34 <bijan> lars: 1) even if their definitions (in the sense of their intension) are they same, they can have *other* distinguishing properties
21:50:38 <DanC> larsbot, you've been thinking of sets for so long that you can't think of things any other way. The set extensionality axiom just DOES NOT APPLY to classes. it just doesn't.
21:50:44 <larsbot> DanC: that's fine, but surely being in a subclass loop must have consequences for the properties?
21:50:49 <DanC> nope.
21:50:54 <danbri> no.
21:51:07 <larsbot> DanC: never heard of the set extensionality axiom :)
21:51:10 <danbri> not of the class, only of its members
21:51:24 <bijan> 2) IIRC, there are modal logics where sameness of extention in every possible world still allows for difference of intension.
21:51:24 <larsbot> ok, you've said this many times now, so presumably it's true
21:51:30 <larsbot> but do you have any arguments for it?
21:51:37 <larsbot> I'd like to understand it, not just know it
21:51:58 <DanC> it's less constraining.
21:52:11 <larsbot> what is?
21:52:15 <DanC> you can fit set theory inside this sort of structure, but not the other way around.
21:52:25 <bijan> lars: do you mean arguments that it's the case or arguments for why it is the case?
21:52:37 <DanC> look at the cyc upper ontology. There's a subclass of Class called SetMathematical (or some such).
21:52:42 <larsbot> bijan: I don't know that there's a difference :)
21:52:54 <bijan> Oh. Really?
21:53:08 <sbp> Would it be O.K. to define a subClassOf relationship where in fact the classes in a loop would be identical in every way, not just in set membership?
21:53:17 <bijan> You don't know that there's a difference between showing you that the spec defines class the way we say
21:53:21 <sbp> er... i.e. as a sub property of rdfs:subClassOf
21:53:22 <dmiles> there is also a CompleteExtentKnown on classes
21:53:38 <bijan> and showing you that it's a *good and wise thing* (or at least justifiable) that it defines things that way?
21:53:43 <DanC> ugh... let's not get into closed-world-ickiness of cyc, dmiles.
21:53:49 <danbri> dc_rdfig:view
21:53:50 <dc_rdfig> C: Propositions, arguments and truth (http://www.arts.ed.ac.uk/philosophy/study_html/vade_mecum/sections/section2/2-1.htm)
21:53:50 <dc_rdfig> D: http://www.w3.org/2001/04/roadmap/org
21:53:51 <dc_rdfig> E: http://www.w3.org/TR/rdf-schema/#s2.3.1
21:53:52 <dc_rdfig> F: Metamodeling Architecture of Web Ontology Languages, by Jeff Z. Pan, Ian Horrocks (http://citeseer.nj.nec.com/450519.html)
21:53:53 <dc_rdfig> G: http://www.sfu.ca/counterfactual/
21:53:57 <larsbot> bijan: if that's what you mean I prefer the latter
21:54:19 <larsbot> DanC: so there is such a subclass; what does that mean?
21:54:49 <larsbot> that not all sets are classes?
21:54:54 <bijan> sean: OK in what way, for what purpose?
21:55:06 <bijan> er, sbp: ...
21:55:11 <DanC> cf http://www.cyc.com/cyc-2-1/vocab/math-vocab.html#Set-Mathematic
21:55:11 <DanC>           al
21:55:22 <DanC> cf http://www.cyc.com/cyc-2-1/vocab/math-vocab.html#Set-Mathematical
21:55:25 <sbp> I meant "O.K." as in it *would* be a sub property of the current rdfs:subClassOf
21:55:39 <hazmat> hazmat has joined #rdfig
21:55:43 <timbl-mit> If you go with the DanC system in which classes can differ but have the same members, then there is an equivalence relation which means having the same members, and its equivalnce classes are classes in the other sense.
21:55:54 <DanC> larsbot: if you know the members of two cyc:Set-Mathematical's are the same, you know they're identical.
21:56:12 <bijan> sbp: what would prevent it from so being?
21:56:17 <larsbot> DanC: true, but you don't know that of classes
21:56:31 <sbp> nothing that I know of, and hence the question. My knowledge is limited
21:56:33 <DanC> ?
21:56:39 <dmiles> i wish or would like to think that equivalence in classes should be based on there subsumption of each others definng axioms
21:56:51 <dmiles> there = the
21:57:32 <DanC> larsbot, you asked why rdfs:Class is the way it is... there's the technical answer: because the spec doesn't say otherwise; then you asked for rationale. I'm giving it to you: set theory fits inside this structure, but not vice versa. This structure is less constraining.
21:57:34 <dmiles> sets should be based on there memberships
21:57:42 <sbp> so :equivalentInSetMembership, timbl?
21:57:58 <larsbot> DanC: I don't understand the relevance. really I don't
21:57:59 <timbl-mit> yes.
21:58:00 <lasDesk> lasDesk has quit
21:58:29 <larsbot> DanC: are you saying that you chose to define the world this way so set theory would fit inside it?
21:58:48 <sbp> and daml:equivalentTo would be a subPropertyOf that weaker equivalence relationship... not vice versa...
21:58:51 <bijan> We're not defining the world, we're supplying a mathematical structure.
21:58:58 <sbp> Gotta run
21:59:09 <bijan> The mathematical structure in question is somewhat more general than set theory.
21:59:19 <larsbot> bijan: ok, so did you define the math struct this way so that...; same difference
21:59:33 <timbl-mit> * timbl-mit wishes larsbot would pls use "," when directing a remark to DanC rather than reporting a remark by DanC
21:59:38 <dmiles> classes euivalence is based on their set (derived closed set) meberships?
21:59:41 <DanC> oops... cyc:Set and cyc:Class have a common ancestor, but they're not related by rdfs:subClassOf.
21:59:42 <bijan> (And i'm being a bit loose about "mathematical structure" :))
22:00:04 <larsbot> timbl: use "," instead of what?
22:00:12 <larsbot> ah, instead of ":" ?
22:00:15 <DanC> * DanC notes, to tim, that IRC clients put the : in there. bummer, that.
22:00:30 <dmiles> timbl, class equivalence is based on thier set (derived closed set) memberships?
22:00:46 <timbl-mit> * timbl-mit oh. didn't know
22:01:07 <larsbot> danc, mine doesn't
22:01:08 <timbl-mit> You can define various sorts of equivalence. 
22:01:11 <sbp> sbp has quit
22:01:25 <larsbot> * larsbot wonders if he did it right this time
22:01:27 <timbl-mit> The cruch comes when you connect things to them.
22:01:45 <AaronSw> yep, larsbot
22:01:50 <DanC> larsbot, relevance to what? I'm not sure why the original decisions about rdfs:Class were made, but I'm telling you why I support them. They follow the principle of minimal constraint.
22:01:51 <larsbot> goodie
22:01:55 <timbl-mit> Like, if you say danc:setmembershipequivalence is the same as daml:equiavelntTo. 
22:02:40 <larsbot> so, danc supports the belief that the membership of a and b in a subclass loop does not imply that a and b are identical, because that is less constraining?
22:02:49 <tav> ehm
22:02:57 <DanC> * DanC wishes timbl would stipulate that one sort of equivalence, identity, is different from all the others.
22:02:58 <timbl-mit> I don't know what "identical" means.  
22:02:59 <bijan> lars, look, these are stipulative definitions.
22:03:16 <dmiles> timbl, i see.. since some ingformation sources have a context of closed world and others don,t
22:03:21 <bijan> DanC is saying nothing about sub classes in general.
22:03:22 <tav> where you guys sending death threats to the president using xena?
22:03:38 <timbl-mit> * timbl-mit does stipulate that one form of identity, URI equality, is different from all the others. This is log:equalTo.
22:03:42 <AaronSw> No, of course not. We wouldn't send them to the President.
22:03:49 <DanC> no, larsbot, I believe that same-membership doesn't imply identical because only those inferences explicitly licensed by the spec are valid.
22:03:50 <timbl-mit> One  form of identity, URI equality, is different from all the others. This is log:equalTo.
22:04:24 <larsbot> danc, ok, I'm arguing more "what should the spec be" than "what does the spec imply"
22:04:31 <DanC> log:equalTo? that's new to me.
22:04:42 <larsbot> danc, actually, I'm asking to have this explained to me in general, independent of RDF
22:04:52 <larsbot> this is probably the wrong place for that, though
22:05:05 <danbri> Let's not have a dozen species of identity. We have is-one-and-the-same-thing-as, or we don't. Not a good place for fuzziness.
22:05:07 <DanC> independent of RDF? that's quite a stretch.
22:05:16 <larsbot> how so?
22:05:26 <danbri> Woolier notions of sameness can live in thesaurus land, 'thes:synonym' etc.
22:05:33 <bijan> Actually, there should be loads of identity :)
22:05:45 <DanC> * DanC wonders what 'so' in "how so?" referred to
22:05:58 <larsbot> to "quite a stretch"
22:06:12 <bijan> http://world.std.com/~pitman/PS/EQUAL.html
22:06:12 <dc_rdfig> H: http://world.std.com/~pitman/PS/EQUAL.html from bijan
22:06:21 <bijan> H:"The Best of Intentions"
22:06:21 <dc_rdfig> commented item H
22:06:26 <larsbot> do you mean that certain assumptions must be in place for this discussion to be meaningful at all, and so discussing it outside RDF is difficult?
22:06:28 <JHendler_> * JHendler_ leave the fuzziness to webont?  [we will need some extension to daml:equivalentTo, or at least some clearer explanation of when not to use it and a suggestion of what to do instead]
22:06:30 <DanC> stretch: well, I don't think the design of RDFS justifies itself outside the context of RDF. I don't think it's some sort of universal truth.
22:06:39 <bijan> H:|Kent Pitman on equality predicates in common lisp.
22:06:40 <dc_rdfig> titled item H
22:07:06 <larsbot> stretch: well, I was asking about subclass loops more than about RDFS
22:07:32 <larsbot> in my naive conception of things there was one truth about subclass loops, and RDFS was forced to reflect it
22:07:36 <bijan> But you keep reading "subclass" as "lars:subclass" :)
22:07:45 <larsbot> who is you?
22:07:50 <bijan> Larsbot :0
22:07:54 <DanC> tim, what the heck does "Equality in this sense is actually the same URI." mean?
22:08:01 <bijan> Actually, that's a tricky question.
22:08:14 <larsbot> so I keep using my own definition of subclass, is that you're saying=
22:08:15 <larsbot> ?
22:08:18 <bijan> Since *I'm* you, in your sentence.
22:08:21 <DanC> one truth about subclass loops: I don't live in that world.
22:08:21 <bijan> Yes.
22:08:34 <larsbot> danc, that's fair enough; maybe I don't either
22:08:57 <bijan> larsbot, rather, afaict, you keep thinking that "subclass" implies a bunch of strong inheritence facts.
22:09:11 <bijan> And in many systems it doesn't.
22:09:14 <larsbot> well, if it doesn't, disabuse me of that notion
22:09:50 <larsbot> someone said circularity of subclass loops implied identicality of extension, but not of intension
22:09:56 <larsbot> that's where I got confused
22:10:05 <DanC> tim? in the log.n3 schema, you wrote that log:equalTo is "A cwm built-in logical operator." I can't find the code. help?
22:10:14 <bijan> That would be true *in a system* where classes inheritent members but not definitions.
22:10:16 <danbri> thats a fair summary. 
22:10:24 <DanC> ahh... found it. in llyn.py
22:10:48 <larsbot> can you rephrase "where classes inheritent members"?
22:11:04 <bijan> "where classes inherit extention"
22:11:14 <DanC> why do you say "Do not confuse with daml:EquivalentTo."? is log:equalTo defined in terms of cwm's implementation?
22:11:23 <larsbot> bijan: aha
22:11:31 <dmiles> "someone said circularity of subclass loops implied identicality of extension, but not of intension" i am interested in where this came from.. it seems wrong
22:11:44 <larsbot> ask logger :)
22:11:52 <bijan> "In RDFS, circularity of..."
22:11:57 <DanC> I think that was me, dmiles, and yes, it's wrong. I wrote "oops" a few lines later.
22:12:17 <dmiles> oh.. just wondering
22:12:49 <DanC> * DanC was hoping to get to the bottom of timbl's ideas about equivalence and identity, but seems to have lost his attention
22:13:08 <timbl-mit> log:equalTo is true iff and only iff the two things have the same URI.
22:13:20 <bijan> Ouch! Really!
22:13:23 <timbl-mit> * timbl-mit had got distracted.
22:13:28 <DanC> er... do you and I have the same URI, Tim?
22:13:38 <timbl-mit> no
22:13:38 <larsbot> timbl: so if person A and person B define different tag URIs for me, they can never be joined up?
22:13:45 <DanC> I'm pretty sure I can make you a liar whatever answer you give, tim.
22:13:50 <bijan> If they have *some name* the same?
22:13:53 <timbl-mit> Not with equalTo.
22:14:01 <bijan> If they have the *same name* they are the same!
22:14:10 <dmiles> Like, if you say danc:setmembershipequivalence is the same as daml:equiavelntTo.
22:14:19 <dmiles> then they could be right?
22:14:25 <larsbot> * larsbot -> laundry
22:14:49 <DanC> hmm... I chose a bad example...
22:14:50 <dmiles> meaning if you asserted some connection between two URIs
22:15:13 <dmiles> you could make then work any way you thought they needed to 
22:15:22 <timbl-mit> this log:forAll  :x.  { :x log:equalTo <foo>  .... }    will bind :x to <foo>
22:15:27 <dmiles> there must be support for that right?
22:15:29 <DanC> do is [ contact:mailbox <mailto:connolly@w3.org> ] log:equalTo [contact:homePage <http://www.w3.org/People/Connolly/>] or not?
22:15:36 <timbl-mit> No.
22:15:36 <bijan> Ok, it has to be in a context. Ok, URI's in a document are logical variables (or something like that). In that context they have to have the same denotation...
22:15:50 <danbri> We say that if Dog is a subclass of Mammal, all resources that are of rdf:type Dog are also of rdf:type Mammal. That's all we say. We don't say: "all things true of the class Mammal are also true of the specialised subclass Dog', where the classes Mammal and Dog are resources in their own right.
22:15:54 <timbl-mit> You need daml:equiavalentTo.
22:16:07 <timbl-mit> This is what cwm shothands to "="
22:16:08 <danbri> (that was to larsbot btw)
22:16:28 <danbri> "log:equalTo is true iff and only iff the two things have the same URI."
22:16:37 <danbri> * danbri falls off his chair
22:16:39 <DanC> so log:equalTo means "somebody used the same URI to tell cwm about this thing in this session"?
22:16:45 <danbri> Is that the only notion of identity cwm undestands?
22:16:55 <bijan> danbri: read my double take :)
22:17:08 <timbl-mit> If you don't like log:equalTo don't use it!!
22:17:17 <bijan> DanC: in a specific context?
22:17:22 <DanC> i.e. tim, you realize log:equalTo depends on the order files are read, cwm internals, that sort of thing?
22:17:26 <danbri> I guess if cwm's happy dispensing its own URIs for things, that's fine. But it waters down the notion of 'being the URI of something'
22:18:18 <timbl-mit> log:equalTo is what you need if you are going to subsitute in a log:forAll.
22:18:41 <timbl-mit> If two things are daml:equaivalent, that doesn't mean that "forAll x" implies "forAll y"!
22:18:48 <DanC> ah... so log:equalTo is special syntax. it's not a relation between things in the real world. whew!
22:18:55 <timbl-mit> yes
22:19:23 <DanC> * DanC scratches head for a bit
22:19:31 <bijan> There must be a better name :)
22:19:32 <DanC> why is it there? what's it for?
22:20:01 <DanC> I guess you answered that already...
22:20:04 <DanC> are there test cases?
22:20:06 <timbl-mit> hacks. See sameThing.n3
22:20:11 <danbri> log:cwmInternalEqualTo ?
22:20:30 <bijan> No :)
22:21:06 <DanC> I see sameThing.n3; it helps me not at all with my confusion.
22:21:17 <DanC> oh... log:notEqualTo
22:21:23 <timbl-mit> hack
22:21:42 <timbl-mit> The query system should not see forAll or forSome ... but at the meoment it does
22:21:46 <danbri> http://www.perl.com/pub/a/2001/11/15/creatingrss.html
22:21:46 <dc_rdfig> I: http://www.perl.com/pub/a/2001/11/15/creatingrss.html from danbri
22:21:48 <timbl-mit> so that hack is necessary.
22:21:57 <DanC> [I think that's a misuse of "hack". I think you mean "kludge". I do that myself]
22:22:09 <timbl-mit> Sorry, kludge.
22:22:21 <dmiles> what is a hack?
22:22:35 <DanC> a hack is a good thing. a nifty piece of code, for example.
22:23:06 <DanC> cf http://www.tuxedo.org/~esr/jargon/html/entry/hack.html "n. Originally, a quick job that produces what is needed, but not well."
22:23:07 <dmiles> cool :) .. I am going to start using them that way in comments
22:23:28 <DanC> 1. n. Originally, a quick job that produces what is needed, but not well. 2. n.
22:23:28 <DanC> An incredibly good, and perhaps very time-consuming, piece of work that produces exactly
22:23:28 <DanC> what is needed. 
22:23:40 <danbri> I:|Create RSS channels from HTML news sites, by Chris Ball (November 15, 2001)
22:23:41 <dc_rdfig> titled item I
22:23:44 <bijan> Hmm. It seems right in the first sense.
22:24:04 <bijan> I often use "dirty filthy hack" :)
22:24:19 <DanC> words take on meaning by use. anyway...
22:24:30 <bijan> Plus, that's the *hackers* definition :)
22:24:37 <bijan> Or connotation.
22:24:43 <DanC> where was I? ah yes... hourglass diagrams.
22:24:59 <JHendler_> * JHendler_ notes that University of Maryland wouldn't let me call my class "Semantic Web Hacking" - they thought it had wrong connotation
22:25:22 <danbri> I:More semantic web screenscraping, though Perl rather than the [XSLT flavour|http://www.w3.org/2000/08/w3c-synd/].
22:25:22 <dc_rdfig> commented item I
22:25:32 <bijan> The Snooty Intellectual or Corporate Snob uses are rather...different :)
22:26:32 <danbri> hack: the trouble with things like log:equalTo etc is I never no if it's a cute hack to make Cwm perform handy tricks, or a proposal for some new web architectural design principle... (ie. a non-cute hack :)
22:26:42 <JHendler_> danbri, stay tuned - a student of mine has turned the "Running SHOE" web scraper into "Running RDF" - he is cleaning up code in next couple of days and we will chump...
22:26:59 <danbri> * danbri stays tuned!
22:27:20 <danbri> Any chance of an XML interface to the old Parka KB?
22:27:36 <timbl-mit> forSome, forAll, equalTo:  kludges.    
22:28:14 <danbri> Last time I tried the binpiq applet (last year I think) the KB was still up and answering queries. It'd be fun to hook it up to the RDFAuthor rdf query client...
22:28:36 <JHendler_> Working on some issues there - yes in terms of working (he's working on that now), but licensing is a mess - we may decide to redo the whole thing to get it back from the people university "sold" it to
22:29:07 <JHendler_> we do have the SHOE search servlet up and running, which implies Parka KB is up - I'll check with Jeff.
22:29:41 <JHendler_> Parka-DB looks like a great back end to webont, so I intend to get it freed somehow.
22:29:56 <DanC> * DanC goes back to diagram for sandro...
22:31:53 <sbp> sbp has joined #rdfig
22:31:58 <danbri> If you've a KB that could answer RDFish queries, I can point you at the SOAP bits and pieces that might be handy for exposing XML Protocol interfaces...
22:32:16 <danbri> (though that's not an entirely fashionable approach in these parts!)
22:33:32 <JHendler_> Now that I have students again, I can pull my own weight in the SW-AD world, starting to get things moving here - SW meets WS something I definitely want to play with, so we're going to have to get SOAP, WSDL etc. literate...
22:34:30 <danbri> The tools are fairly easy to get to grips with, Perl/Python/Java etc. Ping me whenever it'd be helpful to take a look at what's available...
22:35:26 <sbp> * sbp was wrong about "daml:equivalentTo would be a subPropertyOf that weaker equivalence relationship" - equivalentInSetMembership would be a superPropertyOf sameClassAs, of course
22:37:34 <JHendler_> * JHendler_ has to go, Shabbat Shalom all.
22:37:36 <JHendler_> JHendler_ has left #rdfig
22:38:33 <sandro> sandro has quit
22:38:35 <ericP> ericP has quit
22:38:47 <timbl-mit> timbl-mit has quit
22:39:44 <timbl> timbl has quit
22:40:11 <DanC> sean, if you have some time to expand those thoughts to about 3 pages and some demo code, you'd be doing me a big favor. that's been on my todo list forever. (sameclassAs vs. equivalentTo vs. subPropertyOf etc.)
22:40:29 <danbri> * danbri heads off in search of evening remnants; 'nite all
22:41:16 <sbp> I'll certainly research it if I find some time
22:41:47 <danbri> danbri has quit
22:49:01 <sandro> sandro has joined #rdfig
22:53:11 <sbp> sbp has quit
22:58:49 <dajobe> dajobe has quit
23:16:56 <sbp> sbp has joined #rdfig
23:20:08 <sbp> sbp has quit
23:32:41 <sbp> sbp has joined #rdfig
23:33:01 <timbl> timbl has joined #rdfig
23:38:56 <sbp> sbp has quit
23:39:44 <hazmat> hazmat has quit
23:39:46 <kifbot> kifbot has quit
23:41:20 <grove_> grove_ has quit
23:42:45 <kifbot> kifbot has joined #rdfig
23:43:10 <grove_> grove_ has joined #rdfig