Semantic Web Interest Group IRC Chat Logs for 2002-07-31

00:00:10 <danbri> "Hope this helps."

00:00:16 <danbri> hey Guha

00:01:36 * danbri wonder if guha has a solution to the neverending story of rdf datatypes...

00:14:57 * DanC uses the so-called "local" idiom happily with the cyc ontology, btw.

00:27:33 <Guha_> sent out a tongue-in-cheek proposal to rdfcore.

00:29:51 <Guha_> danc ... what is the "local idiom"?

00:54:22 <DanC> ping? am I back?

00:54:52 <DanC> the local idiom is, e.g. someEvent --startingDate-->someDay --xsdt:date--> "2002-10-23"

00:55:15 <DanC> in N3: :someEvent :startingDate [ xsdt:date "2002-10-23" ].

00:55:34 <DanC> i.e. datatypes are properties that map from values to lexical forms.

01:16:00 <Guha_> ah. I see. Not that different from the tongue-in-cheek proposal I sent out ... except, I'd prefer someEvent --startingDate-->someDay -->[2002-10-23 the date]. Otherwise we have way too many b-nodes. This also maps nicely into apis. Need to think of how to do that without requiring rdf syntax doc to list every possible data type.

01:17:37 <Guha_> Bad cut and paste. I meant: someEvent ---startingDate-->[2002-10-23 the date].

02:30:21 <SethR> i came here from a sailor ... sorry, bye

02:31:16 <Morbus> uhh.

02:31:54 <Seth> quick question - a urn is never dereferenced, is it?

02:54:01 <danbri> <Seth> quick question - a urn is never dereferenced, is it?

02:54:08 <danbri> boggle. Why ever not?

02:54:29 <danbri> things named by urns are harder to find representions of. that's all.

03:02:28 <_10bea_freezope> Hi.

03:04:03 <_10bea_freezope> Am geezer-geek. Know about xml. New to and wanting to RSS and such.

03:04:43 <_10bea_freezope> Anyone awake?

03:10:25 <DanC> sorta

03:11:07 <_10bea_freezope> Place for rdf/rss questions/discussions?

03:11:17 <DanC> Guha, [2002-10-23 the date] is pretty much the same as [ xsdt:date "2002-10-23" ]. Yes, functional terms might be nicer than bNodes, eventually.

03:11:25 <DanC> yup, rdf/rss questions welcomed...

03:11:28 <DanC> ... sometimes even answered!

03:12:19 <_10bea_freezope> Great. I program off and on since Fortran in 1966. got some domains registered. Know about xml and write raw html.

03:13:04 <_10bea_freezope> Want to make rdf feeds for various of my pages various places to eat from time to time. Any help?

03:14:10 <_10bea_freezope> Looking for a deeper reference on rdf/rss than the w3c tutorial. Like a K&R for rdf/rss .

03:15:19 <_10bea_freezope> DanC, you still there?

03:27:25 <_10bea_freezope> zzz

04:06:05 <MB_away> MB_away is now known as MarkB

04:06:31 <MarkB> fwiw, I use http://logicerror.com/blogifyYourPage for most of my HTML->RSS needs

04:32:28 <_10bea_freezope> _10bea_freezope is now known as _10bea_freezope_

04:33:11 <_10bea_freezope_> _10bea_freezope_ is now known as _10bea_fz_away

08:11:26 <burtonator> Do any tools actually use RDDL?

10:40:32 <burtonator> danbri: back to our earier RDF conversation. I was thinking about a compromise on the RDF format. What about a nested triple mechanism?

10:41:27 <burtonator> I mean is there any reason:

10:41:32 <burtonator> <item rdf:about="http://www.kuro5hin.org/?op=displaystory;sid=2002/7/30/15328/1925">

10:41:34 <burtonator> <image:item rdf:about="http://www.kuro5hin.org/images/topics/culture.jpg">

10:41:34 <burtonator> <dc:title>Culture</dc:title>

10:41:34 <burtonator> </image:item>

10:41:34 <burtonator> </item>

10:41:37 <burtonator> couldn't be valid RDF?

10:41:55 <dajobe> because it doesn't match the RDF model

10:42:17 <danbri> that isn't, as dave says, valid rd/xml syntax. So the model you have in mind is unclear

10:42:31 <burtonator> I realize it isn't valid RDF

10:42:38 <dajobe> I said model, not syntax

10:42:50 <burtonator> oh..

10:42:52 <burtonator> sorry...

10:42:53 <dajobe> I want to see what model yo uwant, not XML syntax

10:43:07 <danbri> dajobe, we don't know if it matches the RDF model yet, cos we don't know what syntax he's using

10:43:13 <dajobe> indeed

10:43:26 <burtonator> ok... let me create two quick examples... hold on.

10:43:41 <danbri> there's an item, which has a uri and a title. also some sort of relationship to an image.

10:43:43 <dajobe> but nested triples are not in the (current) RDF model, whatever syntax is for it

10:45:31 <burtonator> two secs... I will be quick since it is late.

10:46:23 <danbri> <item rdf:about="http://www.kuro5hin.org/?op=displaystory;sid=2002/7/30/15328/1925"><x:category_logo rdf:resource="http://www.kuro5hin.org/images/topics/culture.jpg"/><dc:title>Culture</dc:title></item>

10:46:38 <danbri> ...is how i'd RDF/RSSize what I think your data amounts to

10:46:52 <danbri> ie a normal item, plus a logo associated with a topic that the item falls under

10:53:37 * danbri thinks burtonator fell asleep

10:54:20 <burtonator> no... here.

10:54:26 <burtonator> (sorry... stupid bug)

10:54:47 <burtonator> the problem with this is that you have to have a triple below to specify any further information.

10:54:56 <burtonator> dc:title won't be included within the RDF graph

11:00:16 <burtonator>http://www.peerfear.org/download/example.html

11:00:17 <dc_rdfig> A: http://www.peerfear.org/download/example.html from burtonator

11:00:27 <burtonator> A: Example of a tested triple model for RDF???

11:00:27 <dc_rdfig> added comment A1

11:01:06 <danbri> missing an end > for: <image:item rdf:about="http://www.kuro5hin.org/images/topics/culture.jpg"

11:01:16 <burtonator> so instead of a <image:item rdf:resource="blah"/> and then another rdf:about triple under it...

11:01:44 <burtonator> oh... didn't try to parse the second one.

11:03:14 <dajobe> burtonator: can you please title the chump things with A:|title ?

11:03:20 <danbri> the second looks close to parse-able RDF

11:03:47 <danbri> ...the first looks tangled cos its making use of rdf:Bag, presuambly because of the rule in RSS's subset of RDF about repetition of elements?

11:03:59 <burtonator> A:|A simple alternate way to organize triples

11:04:00 <dc_rdfig> titled item A

11:04:32 <burtonator> danbri: the second is not valid RDF... at least from what I read and using IsaViz to verify.

11:05:28 <danbri> it still isn't clear what triples you're trying to generate

11:06:09 <danbri> the use of rdf:Bag in the first means the article is linked to a bag that is linked by _1 to the image item

11:06:20 <danbri> it doesn't appear that 2nd eg is an attempt at generating those triples

11:06:27 <danbri> so you're sorta comparing apples and oranges

11:07:06 <burtonator> I want to specify multiple images with different widths.

11:07:15 <burtonator> I can only really do that with rdf:Alt

11:07:46 <burtonator> I just don't see why I can't place the triples right under rss:item

11:24:45 <danbri> are the images per article, or per category-of-articles

11:24:46 <danbri> ?

11:25:12 * danbri not sure if we're tryign to solve a specific problem you have, or fix RSS's RDF subset, or alter the RDF/XML syntax

11:25:28 <danbri> you can specify multiple images with different widths easily in RDF

11:25:45 <danbri> trickyness is the constraint about not allowing multiple subelements in RSS 1.0

11:26:47 <JibberJim> could you not just seeAlso the image item in RSS 1.0 to a different RDF doc which listed the different images - then it would be okay RSS 1.0 for dumb clients, and clever clients could get the seeAlso'd doc.

11:46:53 <pixel-away> pixel-away is now known as pixel

11:47:00 <pixel> pixel is now known as identify

11:59:54 <identify> identify is now known as pixel

12:31:49 <pixel> pixel is now known as pixel-meeting

13:35:14 <danbri> <ex:Example rdf:ID="foo3">

13:35:14 <danbri> <ex:property rdf:parseType="xsd:decimal">10</ex:property>

13:35:14 <danbri> </ex:Example>

13:35:23 <danbri> this has come up a few times lately...

13:35:36 <danbri> as a proposal for rdf datatyping syntax

13:35:42 <danbri> its sort of growing on me...

13:42:46 * dajobe groans

13:42:55 <dajobe> I'd prefer the xsi:type attr, myself

13:45:03 <DanC> xsi:type... the details look really, really hairy to me. Though the general idea is kinda interesting...

13:45:38 <DanC> details, e.g.: is <dc:date xsi:type="my:dateThingy">yesterday</dc:date> OK?

13:46:02 <danbri> xsi:type will barf existing parsers...

13:46:04 <dajobe> I did propose making all literals have a type (URI, not xsd qname) months back, WG didn't seem to want it although Jan and I liked it. Nevermind

13:46:05 <danbri> won't it?

13:46:07 <dajobe> barf: yes

13:46:30 <dajobe> as well lots of burton's stuff

13:46:30 <danbri> Netscape, Adobe etc might be unhappy (though their parsers have issues).

13:46:42 <dajobe> tre

13:46:44 <dajobe> true

13:46:56 <DanC> er... what are NS/Adobe gonna say about rdf:nodeID?

13:47:02 <danbri> why do you prefer xsi:type over rdf:parseType? just the choice of the word?

13:47:18 <dajobe> they are using a subset of the existing rdf/xml, and that'll still wrork

13:47:21 <danbri> I don't know. Hope to find out soon.

13:47:36 <dajobe> danbri: keep me updated pls?

13:47:51 <danbri> oh, I meant: 'cos we'll all find out soon.

13:48:04 <DanC> xsi:type harks to stuff like SOAP encoding. If we're not gonna do what the XML crowd does, our existing syntax is not bad at all: <ex:property xsd:decimal="10"/>

13:48:07 <danbri> Everyone in #rdfig seemed very positive about it

13:48:12 <dajobe> preference - well, unless rdf:parseType is tidied up re what is in the string? qname, uri, set of terms - it's a mess

13:48:32 <danbri> it is pretty messy currently, agreed.

13:48:41 <dajobe> positive - what?

13:48:49 <danbri> about rdf:nodeID

13:49:12 * danbri rebuilds his visor one package at a time

13:49:13 <dajobe> yeah

13:49:28 <dajobe> eikons said it was easy to impl, has done it and is happy with it.

13:49:32 <danbri> (it just wouldn't sync via USB, so I blanked it and restarting...)

13:49:46 <danbri> whee! I spoke to him fri/sat and he was going to do it. That's fast feedback...

13:49:49 <dajobe> nice xml-dev thread on data mining semweb...

13:51:51 <dajobe> rather I meant eikeon and his python http://rdflib.net/

13:54:09 <danbri> what's the current state of art for getting calendar data into palms from Linux? DanC's python script?

13:54:17 <libby> bah, I spent like a *day* trying to find something like this, and now I find it first hit on google: http://www.xngr.org/cmanager/downloads.htm

13:54:28 <danbri> what is it?

13:54:31 <libby> It looks like you can sync with xml files! in java!

13:54:45 <libby> it's a java conduit manager but also some xml conduits :)))

13:55:23 <danbri> it says databook: not ready yet...

13:55:29 <libby> so, I wanted something that kate and the rest of us could use to sync swad-e events with palms, that would work ona ll platforms ... looks like this will do it, though aint tried yet

13:55:32 <danbri> date, rather

13:55:50 <libby> serviocve not ready yet, xip is there

13:56:13 <libby> I guess danc's stuff is the stuff you want...but I'm gonna have a go with this too

13:56:17 <dajobe> I thought jpilot + plugins looked encouraging (at least for linux)

13:56:41 <libby> not looked at it for a bit. I want something cross platform cos really it's most important for kate

13:57:08 <libby> maybe I shoudl chump it. dunno if any good though

13:57:21 <libby>http://www.xngr.org/cmanager/downloads.htm

13:57:21 <dc_rdfig> B: http://www.xngr.org/cmanager/downloads.htm from libby

13:57:45 <libby> B:|C manager - Java conduit manager

13:57:46 <dc_rdfig> titled item B

13:57:59 <libby> B: plus some thuing for syncing XML with 4 mian palm apps

13:57:59 <dc_rdfig> added comment B1

13:58:10 <DanC> B:platforms?

13:58:10 <dc_rdfig> added comment B2

13:58:27 <libby> B:not tried it yet but *just* what I've been looking for for syncing deliverables cross-platform

13:58:27 <dc_rdfig> added comment B3

13:59:03 * libby gets over-excited about datebooks ;)

13:59:08 <DanC> B:seems to rely on Palm's stuff

13:59:08 <dc_rdfig> added comment B4

13:59:12 <libby> ...but it's gonna be great!

13:59:30 <libby> B:dammit: this? Palm JSync Suite

13:59:31 <dc_rdfig> added comment B5

13:59:50 <DanC> danbri, I can (still) recommend my palmagent stuff. I'm reasonably sure it works for getting calendar stuff into your palm/visor

14:00:03 <libby> bah. It'll still do for thr windows crowd. linux types can fend for themsleves

14:00:54 <libby> I see your stuff has been taken forward by Norman? walsh at extreme, Danc

14:01:00 <DanC> I'm more jazzed than ever about open-source. The OS/X Jaguar announcement had me all excited until I saw the price tag. I'm spoiled by open source.

14:01:17 <DanC> Norm's stuff is independently developed, if you can believe it!

14:01:30 <libby> inspired by...right?

14:01:46 <DanC> well, he was exposed to N3 via the circles-and-arrows stuff, but he didn't realize N3 had its genesis in my desire for wiki RDF on a palm

14:02:14 <libby> heh

14:02:20 <DanC> (the N3 implementation, that is; the genesis of the design is TimBL's need to type RDF in IRC/email)

14:02:52 <DanC> are you gonna be at extreme, libby?

14:03:17 <libby> nah. wish I'd thought about it now, but I didnt

14:03:30 <DanC> sigh...

14:03:32 <libby> it's rather topicmaps-y isn;t it

14:03:40 <libby> 'whither topicmaps' and so on

14:03:48 <dajobe> you're in Bristol later this year, DanC right?

14:04:03 <DanC> good lord willin and the creek don't rise, yeah, dajobe

14:04:07 <danbri> DanC, "It's designed to only read your data, but who knows".... can I get data _into_ the palm as well?

14:04:42 <DanC> yes, danbri; less well-documented... but palmagent is an HTTP server that accepts RDF POSTed to your .pdb file

14:04:53 <libby> ooh

14:04:54 * danbri can't stop calling it a palm; visor has too many syllables :)

14:05:01 <libby> didn;t realize that was how it worked

14:05:33 <DanC> are you in a position to test it now, danbri, if I walk you thru it?

14:05:36 <danbri> I'm somewhat risk averse right now, having got the device so confused that I'm starting over. But I do want a mechanical way to add dates.

14:06:05 <DanC> palmagent never actually screws with the device itself. It only writes to files on the host.

14:06:05 <danbri> yes, I could try it.

14:06:35 <danbri> It writes pdb files? Are there any syntax/format checking, validating etc tools available?

14:06:43 <libby> I discovered recently that windows doesnt store pdb files but undocumented, liable to change dba :(

14:06:56 <DanC> yes, writes pdb files. I suppose you could use the palm emulator as a validator.

14:07:14 <danbri> I've always been to chicken to screw with those. Or I guess you're going via the (perl? python?) apis, so that should increase the safety...

14:07:18 <DanC> yeah, the palm desktop tools use MS structured storage. phtphpht.

14:07:32 <danbri> palmagent does so many things I get confused, but that's good :)

14:07:33 <libby> that _so_ sucks

14:08:08 <DanC> $ perl pdkb.pl

14:08:09 <DanC> Can't locate Palm/Datebook.pm in @INC (@INC contains: /usr/local/lib/perl/5.6.1 /usr/local/share/perl/5.6.1 /usr/lib/perl5 /usr/share/perl5 /usr/lib/perl/5.6.1 /usr/share/perl/5.6.1 /usr/local/lib/site_perl .) at pdkb.pl line 34.

14:08:09 <DanC> BEGIN failed--compilation aborted at pdkb.pl line 34.

14:08:21 * DanC starts apt-getting...

14:08:32 <danbri> DanC, here's my goal: to be able to (a) add stuff (b) update all the event descriptiosn that came from some given source.

14:09:10 <danbri> ie. I'm worried about programmatically adding events (eg. all the scraped swad-europe deliverabel dates) and not having any way but by hand to remove just those ones, eg. if they're regenerated, updated...

14:09:19 <DanC> umm... yeah, you can update. with an HTML forms interface, if not POSTed RDF.

14:09:45 <DanC> # apt-get install libpalm-perl

14:09:55 <DanC> $ perl pdkb.pl

14:09:55 <DanC> @@ convert

14:09:55 <DanC> Can't open "": No such file or directory

14:09:59 <danbri> Is there some way one could tag all the events from a certain source...? maybe with a palm category or note field.

14:10:04 <danbri> OK, I'll give it a whirl.

14:10:14 <danbri> Shall I start with the online docs and see how far I get?

14:10:15 * DanC slaps himself for crappy diagnostics

14:10:22 * danbri has debian envy

14:10:25 <DanC> you can watch what I do and try it

14:10:33 <danbri> OK

14:10:45 * DanC UTSLs...

14:10:47 <DanC> # usage:

14:10:47 <DanC> # perl pdkb.pl --serve DATEBOOK

14:10:48 <danbri> I'm here, http://dev.w3.org/cvsweb/2001/palmagent/ reading Installation.

14:11:36 <DanC> hmm... now where do I get a DATEBOOK file to start with... ah... lemme start the emulator...

14:12:59 <DanC> what's it called? ah.. pose

14:14:32 * danbri backs up his visor

14:14:35 <DanC> crap; I need a ROM image, but all my pilots are busted

14:15:38 <danbri> crap; the connection between your handheld computer and the desktop was lost. Some of your data was NOT backed up. Please check your setup and try again.

14:15:46 * DanC finds an old pdb file...

14:15:58 <DanC> are you using windows, danbri?

14:16:07 <danbri> linux/suse

14:16:08 * DanC recommends coldsync under linux

14:16:18 <danbri> Instead of pilot-xfer?

14:16:24 * danbri has sense of a recovered memory

14:16:27 <DanC> $ perl pdkb.pl --serve foo.pdb

14:16:27 <DanC> Record 1269 has same offset as previous one: 155618

14:16:27 <DanC> Record 1270 has same offset as previous one: 155618

14:16:27 <DanC> Serving palm datebook <foo.pdb> at: <http://dirk:35484/>

14:16:36 <danbri> I got all this working on my boston box (debian)

14:16:40 <danbri> ...maybe that was how

14:18:01 <DanC> can you try that? $ perl pdkb.pl --serve foo.pdb

14:18:31 <danbri> Serving palm datebook <DatebookDB.pdb> at: <http://fireball:52146/>

14:18:40 <DanC> ok, you're winning.

14:18:49 * danbri builds coldsync

14:18:58 <DanC> you can now browse around your datebook.

14:19:54 <danbri> workie! thanks :)

14:19:59 <DanC> hmm... dbrec-test.rdf is an example of an RDF file that you can POST.

14:20:03 <danbri> (with my old .pdb file)

14:20:04 <DanC> I think.

14:20:54 <danbri> I do still find your rdf/xml indenting style near impossible to read... but these things are a matter of taste

14:21:23 * DanC recommends tidy, or even cwm, if you want human-readable.

14:21:51 * DanC concentrates on data-integrity. no adding whitespace where the originator didn't ask for it.

14:21:56 <danbri> xmllint maybe too

14:24:24 <danbri> backing up now, w/ coldsync. minor scary warning...: "SLP: bad CRC: expected 0x1c80, got 0xbeef. "

14:24:35 <danbri> ...and it failed. with same error. Time to wiggle wires, I think.

14:24:44 <danbri> same = same as with pilot-xfer

14:25:03 <DanC> er... didn't you already blank it?

14:25:47 <DanC> some coldsync CRC diagnostics can be safely ignored; it'll re-try. But if it craps out, yes, there's a problem.

14:26:29 <danbri> I put some more stuff back on there, bit by bit. If I can't get things so a full backup/restore works over USB, the whole thing's doomed :(

14:27:02 <dajobe> ah USB

14:27:14 <dajobe> I had problems with that, had to hand-compile jpilot and right library

14:27:28 * DanC would expect USB to be more reliable, not less. v. suprised by that.

14:27:45 <danbri> It's a while since this thing had its kernel recompiled, maybe usbserial, visor etc modules are better now?

14:27:57 <dajobe> it was linux usb serial device support that was the problem, had to recognise it via hotplug

14:28:07 <dajobe> all working nice now

14:28:20 * danbri wonders if a reboot might shake some dust from somewhere (kernel stuff... mumble...)

14:28:29 <danbri> yeah, hotplug seems to work.

14:29:07 <danbri> Jul 31 14:46:47 fireball kernel: usbserial.c: Handspring Visor converter now attached to ttyUSB0 (or usb/tts/0 for devfs)

14:29:12 <dajobe> that's it

14:29:14 <danbri> Jul 31 14:47:03 fireball modprobe: modprobe: Can't locate module usb-serial

14:29:16 <dajobe> check perms on that device

14:29:22 * danbri saw that earlier

14:29:28 <danbri> I'm getting some data on and off ok

14:29:38 <dajobe> do lsmod and see if module 'visor' is loaded

14:29:38 <danbri> it is flaking out pretty much at random, mid transfer

14:29:53 * dajobe notes this is now #linux-palm-visor-support

14:30:02 <danbri> visor 8176 0

14:30:03 <danbri> usbserial 17424 14 [visor]

14:30:03 * DanC re-discovers the pdkb.pl swan song: a kludge to spit out iCalendar format, so I could use korganizer (later: evolution) after my palmpilot died

14:30:15 <dajobe> 14!

14:30:24 <danbri> 14 what? copies?!

14:30:28 <dajobe> uses

14:30:42 <dajobe> visor 10156 1

14:30:42 <dajobe> usbserial 17020 0 [visor]

14:30:50 <danbri> it's like a hit count, or there are 14 processes talking to it?

14:30:56 <dajobe> something in your system is using it; 14 processes, yes

14:31:32 <dajobe> catch me later when I'm actually using it, I'll see what i get

14:31:35 * dajobe re-lurks

14:32:05 * danbri mystified

14:32:55 * danbri decides to reboot

14:44:12 <AndyS> " * danbri decides to reboot" : what does the URI "danbri" refer to? The person? The machine? The document in MI5 about DanBri? ;-)

14:44:48 <Morbus> has anyone used apple's new iCal, btw?

14:49:07 * DanC tries to remember some OWL axiom stuff I was supposed to bounce of timbl...

14:56:27 * tim-half notes/grumbles that OWL "axioms" are not axioms for the OWL langauge but bits of the owl languge others are supposed to use in their onologies.

14:58:04 <DanC> meanwhile, I agreed to develop OWL axioms in another sense. But I can't quite recall what sense that was.

14:58:43 <DanC> sandro, do you remember?

15:00:07 <tim-half> Axioms in N3 for the OWL terms?

15:01:04 <DanC> was that all? I thought it was related in some interesting way to the abstract syntax thingy

15:01:37 <DanC> or maybe it was just that if I could specify OWL in N3, and sandro could specify N3 w.r.t. otter, we'd have OWL in otter.

15:02:15 <DanC> so... I don't see how to specify owl:disjointFrom with out not; i.e. without log:Falsehood.

15:02:20 <DanC> is log:Falsehood fair game?

15:03:36 <tim-half> Yes, I think so.

15:03:59 <DanC> tim, I think I'm gonna need a command-line option for which namespace to use for daml/owl. i.e. for list terms and =

15:04:30 <tim-half> ugh!

15:04:58 <tim-half> Currently, cwm accepts everything I have heard of for lists, and generates dpo.

15:05:09 <dajobe> is owl WDs using the new rdf Collection stuff or owl:List etc?

15:05:17 <tim-half> What is the current favorite? rdf: ?

15:05:33 <dajobe> well, we added them to rdf:

15:05:34 * DanC checks owl WDs-to-be...

15:05:48 <dajobe> I think MDean fixed, last time I looked at his draft-draft

15:05:50 <tim-half> I think own uses rdf lists, from memory.

15:05:53 <danbri> DanC, Tim... do we have any good resources on w3.org in defence of 'deep linking'?

15:06:13 <tim-half>http://www.w3.org/DesignIssues/LinkMyths

15:06:13 <dc_rdfig> C: http://www.w3.org/DesignIssues/LinkMyths from tim-half

15:06:21 <tim-half> I wouldn't say good buts my 2c

15:06:43 <tim-half> C:In defense of deep linking

15:06:43 <dc_rdfig> added comment C1

15:07:17 <danbri> Some ILRT friends working on internet catalogues have been getting the occiasional request that something be removed...

15:07:19 <danbri> ta

15:07:46 <danbri> C:see also [http://www.wired.com/news/politics/0,1283,51887,00.html?tw=wn_ascii|deep linking returns to surface] (Wired, April 18 2002)

15:07:46 * DanC is so insulted by the cases against deep linking that he hasn't managed to write anything civil in response

15:07:46 <dc_rdfig> added comment C2

15:08:03 <tim-half> If they are really outrageous reques s I could add a "hall of flame" section.

15:09:09 <danbri> the Times... (who should know better)

15:10:13 <DanC> [[ Several components employ lists constructed using the rdf:List, rdf:first, rdf:rest, rdf:nil, and rdf:parseType="Collection" vocabulary [Ed: identify link] recently added to RDF. ]] -- owl WD-to-be

15:14:11 <tim-half> Do the parseType="collection" lsist ahve any connection with Sequences?

15:14:48 <tim-half> { x rdf:1 y}=>{x rdf:first y} ?

15:15:04 <tim-half> tim-half is now known as tim

15:15:58 <dajobe> tim: No, they are a new thing

15:16:22 <tim> And the bags are still in?

15:16:22 <DanC> nope; in fact, I think I had an action to make a test case that clarified that rdf:_1 didn't entail rdf:first, just in case anybody was wondering.

15:16:29 <dajobe> and we just provide the mechanism - here's a closed collection of things, use them how you like

15:16:38 <dajobe> bag - yes

15:16:42 <DanC> the bags are still there. I'm hoping the primer won't mention them.

15:16:47 <dajobe> but like alt, seen as a bit useless, imho

15:16:47 <tim> :)

15:17:50 <tim> Suppose alt was dumped, and seq declred to be the same as collection (both ordered lists).

15:18:16 <DanC> the class name is rdf:List, not collection, fyi

15:18:49 <dajobe> yes

15:19:07 <dajobe> ah, we don't say the cloesed collection is ordered, just closed

15:19:42 <DanC> how do we say it's closed? I thought we just said: parsers emit these triples. Do we actually say rdf:first is functional?

15:19:52 <dajobe> we, ha ha ha, again thought best not to invent a new collections mechanism.

15:20:13 <Seth> does {A r (b,c,d)} => { {A r b} {A r c} {A r d}} ?

15:20:20 <dajobe> closed in that wehn you get to ->something->rdf:nil, you've got to the end

15:20:26 <dajobe> functional?

15:20:54 <DanC> i.e. that each list has at most one first.

15:21:12 * DanC can't decipher Seth's notation

15:21:21 <dajobe> oh, I dunno

15:21:28 <dajobe> I tried to find a daml defn, wasn't clear there

15:21:32 <Seth> just add : to taste.

15:21:42 <DanC> what's (b,c,d)?

15:21:56 <Seth> isn't that the way you do lists in n3 ?

15:22:03 <dajobe> "DAML+OIL needs to represent unordered collections of items (also known as bags, or multisets) in a number of constructions,"

15:22:07 <dajobe> -- http://www.daml.org/2001/03/reference.html

15:22:12 <DanC> ah... you mean (:b :c :d) . N3 lists have no ,

15:22:20 <dajobe> rather http://www.daml.org/2001/03/reference.html#collection

15:22:45 <bijan> DanC: what do you mean each list has at most one first...

15:22:48 <Seth> ok ... does does {A r (b c d)} => { {A r b} {A r c} {A r d}} ?

15:22:49 <DanC> I think I understand the question now, Seth. and the answer is: absolutely not.

15:22:55 <bijan> Wouldn't that preclude structure sharing?

15:23:08 <bijan> Or do you mean if you have another first you ahve a differnt list?

15:23:13 <DanC> (you've got too many {}s in your conclusion. you just need '.'s)

15:23:55 <Seth> ok ... does does {A r (b c d)} => { A r b. A r c. A r d. } ?

15:24:00 <DanC> no.

15:24:37 <DanC> i.e. if you say a/the dc:creator is (bob fred joe), then you haven't said that bob is a creator

15:25:07 <DanC> that belongs in an RDF FAQ/primer somewhere. I see people using collections as if they worked that way.

15:25:37 <las> frm

15:25:45 <DanC> the specification of dc:creator could say "this property distributes over lists", meanwhile.

15:25:47 <Seth> ok, i see ... but how do we express that r does in fact work that way ?

15:25:50 <tim> You could of course write an application to that effect for dc:craetor specifically.

15:26:24 <Seth> there are many r that do work that way .. would be useful to be able to express that.

15:26:26 <DanC> how? just like you wrote it: {A dc:creator (b c d)} => { A dc:creator b. A dc:creator c. A dc:creator d. }

15:26:44 <danbri> Yes, I see dc:creator pointing at bag, and it worries me

15:26:47 <deltab> is there an eachOf? x dc:creator [eachOf (bob fred jane)]

15:26:58 <DanC> well, perhaps a seth:DestributesOverListsProperty thingy would be useful.

15:27:21 <danbri> Anything in WebOnt's OWL that would do this?

15:27:35 <danbri> (my hunch is 'almost certainly; Ian would know how...')

15:28:20 <Seth> things like & , or , + , - work that way ... its kind of like we say the path (r , seq) is distributive

15:28:37 <Seth> take out - from that list

15:29:08 * DanC tries to think like Ian and do what danbri wants... failes

15:30:40 <danbri> maybe using a class instead of a container?

15:31:02 <danbri> this is the long-postponed pics generic label functionality...

15:31:52 <Seth> danbri, class would not order the list.

15:32:09 <danbri> agreed, you'd have to do that as well

15:32:39 <Seth> easier to just have a DestributesOverListsProperty like Danc suggested

15:33:07 * danbri wonders if Class and List are mutually disjoint

15:33:45 <Seth> more like apples and oranges

15:34:18 <Seth> correction: more like apples and colors

15:34:22 <tim> owl:oneOf links then in owl-land

15:36:20 <Seth> where are we suppos to start growing vocabulary like the 'semref' from yesterday and now this 'DistributesOverListsProperty' from today ?

15:37:03 <sbp> if you've got a Website, and you're able to write and publish RDF...

15:37:39 <tim> Is tere a need for an rdf interest group scratchpad for schemas of new ideas?

15:37:55 <Seth> yeah but if everybody does that we end up with chaos ... hey remember SWAG ... too bad it died before we really needed it.

15:38:17 <danbri> tim, yeah I have a todo to talk to oyu about setting up a wiki at w3.org

15:38:29 * tim wodners whether Danc's wiki could host a schema

15:38:33 <Seth> tim, scratchpad works for me. How do we actually make it happen?

15:38:55 <danbri> ...I'd love to hack it so a one could scribble into wiki and have it generate the rdf flavour automatically

15:39:12 <danbri> I think DanC's w3.org wiki died...

15:39:23 <tim> I ownder whether public write access it possible on w3.org documents.

15:40:04 <Seth> why not make people submit legit rdf/xml to a web page that would parse it and put it in a context called 'scratchpad' ?

15:40:14 <danbri> I wondered that too; you seened to think it was, last we discussed

15:40:19 <DanC> www-archive is a pretty good public-write mechanism/policy, I think.

15:40:44 <danbri> For some reason, haveing a re-editable writable space seems riskier

15:40:54 <DanC> qutie

15:40:57 <DanC> quite

15:41:05 <sbp> get them to PUT; that'd force some level of quality (sadly) :-)

15:41:07 <danbri> I'd prefer a different domain name, to keep things more clearly distinct

15:41:25 * danbri STILL trying to get palm to sync :(

15:41:38 <deltab> or POST? (not using form encoding)

15:41:40 <DanC> sync? or just back up?

15:42:09 <sbp> POST to append, PUT to replace, I suppose

15:42:30 <Seth> so what are we PUTing .. rdf/xml ?

15:42:30 <danbri> 'talk reliably to my computer over usb'

15:42:36 <Guha_> hi danbri ... you use Mac OS X, right? Try apple's new ISync

15:42:45 * danbri doesn't

15:42:55 <Morbus> iSync uses SyncML, right? have you used it?

15:42:57 * danbri grits his teeth and does it the linux way

15:43:10 <danbri> but I do fancy getting a mac

15:43:12 <Guha_> ah, I see. you must like pain.

15:43:21 * DanC suggests Guha disclose price tags on his "try this..." suggestions.

15:43:21 <Morbus> do more than fancy. lust for, danbri, lust for.

15:43:48 <DanC> apple wants $125 for iSync, no?

15:44:12 <Guha_> ah I didn't know that. I thought you got it for free with mac 0S X

15:44:43 <DanC> it's part of OS/X Jaguar, which is $125. No upgrade discount. :-{

15:45:20 <DanC> er... $129. http://www.apple.com/macosx/

15:45:20 <Guha_> greedy bas***ds

15:45:33 <bijan> Oh, pshaw. I'ts free with a new mac!

15:45:34 <bijan> :)

15:45:50 <DanC> it might well be worth $129; dunno.

15:46:00 <Morbus> Guha_: WinXP upgrade was 199.

15:46:04 <Morbus> less greedy than MS ;) ...

15:46:21 <Morbus> the full version was 299. the 129 cost is upgrade/full.

15:46:23 * bijan is a little miffed about it as he has a fairly new ibook that didn't fall into the upgrade grace period :(

15:46:30 * DanC too

15:46:39 <tim> Presumably the cost is to cover support... they estimate that if you upgrade you will take $129 more support.. that probably means you can expect what ... two call-in bugs? ;-)

15:46:55 <Guha_> oh my. you poor things.

15:47:08 <libby> it's rough payign for software

15:47:39 <DanC> I don't doubt the software is worth $129. I'm just spoiled. I haven't paid for software in so long.

15:47:43 <libby> applecare is worth paying for tho - 2 working days they fixed mine, so no tape any more :))

15:47:44 <Guha_> some of us can only envy you folks ... :-( as we fight the blue screen of death

15:47:49 * Morbus states for the record that he takes great massive amounts of effort to preach the greatness of OS X, and that coommnets like "you poor things" drive him flipping mad.

15:48:18 <DanC> I think "you poor things" was written in envy.

15:48:36 <Guha_> ahem yes. What DanC says.

15:48:47 <dajobe> I saw something about amazon.com update (US only) with $50 cash back

15:48:55 * Seth comes up with a plan to give sailors away, but sell cd's with certification and a gilded edge + support and connection to a social network

15:49:31 <DanC> This rendezvous thing has me curious. http://www.apple.com/macosx/jaguar/rendezvous.html

15:49:39 * DanC struggles to focus on webont axioms...

15:49:42 <dajobe> libby: friends ibook still been away >week, dead

15:49:56 <libby> really? mine were just fantastic

15:50:12 <libby> and they gave me anew power adaptor. and fixed my screen as well as the dvd

15:50:15 <Guha_> focus DanC focus.

15:51:25 <bijan> ...on something fun.

15:51:55 * bijan struggles with which is more fun, bemoaning after software one doesn't want to pay for or struggling with axioms one doesn't want to write...

15:52:21 <libby> gah....I can;t get pdb files ontop my mac :(

15:53:35 <Guha_> The knowledge engineering folks once did a survey and analysis of the cost of writing rules/axioms and found a great deal of uniformity in the "average cost per rule". Turned out to be $40 per rule.

15:53:59 <DanC> wild.

15:54:14 <tim> That's three rules get you ann OSX upgrade.

15:54:22 <bijan> Is this is frege era dollers? :)

15:54:34 <Guha_> No, circa 1988

15:54:38 <DanC> hmm... does owl:Thing include strings, dates, and such? or just individuals?

15:55:48 <bijan> Hmm. 1988 dollers...what's that $56 now?

15:55:52 <Guha_> it certainly should include *everything*

15:56:00 <bijan> $48?

15:56:09 <Seth> didn't Doug write zillions of rules already .. why can't we just look up the rule and use it now ?

15:56:22 <Guha_> you should.

15:56:32 * Seth likes to show his innocence

15:56:58 <Seth> well then, DanC .. there you go ..

15:56:58 <DanC> timbl... you were wondering, in the TAG telcon, if the cyc ontology had anything relevant to your notion of document. It does: cyc:ConceptualWork

15:57:24 <tim> cool. Thanks.

15:57:48 <tim> so I can assume cyc:ConceptualWork = doc:Work.

15:57:52 <DanC> timbl, take a look at http://www.cyc.com/cycdoc/vocab/info-vocab.html

15:58:33 <DanC> esp stuff like #$ist-Information , i.e. log:includes

15:58:34 <tim> Of course I won't be able to think that for long, it's have to be cyc:ConceptualWork owl:sameIndividualAs doc:Work.

15:58:43 <tim> 0.004 ;-)

15:58:49 <Guha_> Cyc has a fairly sophisticated notion of "information bearing thing". Karen Pittman and I wrote a paper on it long time ago.

15:59:24 <tim> Good. I think it is an important concept.

15:59:24 <DanC> many of the tricky bits of N3's logic have analogs in info-vocab. log:semantics is ... umm...

16:00:18 <DanC> log:semantics might be #$containsInformation. hmm... I had it the other day. should have made a note in log.n3

16:01:51 * DanC notes the connection between foaf:depiction and #$mtImageDepicts

16:01:54 <Guha_> That page doesn't convey the essense ... the interesting thing is the way a clean distinction is made between an IBT and its propositional content, and the relation between the two, expressed technically using the microtheory mechanism

16:02:35 <DanC> well, the page might not convey it, but I'm pretty sure TimBL came up with the same design. Well, something simlar. N3 tends to use formulas where cyc uses microtheories.

16:02:43 <danbri> <Guha_> Cyc has a fairly sophisticated notion of "information bearing thing". Karen Pittman and I wrote a paper on it long time ago.

16:02:47 <danbri> Is this online anywhere?

16:03:03 <danbri> there used to be a stash of postscript files on the cyc site, but last I looked I couldn't find

16:03:17 <Guha_> I am trying to find it. it was written in 1991 and presented at the AAAI Spring Symposium.

16:03:34 <Seth> DanC, probably because formulas and microtheories are both collection of statements

16:04:26 <tim> I always assumed a microtheory was a huge sub-logic including classes and predicates axioms around them. Whereas a formula is rather simple.

16:05:09 <Seth> aahh .. how many statements can you put in a box of statements

16:05:26 <Morbus> how big is the box?

16:05:27 <Guha_> not really. a microtheory is just a technical artifact for capturing the contextuality inherent in every linguistic occurance, including formulae.

16:05:43 <Guha_> They can be as big or as small as you want them.

16:05:51 <Seth> :)

16:05:59 <DanCon> a microtheory is somewhere between (what cwm/timbl knows as) a document and a formula.

16:06:23 <Seth> a document is just another breed of a box of statements

16:06:25 <Guha_> I have built systems where they were as small a single NL utterance used to disambiguate pronominal references

16:07:01 * DanCon wonders if he got around to citing "Contexts: ..." by Guha in http://www.w3.org/DesignIssues/Logic ... no, but...

16:07:14 <Seth> n3 is wierd cause sometimes it is a box of boxes

16:07:17 <Guha_> hey, please do ;-)

16:07:25 <DanCon> "Guha claimed induction is non-mon"; I still haven't researched that one. I think I lost the specific paper you pointed me to, Guha.

16:08:05 <Guha_> no, I claimed induction is second-order and very closely related to the circumscription axiom, which by itself is not non-mon, but enables non-mon

16:08:14 <tim> It sounds as though Guha should find his entire works and chump the master index!

16:08:50 <Guha_> :-). More fun to create new works.

16:09:15 <Seth> and while were at it, why dont we decide that the semantic web is certainly not monotonic

16:10:03 <DanCon> ah; thx, Guha. that makes more sense.

16:10:32 <Guha_> induction being non-mon would ahem, cause major problems in math.

16:11:59 <danbri> OK, Visor working again, it seems. No idea what was wrong. Back to trying to add calendar data into it...

16:12:07 <DanCon> whew!

16:12:58 <danbri> DanC, where did we get to? ah... more pdkb.pl

16:13:01 * danbri rtfms

16:13:07 <Seth> so a context is either monotonic or it is nonmonotonic ... what is the cyc word for a monotonic microtheory? ... a nonmonotonic microtheory ?

16:13:52 <Guha_> the context mechanism is orthogonal to non-mon mechanism.

16:14:37 <Guha_> Unfortunately, we still don't have even a 10% adequate semantics for contexts, so all we can go by are assorted implementations.

16:14:52 <Guha_> Effectively, all of cyc is non-mon.

16:15:59 <Seth> is {A isa Context; subClass MonotonicContext} a category error? ... sorry i dont know the cyc vocab very well

16:17:29 <Guha_> that is new syntax ... I need to go grok that.

16:18:04 <Seth> just add colons as necessary and you get n3 .. it just builds quads

16:18:29 <Seth> it gives the folowing triples

16:18:35 <Seth> A isa Context

16:18:47 <Seth> A subClass MonotonicContext

16:18:50 <Seth> .

16:21:48 <Seth> actually the only question here, guha, is can we classify contexts are monotonic and nonmonotonic or not ... or is that a catagory error in your grokology?

16:22:00 <danbri> DanCon, I'm trying palmagent. Made some edits via HTML/HTTP interface, it doesn't seem to preserve state.

16:22:12 <danbri> ...is idea that the mini-server rewrites the .pdb file?

16:22:43 <danbri> ie. I added a comment to an event, looked at some other stuff, went back, comment was there. But then I stopped and restarted the server, and comment had vanished.

16:23:54 <DanCon> the mini-server writes *a* .pdb file, but not *the* pdb file. look for ,post.pdb in the directory where you're running it.

16:24:06 <DanCon> I never got confident enough to overwrite the original.

16:25:30 * danbri tries: perl pdkb.pl --serve ,post.pdb

16:26:02 <danbri> seems to work :)

16:26:15 <danbri> I have two entries for each entry in the datebook, for some reason.

16:26:18 <danbri> different IDs.

16:26:36 * Seth wonders why guha fell silent ... was question off wall?

16:26:36 <danbri> this could be bad (old too) data on my part. Not worrying for now.

16:27:14 <pixel-meeting> pixel-meeting is now known as pixel

16:27:19 <Guha_> ah no. distracted. Yes, those two statement together are a category error

16:27:58 <Seth> how about just the one { A subClass MonotonicContext} .. is that a catagory error?

16:28:01 <danbri> DanCon, I'm ready to try batch populating the datebook from scripts. What's the best example to follow?

16:28:23 <DanCon> umm... is there something in the makefile about dbrec-test.rdf ? or in the README?

16:29:47 <Guha_> Seth, in general, with some exceptions instanceOf(A, B) ^ subClassOf(A, B) is often a category error.

16:29:57 <danbri> not that I can see (grepping for 'dbrec' doesnt' find much)

16:30:29 <DanCon> umm... I think there's a little post.py script, no?

16:30:48 <Seth> guha, ok i grok that .. but how about just { A subClass MonotonicContext} ? ... can we classify contexts according to their monoticity or not ?

16:31:17 <DanCon> try: python postrdf.py host port <dbrec-test.rdf

16:31:33 <Guha_> yes you can, question is, how useful is that?

16:32:12 <Seth> well sometimes we want to draw conclusions as if we know everyting about a subject, and sometimes we don't.

16:33:34 <Seth> for example, i know all of the customers at speaktomecatalog.com .. i have them all in a database .. if you ask if you are a customer, i can tell you ... istnt that a monotonic context?

16:33:51 <danbri> python postrdf.py localhost 33010 < dbrec-test.rdf

16:33:55 <danbri> seemed to work

16:34:03 <Guha_> actually, no, its not.

16:34:14 <DanCon> well, then you're winning. is dbrec-test.rdf a clear enough pattern?

16:34:44 <DanCon> I can't remember if <pdu:inserts rdf:resource='#n'/> is still needed

16:35:10 <danbri> workie :)

16:35:14 <danbri> thanks!

16:35:15 <Seth> is that a nonmonotonic ?

16:35:50 <danbri> I wasn't sure that sync'ing the ,foo.rdf file would be enough. My mental model of what's going on is pretty hazy...

16:36:03 <danbri> do I now have two palm databook databases on my palm?

16:36:16 <DanCon> no, pdu:inserts is apocryphal. the addRecords routine in pdkb.pl just looks for all subjects of pd:date

16:36:53 * danbri can't connect your answer to my question

16:37:03 <DanCon> 2 databases: I don't think so, but I need some context: are you using coldsync?

16:37:21 <danbri> I used pilot-xfer -i ,post.pdb

16:37:38 <danbri> ah, that is the entire database. I got it into my head that it was a sole entry for the .rdf file.

16:37:53 <danbri> I should've used coldsync for consistency.

16:39:44 * Seth wonders if he will ever understand monoticity

16:39:46 <danbri> just coldsync synch'd now.

16:41:58 <Guha_> Seth, its one of the easiest things

16:42:46 <Seth> well ive heard PatH describe it, and ive read a book on it, and i still come up with the wrong answers

16:43:07 <Guha_> if A => C but A+B does not imply C, for any statements A, B and C, then the system is not monotonic.

16:44:04 <Seth> and that's all there is to it ??

16:45:07 * danbri tries re-adding the same event, doesn't get duplication, assumes this is good

16:45:41 <danbri> DanCon, if you've any more time for this, I'd like to look at adding repeating events etc.

16:45:49 * Seth feels a mentograph comming on

16:45:50 <danbri> I sohuld take a look at the python server first, probably

16:46:53 <danbri> s/python/perl/

16:47:56 <DanCon> [that answer was to my own question]

16:48:03 <DanCon> danbri, have you got some ESW calendar info in .rdf anywhere public? I'm willing to get you started writing rules to convert to pd vocab.

16:48:05 * tim wonders whether dabri's palm has become non-monotonic.

16:49:56 <DanCon> er... are you surre about re-adding? note pdkb.pl only writes to ,post.pdb. So you have to run it on foo.pdb, then mv ,post.pdb foo.pdb, then run it again to be sure you're not getting a dup

16:51:22 <DanCon> I think repeated events are supported in the POST/rdf case. (though not in the HTML forms interface)

16:52:14 <DanCon> no, I guess not. The routine to fix is fillRecord

16:53:55 <DanCon> odd... why did I bother with exception lists if I didn't do repeat interval?

16:54:15 <DanCon> danbri, see #@@ repeat in fillRecord?

16:54:27 <DanCon> danbri, do you have ESW data somewhere?

16:54:37 <danbri> see RSS c/o http://www.w3.org/2000/06/webdata/xslt?xslfile=http://www.w3.org/2001/sw/Europe/events/home2rss.xsl&xmlfile=http://www.w3.org/2001/sw/Europe/events/Overview.html

16:54:38 <libby> we got data

16:54:49 <libby> what sorta data?

16:55:09 <libby> I've got deliverables in pseudo rss+events too

16:55:10 <DanCon> data I could convert, via N3 rules, to palmpilot vocabulary, to stick in danbri's palm

16:55:16 <libby> ah

16:55:29 * DanCon recovers from rr.com hiccup

16:56:41 <libby> delivs: http://www.w3.org/2001/sw/Europe/plan/workpackages/live/rdf/_esw_rssevents_delivs.rdf

16:57:13 <danbri> ah, right you are DanC. It does add duplicates.

16:57:45 <monokrom> does anyone recall a doc at the w3.org site that addressed the formatting of url querystrings? I've been googling for it, but can't find it and I'm sure that I saw it once.. a ways back

16:58:07 <danbri> What I *hope* to do is have a pile of RDF feeds that I periodically merge into my palm. I wouldn't edit those events in palm (hmm, would I?) but would over-write them with latest version read from the Web.

16:58:18 <libby> there's also conferences: http://swordfish.rdfweb.org/calendar/events/swevents.rss

16:59:20 <danbri> monokrom, nothing that I can remember, sorry.

16:59:37 <monokrom> danbri thanks anyway :D

16:59:51 <DanCon> monokrom, url querystrings: that's in the HTML 4 spec

17:00:11 <monokrom> oh? ok :D i'm going there now, thanks

17:00:57 <DanCon> monokrom, http://www.w3.org/TR/html4/interact/forms.html#h-17.13.4.1

17:01:45 <monokrom> excellent, that's precisely what I needed - thanks !

17:02:46 * DanCon finds that write-up lacking in examples. :-{

17:02:46 * DanCon finally cited Guha '90 from http://www.w3.org/DesignIssues/Logic#CLA so I shouldn't have to hunt for it again

17:02:53 <DanCon> er... Guha '91

17:03:52 <DanCon> heh... libby, that's pretty quick-n-dirty: <title>proj_tech_plan (mailto:danbri@w3.org)</title>

17:04:22 <danbri> we wanted something readable on a palm...

17:05:30 <DanCon> logger, pointer?

17:05:30 <DanCon> See http://ilrt.org/discovery/chatlogs/rdfig/2002-07-31#T17-05-30

17:05:35 * danbri getting mixed up with http://www.w3.org/2001/sw/Europe/plan/workpackages/live/bin/esw-report.rb

17:08:50 * sandro looks seriously at using eikeon's rdflib for some current hacking. hrrrrrrrm.

17:10:18 <danbri> DanC, are you hacking on rssevent2palm converter?

17:10:23 <DanCon> yup; it's working...

17:10:25 <DanCon> [ :date "2002-09-23";

17:10:25 <DanCon> :description "Possible SWAD-Europe face to face meeting";

17:10:25 <DanCon> :note "The second face to face meeting for SWAD-Europe will be held on either 23 September 2002 in Rutherford Appleton Laboratories. (Added 30 July 2002)" ].

17:10:29 <danbri> ooh

17:10:37 * eikeon cheers on sandro :)

17:11:07 <danbri> I don't know enough about the respective schemas to know if I'll hose my data, but since most of my calendar is currently in my head, i've not much to lose.

17:11:42 <danbri> I'm going to spend another half-day on this stuff this week...

17:12:11 <danbri> Is it in cvs?

17:13:07 * danbri asking dumb question; checks; it's not yet

17:14:12 <DanCon>http://www.w3.org/2000/10/swap/test/esw2pd.n3

17:14:12 <dc_rdfig> D: http://www.w3.org/2000/10/swap/test/esw2pd.n3 from DanCon

17:14:34 <DanCon> D:|esw2pd.n3 converter from RSS events to palm

17:14:34 <dc_rdfig> titled item D

17:14:44 <DanCon> D:Invoke ala: python cwm.py esw2pd.n3 --think --purge --rdf=d >,palmStuff.rdf

17:14:45 <dc_rdfig> added comment D1

17:15:40 <DanCon> D:the specific link to the ESW data could be factored out.

17:15:40 <dc_rdfig> added comment D2

17:16:25 <DanCon> D:another approach would be to say { rss:title s:subPropertyOf pd:description } and use rdfs-rules

17:16:25 <dc_rdfig> added comment D3

17:16:55 <tim> [12:58] <danbri> What I *hope* to do is have a pile of RDF feeds that I periodically merge into my palm.

17:17:01 <DanCon> C:|Links and Law: Myths

17:17:01 <dc_rdfig> titled item C

17:17:42 <tim> danbri, the teleconfernce calendar plus a hand-hacked list of the conferences I am in generates dataebook entries

17:18:29 <danbri> in the format that danc's tool consumes?

17:18:51 * DanCon thinks timbl's using my ical vocab, not my palm datebook vocab

17:19:33 <danbri> [:rxd "YYYY-MM-DDTHH:MM:SSZ"; :from "127.0.0.1"; :method :POST; :res </DatebookDB/>].

17:19:33 <danbri> I don't grok anonymous nodes nor rdf:ID at trdf.pl line 124.

17:19:35 <tim> ... /swap/pim/conf2cal.n3 conversts from the Zakim teleconferece booking ontology to the thing Dan's tools take for generating icalendar....

17:19:42 <danbri> ...the py server doesn't like your rdf dialect

17:19:47 <tim> So is tehre a barrier between iCalendar and datebook?

17:19:48 <danbri> hmm, i keep thinking its python

17:19:55 <danbri> perl server...

17:20:43 <tim> I'm surprised danc hasn't agtewayed between ical and palm

17:22:40 <danbri> odd. Running dan's esw2pd.n3, it seems to pass through an about="" without expanding:

17:22:47 <danbri> <> dc:description """

17:22:47 <danbri> converter from RSS events to palm.

17:22:51 <danbri> ...is in the .n3 rules

17:23:05 <danbri> <rdf:Description rdf:about="">

17:23:05 <danbri> <dc:description>

17:23:05 <danbri> converter from RSS events to palm.

17:23:18 <danbri> ...is in the output generated. Is this typical cwm behaviour?

17:23:32 <DanCon> the RDF parser in pdkb.pl is pretty impoverished. I saw a release of palm support for pythong that makes me think I'd re-write it in python if I were to mess with it any more.

17:24:17 <DanCon> yes, that's typical cwm: it believes what you tell it (on the command line) and it spits it back out at you

17:24:33 <eikeon> DanCon: Why not rdflib's rdf/xml parser :)

17:24:39 <danbri> Which seems fine, except relative uris are surface syntax...

17:24:48 <DanCon> eikeon, go for it.

17:24:59 <tim> Danbri, it isn't that it passes though <> without expansion. It carefully translates any sel-referential statements itno the output documnet.

17:25:21 <DanCon> well, one reason is that debian and CPAN don't get along

17:25:24 <tim> This is a questionable area. It maybe should be a command line option.

17:25:35 <danbri> ah, so ../ would be absolutised.

17:25:45 <DanCon> it is a command-line option, timbl: I use --base=bogus: a lot to get cwm to absolutize stuff.

17:26:16 <DanCon> what diagnostic do you get from pdkb.pl, danbri?

17:26:29 <DanCon> is it easy to hack the offending bits out of the .rdf file?

17:27:00 <DanCon> no, ../ would stays relative (unless you use --base) too.

17:27:18 * DanCon could just try pdkb.pl himself...

17:27:32 <tim> Here is the problem: If I write <> a log:N3Document; dc:description "This information is derived from statistical restuls of UN surveys betwen 1956 and 1972".

17:27:40 <danbri> that's such a slippery thing to do. Some of us are using about="" for pointing from a doc to its gpg signature... how would it deal with loading two such docs at same time?

17:27:42 <danbri> Serving palm datebook <DatebookDB.pdb> at: <http://fireball:33149/>

17:27:42 <danbri> [:rxd "YYYY-MM-DDTHH:MM:SSZ"; :from "127.0.0.1"; :method :POST; :res </DatebookDB/>].

17:27:42 <danbri> I don't grok anonymous nodes nor rdf:ID at trdf.pl line 124.

17:27:44 <tim> I want to keep th esecond if I am just translatinbg the file

17:28:06 <danbri> eg http://usefulinc.com/foaf/signingFoafFiles

17:28:24 <DanCon> "I don't grok anonymous nodes nor rdf:ID at trdf.pl line 124."

17:28:36 * danbri can't hack out the offendign rdf; rather, it wants IDs (or URIs?) added in.

17:28:42 <tim> signature - good point. Maybe I should turn it off. I used to need it because this log:forAll x was <> log:forAll x but tat is going and gone.

17:29:25 <danbri> carry a notion of 'this' through data merging is really tough. I don't think you can do it in vanilla RDF.

17:30:26 <tim> Well, maybe when I change "this log:forSome" to "@forSome" we will get rid of "this".

17:30:55 <tim> So I don't se it as a long-term problem

17:31:19 <danbri> ok. i'll use --base=bogus in meantime

17:33:02 <danbri> danc/tim, is there a way (in esw2pd.n3) that we can get rdf:about's generated for each rdf:Description ? or generated rdf:ID ?

17:33:46 <sbp> s/bogus/bogus:/

17:36:48 <tim> I'm not familiar with the problem. You want to get rid of the subject and make it a bnode?

17:37:32 <danbri> <rdf:Description>

17:37:33 <danbri> <pd:date>2002-06-17</pd:date>

17:37:35 <danbri> ...should be

17:37:42 <danbri> <rdf:Description rdf:about="...something">

17:37:52 <danbri> ...or danc's subset-rdf parser barfs

17:38:11 <tim> Oh, you can't handle bnodes and you want an arbitrary URi for it.

17:38:15 <danbri> <rdf:Description rdf:about='#n'> ...being the example

17:38:16 <danbri> yes

17:38:50 <tim> Do you have any unique field identifying the object?

17:38:59 <tim> (like a uid: or a datetimeshoesize?)

17:39:18 <danbri> good question

17:39:44 <danbri> I don't _think_ so, though unsure how the tool behaves if there are two things with the same pd:date

17:40:05 <danbri> ...hm that should be fine. Things happen on same day all the time.

17:40:07 * DanCon plays around with cwm's --rdf=d and --rdf=c flag, trying to make pdkb.pl happy, but then runs into parseType="Resource"

17:41:01 <DanCon> danbri, the scraped-from-HTML data does have URIs for events, but they're wierd:

17:41:02 <DanCon> <rdf:Description rdf:about=":/2002/05/er-swade-f2f">

17:41:20 <tim> (If so you can may an n3 rule of the form {[ uid :u; foo :f; bar :b] . :x log:uri is string:concatenation of ("uuid: :u) } => { :x foo :f; bar :b}. or words to that effect.

17:41:21 <danbri> yeah, I remember seeing those. Sounds like an XSLT bug...

17:41:23 <danbri> libby?

17:41:33 <libby> ugh sorry, lazy my fault

17:41:38 <libby> no base uri

17:41:48 <danbri> fixable?

17:41:49 <tim> You have string:scrape and string:concatenation to unweird that.

17:41:53 <libby> you're supposed to pass it a base uri

17:43:37 <DanCon> D:v 1.2 has esw link factored out (to the command-line) and uses just the one rdfs:subProperty rule

17:43:37 <dc_rdfig> added comment D4

17:43:41 <tim> you can scrape the characters after the colon out with string:scrape and then concatenat some reaonable base

17:43:51 <libby> er, I could hard code it

17:44:09 * DanCon prefers not to mask bugs

17:44:22 <libby> me neither, sorry, I'm in the middle of something

17:44:36 <danbri> D:try [http://www.w3.org/2000/06/webdata/xslt?xslfile=http://www.w3.org/2001/sw/Europe/events/home2rss.xsl&xmlfile=http://www.w3.org/2001/sw/Europe/events/Overview.html&Base=http://www.w3.org/2001/sw/Europe/events/id/|adding a Base param] to get full (if 404) uris for each event. This is a hack.

17:44:36 <dc_rdfig> added comment D5

17:44:59 <libby> is that a hack? I thougvht you were supposed to pass it the base

17:45:07 <libby> if you look at the xslt

17:46:25 <danbri> D:...in that these URIs aren't really names for the events, and don't dereference to anything of interest. But the palmagent rdf parser expects ids on events...

17:46:25 <dc_rdfig> added comment D6

17:47:14 <libby> I think the base is http://www.w3.org/2001/sw/Europe/

17:47:31 <libby> well apart from the f2f2 one and the telcon

17:47:35 <danbri> Do we have a doc for each event? that'd be cool.

17:48:07 <danbri> So anyway, need more hacking... while we have URIs now for the events going into esw2pd.n3, those coming out are different resources, and still nameless.

17:48:34 <danbri> [ ev:startdate :WHEN;

17:48:35 <danbri> rss:title :WHAT;

17:48:35 <danbri> rss:description :DESC;

17:48:35 <danbri> ] } ]

17:48:35 <danbri> } log:implies {

17:48:35 <danbri> [ pd:description :WHAT;

17:48:37 <danbri> pd:note :DESC;

17:48:39 <danbri> pd:date :WHEN

17:48:41 <danbri> ]

17:48:58 * danbri --think --think-some-more

17:48:59 <libby> can you generate the intermediate days as well using CWM, i.e. the ones between start and end?

17:49:07 <danbri> ooh, quite possibly.

17:49:12 <libby> be handy

17:49:42 * libby can;t get a Datebook.pdb on the mac to play with :(

17:50:05 <libby> or rather I have, using sync buddy, but the file is dodgy

17:50:35 <libby> heyho

17:51:56 <danbri> DanoCon, Tim (or other N3 experts), can you advise on how to pass through the URI...? Do we junk the chunky brackets... (and replace with what?)?

17:52:36 <Seth>http://robustai.net/mentography/monotonic.gif

17:52:36 <dc_rdfig> E: http://robustai.net/mentography/monotonic.gif from Seth

17:53:02 <Seth> E: The definition of monotonic ... Guha, did i get it right ?

17:53:02 <dc_rdfig> added comment E1

17:53:26 <Seth> E:| The definition of monotonic ... Guha, did i get it right ?

17:53:26 <dc_rdfig> titled item E

17:53:55 <Seth> E1:""

17:53:55 <dc_rdfig> deleted comment E1

17:54:54 <Seth> E: seeAlso [http://ilrt.org/discovery/chatlogs/rdfig/2002-07-31.html#T16-42-46|today's chatlog]

17:54:54 <dc_rdfig> added comment E1

17:59:10 <danbri> {

17:59:10 <danbri> :E ev:startdate :WHEN .

17:59:10 <danbri> :E rss:title :WHAT .

17:59:10 <danbri> :E rss:description :DESC .

17:59:10 <danbri> } ]

17:59:11 <danbri> } log:implies {

17:59:13 <danbri> :E pd:description :WHAT .

17:59:15 <danbri> :E pd:note :DESC .

17:59:17 <danbri> :E pd:date :WHEN .

17:59:19 <danbri> }.

17:59:23 <danbri> ....I thought this would work. Doesn't.

17:59:25 * danbri tried though

17:59:27 <danbri> I get trying points, right?

18:00:43 * DanCon catches up...

18:00:53 <DanCon> yes, that's right, danbri.

18:01:58 <DanCon> doesn't work? did you add :E to the forAll list?

18:02:54 * DanCon hunts for documentation about rdflib's rdf parser... loses.

18:03:08 * DanCon tries UTSL...

18:03:40 <danbri> did I what now? :)

18:03:42 * danbri adds :E

18:04:28 <DanCon> wow... nodeID support...

18:04:36 <danbri> cwm has a big think... time passes...

18:04:40 <danbri> It worked :) :)

18:04:45 <danbri> I like it when that happens.

18:05:45 <Seth> dancon, for rdflib you should hail eikeon in #redfoot

18:06:49 <tim> [13:48] <libby> can you generate the intermediate days as well using CWM, i.e. the ones between start and end?

18:07:43 <tim> yes, probably with the time: builtins. Make teh start time in seconds, keep adding 86400 until you get past the end time. Convert each value back to an isoddate.

18:07:49 <DanCon> you can, libby, but generating just the ones you want is tedious/tricky

18:08:02 <tim> time:inSeconds is a bidirectional function.

18:08:31 <DanCon> i.e. you can't just write general rules that relate one day to the next; cwm will go on generating an infinite number of days

18:08:37 <libby> could be very useful

18:08:43 <libby> ok

18:08:44 <libby> heh

18:09:09 <danbri> coolness, I can confirm that the long journey of these event descriptoins from html to my palm pilot is complete!

18:09:13 <DanCon> a backward-chaining reasoner could handle the more general rules more easily.

18:09:30 * DanCon gives danbri a big high-5!

18:09:36 <tim> yeah danbri.

18:10:09 * danbri thanks libby, timbl, danc and doubtless others for the fancy tools that made it all possible

18:10:18 * danbri blushes, grins etc

18:11:34 <danbri> So now I need to get this into real use, or i'll forget how it all works. I suspect that means hacking the perl (or redoing in python/ruby) so reloads don't repeat existing entries.

18:11:51 <libby> it's v cool guys

18:11:53 <DanCon> ok, danbri, your next assignment is to get the toothpaste back in the tube: make an edit on your pilot, and have it propagate back to the HTML page. 1/2 ;-)

18:12:02 <libby> heh

18:12:08 <DanCon> is there ruby palm support?

18:12:42 <libby> see ya

18:12:43 <sandro> Isn't rdf:first the exact same thing as rdf:_1?

18:13:09 <DanCon> OH! don't move, danbri! not until you've made a table relating the palm id's to the about= URIs from the .html

18:13:38 * danbri heads off to dinner, wondering how that table would work

18:13:44 <danbri> need to read up on palm stuff

18:13:48 <danbri> oh, ruby palm:

18:13:50 <DanCon> bummer... we should have squirreled away the URIs in the pd:note field somewhere

18:13:54 <danbri> see http://www.osk.3web.ne.jp/~nyasu/palmsync/

18:14:06 <DanCon> (easy to do with string:concatenation)

18:14:29 <dajobe> sandro: no

18:14:31 * DanCon can't read that; suspects it's supposed to be japanese

18:14:46 * DanCon finds http://www.osk.3web.ne.jp/~nyasu/palmsync/e.html

18:15:09 <DanCon> ooh! [[ "PalmSync" is a Ruby(Scripting Language) library for syncing your PalmPilot with DBMS ]]

18:15:41 <sandro> Why not, dajobe? Why should a rdf:List and an rdf:Seq be different, except that an rdf:List may be terminated?

18:16:01 * dajobe isn't going there

18:16:07 <DanCon> because implementations don't treat them as the same

18:16:30 <DanCon> i.e. we don't set expectations that rdf:Seq will behave like rdf:List, cuz it won't.

18:16:42 * dajobe should write that in the spec

18:16:44 <DanCon> well... it could, if we required implementations to do that, but we didn't.

18:17:43 * sandro thinks that's just asking for trouble, long term. RDF has two kinds of sequences -- there's no real difference between them, but they are different. It's just what the WG felt like doing.

18:18:03 <dajobe> no, we were asked to provide closed collections by daml^wwebont

18:18:35 * DanCon finally figures out the rdflib parser callback mechanism. nifty.

18:18:43 <sandro> You could have done that with rdf:seqLastIndex rdf:_7

18:18:54 <dajobe> eikeon win's the most advanced parser award, at least for now ;)

18:19:10 <dajobe> sandro: we went through all those options, DanC objected to some of them like that one ;)

18:19:20 <DanCon> yeah, but seqLastIndex is a pain to deal with. first/rest is so much easier.

18:19:41 <dajobe> (not just Dan)

18:19:45 <DanCon> just let rdf:Seq go, quietly, into that good night, sandro.

18:19:47 <sandro> But having two parallel but different mechanisms is the worst of both worlds.

18:20:00 * DanCon hopes the RDF primer doesn't mention rdf:Seq

18:20:06 <dajobe> I'm sure it does

18:20:10 <dajobe> I'll object if it doesn't

18:20:11 <sandro> If the WG will take it out of the spec, like aboutEachPrefix, then sure.

18:20:44 <sandro> but if the new Rec is going to say we have two different ways to represent sequences, then we have a real problem.

18:20:52 <dajobe> it isn't a sequence

18:20:58 <dajobe> it is a closed collection *no ordering implied*

18:20:59 <DanCon> hmm... how about a mention ala "this syntax is part of the language. Don't expect too much from it, because it doesn't match intuitions in lots of ways," dajobe?

18:21:16 <dajobe> no

18:21:20 <DanCon> huh? first/rest is quite an explicit ordering

18:21:22 <dajobe> I find the lispy first/rest ugly

18:21:53 <dajobe> it is a structure that creates a collection, no ordering

18:21:55 * DanCon wonders if dajobe has done much with rules or deduction

18:22:11 <dajobe> see http://www.daml.org/2001/03/reference.html#collection

18:22:20 <DanCon> er... "no ordering" is argument by assertion, dajobe. I can quite easliy pick out the items of an rdf:List in order.

18:22:20 <dajobe> "DAML+OIL needs to represent unordered collections of items"

18:22:35 <sandro> no ordering implied? WTF? That's completely lame. WebOnt doesn't have to USE the ordering, like LISP doesn't always care about the ordering, but of course the ordering is there. To say that an RDF processor is allowed to reorder any rdf:List (which you're saying) is.... absurd.

18:22:43 * dajobe is just following daml, above

18:22:51 <DanCon> many DAML+OIL idioms aren't sensitive to the order given by first/rest. that doesn't mean the order isn't there.

18:22:57 <dajobe> I've no mandate to do something different

18:23:09 <dajobe> the key word is closed, not ordered

18:23:26 <DanCon> you don't have to put the word "ordered" in the spec to make it ordered. it's ordered naturally.

18:23:42 <DanCon> if you put "not ordered" in the spec, that will be incorrect.

18:23:43 <dajobe> so XML is ordered but we don't use that either (apart from rdf:li rewriting)

18:24:01 <DanCon> yup; the analogy with XML is pretty good.

18:24:15 <dajobe> I put closed

18:24:19 <dajobe> ordering is for the app to decide

18:24:23 <DanCon> ok.

18:24:24 <dajobe> or duplicates, etc, I don't care

18:24:36 <dajobe> I can warn that ordering is not implied

18:24:45 <DanCon> no, you can't. not sensibly.

18:25:37 <dajobe> hmm

18:25:48 <dajobe> I guess it is up to the app; read the triples, it works out what it represents.

18:25:55 <DanCon> ok.

18:26:24 <bijan> Hmm. You can warn that, from the RDF pov, the order implied by first/rest isn't significant.

18:26:46 <dajobe> Pity this wasn't in an RDF core WD before webont wds were approved, we could have got feedback if it was rihgt. Hopefully ok

18:26:52 <DanCon> no, it is significant. I can write tests that observe it.

18:27:04 <sandro> I think we need to either drop rdf:Seq or just say rdf:List is an rdf:Seq which is terminated (with rdf:nil, in the lispy way). So rdf:_1 is car, rdf:_2 is cadr, rdf:_3 is caddr, etc.

18:27:26 <dajobe> can you de-lisp that for me?

18:27:36 <dajobe> car* etc.

18:28:00 <sandro> rdf:_1 is first, rdf:_2 first-of-rest, rdf:_3 is first-of-rest-of-rest, etc

18:28:27 <sandro> (in other words, the obvious meaning of nth)

18:28:44 <DanCon> I wouldn't miss rdf:Seq, but it's used in RSS, which is about the most widely deployed RDF thingy out there. Saying that rdf:List is an rdf:Seq etc. is a disservice unless all the implementors actually put the relevant smarts in their parsers. But what's going to motivate them to do that?

18:29:50 <DanCon> the folks that are happy with :_1 use it happily, and the folks that use first/rest are happy with it; they don't really care about interoperability accross idioms, as far as I can tel.

18:30:03 <sandro> I guess the implementation challenge is that you need to do some inference: when you see an <li> you need to make an rdf:_n triple and two rdf:first/rest triples. When you see an rdf:_n triple, you need to make some first/rest triples. When you see some first/rest triples, you need to make some rdf:_n triples. Maybe maybe some li's, too... I dunno.

18:30:14 <DanCon> yup

18:30:26 <sandro> It's all trivial if you have horn inference handy, but a little annoying if you're writing in C.

18:30:31 <DanCon> quite.

18:30:45 <dajobe> no inference engines in the parser, please ;)

18:30:52 <DanCon> s/;)//

18:31:12 <sandro> I can write that C code in 7 notes, bob.

18:31:15 <sandro> :-)

18:32:11 <tim> [14:20] <dajobe> it isn't a sequence

18:32:11 <tim> [14:20] <dajobe> it is a closed collection *no ordering implied*

18:32:11 <sandro> So the people down the line who want to use data sources which use the two idioms -- they're the ones who need the inference engines. I can live with that cause I think everyone will have such engines.....

18:32:39 <DanCon> well, actually, rdflib looks like a pretty clean basis for parser hacking. You could try supporting the connection between rdf:Seq and rdf:List there, sandro, and report back how easy/hard it is.

18:32:39 <tim> I think it would be fine to merge Seq and lists.

18:32:55 <tim> I think callingsoemthig which ahs the overhead of fisrt and rest unordered is ridiculous.

18:33:06 <DanCon> nobody's calling it unordered, timbl.

18:33:14 <dajobe> I'd fight against rdf:Seq generating different triples than before. Extra ones - hmm

18:33:30 <tim> [14:20] <dajobe> it isn't a sequence

18:33:30 <tim> [14:20] <dajobe> it is a closed collection *no ordering implied*

18:33:37 <DanCon> keep reading, timbl.

18:33:40 <tim> That means to me that (1 2 ) = ( 2 1)

18:33:46 <Seth> why not just store things in pents (with a seq as an attribute of the arrow), then always convert to that structure ... then no big prob, me thinks.

18:33:55 * tim reads a second time

18:34:35 <sandro> logger, pointer?

18:34:35 <sandro> See http://ilrt.org/discovery/chatlogs/rdfig/2002-07-31#T18-34-35

18:35:09 <DanCon> "down the line" -- yes, that's my position.

18:35:17 <sandro> do li's make it through into n-triples?

18:35:31 <DanCon> li's turn into rdf:_n's

18:35:47 * DanCon wrotes some N3 rules for handling rdf:_n's for ericm the other day...

18:35:54 <dajobe> rdf:li's a macro of sorts

18:36:21 <DanCon> ah: http://www.w3.org/2000/10/swap/test/seqiter.n3

18:36:24 <sandro> and then happily disappear themselves.... Hrm. It might make more sense to have the li's stay around (and be created, if the rdf:_3 form appeared).

18:37:37 <DanCon> hmm... dajobe, we talked about a :member thingy, no?

18:37:41 <sandro> Yeah -- axiomatizing rdf:_[n] like that seems quite doable.

18:37:56 <dajobe> DanCon: yeah, we added a new super property of all the rdf:<n>s IIRC

18:38:05 <dajobe> not sure if danbri put it in a doc yet

18:38:18 <dajobe> (rather the rdf:_<n>s)

18:38:24 <DanCon> is that a superproperty in the RDFS sense, or in the parsers-shall-emit-these sense?

18:38:29 * tim finds rdf:_[n] very ugly. In extreme cases we look inside URIs but not for the basics ofthe language.

18:38:29 <dajobe> RDFS

18:38:41 * DanCon agrees; rdf:_n ugly

18:39:05 <larsbot> has anyone suggested alternatives?

18:39:06 * dajobe would rather have done a new container form than datatypes. <sigh/>

18:39:12 <DanCon> meanwhile, probably 90% of the RDF in the world is RSS, which includes <li>

18:39:19 <sandro> heh

18:39:33 <tim> Just like after 4 days there were already 10 people using Makefile syntax.

18:39:41 <DanCon> we not only suggested alternatives, we adopted one: first/rest, and the Container syntax

18:39:49 <tim> Check how many of them use XML tools only, no rdf

18:40:08 <larsbot> hmmm. how does first/rest perform in persistent storage?

18:40:14 <DanCon> dunno

18:40:26 <dajobe> somebody know lisp (engine) literature?

18:40:28 <larsbot> I've been chewing on this, too, and like neither alternative

18:40:29 <dajobe> at MIT, ahem

18:40:31 <bijan> many, infact, iirc, don't use the seq, but use the ordering/presence of items in the file.

18:40:53 <larsbot> bijan: ouch ouch ouch

18:41:05 <bijan> Yep :(

18:41:12 <bijan> maybe they're all fixed by now...

18:41:18 <DanCon> er... well, timbl, the WG spent a lot of time thinking it was supposed to take compatibility seriously. We've veered off a bit, but telling the RSS folks "oops; your documents are no good any more" isn't something we'd do lightly.

18:41:49 <DanCon> ... but if The Director wanted to send that signal about the RDF 1.0 recommendation, well, ...

18:42:17 <larsbot> dajobe: what do you mean by "engine"?

18:42:19 * bijan watches for genies or whirlwinds let lose from rdf bottles...

18:42:22 <tim> Of course we would not want to make things unnecsarily incompatible.

18:42:31 <bijan> Whew! Weather clear!

18:42:31 <tim> Look at it like this.

18:42:44 <tim> There is a cocnpet of an ordered list.

18:42:46 <dajobe> larsbot: I meant machine

18:42:55 <tim> There are two syntaxes at the moment.

18:42:55 <bijan> WEll, the "unnecessarily" is a storm cloud on the horizen...

18:43:00 <tim> One is the old Seq one

18:43:02 <larsbot> dajobe: oh, as in Symbolics and such stuff?

18:43:08 <tim> The otehr is the new first, rest one.

18:43:15 <larsbot> that is, lisp hardware?

18:43:39 <tim> Given an orderd list (a b c) various things are true.

18:43:55 <tim> parsers can decide what facts about a list they will give you

18:43:56 * DanCon gets mail about Yet Another broken link in the OWL WD-to-be's

18:44:22 <DanCon> "parsers can decide" bzzt. there went interoperability out the window. There went log:includes, for example.

18:44:47 <DanCon> there went the whole RDFCore testing framework.

18:44:49 <bijan> Whew! I was about to try to implement it.

18:44:56 <bijan> Er..log:includes that is :)

18:45:12 <tim> I wonder who uses log:imcludes with rdf:_<n>.

18:45:22 <tim> I wonder who uses rdf:_<n>.

18:45:23 <dajobe> (isn't that owl:includes something now?)

18:46:01 * DanCon is kinda boggled at tim's position

18:46:06 <tim> (no, owl:incldues only works betwen owl:formulae, which are formulae which can'tr include any stamenets about other owl:formulae)

18:46:11 * tim is just exploring ...

18:46:27 <tim> sometime compatability is finding a real trasition path for real people in real posistions.

18:46:43 <tim> Introducing twop inon-interoperable versiosn in teh same spec is kind frighteneing too.

18:47:10 <DanCon> the RSS community uses idioms that the RDF specs turn into rdf:_n triples. I'd agree that most RSS consumers probably use XML parsers, not RDF triple thingies.

18:47:30 <bijan> Though, I'll point out, I do use RDF triple things for my RSS processing.

18:47:42 <bijan> And have been encouraging more people to do so.

18:47:46 <tim> In practice, whatever formalization you come up with, cwm users will need log:in and log:notIn to find out whether things are in lists/sets/bags, or oiterate over them.

18:47:51 <DanCon> If you want to say "well, RDFCore, don't worry about RSS support too much" then I wonder what sort of backward compatibility we're interested in at all.

18:47:53 <bijan> (Just a data point, I can adapt.)

18:47:54 <LotR> people on this channel probably don't count :)

18:48:24 <tim> :)

18:48:41 <tim> You have been warned!

18:48:45 <DanCon> well, the question is: do people on this channel count, or not? i.e. do we want to encourage folks to use RDF tools to slurp up RSS (which, btw, helps with evolvability of RSS), or not?

18:48:50 <bijan> WEll, not everyone I tell "Oh use a triplestore to do your RSS processing" are on this channel.

18:49:23 <tim> peole on this channel don'tcount in tat they are finite in number and could maybe be convinced to change their code for teh better future.

18:49:44 * dajobe surprised at tim's junking rdf:Seq

18:49:46 <Seth> DonCon, i read RSS with rdflib. Please don't deimplement on me !

18:49:47 <tim> they do count in that they are in a posistoin to have views on what future would be better.

18:49:48 <DanCon> RSS is interesting because of its momentum, not just its mass. i.e. RSS content tends to move around the network, not just sit there.

18:50:10 <tim> Seth, do you slurp up RDF_<n> ?

18:50:15 <tim> Seth, do you slurp up rdf:_<n> ?

18:50:23 <tim> What do you do with them?

18:50:36 <tim> I know i can't write a cwm rule for rdf:_>n>.

18:50:42 <tim> s/>/</

18:50:51 <Seth> sailor reads it just fine ... give me a rss file and ill post a screen shot.

18:50:59 <DanCon> no? have you looked at http://www.w3.org/2000/10/swap/test/seqiter.n3 timbl?

18:52:02 <DanCon> in fact, the use of <li> in RSS is only to make up for the fact that RDF is unordered, but RSS semantics are ordered. I think most folks use the order in the XML syntax, not the _n, in practice.

18:52:19 <tim> I have. As I said, peeking into URIs should be for special circumstances only, not to detremine list membership.

18:52:23 <DanCon> even folks that use triplestores probably just look for all the items, and don't pay attention to the order.

18:52:27 <Seth> anybody got a quick rss file for me to slupr?

18:52:43 <DanCon> try http://www.w3.org/2000/08/w3c-synd/home.rss

18:53:09 * DanCon notes the file extension... re recent discussions of .owl vs .rdf vs .txt

18:53:24 <dajobe> ahem

18:53:30 <DanCon> Content-Type: application/xml

18:53:44 <LotR> DanCon: yes, why is that?

18:53:49 <DanCon> dunno

18:54:14 * DanCon should grab a bite before 2p meeting

18:54:22 <tim> Actualkly, I have no problem with keeping Seq as syntax .. the problem is with having 2 concepts of list.

18:54:24 * sandro is wondering how log:notIn could work. For that, you need to to know both (1) that the collection is closed and (2) that the subject is not equal to (is distinct from) each item in the collection. Do you expect you'll usually have that distinctness information?

18:55:31 * tim only for constants.

18:55:51 <tim> "2" log:notIn ( "1" "23" "45" ).

18:55:54 <sandro> s/constant/Literals

18:55:57 <sandro> yeah, okay.

18:55:58 <tim> yes

18:56:00 <Seth>http://robustai.net/sailor/screenshots/w3c-synd_rss.html

18:56:00 <dc_rdfig> F: http://robustai.net/sailor/screenshots/w3c-synd_rss.html from Seth

18:56:21 <sandro> (or things which have an unambiguous property which is (transitively) a literal.

18:56:36 <Seth> F:|Sailor screen shot of RSS file showing how rdflib parses li

18:56:36 <dc_rdfig> titled item F

18:57:28 <dajobe> F:see also more RSS 1.0 feeds and [http://www.redland.opensource.ac.uk/rss/|Redland RSS 1.0 viewer] with 5 parsers

18:57:28 <dc_rdfig> added comment F1

18:58:32 <bijan> DanCon: I do believe that RSS processors probably use XML order, but that's considered incorrect :(

18:58:45 <tim> I don't think the RSS vocabulary gains from having a sequence there. But that's by the by.

18:59:00 <bijan> the Seq is also there for indicateing what's actually in the channel.

18:59:17 <bijan> I.e., if it's not in the seq, it shouldn't be consdiered part of the channel

18:59:23 <bijan> Oops. meeting.

18:59:47 <LotR> what's the use of it being there if it isn't in the channel?

19:01:08 <DanCon> consider a file that describes 2 channels.

19:01:31 <tim> Well, how do you know that the "item", say doesn't have a another "link"?

19:01:46 <tim> They linsk from an item don't seem to need a sequence.

19:02:04 <tim> What's there is there, and what ain't ain't.

19:02:34 <tim> The docuemnt is deemed definitive to the extent if it aint ther it aint a w3Crss event.

19:03:35 <tim> But in generalis *is*useful to have a sequence.

19:05:00 * LotR wonders why you'd want to get two channels from the same url

19:09:27 <Morbus> LotR: I've seen that happen with people who misunderstood the whole poitn.

19:09:49 <Morbus> most recently, a feed creator had stuck multiple <channel>s in one file in a fit of "reorgnization".

19:10:09 <Morbus> since, to them, they brought together multiple feeds of thought into the main HTML page, so thus, the same should hold true for the RSS channel.

19:10:40 <dajobe> Morbus: maybe you should note to people, you wrote a big RSS aggregator

19:11:36 <Morbus> uh, I wrote a "big" RSS aggregator called AmphetaDesk.

19:11:39 <Morbus> hello :)

19:12:25 <Seth> F: woops sailor(or rdflib) is not processing the frags correctly on '<rdf:li rdf:resource="http://www.w3.org/News/2002#item108"/>'

19:12:26 <dc_rdfig> added comment F2

19:14:26 <Morbus> before a week ago, I actually never knew the purpose of the :Seq part of rss 1.0.

19:14:33 <Morbus> but, it makes absolutely perfect sense to me now, and I welcome it.

19:14:42 <dajobe> I was just playing with 1000 "RDF" rss.10 feeds. so far 10% of them are not even XML. Next step to see how many are RDF/XML

19:15:18 <Morbus> dajobe: i see the same sorta thing. i read about 120 feeds on a regular basis, and about 10% are broken. sometimes they're fixed the next day, etc.

19:15:28 <dajobe> I expected that

19:15:30 <Morbus> the biggest issue seems to be the whole encoding.

19:15:32 <dajobe> do you do XML linting

19:15:33 <dajobe> indeed

19:15:56 <Morbus> dajobe: I don't, no. I used to, back in AmphetaDesk v0.92, but in v0.93, its a working parse or nothing.

19:16:52 <dajobe> nice, but harsh :)

19:17:30 <LotR> dajobe: I bet it helps get things fixed

19:17:50 <Morbus> yeah, dajobe, there has been a bit of a backlash on it.

19:17:51 <LotR> (if amphetadesk is popular)

19:18:15 <Morbus> i basically said "hey, if the mailing list wants it, I'll put up a you-must-install-yourself path, but if not, then I have no intention of throwing it back in".

19:18:31 <Morbus> i got some poo poos, but ultimately, it was "we should fix the feeds, not patch the parser".

19:18:54 <Morbus> "must-install-yourself pat*c*h", not path.

19:19:13 <Morbus> LotR: I'd say, honestly, that AmphetaDesk is #2 behind Radio.

19:19:26 * dajobe just d/l it BTW

19:21:40 <LotR> heh. [off] should be a switch instead of a prefix

19:22:15 <dajobe> logger, off

19:23:45 <dajobe> not bad 76% are legal rdf/xml

19:23:50 <dajobe> maybe more

19:24:24 <Morbus> that much? wow. i didn't think it'd be that high.

19:24:35 <dajobe> approx

19:24:43 <dajobe> this is of the feeds marked as rdf/xml

19:24:46 <dajobe> of general rss, no

19:24:51 <Morbus> ah, gotcha.

19:25:00 <Morbus> you using redland to parse 'em?

19:25:09 <dajobe> yes but my parser may have bugs, ahem

19:25:20 <Morbus> heh, through Perl? or something else?

19:25:27 <Morbus> just curious. i'm always interested in Perl code.

19:25:28 <dajobe> no, direct from C

19:25:36 <Morbus> i'm trying to get Rael to work on a XML::RDF::Simple

19:25:36 <Morbus> ;)

19:25:47 <dajobe> lol

19:25:50 <Morbus> he had XML::Simple parsing into triples the other day, I think.

19:26:00 <dajobe> I'm putting redland into CPAN sometime

19:26:05 <Morbus> or no, XML::RSS, it was.

19:26:06 <dajobe> just not priority

19:26:11 <dajobe> yeah, that's an existing module

19:26:15 <dajobe> needs some TLC

19:26:25 <Morbus> badly. he keeps murmering about rewriting it.

19:26:29 <Morbus> i agree.

19:26:40 <Morbus> my cheapass XML::Simple code is three times faster than the XML::RSS code.

19:26:50 <Morbus> for reading/parsing.

19:26:57 <LotR> why not fix XML::RDF if you think you need a ::Simple? (/me really doesn't understand perl people's fascination with ::Simple modules)

19:27:04 <dajobe> hmm - perl i/f to raptor in cpan

19:27:09 <dajobe> nah, other things first

19:27:39 <Morbus> there isn't an XML::RDF, thats the thing.

19:27:52 <Morbus> and I'd rather see a threeliner like XML::Simple than a longer one like RDFStore

19:28:37 * LotR thinks XML::Simple is *so* evil

19:28:38 <las> frm

19:28:47 <Morbus> LotR: why's that?

19:29:38 * dajobe is quite pleased at having >700 legal RSS 1.0 feeds *at all*

19:29:44 <dajobe> good test data

19:30:04 <Morbus> dajobe: you getting your feed URLs from Syndic8?

19:30:10 <dajobe> rssengine

19:30:15 <Morbus> ah, that works too.

19:30:23 <LotR> you don't do xml with it, just text foo which may or may not be a valid xml fragment

19:30:37 <dajobe> so these aren't flavo[u]red feeds, but natural ones, IYSWIM

19:30:44 <Morbus> LotR: what are you talking about?

19:30:51 <Morbus> if it's not valid XML, XML::Simple chokes, as its supposed to.

19:31:06 <Morbus> dajobe++

19:31:14 <Morbus> ISWIM, oh yes <g>

19:31:21 <Morbus> er, UM.

19:31:25 <Morbus> ack. you know.

19:47:12 <pixel> pixel is now known as pixel-away

20:11:43 <danbri> danbri is now known as nadia

20:11:55 <nadia> nadia is now known as danbri

20:12:13 <Morbus> short bathroom break?

20:12:26 * danbri tried to register 'nadia's nick while the other nadia was logged off...

20:12:36 <danbri> ...but its been registered; don't know who by

20:13:01 <Morbus> danbri: /msg Nickserv info nadia

20:18:47 <danbri> that's nadia-nadia, not othernadia :)

20:19:11 <Morbus> oh. heh, heh. uh <g>

20:19:42 * danbri wonders if nickserv can store homepage urls and other such factoids

20:19:55 <bijan> Believe so.

20:20:54 <Morbus> yeah, it can.

20:21:07 <danbri> for homepage: do /msg nickserv set url http://rdfweb.org/people/danbri/

20:21:20 <Morbus> yeah. you can also link nicks too.

20:21:54 <danbri> EMAIL, ICQ UIN, GSM, PHONE, MASTER(nick), etc...

20:22:02 <danbri> email and homepage are the most useful

20:36:47 * dajobe remembers $today-1 year was 1st RDF Core f2f - apropos of nothing

20:37:21 <dajobe> boo, 1 day out

20:45:43 * tim responds to Aaron's response to his HTTP-URI page

20:45:57 <sandro> where...?

20:49:18 <tim> On its way to ww-rdf-interest and by revising the HTTP-URI page itself.

20:51:00 <sbp> www-tag

20:51:08 <sbp> http://lists.w3.org/Archives/Public/www-tag/2002Jul/0349

20:53:04 * danbri realises why both his Samba servers weren't reachable from the laptop

20:53:18 <danbri> (I'd pulled out the wireless card)

20:53:36 * tim under pressure to change his mind about an onion not being an information object. "Librarians can catalog anything from an onion to a dog's ear. " http://www.librarianavengers.com/library.html <- from google rssengine

20:53:40 * danbri takes the hint and winds things up for the day

20:54:02 <tim> So where's this master list of RSS feeds?

20:54:20 <tim> ah got it

21:19:51 <danbri> DanCon, can I check in my edited esw2pd.n3 over your old one?

21:20:04 <DanCon> sure.

21:21:25 <DanCon>http://www.w3.org/TR/2002/WD-owl-features-20020729/

21:21:26 <dc_rdfig> G: http://www.w3.org/TR/2002/WD-owl-features-20020729/ from DanCon

21:21:36 <DanCon> G:|Feature Synopsis for OWL Lite and OWL

21:21:36 <dc_rdfig> titled item G

21:21:44 <DanCon> G:W3C Working Draft 29 Jul 200

21:21:44 <dc_rdfig> added comment G1

21:21:52 <DanCon>http://www.w3.org/TR/2002/WD-owl-ref-20020729/

21:21:52 <dc_rdfig> H: http://www.w3.org/TR/2002/WD-owl-ref-20020729/ from DanCon

21:22:03 <burtonator_> burtonator_ is now known as burtonator

21:22:06 <DanCon> H:|OWL Web Ontology Language 1.0 Reference

21:22:06 <dc_rdfig> titled item H

21:22:10 <DanCon> H:W3C Working Draft 29 Jul 200

21:22:10 <dc_rdfig> added comment H1

21:22:13 <danbri> done.

21:22:16 <DanCon>http://www.w3.org/TR/2002/WD-owl-absyn-20020729/

21:22:16 <dc_rdfig> I: http://www.w3.org/TR/2002/WD-owl-absyn-20020729/ from DanCon

21:22:28 <DanCon> I:|OWL Web Ontology Language 1.0 Abstract Syntax

21:22:28 <dc_rdfig> titled item I

21:22:33 <DanCon> I:W3C Working Draft 29 Jul 200

21:22:33 <dc_rdfig> added comment I1

21:22:39 <DanCon> triplets!

21:24:57 <dajobe> 200? :)

21:25:53 <DanCon> G:see also: [diffs and logs|http://lists.w3.org/Archives/Public/www-webont-wg/2002Jul/0343.html], i.e. the staff contact tedium involved in publishing.

21:25:53 <dc_rdfig> added comment G2

21:27:10 <dajobe> G:just noticed rdf:Property and rdfs:Property - easy mistake

21:27:10 <dc_rdfig> added comment G3

21:28:16 <dajobe> G:no references section

21:28:16 <dc_rdfig> added comment G4

21:38:47 <dajobe> I:

21:38:47 <dc_rdfig>http://www.w3.org/TR/2002/WD-owl-absyn-20020729/

21:38:49 <dc_rdfig> OWL Web Ontology Language 1.0 Abstract Syntax

21:38:50 <dc_rdfig> (1:DanCon) W3C Working Draft 29 Jul 200

21:39:00 <dajobe> I:still uses own:List, owl:first etc. - must remember to note that

21:39:00 <dc_rdfig> added comment I2

21:39:19 <dajobe> I2:still uses owl:List, owl:first etc. - must remember to note that

21:39:19 <dc_rdfig> replaced comment I2

21:40:10 <dajobe> I:ooh, SEQ <item1> ...<itemn> maps to that form. ordering, hmm?

21:40:10 <dc_rdfig> added comment I3

21:43:19 <DanCon> I:yes, first/rest is an ordering mechanism. But note the semantics of things like unionOf aren't sensitive to the order.

21:43:19 <dc_rdfig> added comment I4

21:45:29 <dajobe> I:defns need more anchors to hang stuff on, point to. Especially when multiple things defined in a section. But 1st WD, that's how it goes

21:45:30 <dc_rdfig> added comment I5

21:45:42 <tim> I feel that when people model things (in the sense of build an ontology for them) there is great value in distinguishing between ordered and unordered data.

21:46:25 <tim> For example, it can really effect the oprimizations one does, and when comparing things the results are just totally wrong unless you have the right type.

21:46:47 <dajobe> I:multiple 'ntriples' on a single line - not allowed in current syntax

21:46:48 <dc_rdfig> added comment I6

21:47:33 <tim> As a language for expressing data, we need sets and lists. I'm happy to express a set as "the set whose members of there of this list but in any order".

21:47:39 <tim> s/As/In

21:47:42 <dajobe> I:doesn't cite ntriples, or link to it

21:47:42 <dc_rdfig> added comment I7

21:47:57 <DanCon> yeah, the larch LSL handbook has insert(e,C) for building sets, and precat(e,C) for building lists

21:48:17 <dajobe> I:"DAM+OIL"!

21:48:18 <dc_rdfig> added comment I8

21:49:09 <DanCon> ... and rules that insert(e1, insert(e2, C)) = insert(e2, insert(e1, C))

21:50:10 <DanCon> in this list but in any order: sounds like owl:oneOf, no?

21:50:38 <dajobe> H:appendix B, some unprefixed attributes

21:50:38 <dc_rdfig> added comment H2

21:51:41 * DanCon is working on N3 rules for oneOf. quite tricky to specify it completely

21:52:18 <DanCon> e.g. how to say that [owl:oneOf ("3")] owl:disjointWith [ owl:oneOf ("4")].

21:55:25 <DanCon> and that [owl:oneOf (:x :y :z)] s:sameClassAs [owl:oneOf (:z :y :x)].

21:56:46 <tim> butin any order: Well, onOf gives you a class.. I suppose that can be used as an unorded set.

21:57:33 <DanCon> I certainly think that :Set s:subClassOf s:Class.

21:57:47 <tim> So flattening a formula to triples would be F owl:oneOf ( [ rdf:predicateSymbol <x> rdf:objectLiteral "foo" ... ?

21:58:26 <DanCon> yes. (modulo use/mention bug with <x>)

21:58:47 <DanCon> jhendler, in case you're here to chump the new WDs, I did it already.

21:58:55 <DanCon> dajobe's going over them with a fine-tooth comb

21:59:00 <dajobe> lol

22:00:16 <dajobe> hmm, I guess rdfcore the WG should be poked with the announcement

22:00:49 <dajobe> wait for it to get on w3c home page/news I guess

22:01:07 <jhendler> DanCon, just caught up with the chat/chump, saw you did that - thanks, dajobe, also thanks

22:01:25 <jhendler> mainly just coming to decompress and unwind with a few friends :->

22:02:04 <DanCon> :)

22:02:19 <dajobe> post first-WD relief? :)

22:02:34 * DanCon studies Jos's work on oneOf http://www.agfa.com/w3c/euler/owl-rules

22:02:47 <DanCon> { :rule12i1 } log:implies { ?a :includes ( ) } . <- works nicely in a backward reasoner.

22:04:42 * tim squints at the upside-down excalamtion mark betwee ":rule12" and "1"

22:05:13 <DanCon> y'know, I was really poo-pooh on having separate OWL and OWL-lite, but here I am, going thru the OWL feature synopsis, trying to write an N3 specification... I breezed thru the lite stuff, and the first thing in the full-OWL section has me stumped for the whole afternoon.

22:05:18 <dajobe> irc isn't i18n-ized - you could be using different charsets

22:05:50 <DanCon> I think it's rule-twelve-eye-el

22:06:11 <DanCon> no, twelve-eye-one

22:06:37 * DanCon knocks off for the day...

22:08:35 <jhendler> dajobe - we had a first WD a while ago...but I guess this is the first REAL release..

22:08:45 <dajobe> oh, sorry

22:08:52 <dajobe> requirements yeah

22:09:13 <dajobe> I guess OWL(Lite) now is a thing you can point at

22:15:13 <danbri> congrats!

22:15:37 <danbri> what's daml:UnambiguousProperty now? I'll update a schema or two

22:15:56 * danbri shoudl read the spec

22:16:18 <jhendler> danbri - owl:FunctionalProperty (unless it is owl:InverseFunctionalProperty - I forget :->)

22:17:30 <jhendler> one of my students is gonna write a daml+oil to OWL perl script soon - when we have the semantics done, we're going to attempt to write a cwm program to do "real" mapping (i.e. try to get the semantics as close as possible) - unclear how easy or hard the latter is since all the issues are still open (sigh)

22:17:50 <jhendler> syntactically though, we're currently only looking at a few name changes - should be easy

22:18:46 <jhendler> dajobe - I note your comment re our using daml:collection stuff -- our plan is to chage to rdf: as soon as there is an official TR to point at that describes this -- I don't think one has been published yet

22:18:46 <dajobe> are the dozen or so issues in the abssyn likely to be resolved as the editors have put them?

22:19:08 <dajobe> jhendler: no problem, I was just noting for me to remember, and for WD editors to do next draft

22:19:56 <dajobe> I:"at the time of this writing" s/this// ?

22:19:56 <dc_rdfig> added comment I9

22:20:01 <jhendler> ummm, that's hard to answer - by definition those are issues that we are still discussing, so for now it is just editor's opinion.

22:21:51 <tim> It woudl be nice if the schema on line were upadted so that things would validate.

22:22:06 <tim> It would also be ncie if things lime the rdfcore test manifest would daml-validate.

22:23:34 <jhendler> tim - I don't think they will "daml valid" - but they sure should owl validate -- we've got someone working on that validator now. it is a bit dependent on the open issue of namespace names, but I suspect a couple of "if" statements could allow both ways for now

22:23:55 <jhendler> tim - what is the "schema on line" - do you mean the .../owl# document?

22:26:24 <jhendler> the document http://www.w3.org/2002/07/owl goes through the RDF Validator just fine, but gets a warning about the collection parsetype.

22:26:43 <tim> for owl: if we swicth to rdf:first it woudl be nice for the rdf schema to have first and lat in.

22:27:11 <tim> for tests: there is a test schema but the nsmespace doesn't use "#" so it won't validat as far as I can tell as one cnnot find the schema.

22:27:42 <tim> that's the rdfcore tests

22:27:50 <tim> Manifest.rdf

22:29:20 <jhendler> btw, I believe BBN has agreed to produce an OWL validator ...

22:30:01 <dajobe> hmm manifest test: namespace URI ends in a /

22:30:09 <dajobe> does that need Overview.rdf ?

22:30:23 <danbri> it needs a redirect ...

22:30:35 <danbri> some folk like to take a property name and feed it to http_get

22:30:53 <dajobe> actually that dir doesn't even exist

22:31:17 <dajobe> I can add that

22:31:51 <dajobe> uri is http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema/

22:33:55 <tim> I *love* to take a property name and feet it to HTTP GET.

22:34:37 <tim> Just make that http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema# and you'll be fine.

22:34:41 <dajobe> so /, or # - somebody please pick one ;)

22:34:46 * dajobe ducks

22:34:55 <danbri> so http://xmlns.com/foaf/0.1/mbox now redirects to a single schema doc

22:35:02 <tim> pop in a small testSchema.rdf file and you'll validate.

22:35:12 <jhendler> (anyone know a school looking for senior faculty hiring - I have told UM I'll quit if they propagate a new IP policy that disallows open source stuff -- now I gotta see if they'll call my bluff...)

22:35:28 <jhendler> (don't tell bijan :->)

22:37:31 <jhendler> dajobe - don't tell Tim, but I think the daml validator actually has a piece of code that replaces one with the other - forget which direction...

22:37:44 <dajobe> lol

22:37:54 <dajobe> there goes the global information space neighbourhood

22:38:36 <jhendler> dajobe - sometimes pragmatic interoperability has to win...

22:38:43 <tim> my semantic neighbourhood.

22:39:31 <tim> As in -- I use a computer name and it may or may not find eth one I amn interested in depending on the the pahse of the moon, bandwidth etc. Same thing can be sued for uris maybe.

22:39:34 <jhendler> hmm - is that the name of the sequel to Weaving the Web?

22:40:06 <tim> I won't write another bookm - next time I will write a car -- or maybe a donkey.

22:40:36 <jhendler> (JimH remembers the famous phase of the moon bug in MITs multics, but doesn't want to admit that he's old enough to have been a user)

22:40:46 * tim has been arguing about what a work can be for too long

22:41:23 <jhendler> I sort of like the idea of writing a donkey - a number of puns come to mind, but I'll skip them.

22:41:47 <jhendler> more seriously, I'm glad you've been fighting the good fight on that one - leaves some of the rest of us free to do other things...

22:43:16 * jhendler been too busy trying to get my university to acknowledge open source - that's fighting a fight that should have been over long ago - sigh

22:54:43 <jhendler> gotta go - gotta go see Annie again - the joy of fatherhood :-> ciao all

22:55:37 <tim> Hmmm MIT are very positive about open source

23:20:32 <dajobe> so I made http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema.rdf

23:20:41 <dajobe> but either we change the ns, or add a redirect

23:20:46 <dajobe> but not today

23:23:09 <tim> :)

23:23:15 <tim> Thanks for the schema.

23:24:30 <tim> I'm using the manifest anyway, even though it doesn't daml-validate, to generate tests for cwm's parsers.

23:37:53 <JosD> DaveB, I propose we use @prefix test: <http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema#> . instead of @prefix test: <http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema/> .

23:37:54 <Seth> dajobe, why would you want a description in your schema when you already got rdfs:comment or dc:description that means substantially the same thing?

23:41:56 <JosD> ... I meant w.r.t. http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema.n3

23:47:27 <tim> tim is now known as tim-away

23:52:37 <AaronSw>http://creativecommons.org/metadata/

23:52:37 <dc_rdfig> J: http://creativecommons.org/metadata/ from AaronSw

23:52:49 <AaronSw> J:|Creative Commons Metadata Project

23:52:49 <dc_rdfig> titled item J

23:53:26 <AaronSw> J:Uses RDF/XML to store metadata.

23:53:26 <dc_rdfig> added comment J1