Semantic Web Interest Group IRC Chat Logs for 2001-07-20

00:00:47 <wmf> I don't know what you mean; I didn't kill anything. after I rebooted yesterday, top just didn't work

00:01:02 <AaronSw> Oh, I see.

00:01:05 <wmf> it gave me a "no namelist" message and that was it

00:01:22 <wmf> you know something isn't right when even top doesn't work

00:01:45 <AaronSw> That happens on my Linux box when something hogs all the memory.

00:02:02 <wmf> but this was right after booting

00:02:10 <AaronSw> Very strange.

00:02:13 <wmf> indeed

00:03:08 <ajbusy> ajbusy is now known as ajmitch

00:19:47 <wmf> I don't understand why gcc is called cc on OS X

00:24:12 * sbp wonders if it's possible to create a UTM using Python...

00:26:11 * sbp wonders if anyone could make a programming language out of RDF

00:26:15 <sbp> s/could/would

00:27:34 <AaronSw> i'm sure it'd come in quite handy

00:27:40 <ajmitch> ajmitch is now known as ajbusy

00:28:25 <sbp> I doubt it

00:29:41 <AaronSw> The Semantic Web and programming languages are pretty close...

00:30:28 <wmf> do any prolog systems have RDF integration? I'd be surprised if not

00:30:33 <AaronSw> I dare you to implement a turing machine in log:implies :-)

00:30:41 <AaronSw> prolog: yeah, swi-prolog does

00:30:48 <sbp> Well, XSLT is Turing Complete, so that can be used to model any programming language, theoretically

00:30:49 <AaronSw> and there are some interfaces to XSB

00:31:50 <sbp> Hmm... a turing machine using log:implies...

00:34:36 <sbp> { :state :equals "0" . :tape ?place 's' } log:implies { ?place :becomes { ?place :plus "1" } . ?state :equals "1" } .

00:34:54 <wmf> uh oh, he's going to do it... :-)

00:35:06 <sbp> If CWM had a few more built-ins, it could probably be done

00:35:41 <sbp> s/:state/?state

00:40:25 * AaronSw wonders why all the other interesting W3C projects are dead...

00:40:34 <em> ???

00:40:41 <AaronSw> HTTP-NG, Distributed caching, Propagation, etc.

00:41:07 * em was actually partial to the web characterization effort...

00:41:08 <wmf> I've never even heard of Distributed caching and Propagation

00:41:20 <AaronSw> Just stumbled across it again: http://www.w3.org/Propagation/

00:41:48 <em> really? concept is been around for quite some time...

00:41:53 <AaronSw> Addressing, even, for that matter.

00:42:12 * em catches up with back log

00:42:23 <AaronSw> oh, it was mostly off-topic

00:42:29 * em lol re turing machine in log:implies

00:42:41 * sbp is actually working on it right now

00:42:49 * AaronSw has learned a thing or too from em -- always dare people to do impossible things. :-)

00:43:05 <sbp> I've come up with an 8 line turing machine in Python

00:43:13 <sbp> I wonder if it can be converted to RDF?

00:43:16 <AaronSw> Not that it's impossible...

00:44:01 * DanCon peeks in for a bit

00:44:02 <sbp> Here it is, for anyone interested:-

00:44:08 <sbp> t, s, p = [0], 0, 0

00:44:09 <sbp> def go(t, s, p):

00:44:09 <sbp> if s == 0 and t[p] == 0:

00:44:09 <sbp> t[p] = 1

00:44:09 <sbp> go(t, s, p)

00:44:09 <sbp> if s == 0 and t[p] == 1:

00:44:12 <sbp> t[p] = 1

00:44:12 <sbp> go(t, s, p)

00:44:14 <sbp> DanCon, just in time!

00:45:05 * em waves with one hand to danc

00:45:16 <sbp> t = tape (a list), s = state (required for TMs), and p = position, on the tape

00:45:26 * AaronSw wonders if em usually waves with two hands...

00:46:12 <sbp> oops, there should be an extra line in there to start the call in the first place. Bung it under the def

00:46:55 <em> AaronSw, yes usually :)

00:47:35 * em wonders if dajobe is running logger that ralphs had been hacking on

00:47:44 <DanCon>http://www.w3.org/2000/10/swap/test/sameThing.n3

00:47:44 <dc_rdfig> A: http://www.w3.org/2000/10/swap/test/sameThing.n3 from DanCon

00:48:00 <AaronSw> em, he updated it this morning

00:48:03 * em was wondering if danc chumped this yet :)

00:48:13 <DanCon> A:sameThing.n3 -- whee! cwm can merge nodes

00:48:13 <dc_rdfig> commented item A

00:48:39 <DanCon> A:|sameThing.n3 -- whee! cwm can merge nodes

00:48:39 <dc_rdfig> titled item A

00:48:39 * em has to admit this sameThing.n3 has open up new possibilities for applications

00:48:55 <em> hats off to danc and timbl on this

00:49:00 <AaronSw> cwm can merge nodes? awesome!

00:49:11 <sbp> He he he

00:49:39 <em> back in sec...

00:50:15 <DanCon> A:see also [forgetDups.n3|http://www.w3.org/2000/10/swap/test/forgetDups.n3]

00:50:16 <dc_rdfig> commented item A

00:51:15 <sbp> Pffffffff:-

00:51:15 <sbp> log:implies {

00:51:15 <sbp> :y a log:Chaff.

00:51:24 <sbp> why can't N3 have empty contexts? eh?

00:51:28 <DanCon> A:for command-line recipe, see [retest.sh|http://www.w3.org/2000/10/swap/test/retest.sh]

00:51:28 <dc_rdfig> commented item A

00:51:54 <DanCon> empty contexts? what's that got to do with anything?

00:52:10 * em now waves with both hands

00:52:12 <sbp> Then you could do { :x = :x } log:implies { } .

00:52:21 <AaronSw> umm, so cwm doesn't really merge nodes?

00:52:29 <DanCon> er... so what, sbp?

00:52:43 <sbp> so you could filter out the crap that way

00:52:53 <AaronSw> umm, not really

00:53:11 * DanCon doesn't follow sbp at all

00:53:41 * sbp gives in/up/out/shake it all about/+

00:53:54 <AaronSw> sbp, just because something implies nothing doesn't mean it isn't true

00:53:58 <DanCon> and no, cwm doesn't really merge nodes. but sameThin.n3 and forgetDups.n3 accomplish what folks want when they talk about "merging nodes"

00:54:16 <AaronSw> hmmph

00:54:21 <sbp> You can say it's a log:Truth, and then use -filter=x.n3

00:55:17 <AaronSw> Can we get cwm to memorize http://infomesh.net/2001/05/rdflint/drules.n3 somehow?

00:55:32 <AaronSw> hmm, i suppose i'll have to edit my cwm command line tool

00:55:45 <sbp> That'd be good, then you could just flag -drules

00:56:04 <sbp> Don't forget srules too

00:56:15 <AaronSw> ahh, good point

00:56:36 <sbp> python cwm.py srules.n3 drules.n3 > rules.n3

00:57:06 <sbp> python cwm.py x.n3 -rules > y.n3 = python cwm.py x.n3 rules.n3 > y.n3

00:57:13 <AaronSw> hmmph, but then it spits out the rules too ... ick

00:57:21 <sbp> Yeah! You can apply them

00:57:27 <AaronSw> apply?

00:57:36 <sbp> python cwm.py x.n3 -apply=rules.n3 > y.n3

00:57:43 <sbp> ta da, no more rules

00:57:45 <AaronSw> ahh, cool -- didn't know that

00:58:04 <sbp> it's like filter, but it adds the junk, rather than replacing it

00:58:26 <AaronSw> sbp, can you fix ?rules.n3 to use the right namespace?

00:58:28 <em> AaronSw, s/log.n3#/log#

00:58:40 <sbp> ???

00:58:47 <sbp> Oh, sorry

00:58:48 <em> namespaces for log

00:58:59 <sbp> TimBL and his damn namespace changes... :-)

00:59:01 <em> in drules and srules

00:59:01 <AaronSw> em, was that to me or sbp?

00:59:16 <em> err... whoever owns http://infomesh.net/2001/05/rdflint/drules.n3

00:59:20 <sbp> Alright, I'll do it now...

00:59:22 <AaronSw> that's sbp

00:59:29 <em> sbp, is that you?

00:59:32 <em> if so...

00:59:33 <sbp> Yep

00:59:47 <em> s/AaronSw/sbp :_

00:59:49 <em> :)

01:00:11 <sbp> Hmm... that's funny: my local versions are all different to the online stuff. Ugh

01:00:36 * em thinks sbp must have blinked

01:01:34 <em>http://purl.org/net/wilbur/

01:01:34 <dc_rdfig> B: http://purl.org/net/wilbur/ from em

01:01:47 <em> B:|Wilbur RDF Toolkit

01:01:47 <dc_rdfig> titled item B

01:01:56 <em> B: Wilbur is Nokia Research Center's toolkit for RDF and DAML (and XML),

01:01:57 <dc_rdfig> commented item B

01:01:57 <em> written in Common Lisp.

01:02:27 <sbp> Oh, just worked it out. srules and drules were replaced by the -lite versions. See http://infomesh.net/2001/05/rdflint/ for the latest files

01:03:53 <AaronSw> Do you have the right DAML namespace this time?

01:04:18 <AaronSw> Hey, cool! It worked

01:04:54 * sbp adds an obsolete notice on srules.n3

01:04:56 <sbp> What worked?

01:05:06 <AaronSw> Why -lite?

01:05:16 <AaronSw> worked: my little cwm commandline tool now does deductions.

01:05:38 <sbp> Because a lot of the rules were duplicated in the original attempts. I had to wrok out what rules were primitive, and what were derived

01:05:59 <sbp> So srules.n3 just has lots of unecessary stuff in it

01:06:00 <em> B: Wilbur is written in part by Ora Lassila, Nokia, editor of the RDF M&S recommendation

01:06:00 <dc_rdfig> commented item B

01:07:19 * em recommends sbp chump http://infomesh.net/2001/05/rdflint/

01:07:44 <sbp> I did so ages ago, I think

01:07:56 <AaronSw> you can always chump again

01:08:08 <AaronSw> http://infomesh.net/2001/05/rdflint/

01:08:08 <dc_rdfig> C: http://infomesh.net/2001/05/rdflint/ from AaronSw

01:08:11 * em might have missed it but suggests chumping again

01:08:14 <AaronSw> C:|RDFLint Tools

01:08:14 <dc_rdfig> titled item C

01:08:25 <AaronSw> C:By Sean Palmer

01:08:25 <dc_rdfig> commented item C

01:08:37 <sbp> C:Sean B. Palmer

01:08:37 <dc_rdfig> commented item C

01:08:51 * em wandering into other window to sart writing various upcomming talks

01:08:54 <sbp> :-)

01:09:07 <sbp> Go get 'em em!

01:09:22 <AaronSw> C:Tools to find inconsistencies, make deductions, and rules to implement DAML and RDF Schema properties

01:09:22 <dc_rdfig> commented item C

01:09:22 <AaronSw> ;-), sbp

01:09:52 <sbp> C:Mainly in Notation3 (i.e. convertable to XML RDF), with a bonus Prolog file

01:09:53 <dc_rdfig> commented item C

01:10:13 <AaronSw> Well, the contexts aren

01:10:19 <AaronSw> Well, the contexts aren't exactly convertable...

01:10:45 <sbp> log:quote :-)

01:11:15 <AaronSw> Hmm, your rules keep making cwm deduce everything is equal to itself... guess that's not so bad...

01:12:37 <sbp> You can quite easily filter them out

01:12:47 <sbp> But that's what I was saying about the empty contexts

01:13:33 <AaronSw> No, empty contexts don't work

01:15:06 <sbp> How about:-

01:15:07 <sbp> { { :a :b :c } log:implies { :God :does :Exist } } a log:Truth; log:forAll :a , :b , :c .

01:15:46 * AaronSw wonders if he can get cwm to do his geometry proofs...

01:17:54 <DanCon> depends on what sort of proofs.

01:18:15 <AaronSw> hmmph, http://www.w3.org/2000/10/swap/test/forgetDups.n3 uses all sorts of undefined prefixes

01:18:37 <DanCon> really? cwm.py is pretty sloppy about those.

01:19:04 <DanCon> forgetDups.n3 was cut out of sameThing.n3, so I'm not surprised I forgot to get the namespaces right

01:19:05 <AaronSw> yeah, rcs: and contact:

01:23:20 * DanCon checks in fixes

01:24:03 <DanCon> forgetDups.n3 1.2 is now self-contained

01:24:08 <DanCon> and has a bit of usage info.

01:24:17 <DanCon> sameThing.n3 description slightly tweaked.

01:24:24 <DanCon> in sameThing.n3 v1.3

01:38:49 <sbp>http://uwimp.com/eo.htm

01:38:49 <dc_rdfig> D: http://uwimp.com/eo.htm from sbp

01:39:16 <sbp> D:|Semantic Web Primer

01:39:16 <dc_rdfig> titled item D

01:39:23 <sbp> D:By William Loughborough

01:39:23 <dc_rdfig> commented item D

01:39:55 <sbp> D:From 70,000ft. Really good

01:39:56 <dc_rdfig> commented item D

01:42:11 <sbp> Gotta run

02:41:11 Disconnected from irc.openprojects.net (ERROR :Closing Link: logger[tatooine.ilrt.bris.ac.uk] by niven.openprojects.net (Ping timeout for logger[tatooine.ilrt.bris.ac.uk]))

02:42:04 Topic now RDF and Semantic Web Chat (home: http://www.w3.org/RDF/Interest/#irc blog: http://rdfig.xmlhack.com/)

02:42:04 Users on #rdfig: logger ajbusy em tav GabeW jang lasDesk urgen AaronSw ssweens DanC_tst dc_rdfig DanCon

02:42:04 <ChanServ> This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

03:18:38 Disconnected from irc.openprojects.net (ERROR :Closing Link: logger[tatooine.ilrt.bris.ac.uk] by niven.openprojects.net (Ping timeout for logger[tatooine.ilrt.bris.ac.uk]))

03:19:03 Topic now RDF and Semantic Web Chat (home: http://www.w3.org/RDF/Interest/#irc blog: http://rdfig.xmlhack.com/)

03:19:03 Users on #rdfig: logger ssweens_work ajbusy em tav GabeW lasDesk urgen AaronSw ssweens DanC_tst dc_rdfig DanCon

03:19:03 <ChanServ> This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

03:19:08 <DanC_tst> File "../../../2000/10/swap/cwm.py", line 1526, in query

03:19:09 <DanC_tst> existentials.remove(pair[0]) # Can't match anything anymore, need exact match

03:19:09 <DanC_tst> ValueError: list.remove(x): x not in list

03:19:32 <DanC_tst> hi Jan. hacking tonight?

03:20:33 * DanC_tst gathers that's just Jan's machine

04:06:16 * DanC_tst works around cwm bug...

04:25:20 <ajbusy> ajbusy is now known as ajmitch

05:01:40 <fitzix> hello

05:12:58 <ajmitch> ajmitch is now known as ajbusy

09:52:05 Topic now RDF and Semantic Web Chat (home: http://www.w3.org/RDF/Interest/#irc blog: http://rdfig.xmlhack.com/)

09:52:05 Users on #rdfig: logger jang ajbusy em tav lasDesk urgen AaronSw DanC_tst dc_rdfig DanCon

09:52:05 <ChanServ> This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

11:09:53 <dajobe> BLURB:Note on Mozilla impl. of containers

11:09:53 <dc_rdfig> E: Note on Mozilla impl. of containers from dajobe

11:10:03 <dajobe> E:[[All RDF containers (Bag, Seq, and Alt) have the same internal representation. This representation is a numbered list. The container type is a hint to the user about whether the order of elements in the list means anything]]

11:10:03 <dc_rdfig> commented item E

11:10:11 <dajobe> E:(from newsgroup article)

11:10:11 <dc_rdfig> commented item E

11:10:32 <dajobe> now I can find that again easily

11:33:45 <jang> that's the way I've done my containers, except I did the crappy explicit renumbering thing

11:36:17 <dajobe> E:it is more true now that container type is a hint, since RDF Core have hacked out the special grammar parts for containers; they are just nodes with types like any other

11:36:18 <dc_rdfig> commented item E

14:01:07 <barstow> barstow is now known as scribe

15:01:48 <scribe> scribe is now known as barstow

15:07:38 <AaronSw>http://www.msnbc.com/news/601586.asp?0dm=H17JT

15:07:38 <dc_rdfig> F: http://www.msnbc.com/news/601586.asp?0dm=H17JT from AaronSw

15:07:45 <AaronSw> F:|Annotea Makes MSNBC.com

15:07:45 <dc_rdfig> titled item F

15:08:13 <AaronSw> F:"ANNOTEA is an open-source initiative sponsored by the non-profit World Wide Web Consortium (W3C), the closest thing the Internet has to a governing body." *Hmmm.*

15:08:13 <dc_rdfig> commented item F

15:09:00 <AaronSw> Watch out W3C folks! (Will you be charging taxes now?)

15:11:21 <AaronSw> F:Uses Annotea as a starting point to talk about all sorts of Semantic Web stuff (RDF, Metadata, Web of Trust ...)

15:11:21 <dc_rdfig> commented item F

15:22:11 * danbri surfaces an off-channel discussion with DanC

15:22:13 <danbri> =====

15:22:15 <DanCon> danbri? speaking of anonymous nodes... I had something of a breakthru yesterday with cwm and anonymous nodes... I figured out how to deal with contact:mailbox being unambiguous

15:22:24 <danbri> Cool!

15:22:35 <danbri> Let me paste in our disagreement -> agreement...

15:22:45 <DanCon> how about mailing it to www-archive?

15:22:48 <danbri> sure

15:22:50 <DanCon> pasting here is likely to be awkward

15:22:57 <DanCon> (a pointer here is fine)

15:25:17 <danbri> done, http://lists.w3.org/Archives/Public/www-archive/2001Jul/0016.html

15:25:41 <danbri> re cwm, I saw your note on this, haven't dug into detail yet.

15:26:36 <DanCon> if you've got any particular node-merging puzzle, I'm willing to show you how to do it with cwm

15:26:51 <danbri> OK

15:27:02 <DanCon> it's pretty simple. kinda brute-force. the surprising bit is that it finishes before the sun burns out.

15:27:32 <danbri> theres a scenario and some rdf/xml in http://rdfweb.org/2001/01/design/smush.html

15:27:43 <danbri> not 100% sure the example data's well formed...

15:28:03 <danbri> data: http://rdfweb.org/2001/01/design/examples.rdf

15:28:26 * DanCon reads smush.html ...

15:28:35 <danbri> not a particularly inspired scenario, but i wanted some tests for node-folding

15:28:45 <danbri> bogus namespace :(

15:28:57 <AaronSw> wow, your example data makes cwm explode wildly!

15:29:00 <DanCon> good point: [[[ This can be somewhat misleading, in that the resource identified

15:29:00 <DanCon> by such a node in an RDF graph is not in itself anonymous; rather it is the

15:29:00 <DanCon> mention of that resource in some chunk of data that is nameless. ]]]

15:29:40 <danbri> thanks; I think the phrase "anonymous resource" hasn't helped us much in understanding the problem...

15:30:07 <AaronSw> 0016 is a good read

15:30:49 * danbri thinks aaron has a strange taste in literature ;-)

15:30:51 <AaronSw> DanBri, your data isn't even valid xml

15:31:01 <AaronSw> literature: and don't you know it!

15:31:04 <DanCon> danbri, the maximum educational benefit is to be had if you try to work out the smush scenario yourself, given the stuff I chumped yesterday.

15:31:08 <DanCon> or I can do it.

15:31:17 <danbri> You do it!

15:31:25 <DanCon> ok...

15:31:33 <danbri> I'll find other ways to learn...

15:31:36 * AaronSw notes that Sean's tools have made this possible for a long time.

15:31:42 <danbri> Sean's tools?

15:31:47 * DanC_tst switches to faster machine...

15:32:06 <AaronSw> Sean's tools: http://infomesh.net/2001/05/rdflint/

15:32:17 <AaronSw> I've built them into my copy of CWM

15:32:27 <DanC_tst> made what possible? merging nodes? I noticed he has the substitution-of-equals-for-equals rules. I guess my innovation was more in forgetDups.n3 than in sameThing.n3

15:32:43 <AaronSw> right

15:34:01 * DanC_tst grabs smush examples...

15:34:06 <AaronSw> Danbri, which is the daml:UniqueProperty, or whatever?

15:35:32 <danbri> [[

15:35:32 <danbri> . Specifically, we make assume that... personalMailbox, personalHomepage and

15:35:32 <danbri> corporateHomepage are uniquely identifying properties. By this we mean that, for any given value of one of these properties,

15:35:32 <danbri> there exists 'at most one' resource with that characteristic.

15:35:33 <danbri> ]]

15:36:05 * danbri re-reads old smush doc

15:37:54 <DanC_tst> er... AaronSw, can you see why cwm is blowing up on these data? (after fixing the XML WF bug)

15:38:03 <AaronSw> it's not for me.

15:38:11 <AaronSw> do you have the -rdf prefix?

15:38:18 <DanC_tst> duh. thx.

15:43:49 <DanC_tst> ok... I think I have the nodes merged... now lemme encode the query...

15:49:31 <danbri> How's it going? are we waiting for cwm to terminate?

15:49:49 <AaronSw> no, it's failing for some reason...

15:53:04 <DanC_tst> I was debugging.

15:53:07 <DanC_tst> got it now:

15:53:09 <DanC_tst> <http://www.mozilla.org/> a agg:Q1Answer .

15:53:09 <DanC_tst> <http://www.w3.org/RDF/> a agg:Q1Answer .

15:53:09 <DanC_tst> <http://www.w3.org/XML/> a agg:Q1Answer .

15:53:17 <AaronSw> ahh, it didn't work because I can't spell!

15:53:34 <DanC_tst> debugging spelling mistakes is really a pain.

15:53:42 <danbri> wohoo!

15:53:57 * danbri impressed

15:54:11 <DanC_tst> other frequently made mistakes: * getting namespaces subtly wrong. * forgetting to declare variables

15:54:23 <DanC_tst> time python2 ../cwm.py --rdf smush-examples.rdf --n3 smush-schema.n3 sameThing.n3 --think --apply=forgetDups.n3

15:54:24 <DanC_tst> --purge >,xxx

15:54:24 <DanC_tst> # Purged 16 statements with...<#_g2>

15:54:24 <DanC_tst> # Purged 12 statements with...<#_g3>

15:54:24 <DanC_tst> real 0m2.546s

15:54:24 <DanC_tst> user 0m2.510s

15:54:26 <DanC_tst> sys 0m0.020s

15:55:42 * DanC_tst checks in...

15:55:44 <danbri> feel free to checkin my example data for cwm tests

15:55:48 <danbri> :)

15:56:22 <DanC_tst> I credited sources in comments; should have used dc:source. maybe that can be your contribution ;-)

15:56:33 * DanC_tst chacls...

15:57:46 <AaronSw> Hmm, the merging stuff doesn't seem to work with --apply

15:57:50 <AaronSw> is that a cwm bug?

15:58:21 <DanC_tst> apply only applies rules once. merging is likely to need --think

15:58:43 <DanC_tst> note that this merging stuff is not a new cwm feature. cwm hasn't changed substantially.

15:59:10 <DanC_tst> the forgetDups.n3 thingy relies on log:uri and log:lessThan built-ins. So I guess log:uri is a big key to making it work.

15:59:15 <DanC_tst> in practice.

16:00:33 <DanC_tst> ok, danbri, so here's how I translated the "uniquely identifying" bit and the query:

16:00:37 <DanC_tst> . http://www.w3.org/2000/10/swap/test/smush-schema.n3

16:01:00 <DanC_tst> and here's the data, cleaned up: http://www.w3.org/2000/10/swap/test/smush-examples.rdf

16:02:31 <DanC_tst>http://rdfweb.org/2001/01/design/smush.html

16:02:31 <dc_rdfig> G: http://rdfweb.org/2001/01/design/smush.html from DanC_tst

16:02:47 <DanC_tst> G:|aggregation strategies

16:02:47 <dc_rdfig> titled item G

16:02:52 <AaronSw> Hmmph, I want cwm to be able to do this stuff seeminglt natively

16:03:16 <DanC_tst> G:a sort of challenge question to the RDF implementor community by danbri et. al. Jan 2001

16:03:17 <dc_rdfig> commented item G

16:04:30 <DanC_tst> G:solved by DanC et. al. today: [smush-examples.rdf| http://www.w3.org/2000/10/swap/test/smush-examples.rdf] is the data with some XML WF bugs fixed. [smush-schema.n3|http://www.w3.org/2000/10/swap/test/smush-schema.n3] encodes the uniqueness constraints and the query.

16:04:30 <dc_rdfig> commented item G

16:04:37 <DanC_tst> I guess I should publish the output...

16:07:41 <DanC_tst> G: command line: python2 ../cwm.py --rdf smush-examples.rdf --n3 smush-schema.n3 sameThing.n3 --think --apply=forgetDups.n3

16:07:41 <dc_rdfig> commented item G

16:07:41 <DanC_tst> --purge

16:07:49 <DanC_tst> G: --purge

16:07:49 <dc_rdfig> commented item G

16:15:25 * danbri looks at http://www.w3.org/2000/10/swap/test/smush-schema.n3

16:16:18 <danbri> OK, time to install python on my laptop

16:16:59 * AaronSw declares victory!

16:17:48 * danbri tries to clarify his view on anonymous resources: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Jul/0266.html

16:17:54 <sbp> Is it just me, or is the fact that "[ :x :y; :a :b ] = [ :x :y; :c :d ] . :x a daml:UnambiguousProperty ." is smeggingly obvious???

16:18:13 <sbp> And the RDF in http://rdfweb.org/2001/01/design/examples.rdf is wrong according to th IE6 parser

16:18:21 <AaronSw> we know that

16:18:24 <danbri> known bug; will fix.

16:18:35 <danbri> First: which vresion of python ought i to be downloading for win98?

16:18:42 <sbp> 2.1

16:18:49 <AaronSw> they're all compatible, for the most part

16:19:00 <danbri> 2.1.1?

16:19:29 <AaronSw> Oh, is that out?

16:19:41 <danbri> yes, http://www.python.org/ftp/python/2.1.1/

16:19:50 <AaronSw> no, it'sonly in RC

16:19:55 <AaronSw> release candidate

16:20:11 <sbp> 2.1 is the latest stable thing

16:22:05 <danbri> DanC: nice work with the aggregation demo :)

16:22:14 <danbri> the .n3 is pretty readable too

16:22:30 * sbp is slightly miffed that sameThing.n3 clearly re-does the work of RDF Lint

16:22:34 <DanC_tst> G:query is now split out as smush-query.n3

16:22:34 <dc_rdfig> commented item G

16:23:10 <DanC_tst> well, RDF lint and sameThing.n3 both re-do the work of the DAML axiomatic semantics, no?

16:23:37 <sbp> That is true, but the DAML-AS isn't in N3

16:23:42 <DanC_tst> and I didn't see RDF lint before I did sameThing.n3. but I can add a link.

16:23:44 <sbp> Hmm... that's a task!

16:23:59 <sbp> Nah, don't worry about it. I'm suprised you didn't see it though

16:24:08 <DanC_tst> DAML-AS not in N3: not yet. not completely, anyway. I have started automating the translation.

16:24:34 <sbp> Neat. It's a shame it's all in KIF. Are there are really decent KIF to RDF tools around?

16:24:49 <DanC_tst> RDF to KIF: umm.. I started hacking something...

16:25:16 <DanCon> 2001/05dax/KIFgrok.py

16:25:21 <sbp> Love the rant about Anon nodes to www-archive, BTW :-)

16:25:52 <danbri> pat hayes on KIF (to DAML ctte list): [[ Do not put too much effort into KIF until we get the new standard

16:25:52 <danbri> written. ISO-KIF will be considerably easier to interface to than

16:25:52 <danbri> current KIF (UTF-8 character codings, simpler syntax, coherent

16:25:52 <danbri> semantics, much more flexible & expressive notation, better treatment

16:25:52 <danbri> of sequence quantifiers.)

16:25:53 <danbri> ]]

16:25:59 <sbp> 404: http://www.w3.org/2001/05dax/KIFgrok.py

16:26:20 <sbp> ISO KIF - yes, might be sensible to wait

16:26:23 <DanCon> I haven't checked it in, I guess.

16:26:44 <sbp> Not essential

16:27:01 <DanCon> I'm pretty sure all those ISO-KIF features are pretty irrelevant to the DAML axioms.

16:27:08 <DanCon> The DAML axioms are pretty vanilla.

16:27:22 <danbri> having 'coherent semantics' sounds like a must-have

16:27:55 <sbp> Well, they're all there, in the DAML-AS; we just need them in RDF

16:28:07 <sbp> (to use them on a day-to-day basis)

16:29:00 <DanCon> the semantics of KIF are plenty coherent for the job of DAML-AS.

16:29:41 <danbri> coherence: there was some bug to do with lists?

16:40:52 <AaronSw> E:[Google Groups Archive of post|http://groups.google.com/groups?hl=en&safe=off&th=fad686d2703c52f,4&seekm=9ikma9%246qq1%40secnews.netscape.com#p]

16:40:53 <dc_rdfig> commented item E

17:28:07 <AaronSw> re the anonymous resources thing, i'm wondering whether it's wrong to interpret a use of a URI in RDF to mean something like:

17:28:22 <AaronSw> there exists a resource with URI foo, etc.

17:28:36 <AaronSw> And anonymous resources simply omit the part that says "with URI foo" .

17:29:28 <danbri> I'm very symathetic to that approach, it brings URIs out from "under the hood" and surfaces them as "just more information" in the data graph

17:29:48 <AaronSw> Right, exactly -- it's also hard to express with current syntax/mindset

17:30:12 <AaronSw> then the question becomes, do you get/lose anything by making up URIs for those that don't have them

17:30:24 <AaronSw> i think the answer is no

17:31:56 <danbri> you dilute what it means to be a URI, to be a name. If names are arbitrarily invented left right and centre, we'll end up creating a category of 'privileged name' to represent the old concept of 'publically known URI'

17:32:21 <AaronSw> Is that a bad thing?

17:32:55 <AaronSw> It seems rather useful, we've gone and taken all this junk and brought it into the open.

17:32:59 <danbri> it'd be a bad thing if publically understood names and arbitrarily invented names weren't mechanically distinguishable

17:33:23 <AaronSw> But they aren't now -- so lets make then distinguishable instead of pretending that anonymous nodes solve our problems.

17:33:39 <danbri> ??

17:33:45 <danbri> who'se pretending what?

17:33:54 <AaronSw> err, perhaps not pretending

17:34:20 <AaronSw> Instead of saying: anonymous nodes solve the "invented names" problem, make a way to describe invented names

17:34:32 <AaronSw> It will allow us to continue to use the XML/RDF syntax for all sorts of things.

17:35:12 <AaronSw> :a :b _:c . :x :y _:c . for example

17:35:28 <danbri> "anonymous nodes" are the problem

17:36:13 <AaronSw> umm, and i'm proposing a solution: give them URIs.

17:36:49 <AaronSw> and some don't like that solution because they want to use anonymous nodes for other purposes.

17:37:09 <AaronSw> perhaps I'm just chasing you in circles here, sorry

17:37:28 <danbri> I've been using var:34235264364354 sort of approach, it uses URI syntax but its a well known corner of URI space, so I can strip out these pseudo-URIs

17:38:13 <AaronSw> yes, i know -- we really need a better solution, though.

17:38:16 <AaronSw> I

17:38:36 <AaronSw> I've been using local #names and daml properties, as we cwmed earlier this morning

17:39:15 <AaronSw> But I think we need some way of saying "a :TemporaryNode" so that processors can feel free to discard it

17:39:24 <AaronSw> or at least its URI...

17:40:13 <danbri> circles: yeah, not sure we're getting anywhere this way. "give them URIs" is a vague sort of deed.

17:40:37 <AaronSw> Not really, but that's a different issue

17:40:50 <danbri> saying something is a :TemporaryNode sounds like Sergey's old proposal to have nodes in the graph of rdf:type rdf2:AnonymousResource.

17:41:17 <AaronSw> yeah

17:41:29 <danbri> The problem is that it looks like the type is being applied to the thing represented by the node (eg. a person or whatever), and not to the representation itself

17:41:45 <AaronSw> right... so flip it around

17:42:01 <AaronSw> :AnonymousNode :includes "URI-as-string"

17:42:23 <AaronSw> you could even use cwm rules on it...

17:42:52 <AaronSw> {:x log:uri :y. :AnonymousNode :includes :y} log:implies {:x a log:Chaff } .

17:44:06 <danbri> trouble with syntax; is :includes something like a liberal inverse of rdf:type that can put literals into classes (such as AnonResource)?

17:44:09 <dajobe>http://www.ietf.org/internet-drafts/draft-ietf-calsch-inetcal-guide-01.txt

17:44:09 <dc_rdfig> H: http://www.ietf.org/internet-drafts/draft-ietf-calsch-inetcal-guide-01.txt from dajobe

17:44:18 <dajobe> H:|Guide to Internet Calendaring, July 18 2001

17:44:18 <dc_rdfig> titled item H

17:44:55 <AaronSw> includes: I suppose so -- i mean, you can rename the terms and get the same effect

17:46:16 <danbri> The URI needs to be stringified otherwise it risks being picked up by 'sameFooAs.n3' or whatever and replaced with other URIs that denote the same thing. right?

17:46:35 <AaronSw> right

17:48:06 <AaronSw> So it's really more like... :TemporaryURI :includes "URI" .

17:48:21 <AaronSw> if we had literal subjects... :-)

17:48:40 <AaronSw> [ rdf:value "URI" ] a :TemporaryURI .

17:50:09 <danbri> yes

17:51:51 * danbri installs python; goes to fetch cwm goodies

18:50:00 <danbri> cwm fans: where do I get sax2rdf ?

18:50:14 <danbri> can't see it in http://www.w3.org/2000/10/swap/

18:51:15 <danbri> guessed it: http://www.w3.org/2000/10/swap/sax2rdf.py

18:57:15 * danbri gets cwm working on his win98 laptop with danc's rdfweb aggregation code

19:17:08 * danbri wonders what % of http://lists.w3.org/Archives/Public/www-rdf-logic/2001Jun/thread.html is about RDF and "rules"

20:10:42 * danbri updates http://rdfweb.org/2001/01/design/smush.html to point to #rdfig logs and cwm code

20:12:28 * dajobe shuffles, notes log schema will change

20:12:40 <danbri> the old urls safe?

20:12:52 <dajobe> yes

20:12:59 <danbri> :)

20:16:18 <AaronSw>http://www.infotoday.com/il2001/

20:16:18 <dc_rdfig> I: http://www.infotoday.com/il2001/ from AaronSw

20:16:29 <AaronSw> I:|Internet Librarian conference, Nov 2001, Pasadena, CA.

20:16:29 <dc_rdfig> titled item I

20:17:38 <AaronSw> I:I'm not sure what a virtual librarian is, but the conference includes """Navigating the Net,

20:17:38 <AaronSw> filled with experts over 3 days and a special emphasis on virtual reference"""

20:17:38 <dc_rdfig> commented item I

20:19:24 * AaronSw got disconnected somehow... reviews logs

20:19:34 <AaronSw> dajobe, why would the log schema change?

20:19:42 <AaronSw> do you mean the log: schema?

20:20:01 <dajobe> no, the foaf bits and pieces

20:20:17 <AaronSw> I:filled with experts over 3 days and a special emphasis on virtual reference"""

20:20:17 <dc_rdfig> commented item I

20:20:20 <AaronSw> ahh, i see

20:20:37 <dajobe> more irc things, event types etc.

20:21:38 <AaronSw> cool

20:22:05 <danbri> can you send a proposal to rdfweb-dev if there are things to go into the foaf schema?

20:24:44 <sbp> Hmm... there's not many results from http://www.google.com/search?q=%22Semantic+Web%22+RDF+URIs+triples+inferences+trust+Notation3

20:25:22 <danbri> aaron, i was replying to your msg re standardised genid generation, but my ssh connection folded. Does http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Jul/0188.html work for you as a counter-example?

20:25:23 <dajobe> I'm just noting that things will change sometime; Ralph and I want to add more types to do with irc

20:26:23 <AaronSw> danbri, i see them as two separate issues

20:27:01 <AaronSw> The issue you brought up is a good one, but the resolution of it is unrelated to whether or not we can generate ids.

20:28:22 <AaronSw> Do you disagree, danbri?

20:29:18 <sbp> Hmm... (regarding lists, from the chump) is it fair to say that [ a rdf:Bag; rdf:_1 :a; rdf:_2 :b ] daml:equivalentTo [ a rdf:Bag; rdf:_1 :b; rdf:_2 :a ] . ?

20:30:16 <AaronSw> ooh, that's a tricky one.

20:30:27 <AaronSw> I'd say no, seeing Dan Connolly's definition of equivalence the other day

20:30:36 <AaronSw> applying that would result in a contradiction

20:30:47 <AaronSw> Which is why RDF Bags aren't really bags.

20:31:07 <sbp> [ a rdf:Bag; rdf:_1 :a; rdf:_2 :b ] :weaklyEquivalentTo [ a rdf:Bag; rdf:_1 :b; rdf:_2 :a ] .

20:31:12 <DanCon> those are equivalent iff a=b.

20:31:24 <danbri> I didn't claim otherwise. What I show is a problem case for certain kinds of genid-generation strategy. Specifically those that use sameness-of-description to conclude things about identity of things described

20:31:31 <danbri> (replying to aaron)

20:31:33 <DanCon> oh... and only assuming bags are sorta magically closed at the end.

20:31:58 <sbp> Well they're not DAML equivalent then - but funtionally, I think they are

20:32:02 <AaronSw> right, the answer to both danbri and dancon is to timestamp anonnodes

20:32:07 * DanCon doesn't remember giving a defintion of equivalence.

20:32:17 <AaronSw> err, the code in samThing.n3

20:32:25 <sbp> timestamp? That's quite a neat idea

20:32:35 <AaronSw> s/samThing/sameThing/

20:32:35 <DanCon> timestamp?!?!?!? please, aaron, give in. Just let's use (exists (?x) (...)). It's a well-worn idiom. Why do you resist it so strongly?

20:33:17 <AaronSw> Are there any RDF tools that use that?

20:33:22 <DanCon> yes. lots of them.

20:33:23 <sbp> Does that prevent people from using URIs as variables in that way?

20:33:33 <AaronSw> ntriples2kif doesn't count, since it's not really an RDF tool

20:33:37 <AaronSw> I suppose I mean RDF parser

20:33:52 <DanCon> squish, cwm, n-triples2kif, all the existing tools I know of.

20:34:08 <AaronSw> cwm doesn't -- it uses generated IDs and log:forSome.

20:34:19 <DanCon> exactly, log:forSome.

20:34:33 <AaronSw> I have no issue with the use of log:forSome.

20:34:43 <AaronSw> I just don't want a special representation for the nodes.

20:34:50 <sbp> Except it should have been in some W3C recommendation...

20:35:05 <DanCon> (and log:forSome is sorta broken. pretending that existentially quantified variables aren't special is bogus. log:forSome makes the most sense when seen as special syntax, not just another RDF vocabulary term.)

20:35:29 <AaronSw> ok, so then cwm doesn't count either, apparently

20:35:43 <DanCon> yes, cwm does count. The inferences it draws are valid.

20:35:49 <AaronSw> danbri, how does squish do it -- you were saying earlier that you used var:foo URIs.

20:35:51 <sbp> You can't treat existentially quantified variables as being a special type of resource, surely?

20:36:01 <DanCon> aaron, you're just being bull-headed. I don't think I'm gonna try to convince you any more.

20:36:02 <AaronSw> dancon, I have no problem with the inferences (i.e. the semantics) it's the abstract syntax i care about

20:36:52 <AaronSw> I just want to see some running code that outputs this before I signoff on it. Is that too much to ask?

20:37:00 <DanCon> the abstract syntax is that of EC logic. This is a well-worn path. I don't know why you resist following it.

20:37:09 <DanCon> you have lots of running code.

20:37:30 <DanCon> "outputs it"? abstract syntax is, by nature, something that code doesn't output.

20:37:47 <AaronSw> that's true... hmm...

20:38:37 <AaronSw> I was under the impression that EC logic was a model theory/semantics... is that incorrect?

20:41:10 <danbri> aaron: libby's squish stuff ('Inkling') has some form of node-folding, and I think keeps track of genids. My use of var:... has been in gluing parsers (perllib, cara) to on-disk triplestores. I'd rather have an explicit representation of genid-ness but var: works with APIs that don't have a special representation for genids.

20:41:47 <AaronSw> right, that's my question -- what APIs/Parsers have special representations of gendis?

20:41:50 <AaronSw> err genids?

20:42:18 <sbp> I think that anon-nodes are just one of those bits of broken architecture that should have been though about before they were put in. Talk about inconsistent methods for internal representation...

20:42:24 <sbp> s/though/thouhgt

20:42:45 <sbp> ugh... s/thouhgt/thought

20:43:19 <danbri> I believe Jan's does. And EricP's (isGenID()). And mine.

20:43:23 <DanCon> broken? that's FUD. EC logic is quite mature.

20:43:55 <DanCon> it's been independently discovered a number of times; pls re-read CMSMCQ's "information factoring and dublin core" paper.

20:43:57 <sbp> Well, existenitally quantified variables aren't broken, just the lack of explanation about how to handle them in RDf M&S

20:44:16 <danbri> what's broken is RDF M+S's casual exposition of this bit of the technology, not the technology itself

20:44:30 <sbp> Yep

20:44:56 <sbp> s/anon-nodes/the explanation of anon-nodes in RDF M&S...

20:44:57 <DanCon> if you're saying the spec is less than crystal clear, I can accept that. But any 2nd year C.S. student knows how to translate "there's a book; it's title is XYZ; it's author is ABC" into FOL. RDF just puts <>'s around that design.

20:45:24 <sbp> But it isn't clear how processors should handle them; look at the mess SiRPAC made originally

20:45:46 <DanCon> what's not clear? processors can do anything that's logically sound.

20:46:06 <AaronSw> OK, so I'll give in after you answer this question: what is the difference between an anonymous resource and a resource with a URI? Pat Hayes, who I trust on the subject, has said that you can deduce practically the same things from them.

20:46:18 <DanCon> "not clear" is a long way from "broken".

20:46:44 <sbp> Not all that far: if something isn't clear, then it may as well be broken

20:47:24 <sbp> If no one took the time to define what the structure of XML was, do you think the implementations would all be fine?

20:47:28 <DanCon> AaronSw, your question is badly formed; what I think you mean is: what's the difference between and existentially quantified variable and a skolem constant? i.e. two different ways to refer to some resource/thing. There's no difference between the resource/things themselves, right?

20:47:42 <AaronSw> Right.

20:47:52 <danbri> aaron: you're asking the wrong question. Resources aren't anonymous; the anonymity / namelessness is w.r.t. to a resource description not the thing described.

20:48:13 <AaronSw> Yes, I understand -- I was just using M&S' terminology.

20:48:17 <DanCon> ;-)

20:48:27 <AaronSw> What's the difference between an anonymous description and a description with a given URI.

20:49:17 * DanCon wonders how to reduce the whole proof theory discussion between Pat and myself to a line of IRC.

20:49:27 <danbri> a name.

20:49:37 <danbri> gzip.

20:49:40 * danbri ducks

20:49:45 <AaronSw> gzip?

20:49:51 <DanCon> Basically, I think Pat and I disagree on whether it makes any difference to skolemize away existentials. I've given up trying to resolve that in email. I need to phone him or get near a whiteboard or something.

20:50:10 <danbri> nevermind.

20:50:49 <AaronSw> Well, you and Pat resolve that then and I'll stay out of the line of fire.

20:50:59 <DanCon> in short, my position is: while a proof that "2l3kj2 > 5" (where 23lkj is a skolem constant) is widely accepted as a proof that "there exists ?x such that ?x > 5", the converse is not true.

20:51:46 <DanCon> to bring it a little closer to earth: the algorithm sirpac uses to generate names is no good for generating skolem constants *specifically* because it is well-known!

20:51:58 <DanCon> the nature of a skolem constant is that *nothing is known about it*.

20:52:27 <DanCon> i.e. uuids make reasonable skolem-constants. Once.

20:52:34 <AaronSw> I understand where you two disagree -- I've followed the discussion.

20:52:51 <AaronSw> Let me know when you work it out.

20:53:21 <sbp> Let us all know; we can plan street parties and so forth then

20:53:27 <sbp> :-)

20:53:45 <DanCon> basically, if RDF parsers want to use uuids to represent anonymous nodes, that's sort of ok -- internally. But if they write the graph back out, without some log:forSome-like cookie-crumbs, they lose information.

20:54:06 <AaronSw> PatH doesn't think so.

20:54:14 <DanCon> yeah, but he's wrong ;-)

20:55:13 <sbp> What does Pat think? That it's O.K. without the log:forSome quantification?

20:55:46 <DanCon> he seems to.

20:55:52 <AaronSw> Pat says that there's effectively no difference between an existential and a skolem constant, unless you're doing query in which case the existential becomes a universal.

20:55:57 <AaronSw> Or at least, that was my understanding.

20:56:03 <DanCon> but his position is elaborately stated, so don't take any 10-word summary of it from me as accurate.

20:56:24 <sbp> Yeah. On the surface of it, I'd have to say I agree with Dan; but you know Pat :-)

20:56:43 <AaronSw> Why do you agree, sbp?

20:56:46 <sbp> It'll be pistols at dawn if neither of you back down

20:57:24 <DanCon> I'd agree that there's no difference between an existential and a skolem constant, now that he's clarified some fine details about skolem constants; e.g. that generating a new skolem constant involves *changing languages*. My position is that you can't really write skolem constants in RDF documents.

20:57:29 <sbp> Agree? Because it doesn't seem to work intuitively to me if you skolemize quantified variables on the output and lose the information that they were indeed variables

20:58:21 <DanCon> ... because we already know the set of names in the language. you can't put something in the rdf:resource="..." attribute that isn't already in the language.

20:59:45 <AaronSw> yeah, that bugged me too

21:00:04 <danbri> DanC, could you live with something like a skolem:uuid:<big-number> URI scheme for parser output?

21:00:12 <danbri> (or purl subspace)

21:00:32 <DanCon> the information that ?x is a variable is in a skolem constant in that the skolem constant wasn't in the original language. But all URIs are in the RDF language. So there aren't any skolem constants that are URIs.

21:01:13 * danbri has to go

21:01:15 <DanCon> well, skolem: would have to be spelled so that it's not a legal URI prefix.

21:01:50 <DanCon> that's a solution to this issue in isolation, but couldn't live with it as part of the language design as a whole.

21:01:51 <AaronSw> And so we just ignore the stuff that RDF/XML can't represent?

21:02:00 <DanCon> ignore? huh?

21:02:18 <AaronSw> like _:a _:b _:foo . _:x _:y _:foo .

21:02:24 <DanCon> no, we recognize that the "anonymous" syntax in RDF *means something*.

21:02:35 <AaronSw> err :a :b _:foo . :x :y _:foo .

21:02:37 <DanCon> it's novel.

21:02:46 <AaronSw> novel?

21:03:06 <DanCon> n-triple syntax with _:name is fundamentally more expressive than without.

21:03:12 <DanCon> ah... I see what you mean by ignore...

21:03:31 <DanCon> yes, unfortunately, RDF 1.0 is limited in that it can't express: :a :b _:foo . :x :y _:foo .

21:03:32 <AaronSw> ignore as in ignore the issue

21:04:04 <DanCon> just like RDF 1.0 can't express "abc" log:lessThan "def".

21:04:04 <dajobe> note that n-triple doc punts on all meaning of _:name

21:04:22 <DanCon> the n-triple doc punts on all meaning. full stop, no?

21:04:38 <dajobe> yes

21:05:06 <dajobe> so don't use n-triple concepts like _:name to explain meaning!

21:05:17 <dajobe> :)

21:05:20 <DanCon> did I do that?

21:05:25 <AaronSw> we're just using them to represent anonymous nodes, dajobe

21:05:47 <dajobe> OK, and still discussing what they meaan - just checking!

21:06:03 <DanCon> (for the nth time): on meaning, my position is represented by n-triples2kif.pl , combined with the wide-spread undersanding of KIF syntax and FOL.

21:06:14 <AaronSw> we know, we know!

21:06:19 <AaronSw> :-)

21:06:46 <dajobe> I *think* n-triplse might need a grouping construct too, to allow scoping of _:names ?

21:07:16 <AaronSw> I thought the grouping was the document

21:07:25 <DanCon> grouping in n3: nope. Not to capture any RDF 1.0 docs. (I don't think. I haven't 100% convinced myself either)

21:07:28 <AaronSw> Hi sswens

21:07:46 <dajobe> rdf/xml does give you an implict document scope, that's related to the issue about ID/about

21:07:55 <AaronSw> err ssweens

21:08:23 <DanCon> I'm pretty sure (exists (?x) (and (p ?x y) (exists (?z) (p ?x ?z))) can always be re-written as (exists (?x ?z) (p ...))

21:08:26 <dajobe> DanCon: maybe for 1.0 but when you start to talk about Existential, forall, etc. the _:names apply in some scope

21:08:42 <DanCon> yes: the scope of the n-triples doc.

21:08:43 <ssweens> hi aaronsw

21:09:17 <DanCon> throw in other scopes (besides document scope) and you've got N3. with {}s.

21:09:35 <DanCon> (ok, you'd have to throw in log:forAll too.)

21:09:41 <dajobe> indeed - I'm just checking that isn't a requirement for ntriplse test cases

21:10:04 <DanCon> I don' think so. I really should work out the proof that existential quantifiers can always be pushed out without changing the meaning.

21:10:36 <DanCon> i.e. write it down. I've worked it out to my satisfaction on a whilteboard 3 or 4 times now ;-)

21:10:38 <ssweens> aaronsw: how are you doing today?

21:10:45 <AaronSw> pretty good, you?

21:11:05 <AaronSw> what brings you to rdfig?

21:11:09 <dajobe> DanCon: could you update n-triples2kif.pl to handle # comments etc ? I've got to hand edit stuff to get it through

21:11:28 <ssweens> finally got some lunch and finally saying something in this channel ...so good ;)

21:11:37 <DanCon> I probably could. you probably could too, dajobe ;-)

21:12:04 <dajobe> I plan to work on my rdf/xml parser this weekend - *making it smaller by taking out containers*

21:12:08 * DanCon is debugging something else just now

21:12:18 <DanCon> taking out containers: whee!

21:12:25 <ssweens> i've been following rdf a little for a project at MS and am just genuinely interested...

21:13:03 <ssweens> aaronsw: it seems like this is the place to find the latest and greatest, so just watching for now

21:13:14 <AaronSw> cool

21:13:28 <AaronSw> rdf is fun...

21:13:54 <AaronSw> :-)

21:15:16 <DanCon> do check http://rdfig.xmlhack.com , ssweens.

21:15:45 <ssweens> aaronsw: very interesting and very challenging too...

21:15:57 <dajobe> n-triples2kif.pl now handles test.nt

21:16:13 <AaronSw> where do you get the graphics for http://www.idiosync.net/news/ btw, are they entered in by hand?

21:16:19 <DanCon> hunting for some common background... ssweens... what's your programming language/tool of choice? Are you into XML/XSL stuff? C++? perl/python? java? C#?

21:16:26 <ssweens> dancon: i have and will continue to ...btw - i love the blogger bot..has Winer seen it yet?

21:16:37 <DanCon> dunno

21:16:42 <dajobe> oh the irony

21:17:00 <AaronSw> dajobe: correction, your copy of n-triples2kif does.

21:17:24 <dajobe> that much is obvious

21:17:43 <AaronSw> well.... feel free to share

21:18:06 <ssweens> dancon: till now, C#, VB, javascript..but my favorite right now is python

21:18:29 <dajobe> would be emailing it to DanCon now AaronSw but I'm typing this to you instead

21:18:31 <DanCon> ah... impeccable taste, sweens. python is nifty.

21:19:20 <ssweens> dancon: i love it more and more, cross platform and it can do it all, i think!

21:19:33 <ssweens> aaronsw: some are downloaded when the channel is first loaded, but sometimes it takes some tweaking

21:19:57 <ssweens> aaronsw: ...it's an ftp script up to the server

21:20:01 <DanCon> have you followed the .NET integration with python? it looks like a hand-in-glove fit, but I haven't followed up on the details since I read the ActiveState announcement.

21:20:45 <ssweens> yeah, i've used it a bit, but their betas keep time-bombing on me and don't get a good chance to use it

21:20:59 <ssweens> the VS integration stuff that is

21:21:29 * ssweens is getting to know IDLE rather well in the meantime

21:21:42 * DanCon finds debugging non-trivial cwm rules/filter apps a nightmare.

21:24:23 <ssweens> dancon: ...and following up..i'm a big XML/XSLT user as well (although XSLT is on my bad side right now)

21:29:57 <ssweens> <off> aaronsw: how has your w3c expenience been so far?

21:32:17 * ssweens scrambling for irchelp.org :)

21:33:17 <sbp> 21:04:04 <DanCon> just like RDF 1.0 can't express "abc" log:lessThan "def".

21:33:36 <sbp> But as N3 uses data: for its literals, it should be possible

21:33:52 <DanCon> N3 uses data:? huh?

21:34:23 <DanCon> In N3, you just write it as is: "abc" log:lessThan "def".

21:34:23 <sbp> Hmm... no, according to the source, it says "should be using data:,application/n3..." or something

21:34:41 <DanCon> regardless of what N3 says, that doesn't change RDF 1.0 syntax.

21:34:50 * dajobe groans at data: appearing again

21:35:14 <sbp> O.K., but can't RDF use <rdf:Description rdf:about="data:application/xml,abc">...?

21:35:45 <dajobe> yes, since data: is a URI, nothing to do with RDF literals

21:36:03 <AaronSw> data:,dajobe%20better%20get%20used%20to%it

21:36:11 <sbp> Who said "abc" in RDF is an RDFS Literal?

21:36:13 <sbp> s/RDF/N3

21:36:21 <dajobe> RDFS does

21:36:27 <sbp> N3 doesn't

21:36:30 <dajobe> or if it doesn't AaronSw will jump on me

21:36:39 <AaronSw> RDFS doesn't define N3.

21:37:07 <dajobe> but sbp was saying [[ but can't RDF use <rdf:Description rdf:about="data:application/xml,abc">...?]] about *RDF* not N3

21:37:30 <sbp> I was asking about a mapping from "abc" in N3 to something in RDF M&S

21:37:47 <dajobe> Oh, no scrolling back it wasn't, oops. In which case, you are describing n3 not RDF.

21:37:48 * dajobe tunes out again

21:37:55 <DanCon> it's possible that we'll eventually decide that writing { data:,abc log:lessThan "def" } is interchangeable with writing { "abc" log:lessThan "def" }. I give it about 45% odds. Meanwhile, there are things you can't say in RDF 1.0 syntax (if you want most of the RDF readership to grok).

21:38:20 <sbp> Fair enough. 45% is good enough to put a bet on

21:38:46 <dajobe> DanCon: indeed we might, I give it less of a chance since I expect RDF 1.0 literal will be a pair of string, optional xml;lang

21:38:48 <AaronSw> sbp, going into business as an RDF bookie?

21:39:10 <sbp> Aaron: Perhaps

21:39:47 <sbp> RDF 1.0 will be a pair of strings? How so, dajobe?

21:40:08 <dajobe> look at the rdfcore wg logs sbp

21:40:20 <sbp> URI?

21:40:33 <sbp> Actually, I know the URI...

21:43:31 <sbp> So, option 2 from: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Jul/0197

21:43:33 <sbp> Yes?

21:43:33 <DanCon> I expect it'll be a pair of (type, string) where type includes stuff like lang, whether it's XML content or just a string, and maybe xml schema datatypes like integer, etc.

21:44:03 <sbp> Fine

21:48:01 <DanCon> I'd rather we just decided xml:lang was a mistake. But there's a lot of code that treats it as significant. I wonder if anybody depends on that part of the code/language.

21:48:24 <sbp> I very much doubt it at this point

21:48:37 <dajobe> it would be a very much bigger political mistake to take out language support

21:48:50 <sbp> from a WAI POV, being able to express the natural of a node in a DLG is very important indeed

21:49:08 <DanCon> RDF has quite good language support without the xml:lang hack.

21:49:26 <sbp> Yes, because it's extensible

21:49:37 <DanCon> if you want to say "the title of the book is in french" you can darn well say that! And from what I can tell, that's what folks do, in practice, without using xml:lang.

21:50:16 <sbp> It's weird, the WAI kind of stayed away (and still do) from the actual design of RDF, becuse there's not much to consider

21:50:23 <sbp> But we like to use it, of course

21:50:25 <DanCon> I agree that it's important to be sure/demonstrate that RDF has good language support. I disagree that xml:lang is critical or even cost-effective for doing that.

21:50:58 <sbp> Not xml:lang, any language support, as long as it is well noted in the recommendation

21:51:10 <sbp> That's the key - noting the accessibility feature

21:52:43 <sbp> cf. http://www.w3.org/WAI/PF/XML/gl-20010717#g4_6

21:53:40 <dajobe>http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/

21:53:41 <dc_rdfig> J: http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/ from dajobe

21:53:46 <dajobe> J:|RDF/XML Syntax Grammar Experiments

21:53:46 <dc_rdfig> titled item J

21:54:04 <dajobe> J:preview - still hacking on it

21:54:04 <dc_rdfig> commented item J

21:55:06 <sbp> "(Only change is to make it XHTML)", pardon?

21:55:20 <dajobe> sbp: passed through tidy

21:55:48 <sbp> Aha, cool

21:56:52 * sbp does some more WWW history research

21:57:15 <AaronSw> dajobe, you may want to use <del> and <ins> rather than using classes

21:57:39 * dajobe has obviously missed out on some later html tags

21:58:26 <sbp> <del> and <ins> are flow elements, BTW

21:58:50 <sbp> rather, they can be used as block or inline

22:00:14 <dajobe> are they HTML 4.01 - I can't find them

22:00:24 <sbp> Yes, they're in there

22:00:29 <DanC_tst> well, I finally debugged this business of generating an agenda for one meeting from a record of a previous meeting.

22:00:29 <barstow> dajobe - nice work!

22:00:50 <sbp> Hmm... http://www.w3.org/History.html misses out the stuff about TANGLE. I kinda liked the idea of TANGLE

22:01:24 <dajobe> barstow: thanks, I needed it to move on with my syntax issues, hopefully people can see where we are; the grammar is shrinking!

22:01:30 <AaronSw> dajobe: http://www.w3.org/TR/REC-html40/struct/text.html#edef-ins

22:01:39 <sbp> Eek, 4.0?

22:01:54 <ajbusy> ajbusy is now known as ajmitch

22:02:17 <AaronSw> fine: http://www.w3.org/TR/html4/struct/text.html#edef-ins

22:02:41 <sbp> Or: http://www.w3.org/TR/html401/struct/text#edef-ins

22:02:45 <barstow> dabjobe - what do you think about Jon's tree/XDuce-based grammar http://www.openhealth.org/RDF/RDFSyntaxFormal ?

22:03:09 * barstow wonders if he's supposed to be using the [off] prefix ...

22:03:17 <dajobe> nah

22:04:08 <dajobe> J:I'm not sure quite where I am heading grammar-wise; infoset terms seem like a good idea and the grammar getting smaller is encouraging

22:04:09 <dc_rdfig> commented item J

22:05:13 * sbp can't wait for RDF 2.0, personally

22:05:32 <dajobe> re jon's thing: he always makes a lot of reference to XML qnames but the infoset terms mean we can move away from that detail. Of course if we decided qnames/namespaces are in the model, his thing might look more appealing

22:05:41 * AaronSw neither, sbp

22:05:55 <dajobe> so the quicker we finish 1.0 revised, the better

22:06:00 <sbp> :-)

22:06:11 <sbp> Just say it's done, and move on

22:06:14 <dajobe> without loosing all enthusiasm on the way

22:06:25 <sbp> s/loosing/losing

22:06:27 <dajobe> sbp: so all the 30+ issues are easy to decide then?

22:06:35 <sbp> Just flip a coin

22:06:53 <sbp> Aaron can propose the head resolution, Dan the tail resolution

22:06:59 <sbp> you can flip the coin

22:07:10 <DanCon> I think if we're viscious about picking the simplest choice for each of the 30 issues, yes they're easy to decide.

22:07:22 <dajobe> excellent

22:07:47 <sbp> Well, some like "what is a resource in the RDF sense of the word" aren't so easy

22:08:01 <dajobe> so getting a quorate rdf core wg in a room and going through all the issue with, say, 5-10 minute guilloteen might decide a lot of them?

22:08:03 <barstow> dabjobe: A problem with all of these syntaxes is the propAttr rule - how do you say it represents any attr value pair but for e.g. the description rule shouldn't contain another rdf:Description rule. I hate fixing the syntax with prose!

22:08:09 <DanCon> re "what resource": it's not reducable to test cases. Just close it.

22:08:30 <sbp> Good idea. But I'd like to see the documentation in RDF M&S fiddled a little

22:08:36 <AaronSw> guillotine - are you bringing one to the F2F?

22:08:50 <dajobe> syntax - yes, I realise the prose issues. Hopefully we have a lot less but I can't see it being all in a formal way.

22:09:00 * dajobe can't even spell guillotinne

22:09:07 <sbp> Just because it's not something that directly affects code, doesn't mean that people won't get hung up on it. It's useful to know how it relates to the terms in RFC 2396

22:09:12 * barstow can't even spell RDF :-)

22:09:30 <sbp> Just call it R for short

22:09:31 <DanCon> documentation: I wonder if the existing spec is worth salvaging. Implementors need test cases. Users need tutorials. The existing spec has little value to either. Formally, we have to dot some i's and cross some t's, maybe, but I want to spend as little time on that as we can get away with.

22:09:58 <dajobe> that's why I'm building up a new document, not cutting down an old one

22:10:00 <sbp> Yeah. We need more tutorials and test cases, I certainly agree

22:10:12 <DanCon> useful: well, solving world hunger is useful too. But if it's not reducible to a test case, I don't believe the RDF Core WG owes the world an answer.

22:10:31 <sbp> Heh! Dan's quotable notable for the day

22:10:46 <AaronSw> Hmm, we'll get WL to write us a testcase for world hunger...

22:10:59 <sbp> I was just thinking that (or something like it)

22:11:03 <dajobe> WL?

22:11:06 <AaronSw> I suppose it'll have to use anonymous nodes.

22:11:09 <sbp> William Loughborough

22:11:28 * dajobe can spell that; I've even been to Loughborough

22:11:32 <sbp> author of http://uwimp.com/eo.htm

22:11:44 <sbp> You're one of the few people that can

22:12:15 <sbp> People always seem to mess up "Connolly" and "Loughborough"

22:12:24 <DanCon> I'm starting to wonder whether it's worth the time to educate the WG about the need for anonymous resources. I'd really hate to lose the expressive power, but I'm ready to do just about anything to put RDF M&S 1.0 to bed.

22:12:34 * AaronSw too

22:12:39 * sbp agrees

22:13:42 * AaronSw tries to think of what will be most effective for building consensus at the F2F... scribe duities, silly puddy, vats of slime, writing RDF parsers...

22:13:48 <DanCon> actually, if the WG decides anonymous resources aren't like ex vars, I won't lose expressive power. My tools will still treat it that way. I may lose a little interoperability with tools that take a more conservative view, but I'll live with that.

22:14:24 <dajobe> maybe a goal is: all issues get assigned owners

22:14:36 <AaronSw> Sounds like a good goal.

22:14:53 <DanCon> owners haven't proved all that valuable so far, myself included. :-{

22:14:55 <AaronSw> And all issues have writeups for some period afterwards.

22:15:01 <dajobe> and then people are encouraged to drive their issue to closure...

22:15:06 <DanCon> I'd shoot for: all issues get test cases or get closed as "too fluffy".

22:15:22 <dajobe> well that can be done by the owners

22:15:23 <sbp> The last person to close their issue has to pay some penalty...

22:15:26 <DanCon> having 30 issues discussed in parallel isn't likely a recipie for success.

22:15:59 <DanCon> ooh... the "last to close" penalty is interesting. But it dissuades folks from taking on big issues.

22:16:15 <sbp> People should be assigned the tricky issues at random

22:16:28 <DanCon> if we could have a "if nobody claims it, we're closing the issue" rule, I could live with "last one to close is a rotten egg."

22:16:31 <dajobe> did you see Ora's aboutEach writeup - I concluded he agreed with me - implementing aboutEach was hard, and he admitted he didn't get it fully done

22:16:38 <sbp> That's a good idea too

22:16:44 <dajobe> thus proposal: remove aboutEach

22:16:53 <DanCon> aboutEach: I won't miss it.

22:16:59 <AaronSw> me neither

22:17:04 <sbp> s/remove aboutEach/move it to a higher level

22:17:10 <dajobe> remove means that

22:17:17 <DanCon> I really only need RDF 1.0 to be a good interchange language. I don't need it to be easy to author.

22:17:35 <sbp> What you really need and what everybody else...

22:17:43 <DanCon> I want it to be easy to write tools that *read* RDF 1.0.

22:17:47 <AaronSw> DanCon, I'm tending to like N-Triples for that.

22:17:53 <AaronSw> err, interchange, that is

22:18:01 <sbp> Yeah, but it's not XML

22:18:16 <DanCon> n-triples won't work for interchange when the value of a property is, say, the abstract of a math paper.

22:18:25 <sbp> BSWL is surely the simplest for just getting triples from A to B using XML

22:18:31 <AaronSw> dancon: huh?

22:18:42 <AaronSw> oh, you mean embedded XML?

22:18:46 * dajobe nods

22:18:59 <dajobe> but we can update n-triples for changes like that if we decide them

22:19:19 <DanCon> BSWL is... how did TimBL put it... wanton reification. I don't want to have to write "subject dan property haircolor value blue". even in XML. I just want to write "dan haircolor blue".

22:19:45 <sbp> <Dan><haircolor><Blue/></haircolor></Dan>

22:19:59 <AaronSw> BSWL would be much cooler if XML supported </>

22:20:00 <DanCon> that's BSWL syntax?

22:20:09 <sbp> It's abbreviated BSWL

22:20:12 <DanCon> oh.

22:20:17 <sbp> Look at the BNF

22:20:17 <DanCon> I must not have read that far.

22:20:22 <sbp> :-)

22:20:31 <AaronSw> sbp, rearrange the order.

22:20:33 * DanCon grits his teeth over </>.

22:20:52 <AaronSw> is that a yes or a no grit?

22:21:08 <sbp> Hmm... <Dan><haircolor><Blue/></></> - quite neat

22:21:13 <DanCon> the lack of </> is a crime against humanity.

22:21:28 <AaronSw> ok, yes grit :-)

22:21:43 * sbp likes the quoteable notables today

22:21:52 <sbp> s/quoteable/quotable

22:22:36 <AaronSw> It seems to me that the lack of </> just made XML really unusable for all sorts of purposes.

22:23:29 <DanCon> </> is easier to read, write, store, support, grok, etc.

22:23:59 <DanCon> anyway... water under the bridge.

22:24:06 <sbp> Why on earth was it not allowed though?

22:24:19 <AaronSw> XML 2.0, right? :-)

22:24:32 <sbp> Ooh, RDF 2.0 and XML 2.0: great mix

22:24:57 <AaronSw> I just hope that RDF 2.0 is more likely than XML 2.0.

22:24:58 <DanCon> If XML is to succeed, it will never need a 2.0.

22:25:13 <DanCon> parts of XML 1.0 might atrophy, but I don't think we can ever re-deploy XML.

22:25:30 <sbp> True. Same thing with HTML

22:25:53 <DanCon> hmm...

22:25:53 <AaronSw> I wouldn't say that!

22:26:22 <sbp> No, things like <form> are too widely supported to get people to use XForms, for example

22:26:31 <DanCon> there's certainly a sense in which HTML stabilized at 3.2... reached the "good enough" point... to where the value of any change is just dwarfed by the cost.

22:26:37 <sbp> And GIF and JPEG just steamroll over SVG

22:27:05 <sbp> It's a shame that XML wasn't invented sooner

22:27:30 <DanCon> and yet if XML was invented sooner, folks would have said "why bother?". They weren't feeling the relevant pain yet.

22:27:47 <DanCon> or from another view: XML *was* invented sooner. 1986. and before.

22:28:03 <sbp> Yes... que sera sera

22:28:19 <dajobe> see the "father of XML" argument on xml-dev this week

22:28:24 <DanCon> I have high hopes for SVG.

22:28:37 <DanCon> not sure whether to call them expectations yet...

22:28:37 <dajobe> I think SVG has a good chance, since it is cool

22:28:41 <sbp> SVG: Adobe backing it is good, but I don't think that market penetration will go far enough

22:28:41 <dc_rdfig> Label SVG not found.

22:28:59 <sbp> It's not worth embedding in XHTML at this stage

22:29:02 <dajobe> re Adobe: phhht to them after what they did to that russian

22:29:10 <DanCon> the 200KB flash plugin vs. >2MB SVG plug-in is a huge barrier.

22:29:13 <sbp> Russian?

22:29:26 <sbp> Yeah, Flash is mopping the floor with other technologies, and that's a great shame

22:29:35 <DanCon> shame? why?

22:29:35 <sbp> In some ways, I really despise Flash. And JavaScript

22:29:43 <sbp> Because it's so inaccessible

22:29:58 <DanCon> hmm...

22:30:02 <AaronSw> see http://boycottadobe.com

22:30:16 <AaronSw> err http://www.boycottadobe.com/

22:30:33 <sbp> Really; the world would have been much better off without Flash (IMO)

22:30:39 <DanCon> do you think the folks who are creating inaccessable content would somehow create accessable content with the same effort put into SVG? accessability is hard work, and most folks just want candy, no?

22:31:02 <DanCon> creating inaccessible... with flash, that is.

22:31:30 <sbp> Maybe, but SVG makes it a lot easier to create accessible content for those that want to. And it makes it a lot easier to repurpose because people can get at the source

22:32:09 <sbp> Flash is basically a lost cause. They totally abandoned accessibility as a marketing ploy; felt they didn't need to account for it

22:32:25 <sbp> Keyboard interface and so forth is dreadful

22:32:32 <sbp> But I could rant and rant about Flash...

22:33:10 <DanCon> re adobe: the DCMA is to blame; i.e. the american people and our apathy. focussing on this event is perhaps productive, but aiming at adobe seems to miss the mark.

22:33:22 <sbp> - http://www.boycottadobe.com/, scary

22:33:40 <AaronSw> Yes, the DMCA was a major mistake.

22:34:04 * AaronSw wonders why people continually transpose C and M in DMCA and DCMI

22:34:51 <sbp> Heh: "It may surprise you that this website is not sponsored by Adobe."

22:38:30 <sbp> WikiTakeOverByXp

22:39:33 <sbp> Dan: why is your Wiki not working?

22:43:42 <DanCon> entropy of some sort, I expect, sbp. Which one?

22:44:39 <DanCon> btw... I dropped my palmpilot one time too many.

22:44:49 <DanCon> you probably won't see much more palmagent hacking.

22:44:51 <sbp> Don't get me started on entropy. http://208.190.202.42/cgi-bin/wiki.pl?HomePage

22:45:06 <sbp> No more palmagent hacking? Aw, but not much lost since I don't have one

22:46:32 <sbp> Then again, I like entropy: I can blame the state of my room on it

22:47:27 <DanCon> shoal, my desktop machine which once answered to http://208.190.202.42 was unplugged, accidently, last weekend.

22:47:37 <DanCon> various parts of it have not recovered.

22:48:02 <sbp> (any room I enter increases its entropy by a factor of ten, due to various crisp wrappers and plecturms being thrown around)

22:48:15 <DanCon> try http://64.219.128.96/cgi-bin/wiki.pl?HomePage

22:48:15 <sbp> Ugh. You might want to update your Wiki page

22:48:37 <sbp> Ah, neat; thanks

22:49:19 <dajobe> hey, that <ins>, <del> stuff works (taking works on mozilla=works)

22:49:37 <sbp> Heh: "hypertext editing in Amaya feels *so good* until kerpow! it crashes"

22:50:04 <dajobe> I ain't using amaya at this moment, I got burnt yesterday with it

22:50:59 <sbp> Amaya annoys be; apart from all of the minor bugs, the interface just seems like it's held together with bubble-gum

22:51:14 <sbp> When you compare it to the solidity of AOLPress, no comparison really

22:51:24 <sbp> s/be/me

22:51:34 <dajobe> sbp: there is a nicer gnome interface but it is less stable. dunno about win32

22:51:47 <DanCon> once you get used to the way ESC and Enter work in Amaya, it's hard to get by without it.

22:51:59 <dajobe> that bit was puzzling at first

22:52:20 <sbp> I dunno - I used it to edit the XWebL specification, and I could quite easily do short pages without it

22:52:43 <sbp> Not sure about long ones though. It is good for handling large documents. I still don't use it on W3C document though

22:53:11 <sbp> s/document/documents

22:54:00 * sbp moves to use string.replace(lineabove, 'document', 'documents').... O.K., maybe not

22:54:17 <sbp> Ahem...

22:55:44 * dajobe sends syntax doc to rdf core wg

22:55:59 <sbp> Hmm... it's a shame that mIRC doesn't allow you to paste to channels when you're not online, isn't it

22:58:16 * sbp sets up a local IRCD

23:25:20 <AaronSw> Hi there danbri

23:25:25 <danbri> 'ello

23:25:26 * dajobe waves

23:36:18 <danbri> aaron, got a minute to chat about cwm?

23:36:25 <danbri> (or anyone else who knows it)

23:37:10 <sbp> I'm here, but I don't know anything about CWM

23:37:36 <danbri> I'm wondering how to treat its matching capabilities like a query system that returns a table of variable-to-value bindings.

23:37:43 <danbri> eg. like a prolog...

23:38:22 <sbp> Can you give an example of what you want?

23:38:27 <danbri> sure

23:39:01 <danbri> does the .js in http://www.w3.org/1999/11/11-WWWProposal/rdfqdemo.html work for you?

23:40:18 <sbp> Yep

23:41:03 <danbri> so the style of interaction there is: give it a query, get back a table of results based on data + rules

23:41:14 <sbp> Oh, you just want something like -spart?

23:41:25 <danbri> what's that do?

23:41:33 <sbp> Just parse something and include -spart as a flag: you'll see

23:43:02 <danbri> er, tell me what to type! i tried it on the end of the aggregation demo discussed earlier, no visible effect

23:44:01 <sbp> It's meant to type out all the nodes as URIs and s p o

23:44:26 <sbp> Just include it as a flag in the command line: python cwm.py myfile.n3 -spart > outfile.n3

23:44:41 <sbp> -s/-p/-a/-r/-t all do particular things

23:45:52 * sbp wonders about a MeRS generator in Python

23:45:57 <sbp> Anyhoo, gotta run

23:46:17 <danbri> thanks! (belatedly...)