00:02:06 <sbp> @ http://www.bartleby.com/61/17/C0621700.html
00:02:13 <chumpster> A: cool. The American Heritage® Dictionary of the English Language: Fourth Edition. 2000. from sbp
00:02:30 <sbp> A:|Bartleby takes on the word "cool"
00:02:32 <chumpster> titled item A
00:02:36 <bijan> bijan (bparsia@login7.isis.unc.edu) has joined #swhack
00:02:48 <sbp> A::"""Jazz musicians who used the term are responsible for its popularization during the 1940s. As a slang word expressing generally positive sentiment, it has stayed current (and cool) far longer than most such words."""
00:02:50 <chumpster> commented item A
00:02:52 <sbp> wb, bijan
00:02:57 <sbp> *** bijan has quit IRC ("Leaving")
00:02:57 <sbp> <sbp> wow, he leaves really abruptly :-)
00:02:57 <sbp> <sbp> last time, I thought I'd pissed him off or something
00:05:23 <bijan> How's your clone coming?
00:05:32 <sbp> I haven't really worked on it all that much
00:05:46 <bijan> I don't quite understand it's limitations and capabilities.
00:05:47 <sbp> I've been distracted with lots of annoying little things, that I can't even remember doing
00:05:57 <sbp> that's alright: neither do I
00:06:02 <bijan> Heh.
00:06:15 <bijan> Have you benched against CWM?
00:06:46 <sbp> I'm too scared to. Actually, rm ran a large query through it, and it sat there for like ten minutes before he crashed it. CWM ran it in under 5 seconds
00:06:54 <bijan> Heh.
00:06:56 <bijan> Oh well.
00:07:08 <bijan> Get Computing with logic
00:07:37 <sbp> ah, that's what you recommended to me before... well, next time I splurge on some books :-)
00:08:01 <sbp> Hmm... I need a place to scrawl such things. The problem is, give me ten minutes and I will have forgotten it
00:08:20 <bijan> did you write your own n3 parser?
00:08:24 <sbp> OTOH, it's in the #swhack logs now, and I'm used to salvaging various bits of junk from there
00:08:27 <bijan> or crib on of dan's?
00:08:28 <sbp> own N3 parser: yep
00:08:41 <sbp> nope, it's 100% "me". i.e., not all that good
00:08:45 <bijan> Hand rolled or did you use a generator/framework.
00:08:54 <bijan> How are you storing triples?
00:09:01 <sbp> hand rolled. Python has good RegExping stuff...
00:09:06 <bijan> Bah.
00:09:10 <sbp> storing triples: in a Python list
00:09:14 <bijan> Ouch.
00:09:17 <bijan> that will *kill* you.
00:09:24 <sbp> it works remarkable smoothly
00:09:33 <sbp> pff: remarkably
00:10:02 <sbp> of course, I could just extend the built-in Python list type if I wanted too
00:10:51 <bijan> *really* bad idea
00:10:52 <sbp> because really, the store classes that I've seen in other Python APIs are just wrappers for lists. A store is really just a list, with some fancy "add" methods and whatnot
00:11:04 <bijan> eek.
00:11:05 <sbp> how would you do it? how do you store them in your clone?
00:11:16 <bijan> Well, in the clone I'm using rdf_db.
00:11:27 <bijan> WHich ultimately stores them in the prolog db.
00:12:13 <sbp> Hmm... so how does *that* store them?
00:12:31 <sbp> heh, I love J's "todo" list on http://kingprimate.com/
00:12:38 <bijan> Not sure.
00:12:45 <sbp> [http://kingprimate.com/2002/02/21.html#a355 for anyone reading these logs]
00:12:57 <bijan> Pretty much like an indexed rdbms.
00:13:25 <bijan> I.e, something like a bunch of big hashtables.
00:13:40 <sbp> Argh. Something else that I'll be going through once I learn a semi-respectible amount of Prolog
00:13:46 <bijan> (well, probably a more complicated datastructures)
00:13:47 <bijan> hmm?
00:14:08 <sbp> the rdf_db source, that is
00:14:22 <bijan> yes, it puts a fair bit of effort into qname expansion.
00:14:35 <bijan> And builds a cache for lookups.
00:14:44 <bijan> I suspect I screw it up hard, regularly.
00:14:45 <bijan> ;)
00:15:25 <sbp> have you run any performance tests on your clone yet?
00:15:41 <bijan> But, for example, I would definitely look at the various rdbms represenations of rdf proposals.
00:15:43 <bijan> Not really.
00:15:47 <bijan> Do you have a good test?
00:15:52 <bijan> Or even a bad test?
00:16:18 <sbp> I have that cool N3 test... but no really good query tests
00:16:26 <sbp> just a few transitive closure things, etc.
00:16:34 <sbp> * sbp looks through his tests
00:17:19 <bijan> I also point you (again) to kjbuckets
00:17:23 <sbp> Hmm... there are a fair few built-in tests. Are you planning on adding built-ins to your clone?
00:17:32 <bijan> And its SQL implementations.
00:17:45 <bijan> Yes, I suppose.
00:17:51 <bijan> Builtins seem necessary.
00:17:57 <sbp> * sbp reads http://www.chordate.com/kjbuckets/kjbuckets.html
00:18:00 <bijan> I'd like to work out more of the pure logic bits.
00:18:14 <bijan> And get a backward chainer in.
00:18:23 <bijan> Builtins are just predicates, yes?
00:18:27 <bijan> Oh, and lists
00:18:29 <bijan> And...
00:18:30 <bijan> Bah!
00:18:53 <bijan> Builtins are relatively uninterestings implementationally.
00:19:15 <bijan> Well,t he interaction between builtins and query is tricky.
00:19:28 <bijan> But aren't they basically call outs to python code?
00:19:47 <sbp> well, I managed to do it after looking at that old conversation between you and Tim
00:19:53 <bijan> Heh.
00:19:55 <sbp> and going through the Llyn source code a bit
00:20:15 <sbp> basically, it searches *before* the query for any built-in predicates
00:20:16 <bijan> Yes, I remember finally figuring out builtins.
00:20:19 <bijan> Yes.
00:20:22 <bijan> And orders.
00:20:26 <bijan> Do light ones first
00:20:29 <bijan> heavy ones after?
00:20:41 <sbp> then it whips out the triples that contain those predicates, does the query, and performs the tests afterwards, with the queueing, as you mention
00:20:51 <sbp> actually, it doesn't have as many states as CWM
00:20:59 <sbp> 15 seemed a few too many :-)
00:21:01 <bijan> What doesn't?
00:21:05 <bijan> Ah, yours.
00:21:06 <sbp> Eep
00:21:27 <bijan> Right, Eep/
00:21:39 <bijan> evidencing evil python
00:21:45 <sbp> heh, heh
00:22:39 <sbp> Entertainingly (un)elucidating processor, perhaps
00:23:11 <bijan> Eliding essential palmer.
00:23:59 <sbp> * sbp is all-essential
00:25:42 <sbp> * sbp goes to get food and drink
00:37:37 <sbp> @ http://news.bbc.co.uk/hi/english/entertainment/new_media/newsid_1834000/1834510.stm
00:37:45 <chumpster> B: http://news.bbc.co.uk/hi/english/entertainment/new_media/newsid_1834000/1834510.stm from sbp
00:37:49 <sbp> B:|Hundreds queue for Xbox in Japan
00:37:51 <chumpster> titled item B
00:38:57 <sbp> B::But, according to the article, the sales trailed off later in the day. Experts are predicting that it's going to be a tough three-way fight between Sony, Nintendo, and MS
00:38:58 <chumpster> commented item B
00:41:00 <sbp> B::Japan always get cool technological doohickies before we do. It's not fair
00:41:01 <chumpster> commented item B
00:42:01 <sbp> argh: http://news.bbc.co.uk/hi/english/sci/tech/newsid_1831000/1831843.stm
00:42:12 <sbp> they tell us about the map, but where's the link?
00:42:46 <sbp> http://mars.jpl.nasa.gov/mgs/ might help...
00:44:28 <sbp> ah, Malin: http://www.msss.com/
00:45:05 <sbp> got it: http://www.msss.com/mars_images/moc/moc_atlas/
00:46:56 <bijan> Off to eat.
00:47:10 <bijan> If you get into using kjbuckets for your clone, I'll kibbitz ;)
00:47:11 <bijan> bijan has quit ("Leaving")
00:48:08 <sbp> that's one against, then :-)
00:49:24 <sbp> @ http://www.msss.com/mars_images/moc/moc_atlas/
00:49:34 <chumpster> C: Mars Global Surveyor MOC2-299 Release from sbp
00:49:46 <sbp> C:|Malin: Mars Atlas Revisited
00:49:48 <chumpster> titled item C
00:50:56 <Morbus> Morbus (~Morbus@63.173.138.138) has joined #swhack
00:51:05 <sbp> C::An atlas of the red planet, compiled by [http://www.msss.com/|Malin] from hundreds of photos from [http://mars.jpl.nasa.gov/mgs/|MGS], into one big composite. All available over the 'net
00:51:06 <chumpster> commented item C
00:51:23 <sbp> C::via. an [http://news.bbc.co.uk/hi/english/sci/tech/newsid_1831000/1831843.stm|article from the BBC]
00:51:24 <chumpster> commented item C
00:51:27 <sbp> Hi there Morby
00:51:33 <Morbus> sbp, i need your help.
00:51:42 <Morbus> you say this UKish term, and I forgot what it swas.
00:51:46 <sbp> you need *my* help? me?
00:51:53 <Morbus> it's just like "Rubbish!" only different.
00:51:54 <Morbus> Bugger?
00:52:04 <Morbus> Is it Bugger? It's not bugger.
00:52:07 <Morbus> what is it.
00:52:13 <sbp> Bugger, shit, fuck it, smeg, bollocks, nads, arse...
00:52:17 <Morbus> all i keep thinking of is bullcocky, and that's not it.
00:52:20 <Morbus> bollocks!
00:52:29 <Morbus> yes, that was it. thanks./
00:52:31 <sbp> heh, heh
00:52:35 <sbp> np
00:52:38 <Morbus> [[[
00:52:41 <Morbus> <p>Rest assured, I'm not going to force you to do anything you don't want to. Some of you have heard of this compiling junk, and it turns your stomach. You're thinking "Wait, the Mac was easy! They added this old-fangled type-in-the-window thing, and now they want me to quack like a programmer?! Bollocks on you, I'm tellin' Mom!"
00:52:45 <Morbus> ]]]
00:52:54 <Morbus> :)
00:53:05 <sbp> heh, heh
00:53:15 <sbp> what's that going to be in?
00:53:22 <Morbus> part 5 <g>
00:53:46 <sbp> y'know, bollocks is taken by some to be moderate-swearing over here
00:53:54 <Morbus> i know :)
00:53:55 <sbp> [not me, though. I couldn't give a crap]
00:54:06 <sbp> heh, heh. All the better, then!
00:54:09 <Morbus> hehehehe
00:54:17 <Morbus> i hope no one from ora reads swhack
00:54:19 <Morbus> ;)
00:54:23 <sbp> Swhack exclusive!
00:55:10 <sbp> * sbp can't wait for this edition
00:56:21 <sbp> Your articles are becoming a beloved ORA institution, you know. People salivate upon hearing that there's going to be a new article coming from the scary Hemenway dude
00:56:28 <Morbus> i doubt it :)
00:56:33 <Morbus> maybe you, patti, and llllll.
00:56:34 <sbp> well, I do
00:56:40 <sbp> heh, yeah
00:56:45 <Morbus> and a few people who email me every so often.
00:57:25 <sbp> and the millions of comments that you got?
00:57:40 <Morbus> eh
00:57:42 <Morbus> * Morbus waves hands around.
00:57:55 <sbp> fly in your room?
00:58:21 <Morbus> no, i farted.
00:58:26 <Morbus> lol
00:58:28 <sbp> heh, heh
00:58:29 <Morbus> sigh.
00:59:11 <sbp> sbp has changed the topic to: Swhack: the original birthplace of toilet humour
00:59:40 <sbp> and the lipogrammatic /x70 game, of course
00:59:46 <Morbus> yeaaaah.
01:00:19 <sbp> that game was just so funny
01:00:29 <sbp> it'd be banned in most countries
01:05:51 <GabeW> GabeW has quit ("Client Exiting")
01:10:18 <sbp> heh: http://www.ananova.com/news/story/sm_527956.html?menu=news.quirkies
01:25:40 <Morbus> gotta love rousing discussions:
01:25:41 <Morbus> <Morbus> that's mine. since 98, babay.
01:25:41 <Morbus> <Morbus> what's the correct way to spell that elongated, inflected "bahbee!" phrasE?
01:25:41 <Morbus> <jillzilla> BaaaaaaaaaaaaayBEEE!
01:25:41 <Morbus> <Morbus> well, that one reminds me of those old deep bass'd 60s, 70s songs.
01:25:41 <Morbus> <Morbus> i think you'd have to remove a few 'a''s to get the one i'm looking for.
01:28:03 <sbp> heh, heh
02:25:36 <MorbusIff> MorbusIff (~Morbus@s86.terminal3.totalnetnh.net) has joined #swhack
02:31:17 <tansaku1a> tansaku1a (~sam@n144-001.tokyu-net.catv.ne.jp) has joined #swhack
02:40:26 <tansaku> tansaku (~sam@n146-214.tokyu-net.catv.ne.jp) has joined #swhack
02:48:35 <Morbus> Morbus has quit (No route to host)
02:57:49 <tansaku1a> tansaku1a has quit (Read error: 110 (Connection timed out))
03:51:42 <jeremiah_> hello
03:51:50 <jeremiah_> jeremiah_ is now known as jeremiah
03:52:24 <sbp> Hi
03:52:30 <jeremiah> how're you?
03:52:51 <sbp> I seem to be fine, thanks
03:52:53 <sbp> and you?
03:59:44 <MorbusIff> MorbusIff has left #swhack
04:22:57 <jeremiah> good
04:23:00 <jeremiah> sigh... I'm tired
04:24:06 <sbp> .time
04:24:07 <xena> 2002/02/23 04:25:35.1185 Universal
05:00:25 <sbp> Gotta run
05:31:26 <sbp> sbp has quit (Read error: 104 (Connection reset by peer))
05:31:32 <sbp-> sbp- (~sean@63.149.73.20) has joined #swhack
05:32:26 <sbp-> sbp- is now known as sbp
11:49:53 <tansaku> tansaku has quit (Read error: 110 (Connection timed out))
13:15:03 <tansaku> tansaku (~sam@n146-214.tokyu-net.catv.ne.jp) has joined #swhack
15:05:20 <tansaku1a> tansaku1a (~sam@n146-214.tokyu-net.catv.ne.jp) has joined #swhack
15:12:27 <tansaku> tansaku has quit (Read error: 110 (Connection timed out))
15:54:44 <tansaku1a> tansaku1a has quit (Read error: 110 (Connection timed out))
17:13:08 <tansaku1a> tansaku1a (~sam@n146-214.tokyu-net.catv.ne.jp) has joined #swhack
19:00:57 <bijan> bijan (bparsia@login1.isis.unc.edu) has joined #swhack
20:30:57 <xena> xena has quit (Connection timed out)
20:41:48 <Glenn_the_Great-AG> Glenn_the_Great-AG (~glenn@pool1-66.max2.ahn.dialup.athens.net) has joined #swhack
20:43:00 <Glenn_the_Great-AG> Glenn_the_Great-AG is now known as Glenn_the_Great-AGN
20:47:38 <Glenn_the_Great-AGN> System Information 12··> 13<·Operating System/Build: Windows 98 (4.10 - 2222)13·> 13<·Uptime: 5days 7hrs 15mins 21secs13·> 13<·Resolution: 1024x76813·> 13<·CPU Info: 1-AMD K7 (0.18 µm) {4} @~654.63MHz13·> 13<·Memory Resources: Usage: 330/384MB (85.94%) 12[4||||||||14||12]13·>
20:48:18 <Glenn_the_Great-AGN> Glenn_the_Great-AGN has quit ("Get it at http://chat.theliberation.com/ss <·SS4·>")
20:55:55 <sbp> heh, heh. Thanks for that :-)
20:56:00 <bijan> Hey sean.
20:56:09 <sbp> Hi bijan
20:57:38 <bijan> * bijan updates cwmclone
20:57:43 <bijan> Got the consult eof bug done.
20:57:47 <bijan> Fixed.
20:57:52 <bijan> Much less embaressing now :)
20:58:16 <sbp> what bug was that, then?
20:58:34 <bijan> The consulter wouldn't terminate.
20:58:37 <bijan> You had to cancel.
20:58:57 <bijan> If you had a certain sort of whitespace at the end it would infinite loop :)
20:59:03 <bijan> Tail recursively.
20:59:08 <bijan> So no stack blow out.
20:59:09 <bijan> And too fast.
20:59:13 <bijan> So no interrupt :)
20:59:18 <sbp> you mean "@stop." wouldn't work?
20:59:30 <bijan> No, this is in the file reader.
20:59:32 <bijan> I.e., consulting.
20:59:39 <sbp> ah
21:00:08 <bijan> Trying to do the document namespace and empty prefixes.
21:00:29 <bijan> So I can read a larger number of regular n3 files.
21:02:47 <Galahad> Galahad (xena@mewtwo.espnow.com) has joined #swhack
21:03:23 <bijan> If you do come up with a good "speed" test case, I'd love to see it.
21:03:28 <bijan> Preferably without builtins.
21:03:37 <bijan> I need to think about builtins, but they are on the list.
21:03:47 <sbp> Well, rm's test case might be fairly valuable. Let me try to find it...
21:08:06 <sbp> ah, in http://blogspace.com/swhack/chatlogs/2002-02-21.txt
21:08:13 <sbp> at 05:41:33
21:11:48 <bijan> BRB, got to get hot drink.
21:11:51 <sbp> O.K.
21:12:02 <sbp> I'll paste the N3 version into the channel in the meantime...
21:12:13 <sbp> @prefix : <#> .
21:12:13 <sbp> @prefix u: <util#> .
21:12:13 <sbp> @prefix log: <http://www.w3.org/2000/10/swap/log#> .
21:12:18 <sbp> this log:forAll :trk , :trkStruct , :has , :prop , :partStruct , 
21:12:18 <sbp>    :otherProp , :anonLine , :lineText .
21:12:23 <sbp> { :trk <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
21:12:23 <sbp>    <http://www.redmonk.net/rdf/lyric/elements/0.1/lyrics> .
21:12:28 <sbp> :trk <http://www.redmonk.net/rdf/lyric/elements/0.1/structure> :trkStruct .
21:12:29 <sbp> :trkStruct :has :prop .
21:12:32 <sbp> :prop <http://www.redmonk.net/rdf/lyric/elements/0.1/structure> :partStruct .
21:12:32 <sbp> :partStruct :otherProp :anonLine .
21:12:36 <sbp> :anonLine <http://www.redmonk.net/rdf/lyric/elements/0.1/linetext> :lineText .
21:12:36 <sbp> } log:implies {
21:12:43 <sbp> [ u:trk :trk; u:trkStruct :trkStruct; u:has :has; u:prop :prop; 
21:12:44 <sbp>    u:partStruct :partStruct; u:otherProp :otherProp; u:anonLine :anonLine; 
21:12:44 <sbp>    u:lineText :lineText ] } .
21:13:07 <sbp> * sbp had to toss-up between www-archive and #swhack :-)
21:22:49 <monokrom> monokrom (~chatzilla@208-58-246-147.s147.tnt3.atnnj.pa.dialup.rcn.com) has joined #swhack
21:22:57 <sbp> * sbp waves to monokrom
21:23:19 <monokrom> * monokrom waves back
21:24:19 <sbp> if you've not been in here before, I should note that this channel is publically logged to the Web, blah, blah, blah
21:24:58 <monokrom> * monokrom nods
21:25:01 <monokrom> gotcha
21:25:16 <sbp> logs are at http://blogspace.com/swhack/chatlogs/
21:25:34 <sbp> but we don't mid since we're all a bit nutty about recording information persistently...
21:25:47 <sbp> however, you can make comments off-log by prefixing them with "# "
21:41:31 <bijan> I have to say that I don't find n3 very perspicuous as a logic notation.
21:42:37 <sbp> many have noticed that N3 is a particularly quirky logic notation
21:59:03 <Glenn_the_Great-AG> Glenn_the_Great-AG (~glenn@pool1-66.max2.ahn.dialup.athens.net) has joined #swhack
21:59:56 <Glenn_the_Great-AG> Glenn_the_Great-AG has quit (Client Quit)
22:00:49 <bijan> X = user6722592
22:00:49 <bijan> Y = 'http://util#anonLine'
22:00:50 <bijan> Z = 'http://test/#j45234'
22:23:35 <AlexMax> AlexMax (~xircon@clt74-11-127.carolina.rr.com) has joined #swhack
22:23:55 <sbp> * sbp waves to Alex
22:24:00 <AlexMax> hey
22:24:06 <AlexMax> I was just told about this place
22:24:10 <Glenn_the_Great-AG> Glenn_the_Great-AG (~glenn@pool1-66.max2.ahn.dialup.athens.net) has joined #swhack
22:24:15 <AlexMax> We got haxored by one of you guys
22:24:17 <sbp> BTW, this channel is publically logged to the Web etc. To do an off-log comment, use "# "
22:24:20 <AlexMax> Heh
22:24:25 <sbp> haxored? Mmmkay
22:24:26 <AlexMax> Okie
22:24:39 <Glenn_the_Great-AG> Do any of you know a "Li0n7?"
22:24:51 <Glenn_the_Great-AG> I have reason to believe he has hung around here before
22:25:25 <sbp> er... this is not a cracking channel. it's a hacking channel
22:25:36 <Glenn_the_Great-AG> We are talking about hacking
22:25:43 <Glenn_the_Great-AG> Our main page was hacked by a Lion7
22:25:43 <sbp> ah, good
22:25:48 <AlexMax> Yeah
22:25:55 <deltab> he's never been here to my knowledge
22:26:00 <sbp> heh. no, I really do think you mean "cracking"
22:26:09 <sbp> I do recall a Lion7 coming in here briefly
22:26:10 <deltab> or if so under a different nick
22:26:14 <sbp> or something like that
22:26:32 <deltab> ah, Lion7, yes
22:26:43 <sbp> heh. Remind me to kick Aaron when he comes back for choosing a dorky channel name
22:26:49 <Glenn_the_Great-AG> When is the last time you saw Lion7 in here?
22:27:06 <sbp> dunno. I can grep the logs if you really want me to
22:28:19 <sbp> * sbp greps
22:28:19 <Glenn_the_Great-AG> I don't really have time to sort through them all
22:28:19 <AlexMax> Dude
22:28:19 <AlexMax> Grep is a unix command
22:28:19 <Glenn_the_Great-AG> Okay
22:28:19 <AlexMax> That (I think) searches for a specific text string
22:28:19 <AlexMax> in a file
22:28:19 <sbp> indeed
22:28:19 <deltab> 2002-01-27
22:28:19 <Glenn_the_Great-AG> I thought "grep" was supposed to be a cute slang word for "grab"
22:28:19 <AlexMax> Nah
22:28:19 <sbp> it'll take a minute or two, since we have quite a few logs...
22:28:23 <deltab> no, it's an abbreviation
22:28:27 <Glenn_the_Great-AG> Well, thanks anyway
22:28:28 <sbp> 2002-02-10 17:37:00 <Lion7> Lion7 has left #swhack
22:28:28 <AlexMax> Glenn, when you want to hang around these people, you need SOME prerequisites
22:28:33 <sbp> get regular expression
22:28:34 <AlexMax> ^_^
22:28:36 <deltab> g/re/p
22:29:04 <deltab> "for every line matching this regular expression, print"
22:30:17 <Glenn_the_Great-AG> I wouldn't know. I'm a Windows whore......
22:31:15 <Glenn_the_Great-AG> Well, I don't actually run the webpage
22:31:20 <Glenn_the_Great-AG> I'm a forum administrator
22:31:26 <sbp> ah, neat
22:32:52 <Glenn_the_Great-AG> Apparantly this is a routine that Lion7 likes to do to peoples' websites
22:33:15 <Glenn_the_Great-AG> If you want to see what he did, look at the top of www.armageddongames.com
22:33:22 <Glenn_the_Great-AG> At least he didn't do anything harmful
22:33:37 <sbp> well... as I say, we don't know much (if anything) about the guy. You can check for yourself from the logs
22:34:08 <Glenn_the_Great-AG> Well, thanks for your help. I'm just trying to grab whatever little info I can
22:35:32 <Glenn_the_Great-AG> Glenn_the_Great-AG has left #swhack
22:36:16 <sbp> [[[
22:36:16 <sbp> 17:40:03 <sbp> sbp has changed the topic to: Swhack: Where the lamers and sk do roam
22:36:16 <sbp> 17:40:12 <AaronSw> s/lamers/lamerz/ please
22:36:16 <sbp> 17:40:14 <sbp> oops
22:36:16 <sbp> 17:40:19 <sbp> sbp has changed the topic to: Swhack: Where the lamerz and sk do roam
22:36:17 <sbp> ]]]
22:39:08 <AlexMax> * AlexMax is away: (Auto-Away after 10 mins) [BX-MsgLog On]
22:45:11 <deltab> wouldn't it be easier just to turn off the logger?
22:46:05 <sbp> why's that, deltab? :-)
22:46:09 <sbp> [OLM]
22:46:40 <sbp> sbp has changed the topic to: Swhack has now been banned in every country in the world except for Finland. We're not sure why
22:47:58 <sbp> not sure why: I blame Aaron, of course. He'll blame me, so I'm just getting in my shot early
22:58:41 <bijan> <rdf:Description about="user6722785"
22:58:41 <bijan> u:lineText="I sit at my table and wage war on myself">
22:58:41 <bijan> <u:anonLine rdf:resource="http://test/#j45234"/>
22:58:41 <bijan> <u:has rdf:resource="&rdf;_1"/>
22:58:41 <bijan> <u:otherProp rdf:resource="&rdf;_1"/>
22:58:41 <bijan> <u:partStruct rdf:resource="http://test/#j45233"/>
22:58:44 <bijan> <u:prop rdf:resource="http://media.redmonk.net/files/wlp.xml#v1"/>
22:58:46 <bijan> <u:trk rdf:resource="http://musicbrainz.org/track/1a2764f2-3081-485f-b3d5-d1e64ca823b4"/>
22:58:47 <bijan> <u:trkStruct rdf:resource="http://test/#j45224"/>
22:58:49 <bijan> </rdf:Description>
22:59:19 <sbp> user6722785? sound of much barfing
22:59:41 <bijan> Eh.
22:59:44 <sbp> actually, we didn't really resolve how to hack bNode labels into XML RDF
22:59:44 <bijan> works for now :)
22:59:54 <bijan> 27 bnodes
23:00:08 <sbp> since "_" is void as a URI scheme, someone (probably TimBL) suggested using that
23:00:23 <sbp> yep, that's what CWM got
23:00:28 <bijan> Well, that's just what my gen_id does.
23:00:34 <bijan> It's easy to change it to whatever.
23:00:39 <bijan> But it doesn't matter.
23:00:45 <bijan> re: inferencing.
23:00:49 <sbp> is that really unique based upon the current bNodes, or just a random thing?
23:01:03 <bijan> unique
23:01:13 <bijan> user is the rdf_db it's unique to
23:01:23 <bijan> It maintains an integer.
23:01:24 <sbp> wow, that was speedy
23:01:50 <bijan> .641 seconds
23:02:03 <bijan> Er.. or something.
23:02:05 <bijan> That doesn't seem right.
23:02:10 <bijan> I'm not sure about timing in prolog.
23:02:20 <sbp> * sbp wonders how to benchmark CWM
23:02:37 <sbp> there must be a Python benchmarking framework somewhere...
23:02:43 <bijan> profile is built in.
23:02:49 <bijan> can't you do it fromthe command line?
23:02:54 <bijan> How long does CWM take?
23:03:36 <sbp> dunno, can't time it. Less than 5 seconds, certainly
23:03:53 <bijan> How long did eep take?
23:04:01 <bijan> That was the half hour -->crash?
23:04:15 <sbp> erm...
23:04:18 <sbp> *ahem*
23:04:26 <sbp> yeah. About twenty minutes before I crashed it
23:04:38 <sbp> taking up plenty of memory in the meantime
23:04:57 <bijan> Well, I could have told you straight off that this would be trivial.
23:05:04 <bijan> The prolog rule is *not* complex.
23:05:30 <bijan> You probably have a bug somewhere.
23:06:07 <bijan> This isn't a great test if CWM resolves it so quickly, actually, as I would expect.
23:06:09 <sbp> yep - but it's something that I don't particularly want to debug... :-)
23:06:33 <bijan> You were supposed to give me a performance tester.
23:06:36 <sbp> no, it's not a great test - more of a sanity test. I'm sure that there are some decent files in SWAP, but most have builtins etc. - ugh
23:06:45 <bijan> Of course, it's nice that mine works and yours doesn't :)
23:06:45 <sbp> well, that's the best one that I have!
23:06:51 <sbp> blargh! :-)
23:07:10 <bijan> Really? Have you ever run *anything* that takes a long time on CWM.
23:07:19 <bijan> --thinking, likely.
23:07:29 <sbp> of course, plenty of stuff. Under --think, as you say...
23:07:30 <bijan> That isn't obvious like a log:semantics
23:07:44 <sbp> I've been debugging stuff that takes like five minutes per run. That's frustrating
23:07:50 <bijan> So, what I need is something reasonably tortuous, that we can purge the builtins ;)
23:07:53 <bijan> Ok, one of those.
23:08:04 <sbp> well, I'll go find some decent N3 on my hards drive...
23:08:08 <sbp> s/hards/hard/
23:08:14 <bijan> I was *wondering*!
23:08:43 <sbp> * sbp searches for *.n3
23:08:57 <sbp> 1927 files found
23:09:01 <bijan> Er..
23:09:12 <sbp> 10.9 MB
23:09:14 <bijan> Just whip out a python script that runs and times each one! :)
23:09:17 <bijan> Yick, man.
23:09:20 <sbp> heh: I should run the whole lot through CWM
23:09:24 <bijan> You have to give up the n3 habit!
23:09:40 <sbp> It'd probably give me a world peace formula or something
23:09:54 <sbp> N3 habit: heh, heh. It's so difficult to stop!
23:10:22 <sbp> can you parse = and ()?
23:10:27 <bijan> No.
23:10:34 <sbp> ugh
23:10:38 <bijan> Sorry.
23:10:44 <bijan> () is on the list.
23:10:56 <sbp> axioms1.n3 would have been a good test
23:10:57 <bijan> = should be easy if I knew what it expanded to :)
23:11:15 <sbp> aha, I think I've found a good one
23:11:45 <sbp> here: http://infomesh.net/2001/12/map/data.n3
23:11:53 <sbp> it has default prefixes, but then so does most N3
23:12:03 <sbp> s/ :/ x:/
23:12:13 <bijan> yes.
23:12:24 <sbp> 136KB of N3, there
23:12:28 <bijan> x: is taken :)
23:12:34 <bijan> How long does it take on CWM?
23:12:49 <sbp> x: so it is!
23:12:57 <sbp> long: to do what? we need a query...
23:13:22 <sbp> I guess I could run { ?x a:icao ?y } => { ?x a:icao ?y } .
23:13:26 <bijan> Well, think one up! :)
23:13:26 <sbp> how's that?
23:13:31 <bijan> I don't know.
23:14:00 <sbp> * sbp gets to work on it
23:16:32 <sbp> wow, it takes ages. I thought I'd borked it, so I actually stopped it
23:16:35 <sbp> * sbp runs it again
23:18:09 <bijan> It's huge!
23:18:48 <sbp> given that Python can RegExp the necessary information out in about 0.01 seconds flat...
23:19:10 <sbp> heh, it started at 23:17:47, and it still going...
23:19:26 <sbp> running cwm data.n3 --filter=q.n3 > q-out.n3
23:20:14 <sbp> now I think about it, I am running the new CWM, which is meant to be very slow. Perhaps I shouldn't have downloaded it
23:20:20 <bijan> Hmm. I broke n3_repl. at some point.
23:20:24 <bijan> * bijan sighs
23:21:31 <sbp> N.B., the query is just:-
23:21:31 <sbp> @prefix : <#> .
23:21:31 <sbp> @prefix log: <http://www.w3.org/2000/10/swap/log#> .
23:21:31 <sbp> @prefix a: <http://www.megginson.com/exp/ns/airports#> .
23:21:31 <sbp> this log:forAll :x , :y .
23:21:31 <sbp> { :x a:icao :y } log:implies { :x a:icao :y } .
23:21:37 <bijan> Ah, damn, bad syntax.
23:21:39 <bijan> Thanks
23:21:47 <sbp> np
23:22:01 <sbp> CWM is still plodding along...
23:22:09 <sbp> done!
23:22:13 <sbp> $ u timenow; cwm data.n3 --filter=q.n3 > q-out.n3; u timenow
23:22:14 <sbp> 20020223-231747
23:22:14 <sbp> 20020223-232155
23:22:25 <sbp> 4:08
23:22:53 <sbp> that's quite terrible. Where'd my old .tar.gz go to... I'm gonna try the old version
23:23:33 <sbp> * sbp slurps it via. FTP
23:23:58 <bijan> Ok, reading it in.
23:24:09 <bijan> Oh ohoh .
23:24:15 <bijan> I'm going to have a *huge* advantage.
23:24:19 <bijan> Since I wont' have to parse.
23:24:21 <bijan> Hmm.
23:24:39 <sbp> ugh
23:24:54 <sbp> wow!
23:24:55 <sbp> $ u timenow; python cwm/cwm.py data.n3 --filter=q.n3 > q-out.n3; u timenow
23:24:55 <sbp> 20020223-232416
23:24:55 <sbp> 20020223-232438
23:25:04 <sbp> 1.82 is sooooo much faster!
23:25:16 <bijan> indeed.Ok, I did it
23:25:33 <sbp> only 22 seconds under 1.82? I can't believe that
23:25:42 <bijan> Did it get the right results?
23:25:49 <bijan> X = 'http://www.test.com/#MOB'
23:25:49 <bijan> Y = 'http://www.megginson.com/exp/ns/airports#icao'
23:25:49 <bijan> Z = literal('KMOB') ;
23:25:49 <bijan> X = 'http://www.test.com/#BFM'
23:25:49 <bijan> Y = 'http://www.megginson.com/exp/ns/airports#icao'
23:25:50 <bijan> Z = literal('KBFM') ;
23:25:51 <bijan> Etc.
23:26:02 <bijan> I'm under a minute, even with parsing.
23:26:08 <sbp> yeah, that seems alright
23:26:17 <sbp> well, you whip the new CWM, but the old CWM whips you
23:26:30 <bijan> But is the old CWM correct?
23:26:53 <bijan> * bijan notes that CWMclone is a wild prototype :)
23:27:53 <sbp> heh, heh
23:28:21 <sbp> the old CWM is a good stable version
23:29:22 <bijan> Doesn't mean it's correct.
23:29:29 <sbp> sample output:-0
23:29:34 <bijan> I'm just pointing out htat I have *no* optimization.
23:29:49 <sbp> s/0//
23:29:49 <sbp>     :01R     a:icao "K01R" .
23:29:49 <sbp>     
23:29:49 <sbp>     :01T     a:icao "K01T" .
23:30:17 <bijan> It generates a *lot* of output.
23:31:01 <bijan> Ah, it's the identity.
23:31:08 <bijan> Not an interesting test.
23:31:32 <sbp> it is correct
23:32:34 <bijan> How fast is your machine?
23:33:22 <bijan> I put my results in sw/cwmclone/
23:33:24 <sbp> the data seems very consistent
23:33:24 <sbp> Hmm... I suppose there's a way to test - I could pretty print them both and hash the outputs
23:33:24 <sbp> but I'm not going to bother :-)
23:33:54 <bijan> Well, you can dl mine and compare.
23:34:13 <bijan> At http://www.unc.edu/~bparsia/sw/cwmclone/tc2_2_results.rdf
23:34:21 <bijan> How fast is your machine?
23:35:05 <bijan> Actually, I don't know how fast *mine* is.
23:35:15 <bijan> Oh, mine is going to be *really* hurt by this one.
23:35:46 <bijan> I essentially have to touch every inference twice.
23:36:03 <bijan> I mean, I shouldn't *have* to, but I do.
23:36:24 <bijan> So, one should expect to easily double the speed.
23:36:29 <bijan> Without anything tricky.
23:37:47 <sbp> sorry, my connection was lagging
23:38:03 <sbp> it turns out that you were speaking, and the bits of text were only coming through after some delay
23:38:08 <bijan> heheh.
23:38:26 <bijan> Hmm. But no, it won't double.
23:38:30 <bijan> The testing is *very* fast.
23:38:37 <bijan> Looks like the time is in db updating.
23:38:45 <bijan> I have more memory.
23:38:48 <bijan> But I may be slower.
23:38:54 <bijan> 22.9 seconds for the filter.
23:38:57 <bijan> No parsing.
23:39:11 <sbp> not too bad - equals the old CWM
23:39:50 <bijan> We should run it on your machine.
23:39:53 <sbp> * sbp emails the speed tests to Tim/Dan/Norm
23:40:23 <sbp> ta da: http://lists.w3.org/Archives/Public/www-archive/2002Feb/0038
23:40:35 <sbp> run it on my machine: sure. just tell me what to do
23:41:53 <bijan>  ok, got to http://www.unc.edu/~bparsia/sw/cwmclone/
23:42:00 <bijan> Get cwmclone.P
23:42:07 <bijan> And tc2.2.n3
23:42:32 <sbp> the Website seems to be borked
23:42:35 <bijan> Oo, you know what's prolly slowing down the parse...all the info I preint.
23:42:56 <bijan> did I typo? http://www.unc.edu/~bparsia/sw/cwmclone/
23:43:00 <bijan> Is working for me.
23:43:09 <sbp> working for me now too...
23:43:31 <sbp> got 'em
23:44:08 <bijan> Ok, fire up swi.
23:44:18 <sbp> O.K.
23:44:32 <bijan> ['path/to/cwmclone.P'].
23:45:00 <sbp> Yep
23:45:48 <bijan>  Ok, get_time(T1),n3_consult('path to n3', user), get_time(T2), Time is T2 - T1.
23:46:45 <bijan> (This should take a bit. My tokenizer/parser are 1) slow, 2) prolly non-deterministic, and 3) printing lots of info :(
23:47:11 <sbp> after a lot of printing:-
23:47:11 <sbp> S: x:P60 V: y:state O: literal(WY) C: user
23:47:11 <sbp> alldone
23:47:11 <sbp> No
23:47:18 <bijan> Yes.
23:47:27 <bijan> How long did it take, btw?
23:47:35 <bijan> hm. Might not have answered.
23:47:35 <sbp> no idea
23:47:38 <bijan> Ok.
23:47:43 <bijan> compile_rules(user).
23:47:59 <sbp> done it "Yes"
23:48:15 <bijan> get_time(T1), filter(user, test), get_time(T2), Time is T2 - T1.
23:48:47 <sbp> T1 = 1.01451e+009
23:48:47 <sbp> T2 = 1.01451e+009
23:48:47 <sbp> Time = 15.38 
23:48:54 <bijan> Yep, faster than mine.
23:49:00 <bijan> By about what I expected.
23:49:24 <sbp> and quite a bit faster than CWM, but no parsing...
23:49:32 <bijan> I still prolly come in behind CWM the younger.
23:49:37 <bijan> No parsing, and no writing to disk.
23:49:52 <bijan> Though rdf_save/2 can approximate.
23:50:07 <bijan> In *much* less code though :)
23:50:13 <sbp> true
23:50:20 <sbp> * sbp wonders if he dare try Eep on it
23:50:30 <bijan> And note, SWI is not, generall, a *fast* prolog.
23:52:24 <sbp> what's a better alternative? XSB?
23:52:38 <bijan> I don't know off hand.
23:52:42 <bijan> I've note benchmarked.
23:52:51 <bijan> Plus it very much depends on your appication.
23:52:56 <bijan> XSB may be faster for some things.
23:53:56 <bijan> But some speed gains possible:
23:54:12 <bijan> In the "generate applicable rules' bit (espeically for a filter, there isnt' any need.
23:54:38 <bijan> rdf_db may be slowing things down (or speeding them up!)
23:55:19 <bijan> We need a more complex inference to compare, I think.
23:55:25 <sbp> yeah...
23:55:30 <bijan> THis is more of an I/O thing.
23:56:10 <bijan> hmm. Satisfaction testing only takes .03 seconds. Negligible.
23:56:25 <bijan> That suggestst that most of the time is in the assertion.
23:56:34 <bijan> This may be due to rdf_db:rdf_assert
23:57:19 <sbp> what is that doing?
23:57:21 <bijan> Is there an interesting, more complex query to be done on this data?
23:57:33 <bijan> Well, it does some qname expansion.
23:57:37 <sbp> yes, but it involves builtins
23:57:56 <bijan> We'll, I"m trying to think of something that doesn't involve builtins :)
23:58:07 <sbp> the filter's at http://infomesh.net/2001/12/map/distance.n3
23:58:30 <bijan> math, blagh
23:58:53 <bijan> I could hand translate that...
23:59:15 <bijan> Does that take a long time on CWM?