00:00:00 --- log: started haskell/07.10.09 00:00:02 so this "applicative" stuff is a lot like some of the concepts in "funmath" 00:00:10 where you write hats over functiosn and operators to lift them 00:00:57 --- quit: ddarius (Read error: 110 (Connection timed out)) 00:01:32 --- quit: ivanm (Read error: 110 (Connection timed out)) 00:01:49 --- quit: jewel (Read error: 113 (No route to host)) 00:02:08 --- join: iblechbot (n=iblechbo@ppp-62-216-200-249.dynamic.mnet-online.de) joined #haskell 00:03:54 --- quit: jonathanv (Read error: 110 (Connection timed out)) 00:05:03 --- join: FZ (n=chatzill@unaffiliated/fz) joined #haskell 00:05:14 --- join: filp (n=hordf@ip-168-248.sn2.eutelia.it) joined #haskell 00:07:53 --- quit: boyscared (Read error: 104 (Connection reset by peer)) 00:08:11 --- join: boyscared (n=boy@oh-76-5-125-71.dhcp.embarqhsd.net) joined #haskell 00:09:17 --- join: phb2 (n=phb2@c213-100-44-22.swipnet.se) joined #haskell 00:10:30 --- quit: shachaf (Read error: 113 (No route to host)) 00:11:35 --- quit: sutats ("leaving") 00:12:12 --- join: doserj (n=doserj@dewey.inf.ethz.ch) joined #haskell 00:12:30 --- quit: Tela () 00:14:36 --- quit: Plareplane ("Leaving") 00:14:38 --- join: piis3141 (n=piis3141@ip4da169f4.direct-adsl.nl) joined #haskell 00:14:58 --- join: Plareplane (n=Plarepla@216-197-152-148.sktn.hsdb.sasknet.sk.ca) joined #haskell 00:15:45 --- join: boegel (n=boegel@konijn.elis.UGent.be) joined #haskell 00:16:16 --- join: jewel (n=jewel@61.247.251.10) joined #haskell 00:18:45 newsham: yeah, I think there is a list of papers to read on various topics 00:18:59 newsham: I'll find it. 00:19:58 yeah, http://haskell.org/haskellwiki/Research_papers 00:19:59 Title: Research papers - HaskellWiki 00:20:22 That has a list of the research papers relevant to Haskell broken down roughly by topic. 00:20:45 The Functional Pearls section has quite a lot of particularly good ones. 00:21:29 --- join: Kyzia (i=kyzia@195.5.34.126) joined #haskell 00:21:44 hello 00:21:46 http://globalchat.pp.net.ua/ 00:21:46 --- part: Kyzia left #haskell 00:21:47 Title: Глобал_Чат - Главная страница 00:22:04 --- quit: bringert () 00:22:58 --- join: amiddelk (n=arie@melinda.cs.uu.nl) joined #haskell 00:28:14 I wish that as an op, I could edit the past conversation on the channel, so as to remove the effects of spammers ;) 00:28:34 Censorship! 00:28:41 yes 00:28:43 and? 00:28:47 --- join: pulczynski (n=marek@fw.grono.org) joined #haskell 00:28:48 isn't that the point of an op 00:28:52 --- join: takuan (n=takuan@83.101.37.197) joined #haskell 00:29:14 --- quit: pulczynski ("Leaving.") 00:29:33 --- join: pulczynski (n=marek@fw.grono.org) joined #haskell 00:30:16 --- join: Aperculum (n=lauri@unaffiliated/aperculum) joined #haskell 00:30:19 --- join: roconnor (n=roconnor@vhe-540354.sshn.net) joined #haskell 00:30:31 --- part: pulczynski left #haskell 00:30:52 --- join: QplQyer (n=jeroen@state.ugent.be) joined #haskell 00:30:54 So I made this mini-mini-combinator language to make turing maching programs, but it's really ugly looking with nested functions. I wonder if this is a good place to stick a monad. 00:31:56 * jeffz wonders how effective spamming a non-english language website amongst mainly english speakers is. 00:32:14 it's cheaper than it is effective. 00:32:15 maybe its just the get the url into logs? 00:32:25 glguy: ah good point. 00:32:28 Google spamming. 00:32:31 Pagerank. 00:32:50 --- quit: Mr_Awesome (Read error: 110 (Connection timed out)) 00:33:52 Shimei: Well, a monad is just a combinator language which has the right kind of combinators in it. 00:33:58 --- quit: boyscared (Read error: 104 (Connection reset by peer)) 00:34:17 --- join: boyscared (n=boy@oh-76-5-125-71.dhcp.embarqhsd.net) joined #haskell 00:35:51 SEO is a pretty skeevy profession. 00:40:45 --- quit: Korollary ("leaving") 00:41:30 --- join: Rack (n=soue@201.132.207.143) joined #haskell 00:42:13 --- quit: LeCamarade ("":o)"") 00:42:35 --- quit: lstephen666 ("Leaving") 00:43:17 --- quit: oerjan ("leaving") 00:44:01 --- join: goban_ (n=whyso@pool-96-228-43-108.rcmdva.fios.verizon.net) joined #haskell 00:44:50 skeevy executaive officer? 00:45:42 --- join: calvins (n=calvins@c-67-188-118-53.hsd1.ca.comcast.net) joined #haskell 00:46:05 --- join: bparkis (n=bparkis@library-356.wireless.umass.edu) joined #haskell 00:46:26 --- join: bringert (n=bringert@3-1-5-7a.gva.gbg.bostream.se) joined #haskell 00:46:28 oh search engine optimization 00:49:31 --- quit: edwinb (Remote closed the connection) 00:50:30 * mgsloan wonders if there are people who just optimize programs 00:51:13 yes there is 00:51:17 they're called ricers 00:51:29 --- quit: mikael () 00:51:30 There certainly are people whose job it is to take code written by physicists and the like and make it run properly on a supercomputer 00:51:45 hah 00:51:54 that's interesting 00:52:32 I wonder what physicists use.. with some nice syntax haskell could be pretty good for physicists 00:52:47 fortran and matlab? 00:52:49 Fortran, what else? 00:52:53 I suppose so 00:53:46 Fortran is insanely well optimized for it's problem domain 00:53:54 compiler-wise, anyway 00:54:56 --- quit: omnId ("Leaving") 00:55:55 yeah, I seem to recall hearing that in general 00:56:11 --- join: cbe (n=cbe@62.70.2.252) joined #haskell 00:56:15 --- join: therp (n=nil@chello080108077214.34.11.tuwien.teleweb.at) joined #haskell 00:56:26 there were some attempts to make a Lisp fast enough to beat Fortran, and the end result did so. the problem is that there are way more people working on optimizing Fortran than there are optimizing Lisp or Haskell. 00:56:32 pretty ugly language though 00:56:40 so, it's hard to maintain an advantage 00:56:43 right 00:57:02 oh yeah, Fortran is gross 00:57:16 Ironically the first language on my programming language timeline that I have taped to the wall. 1954... 00:57:26 (first and still strong?) 00:57:28 --- nick: clad|sleep -> cladhaire 00:57:38 If I had to use it I'd probably write a language that outputs fortran code 00:57:39 --- part: zappa left #haskell 00:57:45 Lisp was #2 and is still being used. 00:57:49 fortran and lisp are the oldest languages still in use 00:58:03 and yeah, fortran came first 00:58:04 basic is still used! 00:58:06 B-O comes next on my chart. :p 00:58:16 Aperculum: basic is young 00:58:17 With a name like that, no surprise it died. 00:58:23 there were some kind-of languages written for earlier computers, I think the first one was designed by Zuse 00:58:34 lisp was originally intended to be a fortran variant, iirc 00:58:37 ibid, basic is oldest highlevel language after fortran and lisp 00:58:43 plankalkГјl 00:58:49 What about algol? 00:58:52 --- quit: rc-1 (Connection timed out) 00:58:57 Shimei: not in use 00:58:58 BASIC is not that old 00:59:07 Oh, right. High level languages in use. 00:59:28 --- quit: mgsloan (Remote closed the connection) 00:59:42 Given that the "basic" in use is VB.Net, I'm not sure it really passed on much lineage. 00:59:52 it will be interesting to see how Fortress turns out. 01:00:20 Aperculum: hm, right, 1963. i though it was late 60s 01:00:44 yeah, only 5 years after lisp 01:00:48 --- quit: turulu (Remote closed the connection) 01:00:52 --- quit: Adamant (Remote closed the connection) 01:00:55 although 5 years is long 01:01:44 Aperculum: cobol is older than basic, btw, pre-60s 01:01:58 and it's still in use, though mostly in legacy systems 01:06:07 --- join: TwigEther (n=Twigathy@raptor.ukc.ac.uk) joined #haskell 01:06:27 XML is like violence: if it doesn't solve your problem, you aren't using enough of it. 01:06:44 * jeffz wonders how you define a legacy system. 01:07:17 --- quit: scodil ("Lost terminal") 01:07:39 glguy: ;) 01:07:43 --- quit: njbartlett () 01:07:57 jeffz: do you mean that cobol is used in many new projects? 01:08:42 jeffz: by legacy, i mainly mean systems that were built long ago and are in maintenance mode 01:13:36 I would define legacy as "a old system with known flaws or limitations which in an ideal world we would replace, but it's so hardwired in that that is expensive or difficult to do" 01:13:40 something like that 01:14:37 the flaw may just be that "noone knows how to make any changes to it, in the future, even though right now it works perfectly" 01:14:49 cobol was still a mandatory course in the business oriented it program here when i started studying (not on that program, fortunately) 01:14:53 not anymore though 01:15:21 ibid: well, where I work, some new cobol programs are written, nothing very large though. 01:15:38 isn't SAP's ABAP language largely cobol ? 01:15:43 jeffz: "mostly" leaves room for exceptions :) 01:15:51 --- join: Jaak (n=Jaak@193.40.10.182) joined #haskell 01:16:27 I sometimes think I might've had more fun doing COBOL than Java. 01:16:41 quicksilver: it is my impression that most legacy systems by my definition are legacy systems by your definition :) 01:17:07 ibid: I guess =) though, I still think the term legacy system is ambiguous and lends itself to be used where people think they can do a better job, often with some misunderstanding. 01:17:38 legacy can mean something as simple as "not conforming to the standards we use now, because it was written before we had them" 01:17:44 jeffz: :) what i was trying to convey is that its use is winding down 01:20:03 ibid: that's probably just a matter of time, I don't think I know anyone who is under 40 years of age who programs cobol. 01:21:22 --- quit: bparkis ("Leaving") 01:21:29 --- join: goban__ (n=whyso@pool-96-228-43-108.rcmdva.fios.verizon.net) joined #haskell 01:21:58 jeffz: yep 01:22:14 jeffz: though, it was still taught here ten years ago :) 01:22:23 --- join: Adamant (n=wtfmonki@unaffiliated/adamant) joined #haskell 01:23:51 --- join: turulu (i=antiludi@gateway/tor/x-db241b485744e9e7) joined #haskell 01:25:04 --- join: xinming_ (n=xinming@70.132.73.218.broad.wz.zj.dynamic.163data.com.cn) joined #haskell 01:25:29 --- join: masak (n=user@130.238.45.242) joined #haskell 01:28:10 --- quit: goalieca (Remote closed the connection) 01:32:27 --- quit: mathrick (Read error: 104 (Connection reset by peer)) 01:34:05 --- join: mathrick (n=mathrick@users.kollegienet.dk) joined #haskell 01:34:39 --- quit: goban_ (Connection timed out) 01:34:51 --- join: mr_tenor (n=nick@c220-239-33-8.rivrw7.nsw.optusnet.com.au) joined #haskell 01:36:58 --- join: KatieHuber (n=katie@203-97-111-120.cable.telstraclear.net) joined #haskell 01:37:19 --- quit: jewel (Read error: 113 (No route to host)) 01:37:19 --- quit: xinming (Read error: 110 (Connection timed out)) 01:37:52 --- join: goban_ (n=whyso@pool-96-228-43-108.rcmdva.fios.verizon.net) joined #haskell 01:39:06 --- join: tarrybon1 (n=dlc@c-71-60-148-169.hsd1.pa.comcast.net) joined #haskell 01:41:56 --- quit: tarrybone (Nick collision from services.) 01:42:07 --- join: jewel (n=jewel@61.247.251.10) joined #haskell 01:42:08 --- nick: tarrybon1 -> tarrybone 01:43:02 --- quit: nominolo ("Ex-Chat") 01:45:12 --- nick: cladhaire -> clad|brb 01:47:23 so, the indentation matters in haskell 01:47:26 like in python? 01:48:26 --- join: augustss (n=Lennart@63.175.43.10) joined #haskell 01:49:42 quicksilver: I "finished" my code 01:50:14 --- join: ivanm (n=ivan@60-242-0-245.static.tpgi.com.au) joined #haskell 01:50:34 --- quit: goban__ (Connection timed out) 01:51:42 Aperculum: yes, indentation matters in python too 01:51:58 Aperculum: but the rules are different, so it's not exactly "like in python" 01:52:11 ibid: is there a main difference? 01:52:47 masak: it's been a long time since i last looked at python :) 01:53:05 to me, there seems to be a main similarity, namely that things that are semantically subordinate other things are further to the right 01:53:11 masak: but for one, in haskell, layout is lexical sugar for { ; } 01:53:12 yes, the indentation dependent syntax is optional, you can throw in braces and semicolons to get a sensible syntax 01:53:29 --- quit: glguy ("leaving") 01:53:34 @seen quicksilver 01:53:34 quicksilver is in #haskell and #ghc. I last heard quicksilver speak 35m 56s ago. 01:54:07 there's brace syntax in python too 01:54:07 --- quit: Kattana (Read error: 110 (Connection timed out)) 01:54:24 ...I think 01:54:26 oh python has braces too? Well I never! 01:54:35 let me check :) 01:54:36 masak: wasn't when i did python 01:54:47 masak: of course it's possible they've stolen them from haskell since :) 01:54:55 I'm probably wrong, actually 01:55:02 --- quit: mlesniak (Read error: 110 (Connection timed out)) 01:55:22 --- quit: iblechbot (Read error: 110 (Connection timed out)) 01:55:23 masak: and, of course, in python the layout enabler is ':', while haskell uses a number of keywords 01:55:43 right 01:55:58 I was thinking of this piece of code 01:56:00 http://www.freedom-to-tinker.com/tinyp2p.html 01:56:03 Title: TinyP2P 01:56:07 which actually does without braces 01:56:18 so I guess python doesn't have them :) 01:58:31 ibid: yes, I'd say that is the main difference then 01:58:56 which "that"? :) 01:59:47 --- quit: ac ("Lost terminal") 02:00:23 --- join: haraldk (n=harald@dhcp106-38.linpro.no) joined #haskell 02:01:29 --- join: ac (n=abram@dsl231-034-070.sea1.dsl.speakeasy.net) joined #haskell 02:01:49 ac: pleased to hear it :) 02:02:05 quicksilver: it was much easier generalizing it then I was expecting 02:03:18 --- join: Kattana (i=Kattana@modemcable181.114-80-70.mc.videotron.ca) joined #haskell 02:04:46 masak: Python has braces, but they're for dictionaries 02:09:49 --- join: njbartlett (n=njbartle@cpc2-barn4-0-0-cust110.brnt.cable.ntl.com) joined #haskell 02:10:18 --- join: MyCatSchemes (n=rb6822@yilvet.cs.bris.ac.uk) joined #haskell 02:10:53 err, I've also heard of a python brace syntax replacing indentation, but that might have been some web service thingy's own quirk 02:13:10 --- quit: seafoodX (Read error: 110 (Connection timed out)) 02:14:03 --- join: fmardini (n=fmardini@77.245.6.99) joined #haskell 02:14:19 --- quit: fmardini (Client Quit) 02:14:34 Japsu: I think that exists for blind people. 02:15:02 --- join: Rebooted (n=opera@88-104-185-170.dynamic.dsl.as9105.com) joined #haskell 02:15:06 --- quit: idnar (Nick collision from services.) 02:15:09 --- join: idnar_ (n=mithrand@unaffiliated/idnar) joined #haskell 02:15:53 --- nick: clad|brb -> cladhaire 02:16:11 --- part: Rebooted left #haskell 02:16:25 --- join: lnxz (n=Joachim@bjo1-1x-dhcp451.studby.uio.no) joined #haskell 02:19:25 --- join: dibblego (n=nobody@220-245-107-64.static.tpgi.com.au) joined #haskell 02:20:32 --- join: seafoodX (n=sseefrie@alphad261.lnk.telstra.net) joined #haskell 02:24:48 --- join: gkr (n=notme@201-212-49-41.cab.prima.net.ar) joined #haskell 02:24:51 --- quit: jwp_ (Read error: 104 (Connection reset by peer)) 02:28:44 --- join: mrchebas (n=alexey@ster.cs.uu.nl) joined #haskell 02:29:27 --- quit: dibblego ("Leaving") 02:29:40 --- join: jwp_ (n=jwp@mail.icrossing.com) joined #haskell 02:29:45 --- quit: mrchebas (Client Quit) 02:29:53 --- join: mrchebas (n=alexey@ster.cs.uu.nl) joined #haskell 02:32:11 --- join: yitz (n=abba@line236.adsl.actcom.co.il) joined #haskell 02:33:16 --- quit: prb () 02:33:17 --- join: gabkdlly (n=gabriel@dslb-088-073-040-137.pools.arcor-ip.net) joined #haskell 02:34:56 --- join: hkBst (n=hkBst@gentoo/developer/hkbst) joined #haskell 02:35:47 --- join: dibblego (n=nobody@220-245-107-64.static.tpgi.com.au) joined #haskell 02:39:10 --- join: exDM69 (n=rsalmin2@omena.hut.fi) joined #haskell 02:40:18 --- quit: phb2 () 02:42:20 --- quit: dibblego ("Leaving") 02:43:57 --- join: dibblego (n=nobody@220-245-107-64.static.tpgi.com.au) joined #haskell 02:46:32 --- join: malcolmw (n=malcolm@venice.cs.york.ac.uk) joined #haskell 02:47:20 --- quit: Tykom () 02:49:37 Anyone here ever spend time computing lambda expressions? 02:49:54 yes, we do that all day long 02:50:15 yitz: Is that a question? =p 02:51:00 I haven't so much, but I just noticed a weird identity: let c=(.), and let c4=c c c c. Then we have that c c4 == c c4 c c c c. Ever see stuff like that? 02:52:43 @pl let {c = (.); c4 = c c c c} in c4 02:52:43 (line 1, column 5): 02:52:43 unexpected "{" 02:52:43 expecting natural, identifier or "in" 02:53:00 unpl i'd think 02:53:07 right 02:53:10 either way it hates me 02:53:14 --- nick: sior|sleep -> sioraiocht 02:53:16 --- join: mrighele (n=mrighele@metropolis.sci.univr.it) joined #haskell 02:53:25 It was probably the "sleep". 02:53:32 @unpl let c = (.); c4 = c c c c in c c4 02:53:32 let { c a b d = a (b d); c4 = c c c c} in c c4 02:53:42 bah 02:53:45 no help 02:54:03 @type (.) ((.)(.)(.)(.)) 02:54:05 forall a a1 b c a2 a3. (a3 -> a -> a1 -> b -> c) -> a3 -> a -> a1 -> (a2 -> b) -> a2 -> c 02:55:18 Given a func w/ 4 params, it composes on the 3rd, w/ the composing func. specified between params 2 and 3. 02:55:50 @type let{c=(.);c4=c c c c}in c c4 c c c c 02:55:52 forall a a1 a2 b c a3. (a -> a1 -> a2 -> b -> c) -> a -> a1 -> a2 -> (a3 -> b) -> a3 -> c 02:56:53 Same type. In fact, they are equal. Prove it by repeatedly applying the identity x (y z)==c x y z. 02:57:19 --- join: mlesniak (n=chatzill@dhcp244.dcaiti.TU-Berlin.DE) joined #haskell 02:57:41 --- quit: bringert () 02:58:44 In fact, that identity is a semi-associativity condition that makes the entire lambda calculus into a semi-monoid. Apparently with quite interesting structure, as my example shows. 02:59:49 c c2 == c4. c c3 == c7. But c cn cannot be reduced for n>3. More interesting stuff like that if you throw flip into the mix. 02:59:50 --- quit: MyCatSchemes ("Swim, swim, hungry!") 03:02:51 --- quit: dogmaT (Read error: 113 (No route to host)) 03:03:15 Oops, sorry, composing func. specified between 3 and 4. Anyway. 03:05:49 yes, interesting 03:07:46 I discovered these things while playing with @pl, some strange and vary surprising patterns start to appear. 03:08:09 --- quit: dibblego ("Leaving") 03:08:09 --- join: swiert (n=wss@sneezy.cs.nott.ac.uk) joined #haskell 03:09:21 --- join: zx]treads (n=student@195.251.214.51) joined #haskell 03:09:55 --- join: gour (n=Gour@3-180.dsl.iskon.hr) joined #haskell 03:11:47 Maybe I'll post it to the cafe. 03:14:03 --- join: chris2 (n=chris@p5B16A71F.dip0.t-ipconnect.de) joined #haskell 03:16:42 --- quit: boegel (Read error: 110 (Connection timed out)) 03:16:48 --- quit: meryrus () 03:17:32 --- join: merus (n=merus@vlan-125-087.onu.edu) joined #haskell 03:18:01 --- quit: fxr (Remote closed the connection) 03:24:10 --- join: ptx (n=ptxmac@0x5550c853.adsl.cybercity.dk) joined #haskell 03:26:36 --- join: tarrybon1 (n=dlc@c-71-60-148-169.hsd1.pa.comcast.net) joined #haskell 03:27:20 --- quit: Thas (Read error: 110 (Connection timed out)) 03:27:25 --- join: snearch (n=snearch@e178044047.adsl.alicedsl.de) joined #haskell 03:28:48 --- join: misterbeebee_ (n=mike@71-35-157-132.tukw.qwest.net) joined #haskell 03:32:48 --- join: Taejo (n=max@196.210.103.191) joined #haskell 03:33:53 --- join: apfelmus (n=apfelmus@F7301.f.ppp-pool.de) joined #haskell 03:34:19 --- join: gimbo (i=gimbo@mini.theravensnest.org) joined #haskell 03:34:23 --- quit: misterbeebee (Read error: 110 (Connection timed out)) 03:36:32 :'( I'd like to create a Geometry.2D module namespace 03:36:32 Hi apfelmus. We were just discussing something I noticed yesterday. I posted it to the cafe. Do you know anything about that stuff? 03:37:03 but 2D is not a valid module name. any alternative thoughts? 03:37:13 TwoDee ? 03:37:28 yitz: uh, oh? /me looks 03:37:37 --- quit: tarrybone (Connection timed out) 03:37:55 Planar 03:38:25 Planar sounds good. But is there a nice name with the numeral 2 in there? 03:39:28 PlanarInOtherWords2Dimensional. um... 03:39:39 short if possible :) 03:39:44 Space2D? 03:40:07 Plane2D ? 03:40:41 --- quit: Averell (Read error: 110 (Connection timed out)) 03:41:57 --- quit: Cale_ (Read error: 104 (Connection reset by peer)) 03:42:22 I guess the "InOtherWords" is implicit... 03:43:28 --- join: Averell (n=averell@p3m/member/Averell) joined #haskell 03:43:42 :) It's a bit redundant, though since Plane or 2D alone would be enough 03:44:24 --- join: igel (n=igel@brsg-4db2d323.pool.einsundeins.de) joined #haskell 03:45:01 --- quit: snearch ("Nettalk6 - www.ntalk.de") 03:45:12 --- join: boegel (n=boegel@konijn.elis.UGent.be) joined #haskell 03:47:31 --- join: GNU\caust1c (n=caust1c@a207.wh-wieland.uni-ulm.de) joined #haskell 03:51:04 --- join: phb2 (n=phb2@c213-100-44-22.swipnet.se) joined #haskell 03:51:21 --- quit: phb2 (Client Quit) 03:53:41 yitz: my head hums. how do you prove c c4 = c c4 c c c c ? 03:53:53 I just get stuck :) 03:54:20 is it better to install haddock, pfesetup, alex, c2hs, cpphs, greencard before installing happy to compile and install ghc-head ??? 03:54:51 > let reg = mkRegex "(\\d)" in show $ matchRegex reg "d123" 03:54:52 Not in scope: `matchRegex' 03:56:03 apfelmus: Geometry.TwoD 03:56:23 --- quit: KatieHuber () 03:56:26 kaol: no regex in lambdabot 03:56:46 I just wanted to check if it would give Just ["d"] as the result, too 03:57:08 --- quit: boegel ("Leaving") 03:57:41 bah. good for nothing Text.Regex, if it doesn't even understand \d 03:57:47 --- join: VerbalDK (n=jensfrad@port35.ds1-ns.adsl.cybercity.dk) joined #haskell 03:58:15 * geocalc =<< help !!! 03:58:22 text.regex uses yoru system regexp library, I thought 03:58:37 yitz pasted "Proof that c c4 == c c4 c c c c" at http://hpaste.org/3172 03:58:51 kaol: I don't think \d is regexp 03:59:00 kaol: I think that's a perl-only thing? 04:00:05 looks like so... Hmpf. 04:00:35 * kaol considers using FFI to call perl 04:00:57 ! 04:01:13 I have two alternatives to suggest 04:01:30 a. use the regexp notation for that, which is [[:digit:]] 04:01:37 --- join: twanvl (n=twanvl@ip5451f037.direct-adsl.nl) joined #haskell 04:01:51 b. observer that regexps are extraordinarily ugly and use some nicer looking parser combinators 04:02:07 c. let d = [0-9] 04:02:09 or regexp combinators 04:02:19 they aren't quite as ugly 04:03:03 SamB_XP: I've never understood the point, though. why not just use parser combinators? 04:03:13 --- join: fritschy (n=x06180@X715f.x.pppool.de) joined #haskell 04:03:18 The point of regexp is they are a convenient language for expressing matching 04:03:23 I have bunch of stuff in the DB and some of the items have a regexp attached to them. 04:03:36 btu as technoloogy has evolved they no longer seem so attractice 04:03:40 ah 04:03:42 in perl format? 04:03:45 yes 04:04:04 well that does explain it 04:04:08 though I don't think that there's anything in there that I couldn't express as POSIX regexpes too 04:04:10 have you considered Text.Regex.PCRE 04:04:12 quicksilver: they aren't so bad when you are making a two-level parser... 04:04:26 also Yi uses them for key bindings 04:04:28 doesn't it? 04:04:58 SamB_XP: no, they're not that bad. ~But I don't think they're that good, either :) I just don't know why any new system would choose to use them. 04:05:17 quicksilver: I certainly don't like a lot of the things regexps are used for 04:05:18 yitz: ah, i found one, too. interesting, interesting. i don't know anything about it, but digging under "Combinatory Logic" should reveal interesting things 04:05:42 --- quit: gkr (Read error: 113 (No route to host)) 04:06:31 yup, looks like the most exotic thing I've used was \d. Easy enough to replace with [0-9] 04:06:36 yitz: http://en.wikipedia.org/wiki/Combinatory_logic#Combinators_B.2C_C 04:06:47 apparently, c is the B combinator 04:07:14 unfortunately, I've got to run off for lunch 04:07:15 see you 04:07:35 is it better to install haddock, pfesetup, alex, c2hs, cpphs, greencard before installing happy to compile and install ghc-head ??? 04:08:09 I don't think happy requires haddock, cpphs, or greencard 04:08:09 malcolmw: You have 1 new message. '/msg lambdabot @messages' to read it. 04:08:45 ok malcolmw thanks 04:08:48 geocalc: maybe Cabal's generic warnings are confusing 04:09:12 hahhh 04:12:37 --- nick: Eelis- -> Eelis 04:13:25 --- quit: progexp101 (Read error: 104 (Connection reset by peer)) 04:13:54 --- join: MyCatSchemes (n=rb6822@yilko.cs.bris.ac.uk) joined #haskell 04:14:43 hm. Has anyone bootstrapped GHC for FreeBSD7/amd64 yet? 04:15:03 --- quit: apfelmus () 04:17:34 apfelmus: yes, and flip is C. 04:17:46 profmakx-> by looking at ghc state page seems no 04:18:04 well 04:18:10 then i will do that tonight i think 04:18:23 cool 04:24:07 hmm, shouldn't http://darcs.haskell.org/generics/comparison/ have a _darcs directory? 04:24:08 Title: Index of /generics/comparison 04:24:14 --- join: tarrybone (n=dlc@c-71-60-148-169.hsd1.pa.comcast.net) joined #haskell 04:24:22 oh, it's parent has one... 04:25:04 --- join: nominolo (n=nominolo@dhcp-249-5.nomad.chalmers.se) joined #haskell 04:28:17 --- join: progexp101 (i=progexp@gateway/tor/x-10473f2eaa167ca9) joined #haskell 04:31:39 smash your boilerplate is interesting... 04:31:47 --- quit: mlesniak ("ChatZilla 0.9.78.1 [Firefox 2.0.0.7/2007091417]") 04:34:42 --- quit: tarrybon1 (Read error: 110 (Connection timed out)) 04:35:05 --- quit: fritschy ("Leaving.") 04:35:38 --- nick: sioraiocht -> sior|away 04:36:11 --- quit: zx]treads (Remote closed the connection) 04:40:34 --- quit: NichardRixon (Read error: 110 (Connection timed out)) 04:42:28 --- quit: gabkdlly (Read error: 110 (Connection timed out)) 04:42:31 --- join: dogmaT (n=dogmat@r2aj154.net.upc.cz) joined #haskell 04:43:11 --- join: gabkdlly (n=gabriel@dslb-088-073-040-137.pools.arcor-ip.net) joined #haskell 04:43:27 ac: ping, you were having issues with gtk2hs timeouts? what was the problem exactly? 04:43:59 dcoutts_: I think he was struggling to get IORefs working so the timeout could share state 04:44:06 dcoutts_: I believe he got it working in the end 04:44:12 ah ok, good 04:44:21 although I suggested IORefs were overkill and just capturin the state in the closure is easier and more functional :) 04:44:37 --- nick: merus -> meryrus 04:44:55 --- quit: dv^^ (Remote closed the connection) 04:46:20 --- join: dv^^ (i=dv@gateway/tor/x-be71e57c26117f0c) joined #haskell 04:46:36 I noticed somethign about refactoring in haskell, actually 04:46:47 if you make a pretty broad-ranging change to your code 04:46:56 (like changing the data representation of something) 04:47:06 if the old code works, and the new code typechecks, it generally works 04:47:44 this is in contrast to dynamic languages where after a big change you have to check *ever* piece of code which directly or indirectly depends on what you changed... or there may be bugs lurking in rarely used code-paths 04:47:46 --- quit: swiert ("bbl") 04:49:09 quicksilver: I noticed that too, and it gave me a warm fuzzy feeling ;) 04:50:04 ooh, scratch your boilerplate is really neet... 04:50:05 i noticed quite a few positive things about refactoring and changing haskell code 04:51:04 --- quit: yitz () 04:51:23 of course, it's not a rigorous theorem. Clearly you can choose to use wooly dynamic data structures and get wooly dynamic problems :) 04:51:30 but it's an observation about code in practice 04:52:16 --- join: bringert (n=bringert@3-1-5-7a.gva.gbg.bostream.se) joined #haskell 04:52:34 --- join: CindyLinz (i=b91101@bsd7.csie.ntu.edu.tw) joined #haskell 04:53:51 quicksilver: but hopefully you don't use any given one very often in a program 04:54:02 --- quit: bringert (Client Quit) 04:54:13 JHC uses a few dynamic things 04:54:33 I think changing one would still cause the code using it to not typecheck, though 04:54:46 SamB: sometimes I wish for a nice automatic derivation thingy to make wooly structures from wool-free structures 04:55:06 SamB: e.g. given data Foo a = Bar a (Foo a) | Baz (Foo a) (Foo a) 04:55:12 quicksilver: if you could have wool-free ones, why would you need them woolified 04:55:20 --- quit: blarz (Remote closed the connection) 04:55:31 --- join: blarz (n=simon@blarzwurst.de) joined #haskell 04:55:38 or rather if you had already written it that way 04:55:52 wouldn't wool-free be the way you'd want it to stay? 04:55:57 SamB: for generic programming type stuff 04:56:01 oh. 04:56:01 traversing the tree generically 04:56:07 a type for 'current branch' 04:56:15 a type for 'other gunk attached to this branch' 04:56:32 autmatic construction of zipper from data 04:56:34 that sort of thing 04:56:46 It would be sledgehammer to crack nut type stuff, a lot of the time 04:56:47 and very ugly 04:56:53 so it's not a serious proposal :) 04:56:54 --- quit: mr_tenor (Remote closed the connection) 04:57:06 seen Uniplate? 04:57:09 yes 04:57:16 helps somewhat :) 04:57:22 a case in point, actually 04:57:32 suppose you have a structure Foo, and a zipper FooZipper 04:57:37 and you write pretty-printers for both 04:57:48 the pretty-printer for FooZipper does the "same thing" at each actual node 04:57:58 but it passes continuations back up the tree 04:58:09 whereas the pretty-printer for Foo works 'forwards/downwards' like always 04:58:25 so you have a backwards/CPS version, and a forwards/recdescent version 04:58:31 but they do teh same thing at each actual node 04:58:35 and you can't share the code :( 04:58:37 annoying 04:58:47 you sure? 04:58:51 no 04:58:57 I'll hpaste some code if you like 04:59:01 and you can tell me I'm being stupid :) 04:59:14 --- join: zx]treads (n=student@195.251.214.51) joined #haskell 04:59:15 I thought it would be possible to use combinators for traversal which have a different strategy 04:59:34 and you plug in the same action for each node, independent of strategy 05:00:30 quicksilver pasted "zipper pretty print: boilerplate?" at http://hpaste.org/3174 05:00:34 there you are 05:00:47 displayPt for the object itself, displayContext for the zipper 05:00:53 very similar code btu I couldn't see how to unify 05:01:50 --- quit: cbe (Remote closed the connection) 05:01:52 --- join: balodja1 (n=noname@62.117.85.97) joined #haskell 05:02:27 hmm, the types are not isomorphic though, are they? 05:02:35 --- quit: balodja (Read error: 113 (No route to host)) 05:03:10 well zippers aren't normall isomorphic to the object they zip 05:03:12 that's rather the point 05:03:18 --- nick: idnar_ -> idnar 05:03:27 the zipper doesn't have any constructor for the 'Leaf' nodes (Gradient and Noise) 05:03:33 because you can never been 'in' a 'Leaf' 05:03:48 and the zippper has two constructors for binary nodes like Blend 05:03:52 and, of course, the zipper has a Top 05:03:56 --- join: Tychom (n=tychom@m146.net81-64-90.noos.fr) joined #haskell 05:04:25 --- quit: MyCatSchemes ("Swim, swim, hungry!") 05:04:34 --- join: sad0ur_ (n=sad0ur@rb1a240.net.upc.cz) joined #haskell 05:04:46 well, I'm not sure there is much boilerplate in your example really 05:05:08 well the drawStrs and the indents 05:05:09 I mean, the drawStr stuff is different for every constructor 05:05:25 yes, but it's the same for corresponding constructors in type and zipper 05:05:48 the indents look like you could do something 05:06:32 "corresponding constructors" - but how do we know they are corresponding? only by the name I guess 05:07:04 well the correspondance is formalised by the definitions of 'up' and 'down' 05:07:07 (not shown) 05:07:26 e.g: ptc_down' (PTC c (Scale s pt)) = Just$PTC (PTCScale s c) pt 05:07:39 but yes, that is the nub of it 05:07:46 even with the indents, I'm not sure you could generate them automatically, because the pattern is non-regular (cf. PTCBlendRight) 05:07:52 makes me want a third type 05:07:56 'FTNode' 05:08:05 which can then be used in both plain and zipper versions 05:08:54 such that PTCScale s c' becomes (PTC (NodeScale s) c'), while Scale s text becomes PT (NodeScale s) tex 05:08:55 --- quit: gbeshers ("Leaving") 05:09:06 but that doesn't quite work because the arities aren't always the same 05:09:12 and it makes my basic ADT less natural... 05:09:26 hence, my comments about wooly and non-wooly and automatic type generation 05:09:49 type families perhaps? 05:09:58 I wondered that 05:10:07 --- join: Rebooted (n=opera@88-104-185-170.dynamic.dsl.as9105.com) joined #haskell 05:10:08 but I haven't explored them yet 05:10:32 I mean, you could have a totally generic syntax tree type, parameterised by node types 05:10:38 --- quit: Tychom () 05:11:37 --- join: andyjgill (n=andyjgil@c-76-105-238-134.hsd1.or.comcast.net) joined #haskell 05:11:42 data SyntaxTree n = Nullary n | Unary n (SyntaxTree n) | Binary n (SyntaxTree n) (SyntaxTree n) | .... 05:11:55 and then you could have a single generic zipper corresponding to that 05:11:56 --- quit: zx]treads (Read error: 104 (Connection reset by peer)) 05:12:03 and parameterise the whole lot by a node type 05:12:22 ... but, that isn't type safe, because it doesn't force you to only use binary nodes in a binary way 05:12:32 I think you could solve *that* using GADTs 05:13:13 --- part: blueriver left #haskell 05:13:31 --- join: dfeuer_ (n=David@pool-71-163-172-151.washdc.fios.verizon.net) joined #haskell 05:13:59 but whatever you do, it's a whole lot less convenient than your original ADT. 05:14:23 --- quit: dogmaT (Read error: 113 (No route to host)) 05:16:26 --- join: bringert (n=bringert@3-1-5-7a.gva.gbg.bostream.se) joined #haskell 05:17:22 --- quit: nominolo (Read error: 110 (Connection timed out)) 05:17:36 --- quit: bringert (Client Quit) 05:19:37 --- join: Thas (n=robertm@75-172-19-225.tukw.qwest.net) joined #haskell 05:21:57 --- quit: Arnia (Connection timed out) 05:25:03 --- join: boegel (n=boegel@konijn.elis.UGent.be) joined #haskell 05:30:34 --- join: iblechbot (n=iblechbo@ppp-62-216-197-248.dynamic.mnet-online.de) joined #haskell 05:31:10 --- part: Rebooted left #haskell 05:31:18 --- join: matt__r (n=matt__r@CPE-121-209-185-196.nsw.bigpond.net.au) joined #haskell 05:31:19 --- join: Arnia (n=jgeldart@client-82-3-80-61.manc.adsl.virgin.net) joined #haskell 05:32:07 I have a simple cabal script (distribution.Simple) set up and it *seems* to rebuild everything every time 05:34:46 and you're just doing "runhaskell Setup.hs build" between builds? 05:34:51 --- quit: augustss () 05:35:06 yep 05:35:16 I am just getting some empirial evidence now 05:35:32 --- join: fasta (n=fasta@zonnebloem.demon.nl) joined #haskell 05:35:54 time for first build is 26 secs; touch one source file and build again - 27sec. 05:36:30 touch nothing and build takes 24 secs 05:36:31 do you have -fforce-recomp in any file or options anywhere? 05:36:39 give or take a second or two 05:36:46 Why doesn't this work? :t \a -> fst $ runST a By that I mean: is it just an implementation problem or is it a more serious problem? 05:37:04 matthew-_: nope 05:37:31 matt__r: is the filesystem mounted with noatime or anything odd like that? 05:38:02 I can't imagine, this is (fairly) vanilla osx and the files are on my boot disk 05:38:35 :q 05:38:37 matt__r: maybe everything depends on the file you touched? 05:38:39 oops 05:39:03 --- join: augustss (n=Lennart@63.175.43.10) joined #haskell 05:39:05 but what about the untouched one, with no changes between builds, I still get 20+ seconds 05:39:11 matt__r: and you don't use the module system. 05:39:19 (quite unlikely) 05:39:19 you bet I do 05:39:30 matt__r: ok, then file a bug :) 05:39:35 how could you not? 05:40:00 cool 05:40:18 matt__r: how much code are we talking about here? 05:40:32 dumping a few thousand lines of code in a bug report is not a good idea 05:40:46 not huge amounts - ummm, I would guess around 1500 lines ???? 05:41:03 matt__r: you should try to simplify the problem. 05:41:15 and which versions of everything? 05:41:18 A universal simplifier would be so nice :) 05:41:21 you mean before I file the bug? 05:41:24 yep 05:41:51 cabal tells you what the first file is that it's trying to compile 05:41:55 --- join: MyCatSchemes (n=rb6822@yilko.cs.bris.ac.uk) joined #haskell 05:42:00 does that give you any clues? 05:42:08 matt__r pasted "versions of everything" at http://hpaste.org/3175 05:42:37 it just compiles everything everytime 05:42:56 matt__r: when you do what? 05:43:02 regardless of dependencies or even if any of the files changed 05:43:23 ?? 05:43:42 matt__r: you must give the machine some command before compilation begins. 05:43:46 matt__r: what is that command? 05:43:46 do you have another machine you can test on? - will the code compile on a linux or windows box, for example? 05:44:12 runhaskell ./Setup.hs build 05:45:02 matt__r: I have no idea, simplify(test on another platform) and file a bug. 05:45:16 hmmm. does it still do it if you do a clean and configure 05:45:20 --- join: glen_quagmire (n=glen_qua@pool-71-247-255-87.nycmny.east.verizon.net) joined #haskell 05:45:27 Well, I will make a little project that stil exhibits the behaviour and post it up on my public repo and then test it on a few machines tomorrow when I am back at work. 05:45:29 I'm just wondering whether the config files have got confused 05:45:34 thanks for the suggestions 05:45:45 I have tried clean and reconfigure 05:46:05 try another platform if you can to see if it's something odd with your local install 05:46:16 but I'm outta ideas too on that one 05:47:12 --- join: nominolo (n=nominolo@dhcp-248-121.nomad.chalmers.se) joined #haskell 05:47:15 or just install linux on your mac, replacing OSX. Then try again. 05:47:17 * matthew_- ducks 05:48:07 > 2**(-1) - 2**(-55) -- how come when exponent is different by 54, the smaller number becomes 0.0 ? 05:48:09 0.5 05:48:38 @go what every computer scientist should know about floating point 05:48:40 http://docs.sun.com/source/806-3568/ncg_goldberg.html 05:48:40 Title: What Every Computer Scientist Should Know About Floating-Point Arithmetic 05:48:40 explains 05:49:24 > 2^^(-1) - 2^^(-55) 05:49:25 0.5 05:49:32 > 2^^(-1) - 2^^(-55)::Rational 05:49:33 18014398509481983%36028797018963968 05:49:59 > 2**(-1) - 2**(-30) 05:50:00 0.4999999990686774 05:50:04 > 2**(-1) - 2**(-30)::Float 05:50:05 0.5 05:50:20 for Float it's different by 25 05:50:28 fasta: did anyone answer your $ runST question? 05:50:32 quicksilver: no 05:50:46 --- join: swiert (n=wss@sneezy.cs.nott.ac.uk) joined #haskell 05:50:46 > 2**(-124) - 2**(-149) :: Float 05:50:48 4.7019774e-38 05:50:48 --- join: cmarcelo (n=cmarcelo@200.184.118.132) joined #haskell 05:50:50 fasta: it's because $ is polymorhic 05:50:57 > 2**(-124) :: Float 05:50:58 4.7019774e-38 05:50:59 fasta: but polymorphism is only over rank-1 types 05:51:15 fasta: so a polymorphic function can't be instantiated at a rank-2 type like runST 05:51:15 even for denormalized numbers, it's number of fraction bits + 2 05:51:23 quicksilver: that doesn't answer the question, does it? 05:51:26 > floatDigits 0 05:51:27 53 05:51:30 fasta: I'm getting there :) 05:51:36 > floatDigits (0::Float) 05:51:37 24 05:51:40 quicksilver: (since that part I already knew) 05:51:42 fasta: the reason for this restriction is to make type inference work 05:51:51 Float has 23 bits for fraction . So, exponents should be different by 24 or more. For Double, 52 + 2 = 54 05:51:57 I can't find why +2 is needed 05:51:59 --- quit: iblechbot (Read error: 110 (Connection timed out)) 05:52:10 s/24/25 05:52:16 glen_quagmire: It will always be one more than the numbe returned by floatDigits 05:52:24 fasta: the augmented H-M algorithm used for type inference makes certain assumptions which wouldbe violated by instantiating polymorphics at rank-2 types. 05:52:29 at least assuming isIEEE is true 05:52:35