00:00:00 --- log: started haskell/02.01.09 00:11:05 --- join: shapr (~user@195.156.199.178) joined #haskell 00:18:13 --- quit: shapr (Remote closed the connection) 00:29:44 --- join: julien (~julien@uu212-190-122-70.unknown.uunet.be) joined #haskell 01:02:51 --- quit: xbill ("Leaving") 01:09:58 --- join: ktk_ (~ktk@jeremeydsl-7.mylinuxisp.com) joined #haskell 01:11:56 --- quit: ktk_ (Client Quit) 01:14:01 --- quit: jlb-wireless (Read error: 104 (Connection reset by peer)) 01:22:07 --- join: jlb-wireless (~ktk@jeremeydsl-7.mylinuxisp.com) joined #haskell 01:23:16 --- quit: jlb-wireless (Read error: 54 (Connection reset by peer)) 01:28:24 --- join: teek (teemu@k72.viikki.hoasnet.fi) joined #haskell 02:55:35 --- join: shapr (~user@195.156.199.178) joined #haskell 02:58:05 hi again 02:58:30 anyone awake who has a clue about the ST monad? 03:08:25 only part one 03:08:35 heh 03:08:38 hi djul 03:09:24 hey djul, have you linked djul to jewel? 03:10:30 --- join: ski (~md9slj@fraggel84.mdstud.chalmers.se) joined #haskell 03:11:18 --- nick: djul -> jewel 03:11:31 it is just my alternate nick when the other is used (when servers die) 03:12:13 hi 03:13:37 anyone asked about ST monad ? 03:19:53 not much talk here now, it seems 03:30:00 shapr asked 03:30:35 i saw that on the log on tunes, which i just discovered 03:30:48 nice thing 03:31:11 read a little recorded from #compilers 03:39:14 arg, which package has ord and chr| 03:39:44 * shapr tries to figure where ord went 03:39:53 aren't ord in Prelude 03:40:14 probly so, but I can't seem to use it in ghci 03:40:22 do I need to import a module in my code? 03:40:38 i'll start up ghci and look 03:42:17 i seem to recall that hugs has conflated some things into Prelude that don't actually belong there, let's see if i can find ghci's Prelude 03:42:21 * shapr tries :set -module text 03:42:37 er, that should be -package 03:43:36 doesn't seem to work 03:44:04 hrm 03:44:08 I wonder where it went 03:44:11 I saw ord and chr last night 03:44:32 hi again ski! 03:44:37 i'm not so used with ghci, have it only at school 03:44:49 can you use ord and chr in hugs? 03:45:02 well, now you recognised me. hi shapr ! 03:45:43 oh, I have to ask if you know John Hughes.... isn't he the guy who came up with the idea of arrows, and is teaching at chalmers? 03:46:14 jewel: you can link djul to jewel and use both 03:46:19 (well i tried to install ghc at my brothers computer, hasn't got the time to try on my own,ghci seemed to work, but not ghc. i don't have cygnus, but i have djgpp, maybe cygnus is required) 03:46:34 I don't think so 03:46:41 but then I use linux, I don't really know 03:46:55 shapr : yes, i can use ord and chr in hugs 03:47:16 hm, weird 03:48:46 shapr : yes i have met John Hughes and talked with him a little (cool !!), there are talks here in a research group many thursdays, but anyone interested is welcome, so i attend sometimes :-) 03:49:18 wow, awesome 03:49:33 i ought to have install linux a long time ago, but i'm so lazy ;-( 03:49:37 I must show up sometime and listen to some of the talks 03:50:15 I wish ghci had a dir() function like the Python interpreter 03:51:51 do so, if you have the opportunity (of course there are various speakers at different meetings (it's called multi meetings, because it's a metting attended by research figures from many of the research groups here (functional, formal method, ...))) 03:52:27 I'd really like to hear John Hughes speak 03:52:41 are the talks often in English? 03:53:58 most often, i seem to recall (i think John can speak at least some swedish but sometimes there are attendants from other countries there (or studying persons at chalmers from other countries)) 03:54:45 * shapr finds chr in PrelBase 03:54:51 seems I should be able to find it in ghci then 03:55:19 hugs' Prelude seem to be divided in sections : {Prelude},{PreludeRatio},{PreludeList} 03:55:31 --- join: clausen (yloyka@2-088.ctame701-1.telepar.net.br) joined #haskell 03:55:51 it's also part of Char 03:55:54 hmmm 03:55:59 PrelBase ... , that could be it. but shouldn't it be automatically imported 03:56:38 * shapr grumbles 03:56:45 how do I cast an Integer to an Int again? 03:57:04 I found out I can import Char, and then call Char.chr 03:57:06 isn't it fromInteger ? 03:57:22 * shapr tries that 03:58:08 oh 03:58:09 cool! 03:58:23 what ? 03:58:28 it works 03:58:43 how to import Char in ghci, me not understand ? 03:58:47 but I have an error in my program somewhere... 03:58:58 I did "module Main where import Char" 03:59:03 then later I called Char.ord 03:59:07 dunno why that works 03:59:13 oh, of course you imported it in your source file 03:59:24 it also works in ghci 03:59:29 :set -package Char 03:59:33 Char.ord 03:59:33 i tried both :l Char and :s -package Char in ghci 03:59:43 still didn't work? 03:59:47 unknown package name: Char 03:59:51 whoa, werd 03:59:52 weird 04:00:06 maybe the debian package has some extras installed? 04:00:39 maybe, or maybe the libraries aren't correctly configures at school, but i wonder ? 04:01:07 I dunno 04:01:13 there's probably a sensible answer 04:01:16 I just don't know it 04:01:20 gotta reboot, brb 04:01:38 --- quit: shapr (Remote closed the connection) 04:14:05 --- join: bsh (~MasterBot@dsp-202-72-158-119.manjimup.westnet.com.au) joined #haskell 04:14:49 hey 04:14:54 hello bsh 04:16:47 we talk more for a few minutes ago, but shapr had to reboot his computer 04:17:02 ah ok ;) .. what is haskell exactly? 04:17:10 * dmiles is back (gone -23:54:44) mIRC xxxScript 04:17:37 oh, haskell is an interesting (IMO) functional programming language 04:18:24 ok.. is there an official site I can get some information about it from? 04:19:07 it's main features distinguishing it from other functional PLs is monads,purity and laziness 04:19:24 well http:/www.haskell.org is a good place to start 04:19:50 oh ok.. :p 04:19:54 also un usenet : comp.lang.functional (amongst other functional PLs) 04:20:16 there seems to be a few mailing lists (that go way over my head) 04:20:22 * clausen really should do some reading! 04:20:28 and wiki : http://www.haskell.org/wiki/wiki 04:20:52 ok, thanks.. i've been looking into alot of PL development lately 04:21:24 yes, i've looked at some history of it somewhere on www.haskell.org but iv'e not participated myself (yet atleast) 04:21:49 ok.. What languages can you code in? 04:22:03 and of course at least this IRC channel :-) 04:23:02 ;) 04:23:13 so if you have any direct questions right away, we could try to answer them right now, if you want to/have time/... that is 04:23:49 What language is haskell developed in? 04:24:48 me ? well i can code in PB-100F BASIC, C64 BASIC, 6502 Assembler, C, Java, Haskell and a little Scheme (+Lisp) and ML (SML+OCaml), that's it i believe 04:25:29 Cool.. I'm just into C .. I know a little BASIC 04:25:40 well many more or less developed PLs have a tendancy to prefer being developed it itself 04:26:22 so one haskell compiler, GHC, is coded in haskell, but Hugs, the haskell fast-cycle interpreter is coded in C 04:26:44 ok 04:27:41 well C is good in it's own way for certain tasks (and if you accept the design decisions), though it has some warts and new ideas have been developed since C was invented 04:28:38 Yeah.. i've been working on an interpreter ... not a very good one - but yeah 04:28:40 BASIC some say isn't a real progemming language, firstly there aren't one BASIC but many many dialects 04:29:01 Yeah true 04:29:45 and also libraries are hard in BASIC, and i consider them important. BASIC (more or less original) lacks any uniform way of calling functions/procedures 04:30:11 interpreter written in what language, for what language ? 04:30:32 I wrote it in C... it was more of a personal language I was writing.. 04:31:26 is it a interpreter for C or for any other lang (or own invented) ? (curious) 04:31:32 ok 04:32:11 own invented :p 04:32:33 (i forgot, i know a little 680x0 assembler also, nice assembler) 04:33:01 I also know a little C++.. just dont normally use it 04:33:02 what type of language (imperative, i presume ?) 04:34:11 yeah, imperative 04:34:19 well, i took a C/C++ course some years ago, and learned not C and not C++ but some weird mixing of concept. since then i have managed to learn C properly on my own, C++ i only know a little 04:35:00 yeah.. All C/C++ i know i've taught myself.. I'm also into Linux/*bsd administration 04:36:38 you'd like to elaborate on the language and interpreter a little, or you'd rather talk about haskell or functional programming (FP) or connected ? 04:37:53 Well the language I was writing was much like GWBASIC 04:40:38 any user-defined functions/procedures ? 04:41:20 nope, not yet implemented 04:42:11 good luck (you also have to decide whether to support recursion or not there) 04:42:20 bsh: you might find the "Dragon Book TM" inspiring / helpful 04:42:39 clausen: interesting.. thanks 04:42:48 bsh: the real name of the book is 04:42:56 * clausen googles 04:43:06 hehe :p 04:43:21 well, it's by Aho, Sethi and Uhlman 04:43:26 that I remember (spelling?) 04:43:29 iv'e heard that book name, but i'm not sure of what written in it 04:43:52 i'm on holidays at the moment.. I live in Sydney, Australia - Im in Western Australia right now.. 04:44:27 * clausen lives in Melbourne / Australia 04:44:31 * clausen in Curitiba / Brazil ATM 04:44:45 okeydokey, so you are from "down under" 04:44:50 Ah cool.. i've never been to Melbourne 04:45:12 ski lives in Gothenburg / Sweden 04:45:18 http://www.amazon.com/exec/obidos/ASIN/0201100886/ref%3Dnosim/pineapplesoftonl/002-4289326-7069634 04:45:21 ah ok ;) 04:45:25 clausen: thanks 04:45:56 clausen: why are you in Brazil? 04:46:07 bsh: holiday 04:46:12 bsh: I used to work for a .br company 04:46:16 ahh, it's about compilers :-) 04:46:20 bsh: but, now I'm just meeting friends 04:46:21 ah ok 04:46:32 anyway, dragon book is mainly about: 04:46:41 lots of research about parsing 04:46:48 (lexical analysis, different types of parsers, etc.) 04:46:52 and lots of practical advice too 04:46:59 but lots of cool theory :) 04:47:01 (really cute!) 04:47:14 clausen: cool :) 04:47:20 and also, stuff about code generation, optimizing code, etc. 04:47:42 clausen : iv'e got one compiler book : Compiler implementation in ML 04:47:53 Have you ever looked into OS development? 04:48:03 bsh: yes 04:48:28 * clausen maintains GNU Parted, and has to talk lots to Linux kernel people... 04:48:38 --- join: shapr (~user@195.156.199.178) joined #haskell 04:48:44 bsh : not seriously, just surfed and looked at some pages a little (e.g. tunes) 04:49:07 (and hurd people) 04:49:16 * clausen also read a few books 04:49:26 clausen: cool.. I've been looking into a fair bit of OS development 04:49:40 * clausen would like to go back to compilers 04:49:43 (haskell!) 04:49:43 shapr : iv'e saved some (perhaps trivial) replies for you but, i think i wait a little with them 04:49:48 ghc looks rather heavy 04:49:48 ok :) 04:49:53 ghc is nifty 04:49:53 * clausen will need a lot of time to understand it 04:50:10 and I'm stil spending lots of time on Parted 04:50:16 (grrr, it's hard to decide what to DO with life!@) 04:50:23 program! 04:50:24 write code! 04:50:38 imho, code is second only to women ;) 04:50:39 absolutely ! 04:50:39 shapr: yeah, but there's always another thing to write on parted 04:50:47 and sleep is way down the list %-D 04:50:48 hehe 04:50:57 * clausen can't program when he's tired 04:51:04 I can, though not very well 04:51:10 hehe :p 04:51:18 he 04:52:26 do you dream about code or coding or coding problems sometimes ? 04:52:54 i do. often weird abstract dreams ;-) 04:53:04 * clausen dreams about code 04:53:07 particular when hacking compilers 04:53:11 particularly 04:53:12 yah I dream about symbols lots 04:53:26 I have programming dreams all the time 04:53:52 but it's usually stuff like a chess board that I can place pictographs on to represent a programming problem 04:54:08 sometimess it seems that i reason about a problem in sleep, but when i awake most of it's gone/forgotten and the rest don't often make sense 04:54:25 or sometimes rosewood colums, rails, and blue marbles that form a physical model of a process 04:55:14 has everyone here seen the movie named memento? 04:55:21 that's a programmer's movie in my opinion 04:55:38 yep 04:55:40 well, i ought to try to remember dreams more, but it's difficult, especcially if you have irregular sleeping habits and are not accustomed of recalling dreams, bwahaa 04:55:44 shapr: I agree 100% 04:55:53 * clausen thought: that's me! 04:55:57 ski: keep a notebook and a pen next to your bed 04:55:57 * clausen has a terrible memory 04:56:36 i have succeeded with that thrice i think, have to recall/remember to try it more often 04:56:39 last night I dreamed about using Monads to sequence actions in Tribes2 %-O 04:56:47 that was weird 04:56:50 cool 04:57:58 often i at least try to mnemonically induce my (hypothetical ?) sub-conscious to recall dreams and perhaps even have lucid dreams 04:59:42 where a lucid dream is defined as a dream where you are aware that you are dreaming when you are dreming. this often means that you can willfully control and change your dream (and fly , etc ...) 05:02:17 * jewel dreamt about creating a StarCraft type game where players can write their own ai logic for each of the unit types 05:02:39 i've had (at least one for sure) such dream, but it only lasted half a minute or something, the damn thing is unstable ! like balancing on an edge, if you get too emotionally involved (yess, i've finally succeeded) then you most often cause yourself to awake, to the other extreme lies the danger of losing lucidity and the dream becomes an ordinary dream again (with you non-lucid following your act role in the dream) 05:04:17 hmmm, iv'e thought of some type of game where one should be able to define "procedures" for common actions and combining them, perhaps even tune their own ai so that they don't have to comman all detailed actions explicitely 05:04:34 yah, that kind of thought goes through my head really often 05:04:50 the first "dark reign" had a good chunk of that ability 05:05:04 (sorry for rambling about lucid dreaming so much, but the concept makes me excited) 05:05:10 you could set pursuit, heal, and reaction 05:05:12 no worries 05:05:20 I've had lucid dreams myself 05:05:36 it's easy for me to do lucid dreaming because I always know I'm dreaming 05:05:56 what type of game is dark reign ? strategy ? real-time ? 05:05:57 but usually I try to follow the role so I can listen to what my subconscious is saying 05:06:08 yup, dark reign is real time strategy 05:06:14 it's a bit older 05:06:21 I ran it on my p90 with no problem 05:07:09 mindrover takes that idea to the other extreme 05:07:20 you can only program your unit 05:07:29 okeydokey 05:07:36 you get to build it out of components and wire them all together, it's quite nifty 05:07:56 interesting 05:08:04 yah, and I got the linux version of course ;) 05:08:27 good, good 05:08:45 personally I'd like to write a 'life' game that lets you build your own automata 05:09:07 hmm, that shouldn't perhaps be to hard 05:09:15 that way you'd end up with more 'memetic' programming, rather than just scripting existing units only 05:09:58 mindrover gives you some small ability to do the same thing 05:10:19 in some parts of it, your two robots are on one team, playing soccer against another team 05:10:37 and one of the components is a radio that sends a range of integers to your team 05:11:10 shapr : all your dreams (at least recalled ones) are more or less lucid. envy,envy ;-) 05:11:51 the integers denote what ? 05:12:20 whatever you want them to denote 05:12:25 you just receive an integer 05:12:43 you can use it to change modes from offense to defense or whatever 05:12:46 ahh, so you can decode it in each unit, building an own mini-protocol ? 05:12:52 right 05:13:21 I haven't built a team yet, but my first thoughts were to use it to switch from offense to defense 05:13:43 I had also considered passing some info about enemy location that way 05:14:09 so the orders arn't directly automagically transferred to the units at "run time", you have to distribute it over some channel (the radio integer stream), interesting idea 05:14:52 psst, try the mindrover demo... http://www.lokigames.com/ 05:14:58 * shapr pretends to be a drug dealer 05:15:02 the demo is free ;) 05:15:26 thanks 05:16:01 mindrover is good because it's a really fun game that has a deterministic ending 05:16:19 aka, you can get over it, you don't stay addicted 05:16:23 oh, about the ST monad... 05:16:45 well, what about it ? speak up, man ! 05:16:51 I recently read the paper about arrows 05:16:59 and in there he shows this: 05:17:00 yes ? 05:17:03 tick :: STRef Int Int 05:17:04 tick = readSTRef >>= \n -> 05:17:04 writeSTRef (n + 1) >>= \() -> 05:17:04 return n 05:17:17 well, he uses StateMonad in place of STRef 05:17:21 and he uses fetch and save 05:17:38 how can I turn the proto code he used into working code? 05:18:47 well, in the "real" ST monad the type of tick would be ST s Int or something similar, i think 05:19:30 his monad "STRef" seems here to only contain a single state (variable), in this case of type Int (STRef *Int* Int) 05:19:41 he actually just calls it StateMonad 05:19:45 I was trying to make it real 05:19:52 so I put in STRef 05:19:57 and his read and write seems to have type : 05:20:06 readSTRef :: STRef s s 05:20:20 writeSTRef :: s -> STRef s () 05:20:59 ok StateMonad (whatever :), that sound like a more familiar/convetional name for such a single state monad 05:21:46 which module has ST? 05:21:48 but in the ST monad in hugs (i'm almost sure this is correct for GHC too, and perhaps for nhc and hbc (i don't know)) 05:22:03 these operations have type : 05:22:24 newSTRef :: a -> ST s (STRef s a)) 05:22:43 readSTRef :: STRef s a -> ST s a 05:23:02 writeSTRef :: STRef s a -> a -> ST s () 05:23:08 that's it i think 05:24:05 which module has ST? well in hugs it's ST (or LazyST if you want lazy state) 05:24:25 tick :: ST s Int 05:24:25 tick = readSTRef >>= \n -> 05:24:25 writeSTRef (n + 1) >>= \() -> 05:24:25 return n 05:24:31 hm, still has compile problems 05:24:41 of course i forgot the most important ST function : 05:24:44 Couldn't match `ST s' against `(->) (STRef s1 a)' 05:24:44 Expected type: ST s a1 05:24:44 Inferred type: STRef s1 a -> ST s1 a 05:24:44 In the first argument of `(>>=)', namely `readSTRef' 05:24:44 In the definition of `tick': 05:24:44 readSTRef >>= (\ n -> (writeSTRef (... + 1)) >>= (\ () -> ...)) 05:24:49 * shapr does a minor flood 05:25:04 runST :: (forall s . ST s a) -> a 05:25:58 well writeSTRef takes an STRef as an argument also (readSTRef too) 05:26:13 what's the equivalent for ST ? 05:26:15 you have to create it with newSTRef 05:26:24 oh.. 05:27:37 if it's supposed to be a "global" STRef you can use unsafePerformIO (horror !!), John Hughes had a talk about ways to overcome this for a while ago 05:28:16 or you can make it a parameter to tick (or define tick as a local function in a where/let where you already have called newSTRef) 05:29:08 i think a parameter to tick would be nice in this case (so it can tick any Int STRef-Box you point it at) 05:29:27 --- quit: bsh ("Its better to be hated for who you are than to be loved for who you are not.") 05:29:58 but perhaps you are only going to use one box anyway so perhaps some of the other ways are better in your case 05:30:26 let's say you choose to make the STRef a parameter to tick 05:30:45 then tick might perhaps be defined as follows : 05:31:02 tick :: STRef s Int -> ST s Int 05:31:18 tick ref = readSTRef ref >>= \n - 05:31:32 (should be > at end of previous line) 05:31:55 writeSTRef ref (n + 1) >>= \() -> 05:32:03 return n 05:32:18 does that work ? 05:32:38 * shapr tries it 05:33:26 in other cases tick wouldn't have a ref parameter, but ref would be defined at toplevel (with unsafePerformIO) or in an enclosing context (innner definition) 05:34:02 what's up with unsafePerformIO? 05:34:14 I keep hearing about it, but I have no idea what's unsafe 05:34:17 maybe iv'e confused things a bit : it may be that the unsafePerformIO option only works with IORef and not STRef, not sure :-( 05:34:33 * shapr has negative clue 05:34:51 well the type of it is : 05:35:01 unsafePerformIO :: IO a -> a 05:35:48 one way to look at the IO type is as if it had been defined as (i think GHC does something like this) 05:36:07 newtype IO a = IO (World -> (a,World)) 05:36:46 where a value of type World represents the state of the entire outside world (including filesystem,graphics, etc ...) 05:37:26 hm 05:37:31 that almost makes sense to me 05:37:52 (the only crux is that values of type World shouldn't be duplicated or discarded, otherwise we'd have to discard the nice efficient implementation of World as side-effects on the outside world) 05:38:14 so essentially unsafePerformIO has this type : 05:38:30 unsafePerformIO :: (World -> (a,World)) -> a 05:39:03 oh 05:39:25 so it gets a functionm invents a new World, sends it to the function, recieves the pair of the result, and the possibly updated world, and discards the new world and returns the result 05:39:39 it breaks referential transparency 05:40:06 I'm still not clear on referential transparency 05:40:09 with it's help you can call side-effecting code from within "non-sideeffecting pure code" 05:40:16 intuitively I understand it, but not all its implications 05:40:28 say you define : 05:41:04 foo :: () -> Char 05:41:19 foo () = unsafePerformIO getChar 05:41:35 then foo () isn't the same as foo () !! 05:42:37 (we have to make it a function, in this case by adding a dummy unit argument, because otherwise it would be evaluated once and then cached (it's lazy !)) 05:42:57 (i.e if it's defined at toplevel) 05:43:34 right, I understand that 05:43:36 BUT 05:43:48 why does wrapping it in a Monad suddenly make everything ok?? 05:43:55 and why do you say "lifted" into a Monad? 05:44:27 in some cases one can defined a library more efficiently with side-effects than without, but the libraries interface to the outside is pure, only local side-effects doesn't harm any body (if it's not detectable from outside) 05:44:41 the only thing I can think of is that Monad means "safely functional, plus a side effect we're safely hiding" 05:45:18 --- join: xbill (wli@DOMINIA.MIT.EDU) joined #haskell 05:45:24 because calling a (pure) function returning some IO type doesn't perform any side-effects (as opposed to e.g. C) 05:46:06 it only return an object (call it encapsulated-side-effect or side-effect-object if you wish) 05:46:17 an object that represents an action 05:46:36 oh 05:46:48 then >>= and >> combines two objects representing actions to a combined object 05:46:52 so how do get rid of that object and just get the result? 05:47:04 yah, >>= makes me think of rings on a string 05:47:14 you define main to be of an IO type (maybe only IO () is permitted) 05:47:39 eg. all parts of a haskell program have unknown order of resolution except the guys with >>= between them 05:47:54 they require that one happens after another 05:48:08 then conceptually the pure functions calculate the final (main) side-effect object which then magically (by the ren-time system) gets executed and side-effects performed 05:49:16 yes >>= and >> sequences the side-effects (so they are not indeterministic) 05:49:28 but only relative to each other 05:49:37 yes 05:49:38 which still allows the Haskell implemenation a lot of leeway 05:49:56 nifty 05:50:11 yes, it the "as if" rule, if it behaves as it should, another implementation is allowed 05:50:35 I thought it was monad composition 05:50:41 i.e. one function takes the world 05:50:44 and returns a world 05:50:47 which the second function uses 05:50:50 and returns another world 05:50:52 i.e. 05:51:03 so in practice the pure and side-effecting calculations/computations are done intermixed with each other (and possibly parallell on such architectures) 05:51:09 type Monad :: World -> World 05:51:25 (data?) 05:51:31 (my haskell rather rusty...) 05:51:44 a computation can both affect the global world state and return a result 05:51:55 right 05:52:04 newtype IO a = IO (World -> (a,World)) 05:52:12 (or data is well too) 05:52:16 ok, so your definition of tick worked, but I can't figure what to with newSTRef 05:52:35 anyway, doesn't this define a strict ordering? 05:52:40 well, how are you trying to use tick in the program ? 05:52:44 s/ordering/order/ 05:52:49 of evaluation? 05:52:50 anyway at all :) 05:53:57 clausen : you mean strict ordering of side-effects (but the "as if"- rule can sometimes apply here to think of disjoint state-parts of the world as filesystem and graphics can safely 05:54:32 (can often safely change actual execution order, they are commutative) 05:54:58 ski: can you show me code that would create an STRef and then a function to increment its state? 05:55:32 ski: but, aren't they operating in the sae world, anyway? 05:55:41 (does haskell have a way of separating worlds?) 05:55:47 well, if you have some main test function (say main or test), use newSTRef in that one to create a new (initialized) box and pass it to tick, und so weiter 05:55:50 * clausen has a dogey M key 05:56:19 I guess I should just read more 05:56:52 --- quit: teek ("Coffee break") 05:56:57 clausen : well if they don't affect one another (security breach : the graphics affect humans which affect file-systems ;-) 05:57:18 ski: but, how does HASKELL do it? 05:57:55 ski: indeed: the file was successfully copied 05:58:37 clausen : it should be no problem it think (Clean does something like this, i think. Clean is lazy and has typeclasses like haskell. but it has an explicit (not hidden if present at all) World type, because, their type-system can express that some values are linear, i.e. can't be used twice, can't be discarded in any way 05:59:06 basically the problem is: 05:59:09 main is a monad 05:59:16 poor ski trying to have two conversations at the same time ;-) 05:59:18 how do you join two monads of different worlds? 05:59:23 (s/join/compose/) 05:59:25 hehe 05:59:48 * shapr humorously guesses: zipWith (++) [monad1,monad2] 05:59:49 wait a minute, so i can read thru what you are saying, please :-) 06:00:53 data World = (part1, part2, part3) ...? 06:01:02 and only "use" part1? 06:01:46 (actually, you could make a list, and allocate states from the world) 06:01:51 * shapr is able to have many irc conversations and work at the same time 06:01:54 clausen : main is just magically executed at top-level execution,one can then of course speculate about the philosphical implications of this, and the meaning of two communicating haskell programs with one main each ... 06:02:24 ski: well, imagine trying to do: 06:02:47 ski: copy_file >>= paint "file arrived" 06:02:52 data World = *(*Part1,*Part2,*Part3) 06:03:10 ski: (copy file1 file2) >>= (paint "file arrived") 06:03:11 or whatever 06:03:15 * clausen forgets the notation 06:03:15 shapr : lucky you 06:03:26 what does * mean? 06:03:43 anyway, the two different monads 06:03:47 are from different worlds 06:03:57 (the FS world, and the graphics world) 06:04:01 well it's clean notation of saying that the type must be used linearly 06:04:24 linearly? 06:05:24 a linear value is one that must be named (used) exactly once, no more, no less (bindings don't count as using but getting) 06:06:02 interesting 06:06:11 it's built-in in Clean's type-system (based on linear logic developed by Girard) 06:06:12