00:01:20 -!- MDude has joined. 00:02:10 elliott_: also, foldr performs tail call elimination. 00:02:14 kallisti: You're an idiot. 00:02:23 NEXT PLEASE. 00:02:43 elliott_: it does. that's precisely why it's constant space. no stack frames are allocated. 00:02:50 aside from the first one 00:03:19 ... oy. 00:03:20 fungot: So tell me about recharging Moon Stones? Is it true that it takes aeons? Will we see that gorgeous glow again? 00:03:20 fizzie: to the northwest of this cape. he took back the medal from the frog king. and i'd like to see that mystical sword for myself! geez! 00:03:36 foldr does not /perform/ tail call elimination, foldr /benefits/ from tail call elimination. 00:03:39 -!- itidus21 has quit (Ping timeout: 244 seconds). 00:03:51 Gregor: sooooo pedantic. for fucks sake. 00:03:52 Gregor: It doesn't do that, either. 00:04:00 kallisti has no idea how the call stack works in Haskell. 00:04:09 And I for one am not going to be the one to teach him how it does. 00:04:22 Ohwait, foldl would actually benefit, wouldn't it, not foldr ... 00:04:25 Gregor: I mean, yes, it's guarded recursion, i.e. the recursion is a non-strict field of a constructor in WHNF. 00:04:28 But that's not tail recursion. 00:04:39 Tail recursion is a purely syntactic property that is roughly 100% irrelevant in Haskell. 00:05:17 elliott_: but it's the same optimization as tail call elimination. if it weren't, the stack would overflow as the list is traversed. 00:05:53 kallisti: Have I mentioned you're an idiot? 00:06:03 no. 00:07:20 See, usually when I call people idiots, I make some attempt to explain to them why they're wrong. Also fun fact: This offer is voided when said person has a history of being adamant and sarcastic in asserting 1they're correct in the face of someonew ho knows more about them about a topic! 00:07:28 "The fundamental cause of the trouble is that in the modern world the stupid are cocksure while the intelligent are full of doubt." --Jesus 00:07:53 elliott_: perhaps you think I'm talking about tail recursion in a strict language. 00:07:58 that would be silly, since I'm talking about Haskell. 00:08:42 Perhaps I'm not making any unwarranted or unreasonable assumptions, know exactly what you're trying to say, and you're just wrong. 00:08:48 fizzie: Thx for -c btw, it's best. 00:15:10 elliott_: ...I 00:15:15 okay, you're not going to give me a chance. 00:15:21 waste of time. 00:16:04 kallisti: I've given you about a hundred chances, it's called every single time you've done this. 00:16:28 I'm just stupid enough to keep biting the bait. 00:16:55 elliott_: first of all, I know how Haskell's "call stack" works. 00:18:09 That's what you say literally every time. 00:18:22 I've made multiple claims to knowing about Haskell's call stack? 00:18:38 You understand everything perfectly, you're just having a perfectly reasonable and valid disagreement among peers, until a few days later someone manages to bang into your head why you've made a fundamental misunderstanding, which you then claim was minor. 00:18:43 Seriously, I'm not interested. 00:19:08 It's even more a waste of time than the rest of this channel's topics but at least those are entertaining. 00:19:42 elliott "too good for #esoteric" hird 00:20:46 Yes, that's absolutely what I said. 00:21:32 Y'know, I'll bet if you /generated/ timelines instead of having presets (or as well as having presets), you could actually do an arbitrary-date time travel game reasonably ... 00:23:01 Feh, generating arbitrary timelines would be tough though X-D 00:23:15 Probably not impossible >_> 00:23:45 You'd have to find a way to do it on a massive scale and then sub-divide for individuals without making the whole no longer equal the sum of the parts. 00:23:58 Plus of course certain people are more important to history than others ... 00:25:27 Gregor: Remember when you had that awesome time raytracing thing and it turned out to be boring? :( 00:25:42 * elliott_ still bitter 00:25:57 elliott_: in fact I'm also aware that writing a function with "tail calls" and an accumulating list can be slower than an equivalent function that simply constructs the list directly. 00:25:59 -!- DeadlyFugu has quit (Read error: Connection reset by peer). 00:26:24 elliott_: because I've written both of them and observed that constructing the list directly is faster. 00:26:32 -!- DeadlyFugu has joined. 00:27:24 elliott_: I'm not talking about a syntactic tail call. 00:27:48 Tail calls are about complexity, not speed. And the definition of "tail call" is syntactic (consult e.g. the Scheme specification if you disagree). And I'm not arguing this so why am I even typing this. 00:27:51 Please stop. 00:32:19 elliott_: I really don't think there's anything to argue about. 00:32:44 another thing I "always" do, therefore: the result must always be the same because lol 00:43:57 I think I will go to bed. 00:49:48 hmm interesting. 00:50:02 so if I use Word -> [Word] for my factoring function, it runs extremely fast. 00:50:09 ah wait nevermind 00:50:44 okay. sanity restored 00:51:42 -!- elliott_ has quit (Ping timeout: 244 seconds). 00:59:53 @tell oerjan do you think there could be a faster way to write integer squareroot than floor . sqrt . fromIntegral 00:59:53 Consider it noted. 01:02:44 What's wrong with that? 01:03:08 NOT FAST ENOUGH 01:03:14 I think I've found a faster intsqrt 01:03:30 in Math.NumberTheory.Powers.Squares 01:05:15 Sgeo: nothing is wrong with it. 01:05:55 in fact since I only do one sqrt this is entirely a waste of time. 01:06:39 Clearly, you should use the magic number thingy 01:07:04 Oh, that's the reciprocal of the square root 01:07:38 Also, not integer I guess 01:10:09 wow amazing 01:10:17 no significant improvement in speed 01:10:28 from optimizing one square root operation. 01:10:51 ~.53 to ~.52 01:14:03 (seconds) 01:14:12 (Use ministat (from e.g. FreeBSD) or something equivalent for your benchmarking before calling something significant) 01:14:36 Deewiant: what? I can't just spam time over and over and guesstimate? 01:14:53 anyway I don't think this is measurably significant. 01:14:56 http://www.independent.co.uk/travel/news-and-advice/airlines-cleared-to-use-santas-shortcut-6281263.html What will this do to the Flat Earth society? 01:14:57 Generally speaking no 01:15:17 http://sprunge.us/LBOd any suggestions? 01:16:30 EQ | m == 0 -> [i]; _ -> [] 01:16:36 Not that it should matter 01:17:18 ...why would I do that? 01:18:27 looks to be about the same. 01:18:33 Wait, they consider the edge to be in the Antarctic, right? 01:19:08 I think to make this any significantly faster I'd need to change the algorithm. 01:19:19 I've pretty much optimized this one as much as possible. 01:19:30 Branch prediction should handle most of this kind of thing anyway 01:20:18 You can try using 'if m == 0' as a higher-level check if you want, though 01:21:16 I don't think that will do anything, and will make the code uglier probably. 01:22:11 if' t f True = t 01:22:17 if' t f False = f 01:22:33 Hmm, wait 01:22:57 That makes it hard to do an if based on (==0)\ 01:22:59 I think 01:23:12 ...what. 01:23:38 heh, I could make the code really ugly and unroll the loop a little. :P 01:24:20 for probably absolutely no gain. 01:27:55 -!- MDude has quit (Ping timeout: 240 seconds). 01:36:36 -!- zzo38 has joined. 01:41:51 zzo38: how do you print a Word#? 01:43:32 kallisti: Maybe you use a constructor, is it called W# or something like that 01:43:59 hm 01:43:59 ? 01:44:11 @hoogle W# 01:44:12 GHC.Exts W# :: Word# -> Word 01:44:19 oh, I have to do that. 01:44:19 okay. 01:47:33 @hoogle Word# -> Ordering 01:47:33 Unsafe.Coerce unsafeCoerce :: a -> b 01:47:33 Data.Ord comparing :: Ord a => (b -> a) -> b -> b -> Ordering 01:47:33 Prelude ($) :: (a -> b) -> a -> b 01:47:36 hm 01:48:10 hmmm no wordCompare# 01:56:35 going from about 5 seconds to .5 is a pretty big improvement. 01:56:44 I'm too lazy to try unboxing though 02:02:04 -!- nooga has quit (Ping timeout: 252 seconds). 02:02:44 @hoogle Word# -> Float# 02:02:44 Unsafe.Coerce unsafeCoerce :: a -> b 02:02:44 Prelude ($) :: (a -> b) -> a -> b 02:02:44 Data.Function ($) :: (a -> b) -> a -> b 02:02:47 hm 02:03:40 @hoogle Word -> Word# 02:03:40 Unsafe.Coerce unsafeCoerce :: a -> b 02:03:41 Control.OldException throwDyn :: Typeable exception => exception -> b 02:03:41 Control.DeepSeq ($!!) :: NFData a => (a -> b) -> a -> b 02:04:22 -!- derdon has quit (Remote host closed the connection). 02:13:40 :t 2## 02:13:41 GHC.Prim.Word# 02:13:45 excellent 02:17:17 > float2Int# 3.2# 02:17:17 Not in scope: `float2Int#' 03:03:40 -!- Klisz has joined. 03:06:16 telnet miku.acm.uiuc.edu 03:11:49 :( 03:13:24 turns out using explicit unboxing is slower because I have to reconstruct the boxed types in order to print them. 03:19:40 oh, hmmm, wait. 03:19:52 maybe since IO is strict I should use an accumulating parameter. 03:21:57 -!- incomprehensibly has quit (Ping timeout: 255 seconds). 03:28:50 -!- incomprehensibly has joined. 03:30:31 -!- incomprehensibly has quit (Client Quit). 03:43:39 -!- DCliche has joined. 03:47:24 -!- Klisz has quit (Ping timeout: 276 seconds). 04:03:21 -!- azaq23 has quit (Quit: Leaving.). 04:16:58 -!- Gregor has quit (Ping timeout: 252 seconds). 04:20:43 -!- Gregor has joined. 04:24:50 What's a 10th doctor episode I should watch? (Not Blink) 04:24:59 And not SitL/FotD 04:25:38 Why don't you just watch them in order ... 04:25:52 I've heard that they're mostly bad 04:26:09 So? 04:26:12 Suck it up. 04:28:28 yes being entertained should feel like running cross country. 04:28:38 you must endure it. 04:29:28 * Sgeo has yet to watch the rest of DS9 04:29:48 you know what I find really bizzare about American television? aside from the material itself... 04:29:51 canned laughter. 04:30:35 Yeah, laugh tracks are stupid. 04:30:58 But there is a lot of honest-to-goodness recorded-audience too, and laugh tracks are getting much less common. 04:32:53 I would like a window mode that's "always on top of maximized windows" 04:33:43 or "always on top of larger windows" 04:34:00 Hmm, I'm trying to think of what sit-coms I like that actually /use/ a laugh track ... almost none ... 04:34:09 Sgeo: bli-- oh 04:34:14 uh how about silence in th--- oh 04:34:22 uh 04:35:00 how about the season 4 finale. 04:35:10 Oh wait, duh, Seinfeld. 04:35:46 man whats the deal with seinfeld 04:36:00 It's the greatest sit-com ever made? 04:36:15 -_- 04:36:32 All those who don't like Seinfeld are bad people. 04:36:41 apparently subtle seinfeld references made to seinfeld snobs go unnoticed 04:36:42 * Sgeo is not a bad person. 04:37:14 Gregor: if i could ban you from life i wold 04:37:42 augur: That reference, it was way too subtle, I didn't get it X-D 04:38:10 Gregor: Seinfeld is good 04:38:12 Since it's, y'know, a thing people say. 04:38:30 I also enjoy How I Met Your Mother but I don't remember if it has a lauhg trakc or ont I am dontr emeber 04:38:38 kallisti: It definitely does. 04:38:55 I've basically removed all television from my daily mind diet. 04:39:05 for... like 3 years now. 04:40:13 I like some classique sitcoms too, I wonder if I Dream of Jeannie has a laugh track ... probably, but it may have been live. 04:40:29 I don't think that one was. 04:40:34 I like All in the Family, which is live. 04:41:29 Oh well. Silly laugh tracks. 04:44:27 -!- Darth_Cliche has joined. 04:46:04 I should make a laugh track bot. 04:46:20 And use some terrible heuristics to decide when it should laugh. 04:46:51 -!- I0ther has joined. 04:47:11 `welcome I0ther 04:47:22 I0ther: Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: http://esolangs.org/wiki/Main_Page 04:47:31 Hi Gregor 04:47:38 Thanks HackEgo 04:47:45 -!- DCliche has quit (Ping timeout: 252 seconds). 04:48:14 Meet the bots! 04:48:24 Erm, so many bots I'm not quite sure which do what, tbh 04:48:34 Gregor: use a weighting factor that looks at different patterns and totals a weight, which has to pash a threshhold. 04:48:36 > text "Hello, I'm lambdabot. I do Haskell stuff." 04:48:37 Hello, I'm lambdabot. I do Haskell stuff. 04:48:42 Gregor: :P would be such a pattern. 04:49:16 `echo Hello. I'm .. some bot. I do Linuxy stuff. Shell commands can be run on me. 04:49:19 Hello. I'm .. some bot. I do Linuxy stuff. Shell commands can be run on me. 04:49:25 -!- I0ther has left. 04:49:27 ... 04:49:28 kallisti: I was thinking I'd make a decision tree and train it off of the channel logs, with lines followed by "lol" or "rofl" etc having high value. 04:49:40 fungot: introduce yourself 04:49:41 kallisti: time to shove off! the name's bandeau. here to build the ocean palace? when do i work! the ultimate in defense for my, daughter! just like crono! 04:49:45 Sgeo: Nobody ever thought he was going to stay. 04:49:58 I didn't mean to scare him off 04:49:59 Gregor: ah that's a good idea. You could also use "heh" or "(ha)+" 04:50:07 Indeed 04:52:23 (b|mw|mua)?a*(h+a+)+ 04:52:37 Nah, people don't muahahaha when something is funny. 04:52:56 They muahahaha right after taking a bite out of a baby's skull while The Good Guy watches helplessly. 04:52:57 sure they do. 04:52:58 muahahahahahahaha 04:55:57 baa 04:56:10 Oh, that doesn't fit 04:56:22 mwha 04:56:57 Actually baa does fit. 04:57:00 As does baaaaaaaaaaaaaaaaaaaaa 04:57:06 also, a 04:57:12 Err, aa 04:57:17 But the h is unavoidable, isn't it? 04:57:21 + is one or more? 04:57:23 Nowait, yeah, need an h. 04:57:25 I missed that. 04:57:29 Thought it was * for some reason. 04:57:49 That's actually a pretty good RE :( 04:58:16 * kallisti is master of regex. 04:58:25 * kallisti master 05:01:20 -!- Darth_Cliche has quit (Quit: You are now graced with my absence.). 05:11:04 -!- copumpkin has quit (Ping timeout: 240 seconds). 05:18:09 -!- copumpkin has joined. 05:26:15 http://0taku.livedoor.biz/archives/4055139.html On the WTF scale of 1 to 10, this somehow overflows int. 05:26:37 (Japanese language, understanding not necessary) 05:27:42 (Context: They are celebrating Christmas. ... With characters from animation and animated porn.) 05:28:21 sounds like good fun 05:28:32 I think in Japanese it's something like 「なんでやねん！？」 05:29:35 Yeah, that would be a translation for WTF. 05:33:39 I may have inadvertantly lead Haskellers here, just a headsup 05:34:41 Does this mean anything meaningful to you? newtype T v f x = T (T x f v -> f x); 05:35:27 Sgeo: Meh, no worries. 05:38:21 zzo38, what 05:38:52 Madoka-Kaname: I don't really know what, it is why I asked 05:43:40 There exists a paper about PerlNomic 05:44:26 perlnomic? 05:44:41 ...you've never heard of perlnomic. 05:44:44 nope 05:44:47 is it context based? 05:44:54 Huh? 05:44:55 do you evaluate laws in list and scalar context? 05:45:03 The code is the law 05:45:11 ah, so yes. 05:49:36 -!- PiRSquared17 has joined. 05:50:25 I wonder whether the type I specified is a functor or monad or category (with type arguments swapped) or anything else at all 05:52:17 Sgeo, let's play Python nomic 05:52:40 elliott apparently thought about HaskellNomic back in 2008 06:01:24 I propose the following be made law: 06:01:25 import zlib;exec reduce(lambda x,y:x.decode("hex"),[zlib.decompress("eJxtUsuOwjAMvPtrYLsc94AEhx6SqlJ74MwBKXwANF+/sROPnYIq5VHb45lxQg45pvAT8lxOvJdv4fXOp5fcE6+PFr9l+Vf3d3D1BICEhENLKOuV75vEL2jwW8G4TuID36YSL2AjFx4FYEHnI+/xa4OnKchoIHH6DuAaJJTtJbp4ZUgmcTQJxsBJFICTAdw+LKAe4IrOURJmz1CZyR6XniHXF7AzPNAElRbhi4+PmHJpWP6vaEBKufdobRLUo7Ux64Yk98kIHMgYnE1CwpQ2i8+7IVkDfYcEEw3ASXziHdVvzH2+Dql6SErx0+TQhnD3z0Dfl/PQCFBIMHHYjf29Axj8kNDAhrSJZ9MFEk5Owh/9AyvTW+4=".de 06:01:25 code("base64")).decode("base64")]*4) 06:02:35 o.O 06:03:02 (This is the code version of legalese, right?) 06:07:24 Or, if you want a slightly less line-overflowing version: 06:07:25 import zlib;exec reduce(lambda x,y:x.decode("hex"),[zlib.decompress("eJxVkT2uwkAMhPs9TX4eB0CCIsUYPSkptk6BtBwAmNPjsZMIimi1a3u+GQesg10wgFca8QQn3vxuDfTvhaM+6d75nZjxBs/q/wO9NmME7yxotQcXDffW6jFwmyUADXTf9ez3kwew3wBjERFNhGkTeKghnOCSRIRDyGFnbfF6HcJx9C+qPyVY9BAEsfzBIkIAdMpdOp5DIIDW/reVVM2OsQJPUH4jpENEQwBO4CrMDjgJYLtgOqRFgsWdSSQdhUDuZN2iXpn1lXukTBA/6a2VCFA+a7ib1g==".decode("base64")).decode("base64")]*3) 06:09:09 I don't trust that 06:10:56 Sgeo, has anybody done something like that in Perl Nomic? 06:11:25 Get rid of the exec and it should be .. n/m 06:22:48 http://paste.strictfp.com/23213?key=1acffb6a1f52d873c798284b55fc730e 06:22:53 Sgeo, this is the final version of the "law" 06:23:17 Madoka-Kaname, "helpful" 06:23:55 What does that decode to 06:24:03 (to scared to try) 06:24:04 ? 06:24:07 *too 06:25:52 PiRSquared17, replace the "exec" with a "print" 06:25:52 =p 06:26:37 http://codepad.org/X4JZieG4 06:28:55 The trick here is this: 06:28:55 __builtins__.__dict__["X19pbXBvcnRfXw==".decode("base64")]("X19idWlsdGluX18=".decode("base64")).__dict__["6576616c".decode("hex")] 06:29:00 This is the actual exec call 06:29:03 The first one is a decoy 06:45:53 -!- DeadlyFugu has quit (Ping timeout: 252 seconds). 06:47:30 Make a computer game based on the story of Dungeons&Dragons game I have been playing in (you can read it in order to do so, because I recorded everything in this story so far) 07:09:41 -!- PiRSquared17 has changed nick to PiRSquaredAway. 07:34:03 The type I specified above, it seem to me, x is always going to be covariant and v always contravariant, if f is functor. But I don't know how well they will work 07:37:46 zzo38, what's the difference between the barrier monad and http://hackage.haskell.org/package/monad-coroutine ? 07:40:06 Sgeo: Barrier monad is really a somewhat more specific version of that. I did not see that until afterward, though, and barrier monad library also has some of its addition functions 08:05:12 -e expr Evaluate expr; see for details. 08:05:15 thanks ghc! 08:11:41 Hmm. I wonder. What is the monetary value of Vatican City? 08:11:55 Thus far, estimates I've found have stated NAN. 08:12:33 * Sgeo wonders how to coerce quickcheck into doing what he wants 08:12:49 This primarily involves learning to get to grips with quickcheck 08:12:59 At least, the quickcheck on lambdabot 08:25:01 -!- oerjan has joined. 08:31:35 -!- Patashu has joined. 08:33:16 What might the type I specified to make? newtype T v f x = T (T x f v -> f x); Do you have any idea about this? 08:33:50 Is it any kind of data structure or control structure? 08:35:29 i don't know but it reminds me of the "haskelljoust" game elliott and i discussed, and the type i made for it which he seemed not to like: newtype HJ = HJ (HJ -> Bool) 08:35:30 oerjan: You have 1 new message. '/msg lambdabot @messages' to read it. 08:37:30 Is it any kind of... Functor? Category? Monad? Applicative? Transformer? (Of course to be a category you would need to flip the type arguments, if it can make any category at all!) 08:37:53 oerjan: O, what were you going to use that HJ type for exactly? 08:38:08 I don't know what a "haskelljoust" game is 08:39:29 But, yes I can see what kind of similarity there is 08:42:50 well i mean that as an analogy to bfjoust: each player makes a "warrior", a haskell expression of type HJ -> Bool. they are then tested by calculating w1 (HJ w1), w1 (HJ w2), w2 (HJ w1) and w2 (HJ w2). each warrior must return True when passed itself, and to win it should return False when passed the other, while the other fails at the reverse (by either returning True or not returning a Bool at all within the time limit) 08:46:01 What happens if they both succeed or both fail? 08:46:03 I don't know how well that would actually work. 08:46:07 Sgeo: a tie 08:46:30 Also, Haskell functions are opaque, aren't they? 08:46:31 Do you have an example of a pair of functions that both succeed? 08:46:39 Well, I guess you could pass them data to see how they act 08:47:47 Hmm 08:47:48 However, the type I specified is a bit different, in a few ways 08:47:50 yes, basically you can start with const True (a warrior which never wins) and const False (always disqualified) and build more complicated HJ values from that 08:48:20 * Sgeo wonders if it's possible to make a warrior that always forces a tie 08:48:37 Not sure where I got the idea from 08:49:27 Trivial warrior that's tough to defeat unless a human writes a specialized warrior to defeat it: 08:49:32 ...n/m 08:49:33 Sgeo: i don't know. you can do something like \w -> w (HJ w) which will tend to cause loops, but then you get disqualified for not return True for yourself 08:49:58 I was thinking some sort of warrior that could do a signature thing, but that makes no sense 08:49:59 *returning 08:50:16 Sgeo: actually i think that makes quite a bit of sense. 08:50:26 Unless you can make this idea work: A signature that the warrior checks for, if it means the signature it returns true, otherwise false 08:50:38 I have no idea how you'd pass in a signature though. 08:51:01 Sgeo: well think of const False and const True as the simplest possible signatures. 08:54:36 Oh no, Satan is giving presents to Vorpal now. <-- is he? 08:55:18 For my type: lift = T . const; empty = T $ const empty; T x <|> T y = T $ \z -> x z <|> y z; And I am not sure what else, if any 08:59:57 imagines a world in which wolfram alpha is actually kind of good 09:01:26 Sgeo: warrior for what? 09:01:42 haskelljoust 09:01:45 Read scrollup 09:01:46 heh 09:04:15 Can any Alternative or MonadPlus be made backward? I would think so, by swapping the order of <|> 09:04:40 This might affect functions that use <|> such as optional and some and many 09:08:04 zzo38: does this work for Functor f => Functor (T v f x): fmap g (T h) = T (\t -> fmap g . h $ fmap' g t); fmap' g (T t) = T (\h -> t $ fmap g h) 09:09:47 or pointfree, fmap g (T h) = T (fmap g . h . fmap' g); fmap' g (T t) = T (t . fmap g) 09:10:09 I think I'm in love with enumerator 09:12:12 fmap' :: (x -> y) -> T y f v -> T x f v 09:12:29 *fmap' :: Functor f => (x -> y) -> T y f v -> T x f v 09:12:53 oerjan: Maybe; I am going to sleep now, though. Try it yourself or keep writing message if you want to, which I will read tomorrow or the next day. Or just wait until I am back on 09:12:57 -!- zzo38 has quit (Quit: zzo38). 09:14:02 -!- itidus21 has joined. 09:14:50 `haskell newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f x) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "Yay it types!" 09:15:00 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: haskell: not found 09:15:04 AAAAAAAAAAAAAAAAAAAAAa 09:15:34 !haskell newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f x) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "Yay it types!" 09:15:41 oerjan: why not use lambdabot? 09:15:41 runhaskell: syntax: runghc [-f GHC-PATH | --] [GHC-ARGS] [--] FILE ARG... 09:15:58 Vorpal: HAVE YOU EVER SEEN LAMBDABOT ACCEPT A TYPE DECLARATION? 09:16:06 oerjan: now that you mention it, no 09:16:14 which is kind of silly 09:16:16 it should 09:16:59 bbiab 09:17:06 -!- Vorpal has quit (Quit: ZNC - http://znc.sourceforge.net). 09:18:29 -!- Vorpal has joined. 09:18:29 `run ls bin h* 09:18:32 ls: cannot access h*: No such file or directory \ bin: \ ? \ @ \ No \ _tmpe \ addquote \ allquotes \ calc \ define \ delquote \ etymology \ forget \ fortune \ frink \ google \ hatesgeo \ json \ k \ karma \ karma+ \ karma- \ learn \ log \ logurl \ macro \ marco \ paste \ pastekarma \ pastelog \ pastelogs \ pastenquotes \ pastequotes \ pastewisdom \ pastlog \ ping \ prefixes \ qc \ quote \ quotes \ roll \ toutf8 09:20:25 @tell elliott so let me see, i complained the other day that we didn't have a bot which accepted haskell modules because EgoBot's was broken; kallisti helpfully put one in HackEgo despite your best attempts to sabotage him; you sent a fix for EgoBot and then considered the fact you had _sent_ the fix enough reason to DELETE kallisti's version; which with Gregor's usual responsivity means we _still_ don't have a bot capable of haskell modules. 09:20:25 Consider it noted. 09:22:21 `which runhaskell 09:22:24 /opt/ghc/bin/runhaskell 09:23:10 @tell elliott oh well i guess can use runhaskell directly. i still have this urge to ban you though. merry christmas. 09:23:10 Consider it noted. 09:23:30 oerjan: lol 09:24:11 `run echo 'newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f x) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "Yay it types!"' >typetest.hs 09:24:14 bash: -c: line 0: syntax error near unexpected token `)' \ bash: -c: line 0: `echo 'newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f x) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "Yay it types!"' >typetest.hs' 09:24:22 goddammit 09:24:43 `run echo "newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f x) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "'"Yay it types!"' >typetest.hs 09:24:46 No output. 09:24:55 `cat typetest.hs 09:24:57 oerjan, use `revert to get kallisti's code? 09:24:58 newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f x) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "Yay it types!" 09:25:40 Vorpal: i am not proficient with vcs'es; won't that revert everything else done since then? 09:26:04 oerjan, well yes, so extract the file only then 09:26:21 well it would be trivial to rewrite I believe. 09:26:44 `runhaskell typetest.hs 09:26:49 oerjan: noep 09:26:50 oh 09:26:52 \ typetest.hs:1:70: \ Kind mis-match \ The first argument of `Functor' should have kind `* -> *', \ but `T v f x' has kind `*' \ In the instance declaration for `Functor (T v f x)' 09:26:59 oops 09:27:10 `run echo "newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "'"Yay it types!"' >typetest.hs 09:27:13 No output. 09:27:16 `runhaskell typetest.hs 09:27:21 "Yay it types!" 09:27:24 whew 09:29:43 `help 09:29:45 Runs arbitrary code in GNU/Linux. Type "`", or "`run " for full shell commands. "`fetch " downloads files. Files saved to $PWD are persistent, and $PWD/bin is in $PATH. $PWD is a mercurial repository, "`revert " can be used to revert to a revision. See http://codu.org/projects/hackbot/fshg/ 09:31:04 how can it take minutes to load the system event log in windows.... 09:32:05 @tell elliott i suppose you helped kallisti with it, too... 09:32:05 Consider it noted. 09:32:40 ah there we go 09:32:44 `run (echo '#!/bin/sh'; echo 'runhaskell <<< "$1") > bin/haskell 09:32:47 bash: -c: line 0: unexpected EOF while looking for matching `'' \ bash: -c: line 1: syntax error: unexpected end of file 09:32:55 `run (echo '#!/bin/sh'; echo 'runhaskell <<< "$1"') > bin/haskell 09:32:58 No output. 09:33:01 `haskell main = print (2+2) 09:33:04 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: /hackenv/bin/haskell: Permission denied \ /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: /hackenv/bin/haskell: cannot execute: Permission denied 09:33:12 `run chmod +x bin/haskell 09:33:15 No output. 09:33:18 `haskell main = print (2+2) 09:33:21 /hackenv/bin/haskell: 2: Syntax error: redirection unexpected 09:33:25 oh 09:33:27 kallisti: you know i just looked it up on http://codu.org/projects/hackbot/fshg/index.cgi/rev/d30da6e6e1c6?revcount=240 09:33:34 `run (echo '#!/bin/bash'; echo 'runhaskell <<< "$1"') > bin/haskell 09:33:36 No output. 09:33:41 `run chmod +x bin/haskell 09:33:43 No output. 09:33:45 `haskell main = print (2+2) 09:33:50 4 09:33:57 oerjan: that's pretty much what it was. 09:34:57 oh well. EgoBot's version still works in ghci mode, anyway. 09:35:41 `haskell newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f x) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "Yay it types!" 09:35:47 \ /tmp/runghcXXXX279.hs:1:70: \ Kind mis-match \ The first argument of `Functor' should have kind `* -> *', \ but `T v f x' has kind `*' \ In the instance declaration for `Functor (T v f x)' 09:35:57 * oerjan facepalm 09:36:05 `haskell newtype T v f x = T (T x f v -> f x); instance Functor f => Functor (T v f) where { fmap g (T h) = T (fmap g . h . fmap' g) }; fmap' g (T t) = T (t . fmap g); main = print "Yay it types!" 09:36:10 "Yay it types!" 09:37:39 oerjan: this program is a proof that it types. 09:37:43 beautiful 09:38:30 so i think. there are some cases where ghc only looks things up when used, but i that's mostly to check instance overlap. 09:38:39 *i think 09:38:59 oh and import overlaps too 09:42:24 hm making that pointfree actually made checking the Functor laws easier. 09:44:22 which is sort of expected for category theory, come to think of it. 09:45:36 @tell zzo38 btw, fmap' shows that T is also a contravariant functor in its first argument. 09:45:36 Consider it noted. 09:46:59 -!- ais523 has joined. 09:47:18 merry christmas everyone 09:51:59 ho ho ho 09:53:27 oerjan: is it Christmas today in Norway? 09:53:31 or is it on some other day of the year? 09:53:35 -!- Taneb has joined. 09:53:38 yesterday, mainly 09:54:11