00:03:49 elliott_: how does ghci handle the IO monad? 00:03:54 I cannot remember. 00:05:47 CakeProphet: Whaddya mean? 00:06:02 does it execute IO actions if they're top level? 00:06:23 Yes, it acts like a "do" block in the IO monad. 00:06:27 for instance, will it modify IOrefs if you tell it to. 00:06:29 ah okay. 00:06:35 Except you can also put expressions in. 00:06:42 > x <- ... 00:06:42 gotcha 00:06:42 : parse error on input `...' 00:06:43 works, etc. 00:11:25 hmmm 00:12:02 I was thiking one improvement to these demonstrations would be a way to pass around mutable IOref parameters to these sound generators. 00:12:05 and then change those. 00:12:20 but I think that would greatly complicate the syntax. 00:13:16 @hoogle (:=) 00:13:16 Control.Parallel.Strategies (:=) :: a -> b -> Assoc a b 00:13:28 @hoogle (::=) 00:13:28 No results found 00:13:56 hoogle (~+) 00:15:14 @hoogle (~+) 00:15:14 No results found 00:15:59 CakeProphet: what are you looking for 00:16:11 and what you demonstrated looked totally pure 00:16:15 why force it into the IO monad? 00:18:34 elliott_: no I was referring to yaxu.org 00:18:45 oh 00:18:49 I hope what I just posted to the LSL list wasn't downright stupid 00:19:02 Sgeo: the odds are not on your side. 00:19:05 elliott_: and I was seeing what other operators looked like that, as I was thinking of my own operators to use. 00:20:40 -!- aloril has joined. 00:20:55 http://pastie.org/private/60bknz58chvetmttccch0w 00:21:48 elliott_: though with his emacs interface he really doesn't need to use IOref. He just changes the parameters by hand and reloads it via some black magic. 00:21:59 s/interface/mode/ 00:23:33 I wish Google could search for non-alphanumeric characters better. 00:23:52 "haskell ($)" gives me nothing about ($) 00:24:27 CakeProphet: it's IORef not IOref btw 00:24:29 and use hoogle :P 00:24:36 elliott_: and I was seeing what other operators looked like that, as I was thinking of my own operators to use. 00:24:39 you can't use : 00:24:43 it's reserved for types and type constructors 00:24:46 kind of like Uppercase is 00:24:57 s/type constructors/constructors/ 00:25:01 ah that's right. 00:26:29 elliott_: Hoogle doesn't show the fixity declaration though, which is what I'm looking for. 00:27:10 oh but hey 00:27:11 I foudn it 00:27:15 > (0$0$) 00:27:15 The operator `GHC.Base.$' [infixr 0] of a section 00:27:16 must have lower prec... 00:27:24 the actual hoogle website has it. 00:27:31 oerjan knows the proper trick for that 00:27:33 > (0$0+) 00:27:33 The operator `GHC.Num.+' [infixl 6] of a section 00:27:33 must have lower prece... 00:27:37 hm maybe this is the trick 00:27:37 yeah it is 00:27:39 > (0$0/) 00:27:40 The operator `GHC.Real./' [infixl 7] of a section 00:27:40 must have lower prec... 00:27:48 CakeProphet: also 00:27:52 Prelude> :info ($) 00:27:52 ($) :: (a -> b) -> a -> b -- Defined in GHC.Base 00:27:52 infixr 0 $ 00:28:14 I wonder if Haskell will let me define a negative precedence level. :3 00:28:49 for fancy assignment-like operators 00:29:41 no 00:29:48 actually using 0 is a bad idea... 00:29:53 because it will stop people using $ 00:29:55 use one at the most :P 00:29:58 or... at the least 00:30:13 but I want 00:30:33 "name" := someUngodlyFunction $ someUngodlyParameters ... 00:30:48 replace the := with something else 00:31:06 ah well I guess I'll just have to settle with parenthesis 00:31:13 or, you know, I might not even implement something like that at all. 00:31:14 "name"? 00:31:18 seriously? /strings/? 00:31:30 that's how I roll, dog. 00:31:33 * Sgeo is impatient for a reply 00:31:44 CakeProphet: enjoy creating the worst haskell library of all time :x 00:31:49 type safety, who needs it 00:32:16 don't worry, I haven't created anything yet. 00:32:33 Is Haskell's printf type-safe? I mean, it type-checks just fine, but 00:33:13 Sgeo: no 00:33:16 nobody uses it though :) 00:33:20 and there's a way to do much the same thing safely 00:33:40 it just looks like: printf ("foo " $$ str $$ " bar") "abc" 00:33:42 instead 00:33:42 or e.g. 00:34:00 the thought was that in a live coding environment the less typing you have to do the better. Also, being able to put everything in one line is a place. Anything besides string requires previous setup of some kind. If I can do that and have it not be a pain then I'll use that approach, but otherwise I am using strings and you will hate it. :P 00:34:02 printf ("foo " $$ (int % pad0 9) $$ " bar") "abc" 00:34:04 erm 00:34:06 printf ("foo " $$ (int % pad0 9) $$ " bar") 9 00:34:07 etc. 00:34:10 which is easier to read anyway. 00:34:29 CakeProphet: then don't use haskell 00:34:40 CakeProphet: besides 00:34:43 CakeProphet: variables need initial values 00:34:46 what will you do? 00:34:57 default all integers to 0 and lists to [] and the like with a typeclass? ahahaha 00:35:17 er, no. The idea is that the names refer specifically to one type. a sound generator of some kind. 00:35:26 so what initial value 00:35:39 whatever you set it to initially? 00:35:45 so you need to declare it? 00:35:51 i thought you didn't want that which was why you were using strings. 00:36:52 I just wanted to use strings as names in, perhaps, a mutable map or something, instead of passing around IO refs. It's one option of many. 00:37:12 I'm just saying that strings, apart from being totally type-unsafe and ugly and not even giving you any kind of error for using an undefined variable, 00:37:20 don't absolve you of the need to initialise -- i.e. declare -- variables 00:38:02 right, but they save you effort in that they're effectively global throughout IO, without the need to pass around the reference. I'll probably end up using IORef if anything for mutable parameters. 00:38:08 Man. For coming from the guy who brought us PulseAudio, systemd actually seems like a sane, intelligent design. 00:38:13 pikhq_: it's not 00:38:15 fucking ini files 00:38:18 no sh 00:38:42 elliott_: Are there any non-aesthetic complaints you have? 00:38:52 those are technical complaints 00:38:58 ini files are terrible, especially the way it uses them 00:39:02 How so? 00:39:13 and the only reason it exists is because Poettering hates sh 00:39:32 if Lennart Poettering is behind it, you can basically assume it'll be horrible shit and also, that he'll be a jackass about it 00:39:59 IIRC he responded to a criticism of -- was it systemd? -- at a presentation, by accusing the criticiser of hating disabled people :) 00:40:19 At least how sh is used *currently* in init, sh seems to suck for the purpose. 00:40:31 That is a motherfucking *fuckton* of boilerplate. 00:40:39 when did i say that was a good idea 00:40:53 but his whole goal is "omg sh should never be called during boot" 00:41:00 You seemed to say that "no sh" was a technical complaint against systemd. 00:41:15 yes, because it means you can't do tons of shit easily 00:41:16 i never said 00:41:21 "i want sh exactly like it is in current init omg" 00:41:46 Admittedly, it's probably a good idea to also have a more flexible language in your init system. 00:41:55 By all means, make it so the common case is easily configurable. 00:41:57 like Java. 00:42:02 CakeProphet: lol 00:42:12 all init should be in java. 00:42:18 But you should make the easy bits easy and the hard bits possible. 00:42:18 pikhq_: the people in #musl know more about systemd than me anyway, but i heard enough to make me dislike it 00:42:28 (note: going in to #musl to ask about systemd may not be a good idea :P) 00:42:29 Rather than the easy bits easy and the hard bits akin to getting raped. 00:42:31 pikhq_: so you're saying init should be in Perl? 00:42:45 CakeProphet: No, no, just in *a* flexible language. 00:42:51 sh is not terrible for the purpose. 00:43:05 If you can get rid of the pointless, irritating boilerplate. 00:43:14 -!- Gregorzilla has quit (Quit: Leaving). 00:43:18 how is there any boilerplate in starting e.g. an httpd with sh 00:43:21 it's one line 00:43:32 nobody is advocating /etc/init.d here, that is a strawman 00:43:54 elliott_: Obviously, the means of getting rid of boilerplate is not necessarily getting rid of sh. 00:44:07 there is no boilerplate to get rid of 00:44:17 that's assuming we're all talking from the "systemv init" frame of reference 00:44:19 which guess what, we're not 00:44:36 Just making it so that you literally don't have to stick the basic, barebones logic of the *entire* init system in the configuration files for it. 00:44:42 so, I'm little murky on the differences between sh, bash, and dash. Can someone enlighten me? 00:44:47 elliott_: I have been talking from a systemv init frame of reference. 00:44:56 well that's your problem 00:45:04 If we talk from other frames of reference, there may well be little to no boilerplate. 00:45:13 CakeProphet: sh is too vague to define, be more specific; bash is GNU's shell 00:45:27 dash is the Debian Almquist shell 00:45:29 Certainly, there's no boilerplate to: for /etc/services.d/*;do sh $i;done 00:45:39 as in, is sh just the name for the system's shell program, whatever it may be? 00:45:39 a descendent of NetBSD's version of the Almquist shell 00:45:55 CakeProphet: yes. POSIX defines what sh has to be, but it's a rather minimal specification. 00:45:55 CakeProphet: The system's *Bourne-like* shell. 00:46:05 ah okay. 00:46:21 CakeProphet: There's a few requirements on that beyond "it can spawn processes"... In practice, though, that means "any shell but csh and tcsh". 00:46:26 I notice when I "man sh" I get a page for dash, when originally I thought sh was bash on Ubuntu. 00:46:44 sh is now dash on the Debian family distros. 00:46:54 is there like, a huge difference I should note? 00:47:00 CakeProphet: No. 00:47:08 Unless you use bash's advanced features. 00:47:14 (Hint: You probably don't. Not in your scripts anyway.) 00:47:26 yes I doubt it. I don't even know what those are. 00:47:29 ((I mean really advanced, not "the advanced section of My First Bash Script".) 00:47:30 ) 00:48:00 Yeah, a barebones Bourne shell offers like 99% of what *anyone* uses in shell scripting. 00:48:10 I think the most obscure thing I've used so far is the ${foo%bar} thing 00:48:45 Yeah, that's Bourne, I'm pretty sure. 00:48:46 to rename a file extension 00:49:10 so, bash > dash? 00:49:21 or just... different? 00:49:29 in some tiny ways. 00:49:38 dash is quicker and leaner for scripts 00:49:42 Bash is a superset of the capabilities of the Bourne shell. 00:49:44 and using bash for scripts leads you into the trap of using bash-only features 00:49:47 which makes your scripts non-portable 00:49:52 pikhq_: so is dash. 00:50:07 elliott_: Oh, is it? Hmm. 00:50:14 CakeProphet: basically: obviously you don't want to use dash for interactive use. 00:50:24 but if you are writing scripts then dash will probably do you fine, and will produce more portable scripts. 00:50:29 i say probably 00:50:31 it really will do you fine 00:50:35 you will not notice any difference 00:50:40 when I'm using terminal in Ubuntu I'm using dash right? 00:50:46 um 00:50:47 no 00:50:55 Aaah, the language that dash supports is POSIX *and* a few BSD extensions. 00:50:56 oh, well good. 00:51:18 $ help 00:51:19 GNU bash, version 4.1.5(1)-release (x86_64-pc-linux-gnu) 00:51:31 (which, being extensions that made their way into Korn, are essentially ubiquitous, anyways) 00:51:56 okay, thank you for the enlightenment. 00:52:15 The Bourne shell itself, BTW, is basically the first UNIX shell that was reasonably useful for programming. 00:52:39 And replaced the Thompson shell in V7. 00:53:26 Which itself was the first shell ever, and only really had pipes and file redirection. 00:54:47 Huh. rc and mk were originally written for UNIX. 00:54:54 And part of V10. 00:54:56 yes 00:55:04 Neat. 00:56:19 Shame rc's not ubiquitous; it's got such a better language. 00:57:00 hrm 01:00:50 ?unpl (f . g) x 01:00:50 (f (g x)) 01:01:39 And would you look at that. Sabotage, that distro using musl as its libc, builds with mk. 01:04:34 isn't it a heap of shell scripts mostly :) 01:04:52 /o/ 01:04:52 | 01:04:52 /< 01:05:03 \o/ 01:05:03 | 01:05:04 /'\ 01:05:08 elliott_: Well, yes, probably. 01:05:16 That ' in the middle looks suspicious. 01:05:28 I can't help but laugh when myndzi does that to Perl code. 01:05:38 elliott_, do you think CoffeeScript is a decent language? 01:10:46 elliott_: Inquiry: what do you think is an actually decent init system? 01:10:53 (if anything) 01:11:49 pikhq_: @ :P 01:12:03 But uh, runit is pretty good. 01:18:30 Hey, non-revolting use of sh. 01:19:45 -!- pikhq has joined. 01:23:19 -!- pikhq_ has quit (Ping timeout: 260 seconds). 01:26:39 -!- Mannerisk has joined. 01:27:12 -!- Mannerisk has left. 01:27:32 -!- jwest has joined. 01:28:29 Hi, could anyone give me an idea on how you'd, go backwards in the data list in brainfuck until a counter is 0? 01:28:42 not sure where else to ask 01:30:31 you mean like 01:30:35 n=9 <<<<<<<<< 01:30:40 where n is in a cell? 01:30:54 i.e. "repeat < the number of times as indicated in the current cell" 01:30:59 yep 01:31:02 but n being an input 01:31:04 not constant 01:31:20 that is not possible 01:31:24 oh... 01:31:27 kk 01:31:31 that's a shame 01:31:33 jwest: you may be interested in an array type solution 01:31:39 cells look like 01:31:47 1 | element | 1 | element | 1 | element | 0 01:31:51 starting on the first one 01:31:52 you can do 01:31:53 [>>] 01:31:56 to seek to the end 01:31:57 and things like that 01:32:03 making it very convenient to seek around, etc. 01:32:11 and you can use these "indicator" cells to pass around a counter 01:32:16 letting you move N places in the array, etc. 01:32:29 ooh 01:32:31 right 01:33:01 yeah i'll look into that 01:33:39 http://1.bp.blogspot.com/-HU3XrGrDAZ4/TeVGS5sea-I/AAAAAAAAADk/r4hnr1X5YKQ/s1600/Netflix%2BPerformance%2Bon%2BTop%2BISPs%2B-%2BUSA.png This makes me sad. 01:33:44 thanks elliott_ 01:33:45 Note the units on the graph. 01:34:15 jwest: http://esolangs.org/wiki/Brainfuck_algorithms is sometimes helpful 01:34:33 oh wow, the netfuck saga continues 01:34:49 he can write interpreters in qbasic and has developed simple chats over nullmodem :o 01:35:08 elliott_: ... Wut? 01:35:08 what do the temp1 etc things in the example code mean? 01:35:44 jwest: see top of page 01:35:53 pikhq: "I still don't get your problem. This is so petty and you search every error in my sentense to critisize me... I have already written a full functional interpreter in QBasic and developed simple chats over nullmodem, so I don't really see why NF should be so worse in your opinion. --87.165.166.97 00:34, 1 June 2011 (UTC)" 01:36:14 oh right, sorry 01:36:18 thanks for your help 01:36:24 jwest: :) 01:36:37 elliott_, link? 01:37:31 Lymia: talk:netfuck on the wiki 01:38:13 Hmmm. Maybe I should actually upload all my code somewhere again. 01:38:38 elliott_, link? 01:38:44 I've failed to find it. 01:38:51 see recent changes, i have a headache and closed the tab 01:38:53 pikhq: I have a VPS I'm doing nothing with that runs thttpd :P 01:39:05 (Note: I am an unreliable person.) 01:39:11 pikhq: ooh, do the NetFuck challenge and you'll get 01:39:13 FREE 01:39:13 AD-FREE 01:39:14 WEB 01:39:14 SPACE 01:39:17 the best prize 01:39:22 elliott_: I was thinking more like one of those git hosting sites. 01:39:34 HOW BORING AND MODERN 01:39:51 elliott_, uh. 01:39:51 So. 01:39:52 Be nice to have dimensifuck up somewhere. 01:39:58 And PEBBLE and PFUCK. 01:40:02 This is Brainfuck... with support for two IO channels? 01:40:03 pikhq: I was going to go "GIT :( USE DARCS" but then I realised that the darcs hosting sites suck. 01:40:07 And other random snippets. 01:40:13 Well, darcsden is alright but it's... barebones. 01:40:28 oh hm, it seems to have been made less barebones as of late, nice 01:42:36 Opinions on git hosters? 01:43:17 pikhq: If you're looking to be ais-style obnoxiously legalistic and principled about things, something that isn't GitHub. Otherwise, seriously, just use GitHub. 01:43:45 elliott_: What negatives exist at all? 01:44:37 pikhq: Uhh... their terms of use once disallowed people to block ads on the site (note that GitHub has never even had ads, and this clause is now long-gone). Also, things like issues, the wiki system, and pull requests are "lock-in" features in that they're not implemented with any standard protocol and don't interoperate with non-GitHub sites. 01:44:48 Oh, and the issue system requires JS. 01:45:02 But it's a well-designed, fast site with lots of good functionality, and it's incredibly popular /shrug 01:45:22 And the inability to use the pull request system directly if the person who made the changes doesn't use GitHub is... not the biggest issue ever. 01:45:38 Mmkay, then I suppose GitHub, then. 01:45:46 GitHub is the only host which offers dynamic tarball downloads of the tip too I think, which is nice because you can link to them. 01:45:47 If I don't like it, well, it's not hard to migrate elsewhere with git. 01:46:28 Honestly, in my mcmap development I look at exactly two things on GitHub: 01:46:35 mcmap? 01:46:36 - The commit list (and the commits themselves) 01:46:37 - The issue tracker 01:46:40 coppro: Minecraft thing. 01:46:46 ah 01:47:14 Developed by fizzie; I wrote some parts. 01:47:31 Un momento; time to use ssh-agent again. 01:47:32 -!- pikhq has quit (Quit: leaving). 01:48:11 -!- pikhq has joined. 01:52:49 Hey, just one more question, does the fact that brainfuck is turing complete mean that anything written in another turing complete language can be written in brainfuck? I suspect not. 01:53:01 jwest, it does. 01:53:23 :o 01:53:28 jwest: That is *precisely* what Turing completeness means. 01:53:32 haha, good 01:53:36 it's just hard to believe 01:53:38 Note that Turning Complete only considers computational power. 01:53:48 Things like IO, Networking, etc are not covered. 01:53:52 i guess i should look it up and study it a bit 01:53:52 yeah 01:54:30 jwest, you could, for example, write an x86 emulator in Brainfuck. 01:54:34 This has obvious implications. 01:56:39 finding a well-documented low-level audio playback library in Haskell is proving difficult. 01:56:45 -!- TeruFSX has quit (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.). 01:57:01 portaudio 01:57:08 ohh, kk 01:57:17 Lymia: does the 30000 byte limit stop this at all? 01:57:26 (assuming that is taken as necessary) 01:57:36 elliott_: perfect :) 01:57:37 jwest, yes. 01:57:43 Turning Complete, assuming infinite memory. 01:57:47 ah kk 01:58:04 Actually. 01:58:14 Does the C spec break down with infinite memory, and infinite word size? 01:59:14 break down in what way? 02:00:33 I assume the only problem the C spec would have with infinite memory would be addressing it all. With infinite word size, that wouldn't be a problem. 02:01:16 Infinitely long typecasts would be a problem 02:01:33 Wait. 02:01:36 With infinite word size... 02:01:38 What is sizeof(int) 02:02:27 Lymia: presumably infinity. 02:02:29 If everything is infinitely large then sizeof() is useless 02:02:48 i gotta go, thanks for the help and interesting info, see ya 02:02:51 -!- jwest has quit (Quit: Page closed). 02:03:22 Patashu, but what would sizeof(int) return? 02:03:28 All ones? 02:03:59 How would you express negative numbers? The most infinite bit is set to 1? But infinity has no last bit 02:04:19 You don't use twos compartment. 02:04:31 You could use sign and magnitude, right? 02:04:42 malloc(sizeof(int)) would return the starting address of the infinite memory you just allocated in your infinite memory heap... which I'm pretty sure wouldn't terminate. 02:04:42 Sign could be first, sure 02:05:07 but then again, I don't really know how malloc works. 02:05:07 Would infinite memory C be a Deathstation 9000? 02:05:34 Patashu, wait. 02:05:43 C with infinite memory is possible, provided you address memory with bignums. However, infinite word size makes things difficult, it seems. 02:05:46 What about the various undefined interactions? 02:05:57 undefined behavior, etc. 02:07:54 How are you going to make this work with the guarantee that you are able to touch every bit in memory, e.g. chars are consecutive and leave nothing out 02:08:31 I don't really understand what you mean exactly. 02:08:37 Me or lymia 02:08:42 you 02:09:36 hmmm, well. you could have a floating point addressing scheme. 02:09:46 (I think I understand what you mean now) 02:09:55 Let me see if I can find it 02:11:04 Hmm I don't know where it is, but I know that the C programming language guarantees that chars are consecutive and allow you to touch every bit of memory 02:11:07 ptr+1 points to the next char in the string, and all of the bytes of the infinite width char are represented as decimal points (but then they become uncountable, which is another problem I think) 02:11:14 How do you reconcile this with words being infinitely large? 02:11:24 It doesn't make sense to have an infinity after an infinity, I guess 02:11:30 sure it does. 02:11:42 that's how the real numbers work 02:12:09 Hmmmm. What else should go up? 02:12:26 Question is. 02:12:32 Does C break down when you do crap like that? 02:13:49 Can I address memory in smaller than infinitely large chunks, e.g. 8 bit chunks? 02:14:04 If so, what is the address of the 8 bit chunk at the start of the third infinitely large chunk? 02:14:09 well, yes. as it currently stands, I think every implementation of C would not be able to handle infinite memory. Especially considering that no physical computer can do that. 02:14:38 -!- copumpkin has quit (Quit: Computer has gone to sleep.). 02:15:21 So. 02:15:28 Basic question is, does the C spec hold up under infinity. 02:15:35 Patashu, I dunno. 02:15:44 The word size is infinitely large, right? 02:17:01 to have infinite memory and word size, the cardinality of our set of memory addresses has to equal that of the real numbers. For finite word size and infinite memory, that would be the same cardinality as the natural numbers. 02:18:13 so basically, if you ask "what's the second bit in the fourth word of our computer memory", there is no answer... 02:18:29 it would be like asking, what's the next real number after 4? 02:18:40 (I think) 02:18:59 Guh. 02:19:01 Brain hurt. 02:19:08 Mmkay, a few bits of code are now up on git. Whooo. 02:19:18 Erm, github. XD 02:20:25 Lymia: but, let's assuming you have a arithmetic processing unit that can handle addition of infinite words. Then you don't need to reference those bits individually, you just have your miraculous device perform basic arithmetic for you. 02:20:57 and to get the next word in memory, you just say char_ptr+1 02:22:11 https://github.com/pikhq A handful of things. 02:22:14 and because you can refer to consecutive words, you can make assertions such as "an integer is 4 words long" 02:23:14 and sizeof(int) would simply be 4 words. 02:23:40 which malloc can happily allocate, most likely. 02:24:10 with its magic algorithm, operating on magical infinite-sized words. 02:24:16 But what do I do if I want to reference memory in smaller chunks than a word 02:24:17 I can't? 02:24:21 I guess I wouldn't really need to 02:24:23 not really no 02:24:26 and yes, no need. 02:24:27 If I needed byte arithmetic I can fake it with modulus 02:24:33