Available formats: content-negotiated html turtle (see SIOC for the vocabulary)
Back to channel and daily index: content-negotiated html turtle
These logs are provided as an experiment in indexing discussions using IRCHub.py, Irc2RDF.hs, and SIOC.
| 00:00:15 | <sorear> | How do I do that? |
| 00:00:31 | <sorear> | I thought that fell under the capacity of "never ask for power" |
| 00:00:50 | <sorear> | you see, I'm so helpless in lift I have to actually READ netiquette guides :) |
| 00:00:54 | <sorear> | s/lift/life |
| 00:01:42 | <dons> | its ok being helpless in the lift. trained emergency staff will be there shortly to assist. |
| 00:03:09 | <sorear> | If I'm in a lift I'm already hundreds (thousands?) of miles from home... will they help me get back? |
| 00:04:15 | <dons> | they will do their best. our trained response team will help you catch a bus. |
| 00:04:33 | <dons> | you don't have lifts in california? |
| 00:04:41 | <sjanssen> | elevators. |
| 00:05:06 | <dons> | ah, of course! |
| 00:05:29 | <dons> | and you live in apartments, not flats, I presume? :-) |
| 00:05:35 | <sjanssen> | naturally |
| 00:05:46 | <sorear> | What are the naming conventions for trac user names? |
| 00:05:54 | <dons> | i'm 'dons' on trac. |
| 00:06:25 | <sorear> | that only makes is more confusing, since they have a PhilippaCowderoy |
| 00:06:45 | <dons> | jcreigh: your window border patch cleans things up nicely for me. |
| 00:06:46 | <sjanssen> | choose whatever makes you happy |
| 00:06:51 | <ddarius> | Several thousands. |
| 00:06:53 | <sjanssen> | I'm sjanssen |
| 00:06:58 | <dons> | jcreigh: want to see whether we can make the border colour a config setting? |
| 00:07:38 | <ddarius> | Wie geht's, ekidd? |
| 00:08:05 | <ekidd> | ddarius: Hello! I'm in the process of replacing my iBook battery. |
| 00:08:14 | <sorear> | sjanssen: I don't want a repeat of the Haskell Wiki, where I registered 'Stefan' than discovered of Ljungstrand, Monnier, Holdermanns, ... the next day :) |
| 00:09:03 | <ddarius> | So use sorear. |
| 00:09:24 | <ddarius> | ekidd: Sounds like fun for the whole family. |
| 00:10:26 | <ekidd> | ddarius: If all goes according to plan, this should boost my battery life from 1.5 to ~8 hours. More Haskell hacking! |
| 00:15:22 | <procyon112> | Why don't my debug traces finish printing? Is there a way to flush them? |
| 00:17:13 | <dons> | hFlush stderr |
| 00:17:39 | <sjanssen> | or hSetBuffering stderr NoBuffering |
| 00:17:41 | <dons> | or set hSetBuffering on stderr to HotFlushes , sorry, unbuffered, at the start |
| 00:18:35 | <procyon112> | ?hoogle hFlush |
| 00:18:35 | <lambdabot> | IO.hFlush :: Handle -> IO () |
| 00:19:22 | <sorear> | procyon112: did you get my reply? |
| 00:21:08 | <procyon112> | yes |
| 00:22:11 | <procyon112> | hmm... shouldn't "trace (show bigList) (seq (hFlush stderr) doInfiniteLoop)" flush the bigList before looping? |
| 00:22:35 | <sjanssen> | no |
| 00:22:42 | <sorear> | seq (hFlush stderr) x === x |
| 00:23:01 | <sorear> | since hFlush stderr is a partially applied function, thus in WHNF |
| 00:23:28 | <sorear> | you need to pass it a State# RealWorld for the action to occur |
| 00:23:44 | <sorear> | the portable way of doing that is unsafePerformIO |
| 00:23:59 | <SamB_XP> | hooray dangerous portability! |
| 00:24:16 | <sorear> | unsafePerformIO is actually pretty portable. |
| 00:24:16 | <procyon112> | I'm debugging, so I'm not worried about portability :) |
| 00:24:38 | <sjanssen> | procyon112: just do hSetBuffering stderr NoBuffering as the first step in your main |
| 00:24:45 | <sorear> | SamB_XP: unsafePerformIO has been formally adopted by the haskell-98 committee. Hierarchal modules haven't. |
| 00:24:59 | <SamB_XP> | sorear: oh? |
| 00:25:20 | <dons> | sorear: hmm? |
| 00:25:25 | <sjanssen> | I've always wondered about that. Why hasn't it been accepted |
| 00:25:27 | <SamB_XP> | there is a non-hierarchical FFI API? |
| 00:25:29 | <sorear> | SamB_XP: you need unsafePerformIO to make good interfaces to most foreign functions, so they rolled unsafePerformIO into the FFI spec |
| 00:25:30 | <dons> | sorear: usafePerformIO is part of the FFI standard. |
| 00:25:39 | <procyon112> | sorear: I'm not convinced that redirecting the arrow's left and right hand sides to a map lookup will solve the infinite loop... It seems it would just make it faster... |
| 00:25:40 | <dons> | hier modules are part of the hier modules (draft) standard |
| 00:25:42 | <sorear> | SamB_XP: yeah, import Foreign |
| 00:25:55 | <SamB_XP> | sjanssen: I though the reason was "because there is no Haskell 98 committee" |
| 00:25:58 | <sorear> | FFI is an accepted addendum iirc. |
| 00:26:00 | <dons> | its not really anything to do with the h98 comittee, which doesn't exist. |
| 00:26:21 | <dons> | ffi and hier modules are the only accepted addendums :-) |
| 00:27:00 | <sorear> | procyon112: it lets you notice when you are revisiting visited nodes, which you'll do infinitely often in a cyclic graph |
| 00:28:26 | <dons> | sorear: so do you have any patches to commit back into the binary builder code? |
| 00:28:37 | <dons> | dcoutts and I were pondering your discoveries yesterday. |
| 00:28:44 | <dons> | some quite interesting ideas there. |
| 00:28:48 | <dcoutts> | ah yes |
| 00:30:02 | <procyon112> | hmm. even with hSetBuffering stderr NoBuffering, I only get a partial list printout. |
| 00:31:49 | <procyon112> | sorear: where am I revisiting nodes though? I'm not seeing the line where that can happen. |
| 00:32:02 | <kolmodin> | dons: try 'runGet getWord8 empty' |
| 00:32:27 | <dons> | it fails, doesn't it? |
| 00:32:33 | <dons> | since that's unchecked? |
| 00:32:37 | <kolmodin> | dons: so hard you c'n belive it |
| 00:32:39 | <kolmodin> | segfault |
| 00:32:44 | <kolmodin> | aye |
| 00:32:53 | <dons> | so that's a little surprising |
| 00:32:53 | <kolmodin> | got a patch about it earlier today |
| 00:32:57 | <dons> | cool |
| 00:33:05 | <kolmodin> | yeah, but the patch suprised me |
| 00:33:12 | <dons> | oh? |
| 00:33:18 | <kolmodin> | - when (B.length now < n) $ |
| 00:33:19 | <kolmodin> | - fail "too few bytes" |
| 00:33:19 | <kolmodin> | + () <- when (B.length now < n) $ |
| 00:33:20 | <kolmodin> | + fail "too few bytes" |
| 00:33:25 | <kolmodin> | wtf? |
| 00:33:32 | <dons> | hmm? |
| 00:33:43 | <dons> | someone trying to force evaluation? |
| 00:33:56 | <procyon112> | sorear: unify [] = []. unify ((t1 t2):cs) | t1==t2 = unify cs... no revisits there. |
| 00:34:14 | <kolmodin> | dons: yes |
| 00:34:18 | <dons> | kolmodin: you should just fix it to do this test once at the start. |
| 00:34:27 | <kolmodin> | dons: the weird part is that it worked |
| 00:34:36 | <dons> | hmm. |
| 00:34:45 | <dons> | lazy monads eh? |
| 00:35:13 | <kolmodin> | something like that. my world was rocked a bit |
| 00:36:02 | <dons> | ah well, better now than during the haskell workshop ;-) |
| 00:36:06 | <dons> | (can you add a test for this too?) |
| 00:36:09 | <kolmodin> | I really didn't they it'd make a difference |
| 00:36:25 | <kolmodin> | dons: sure, I'll add one the next days |
| 00:37:18 | <procyon112> | sorear: I may have just pegged it. |
| 00:38:07 | <kolmodin> | the patch I got did enforce 'fail' on another place too, but I don't see how that's required |
| 00:38:10 | <kolmodin> | this was on getBytes |
| 00:38:39 | <sorear> | back. |
| 00:39:58 | <procyon112> | sorear: nope. I made it halting.. but incorrect :P |
| 00:40:22 | <sorear> | procyon112: remember == probably doesn't do what you think it does |
| 00:40:25 | <kolmodin> | dons: I'll look more into it when I'm more.. um... never mind |
| 00:40:30 | <kolmodin> | ?localtime kolmodin |
| 00:40:33 | <lambdabot> | Local time for kolmodin is Thu Apr 5 02:40:31 2007 |
| 00:40:35 | <sorear> | > let x = [0..] in x == x -- this is true, right? |
| 00:40:40 | <lambdabot> | Terminated |
| 00:40:44 | <kolmodin> | (shit!) |
| 00:41:07 | <procyon112> | sorear: oh.... that would be bad |
| 00:41:57 | <sorear> | dons: that [patching builder] would be in direct conflict with my sinister plans. Why speed binary when you can give a bigger number for Text.PrettyPrint.Leijen_FPS_CPS is x% faster than Data.Binary? (kidding of course) :) |
| 00:43:14 | <dons> | heh |
| 00:44:17 | <dcoutts> | dons: you'll not make it in by 11... ;-) |
| 00:44:30 | <dons> | grr. |
| 00:44:35 | <dons> | ghc had to be restarted. |
| 00:44:41 | <dcoutts> | :-( |
| 00:44:50 | <sjanssen> | dcoutts the slave driver? |
| 00:44:52 | <dcoutts> | you can let it build while you're on the train |
| 00:44:54 | <dons> | (-O2 is not good for partial builds getting restarted) |
| 00:44:54 | <dcoutts> | @yarr! |
| 00:44:55 | <lambdabot> | Swab the deck! |
| 00:45:00 | <dons> | dcoutts: indeed |
| 00:45:11 | <dcoutts> | sjanssen: I'm the paper driver apparently |
| 00:45:22 | <dcoutts> | ChilliX told me so :-) |
| 00:45:26 | <dons> | dcoutts: go chase rl to write section 3. |
| 00:45:31 | <dons> | yes, dcoutts is the official paper driver. |
| 00:45:39 | <dcoutts> | dons: I will as soon as he gets in! |
| 00:45:41 | <sorear> | you can't just build with: ssh serenety.cse.unsw.edu.au ghc --make -j20 ? :) |
| 00:45:46 | <dons> | ACTION is merely a bulid slave. |
| 00:45:50 | <sorear> | that would be fast! |
| 00:46:14 | <sorear> | (too bad we don't have -j) |
| 00:46:31 | <sjanssen> | sorear: there's always ghc -M |
| 00:47:01 | <dons> | yeah, ghc -j could just be a wrapper that dumps ghc -M and calls gnu make. |
| 00:47:03 | <sorear> | The bad news is file-at-a-time is twice as slow as --make |
| 00:47:10 | <dons> | hmm? |
| 00:47:18 | <sorear> | the good news is it uses sixteen times as many cores! |
| 00:47:28 | <dons> | it should be a bit faster, since the compiler space doesn't grow so badly |
| 00:47:31 | <procyon112> | sorear: just fixing the == guard by expressing it with explicit pattern matches did not change anything. |
| 00:47:51 | <sorear> | <*procyon112> sorear: unify [] = []. unify ((t1 t2):cs) | t1==t2 = unify |
| 00:48:07 | <sorear> | if t1 and t2 are infinite, and the same, that guard will diverge |
| 00:48:13 | <sorear> | > [0..] == [0..] |
| 00:48:17 | <lambdabot> | Terminated |
| 00:48:58 | <sorear> | ghc does NOT take advantage of pointer equality in == |
| 00:49:28 | <sjanssen> | that'd be evil evil evli |
| 00:49:44 | <procyon112> | sorear: right. I got rid of the guard and replaced it with ((TVar, x):rest) | x==TVar = unify rest and ((TInt,TInt):rest) = unify rest |
| 00:50:11 | <resiak> | sjanssen: why would it be so bad? |
| 00:50:19 | <sorear> | procyon112: you need more than one TVar, else it's hardly a type system |
| 00:50:40 | <procyon112> | sorear: I do :P I was being concise |
| 00:50:59 | <sjanssen> | resiak: because "let x = [0..] in x == x" and "[0..] == [0..]" should be the same thing |
| 00:52:00 | <sorear> | procyon112: the fundamental problem is that you aren't checking for revisiting nodes |
| 00:52:22 | <sorear> | if you try to unify (a -> a)@a and (b -> b)@b |
| 00:52:30 | <sorear> | it sees that -> == -> |
| 00:52:34 | <dons> | resiak: yeah, it would make equational reasoning much harder, since you can't just replace variable names with their definitions |
| 00:52:53 | <sorear> | then it has ( (a -> a)@a, (b -> b)@b ) ( (a -> a)@a, (b -> b)@b ) |
| 00:52:57 | <sorear> | then it has ( (a -> a)@a, (b -> b)@b ) ( (a -> a)@a, (b -> b)@b ) ( (a -> a)@a, (b -> b)@b ) |
| 00:53:06 | <sorear> | then it has ( (a -> a)@a, (b -> b)@b ) ( (a -> a)@a, (b -> b)@b ) ( (a -> a)@a, (b -> b)@b ) ( (a -> a)@a, (b -> b)@b ) |
| 00:53:16 | <sorear> | and so on ad infinitum |
| 00:53:22 | <resiak> | sjanssen, dons: doh, yeah, that makes sense |
| 00:54:56 | <procyon112> | sorear: what does @ mean in (a->a)@a? |
| 00:55:34 | <sorear> | I'm alluding to as-pattens |
| 00:55:42 | <sorear> | (a -> a) @ a is: |
| 00:56:37 | <sorear> | lhs /-\ |
| 00:56:37 | <sorear> | | V |
| 00:56:37 | <sorear> | ROOT => FUN |
| 00:56:37 | <sorear> | | ^ |
| 00:56:40 | <sorear> | rhs \-/ |
| 00:56:48 | <sorear> | get the picture? :) |
| 00:57:17 | <QtPlatypus> | ACTION doesn't. |
| 00:57:57 | <Lemmih> | Is it some sort of rocket? |
| 00:58:01 | <QtPlatypus> | How would I read "@" |
| 00:58:09 | <procyon112> | sorear: yes |
| 00:58:16 | <procyon112> | sorear: but then, shouldn't |
| 00:58:20 | <sorear> | alternatively, M.fromList [(0, 0 :-> 0)] |
| 00:58:20 | <mauke> | "as" |
| 00:58:37 | <sorear> | procyon112: shouldn't? |
| 00:58:53 | <procyon112> | Here's a constraint set then, that should send my unifier into non-halting: |
| 00:59:22 | <procyon112> | [(a,a->a),(b,b->b),(a,b)] |
| 00:59:33 | <procyon112> | But it doesn't... It's well behaved. |
| 00:59:35 | <sorear> | let anInfiniteType = anInfiniteType :-> anInfiniteType in [(anInfiniteType , anInfiniteType)] -- should do it |
| 00:59:43 | <sorear> | ACTION resolves |
| 01:00:34 | <procyon112> | sorear: express that as a list of constraints... my solver doesn't do lets ;) |
| 01:01:41 | <procyon112> | sorear: or native lists |
| 01:03:13 | <procyon112> | sorear: got one that does it |
| 01:03:28 | <sorear> | cool |
| 01:04:13 | <procyon112> | unify [(a,b),(b,a),(a,a->a),(b,a->b)] |
| 01:04:33 | <procyon112> | I'll try to diagnose that |
| 01:04:55 | <sorear> | your checker has a very interesting approach ... not like mine in some important ways |
| 01:05:15 | <procyon112> | sorear: Mine is pretty much straight out of TAPL chapter 22. |
| 01:05:48 | <sorear> | Oh. Don't have that :) |
| 01:05:53 | <procyon112> | Great book. |
| 01:06:05 | <sorear> | and I've yet to understand a single description of a type inference algorithm |
| 01:06:13 | <procyon112> | ^^ the (b,a) constraint is not needed to make it blow stack. |
| 01:06:15 | <sorear> | I've invented plenty tough |
| 01:07:15 | <procyon112> | sorear: TAPL makes it quite simple... It's not *easy*, but it's the best book/paper I've seen to explain all things type related. |
| 01:08:12 | <procyon112> | I figure once I get through "The Art of Prolog" with TAPL under my belt, I will be ready to change my name to Oleg :) |
| 01:11:19 | <procyon112> | Hmmm... [(a,a->a),(b,b->b),(a,b)] unifies, but if I move (a,b) to the head, it blows the stack.... |
| 01:12:05 | <sorear> | [(a,a->a),(a,a->a)] does NOT |
| 01:14:05 | <procyon112> | sorear: does NOT unify you mean |
| 01:14:22 | <procyon112> | sorear: cuz it *does* blow the stack ;) |
| 01:14:30 | <sorear> | right :) |
| 01:14:53 | <sorear> | it steps to [(a->a, (a->a)->(a->a)] |
| 01:15:06 | <sorear> | which steps (by decomposition) to |
| 01:15:20 | <sorear> | [(a,a->a),(a,a->a)] |
| 01:15:25 | <sorear> | look familiar? :) |
| 01:15:40 | <procyon112> | yes |
| 01:19:05 | <sorear> | how hard is TAPL? harder than the implementation of functional programming languages? |
| 01:20:19 | <procyon112> | sorear: Implementation of PL: a tutorial, or his other implementation of programming languages that has type inferrence? |
| 01:20:46 | <sorear> | his other implementation of functional programming languages |
| 01:20:52 | <sorear> | the 500 page one |
| 01:22:01 | <procyon112> | easier than the 500 page one.. harder than the 300 page one ;) |
| 01:22:35 | <procyon112> | (the 300 page one, IFPL:A Tutorial doesn't deal with type inference) |
| 01:23:14 | <procyon112> | TAPL is a gentler learning curve, and MUCH more comprehensive when it comes to type issues. |
| 01:24:31 | <procyon112> | And covers Existential, dependant, subtyping, sum types, product types, F, Fc, Fw, and more. With implementation code in OCaml |
| 01:25:58 | <sorear> | Fc?! |
| 01:26:10 | <sorear> | that's a bit new |
| 01:26:13 | <sorear> | >2000 iirc |
| 01:26:18 | <sorear> | isn't tapl 1996? |
| 01:26:55 | <procyon112> | er.. not Fc. and no, it' |
| 01:26:59 | <procyon112> | it's 2002 |
| 01:27:18 | <procyon112> | F, Fw, F<: and F<:w |
| 01:28:21 | <procyon112> | Where F<:w has <: as a subscript and w as a superscript to F :) |
| 01:29:27 | <Pseudonym> | @yow |
| 01:29:28 | <lambdabot> | My CODE of ETHICS is vacationing at famed SCHROON LAKE in upstate New |
| 01:29:28 | <lambdabot> | York!! |
| 01:29:35 | <Pseudonym> | IRC + typography don't mix. |
| 01:30:01 | <sorear> | well, it's a book on types, so... |
| 01:30:02 | <mwc> | Pseudonym, use the latex notation |
| 01:31:37 | <sjanssen> | > null [undefined :: Integer ..] |
| 01:31:39 | <lambdabot> | False |
| 01:32:56 | <sjanssen> | > null [undefined :: Int ..] |
| 01:32:58 | <lambdabot> | Undefined |
| 01:34:09 | <sjanssen> | @tell dcoutts how does the new list stuff handle enumFrom undefined, where undefined is Int or Integer? |
| 01:34:10 | <lambdabot> | Consider it noted. |
| 01:34:50 | <Pseudonym> | > [ 2^31-2::Int ..] |
| 01:34:52 | <lambdabot> | [2147483646,2147483647] |
| 01:35:06 | <Pseudonym> | sjanssen: Does that answer your question? |
| 01:35:33 | <sjanssen> | Pseudonym: yes, I understand why Int is that way |
| 01:35:51 | <Pseudonym> | And you understand why Integer isn't. |
| 01:35:51 | <sjanssen> | I'm wondering whether it's reasonable to give Integer the same strictness |
| 01:36:29 | <sjanssen> | making enumFrom :: Integer -> [Integer] lazy is just a space leak waiting to happen |
| 01:36:34 | <mwc> | Hmmm, Haskell is 17 now? |
| 01:36:39 | <sorear> | No, it happened. |
| 01:36:58 | <sorear> | I used last [0..] as a do-nothing-in-constand-space-forever op. |
| 01:37:07 | <sorear> | Kaboom. |
| 01:37:28 | <sjanssen> | I've seen it before, for what it's worth |
| 01:37:39 | <dcoutts> | sjanssen: you mean is it strict or not? |
| 01:37:39 | <lambdabot> | dcoutts: You have 1 new message. '/msg lambdabot @messages' to read it. |
| 01:38:05 | <sjanssen> | dcoutts: yes, is enumFrom :: Integer -> [Integer] strict? |
| 01:39:57 | <beelsebob> | ==================================================================================================================================================================================================================================================================================================================================================================================================================================================== |
| 01:40:02 | <dcoutts> | sjanssen: the current base definitions are strict for Int, our stuff doesn't change that but we coulld do easily |
| 01:40:49 | <Pseudonym> | Honestly, though. Generally, if you use an enumFrom* operation, it's generally because you want the elements in the list. |
| 01:40:53 | <sjanssen> | dcoutts: yes, Int should be strict. But what about Integer (note that base makes it lazy)? |
| 01:41:06 | <Pseudonym> | I don't see any advantage to making it strict. |
| 01:41:09 | <dcoutts> | sjanssen: technically it should be non-strict for Int too |
| 01:41:16 | <dcoutts> | Pseudonym: makes it a tad quicker |
| 01:41:30 | <Pseudonym> | dcoutts: I guess. |
| 01:41:30 | <dcoutts> | Pseudonym: for Int at least since it can be unboxed |
| 01:41:36 | <sjanssen> | dcoutts: I don't see how that is possible, remember it has to do bounds checking |
| 01:41:41 | <Pseudonym> | OK, exercise: |
| 01:41:41 | <dcoutts> | for Integer there is no such advantage |
| 01:41:54 | <Pseudonym> | Come up with an Enum type for which it doesn't make sense to make enumFrom* strict. |
| 01:41:58 | <bd_> | dcoutts, there's a reason for it to be strict |
| 01:41:58 | <dcoutts> | since it's not unboxable |
| 01:42:09 | <bd_> | > enumFrom 1 !! 1000000 :: Integer |
| 01:42:11 | <lambdabot> | Exception: stack overflow |
| 01:42:14 | <bd_> | > enumFrom 1 !! 1000000 :: Int |
| 01:42:16 | <lambdabot> | 1000001 |
| 01:42:25 | <Pseudonym> | bd_: If you write that, you deserve what you get. |
| 01:42:26 | <sjanssen> | dcoutts: not unboxable today. It might be unboxable in the future |
| 01:42:34 | <dcoutts> | sjanssen: true |
| 01:42:40 | <bd_> | Pseudonym: True, but people have been surprised in the past |
| 01:42:45 | <sorear> | dcoutts: Integer *can* be unboxed |
| 01:42:54 | <sorear> | not automatically |
| 01:43:03 | <sjanssen> | as far as I can tell Integer is the *only* Enum instance where enumFrom isn't strict |
| 01:43:12 | <bd_> | sjanssen: Also Double I think |
| 01:43:15 | <bd_> | > enumFrom 1 !! 1000000 :: Double |
| 01:43:17 | <Pseudonym> | bd_: Remember this only came up when sjanssen wanted a constant-space infinite loop, and was surprised when that didn't happen. |
| 01:43:17 | <lambdabot> | Exception: stack overflow |
| 01:43:27 | <bd_> | heh |
| 01:43:30 | <sorear> | Pseudonym: actually sorear |
| 01:43:33 | <sjanssen> | > null [undefined :: Double ..] |
| 01:43:34 | <lambdabot> | False |
| 01:43:35 | <Pseudonym> | Ah, yes. |
| 01:43:37 | <dcoutts> | sjanssen: anything that doesn't override the enumFromTo but uses the default will be lazy |
| 01:43:41 | <Pseudonym> | Sorry. |
| 01:43:48 | <dcoutts> | since the Enum default impl is lazy |
| 01:44:02 | <Pseudonym> | OK, I got one. |
| 01:44:08 | <Pseudonym> | data Peano = Zero | Succ Peano |
| 01:44:09 | <sjanssen> | dcoutts: I think you're incorrect. derived instances of Enum do bounds checking |
| 01:44:15 | <Pseudonym> | Define the "obvious" enum instance. |
| 01:44:38 | <Pseudonym> | It doesn't always make sense for enumFrom* to be lazy on Peano. |
| 01:44:39 | <dcoutts> | sjanssen: hmm ok, I'd have to look in more detail |
| 01:44:48 | <sjanssen> | > null [undefined :: Bool ..] -- proof for dcoutts |
| 01:44:49 | <lambdabot> | Undefined |
| 01:44:50 | <Pseudonym> | Because you can peel off Succs from an undefined just fine. |
| 01:45:41 | <dcoutts> | sjanssen: ah, that's because they're defined in terms of enumFromTo for Int, and that is strict |
| 01:46:05 | <sjanssen> | dcoutts: they have to be strict, otherwise they wouldn't work |
| 01:46:13 | <sorear> | dcoutts: |
| 01:46:13 | <sorear> | enumFromLargeInteger :: Int# -> ByteArray# -> [Integer] |
| 01:46:13 | <sorear> | enumFromLargeInteger l# ba# = case int2Integer# 1# of (# l2#, ba2 #) -> J# l# ba# : (case plusInteger# l# ba# l2# ba2# of (# l3#, ba3# #) -> enumFromLargeInteger l3# ba3# |
| 01:46:33 | <dcoutts> | sjanssen: I don't see why |
| 01:47:36 | <sjanssen> | dcoutts: enumFrom terminates after the largest element, how can you do this lazily? |
| 01:47:59 | <sjanssen> | well, I guess the implementation could always yield at least one cons cell, but that is pretty useless |
| 01:50:35 | <sjanssen> | enumFrom x = x : if isMaxValue x then [] else enumFrom (succ x) -- is a lazier option |
| 01:51:20 | <dcoutts> | sjanssen: hmm, good point |
| 01:52:04 | <dcoutts> | yeah so it only need to check for the max bound when it increments |
| 01:52:07 | <sjanssen> | that extra laziness is worthless though |
| 01:52:23 | <dcoutts> | I'm not sure what the h98 report specifies for the strictness of enumFromTo |
| 01:52:38 | <dcoutts> | the base has a note that the Int instance is indeed to strict |
| 01:52:54 | <procyon112> | sorear: I think I've got it! |
| 01:54:41 | <procyon112> | sorear: The occurs check needs to stay, but instead of failing, if moves the constraint to the end of the list. This makes the constraint still available for substitutions, but doesn't force it to traverse the constraint indefinitely. |
| 01:54:55 | <procyon112> | sorear: so the guard becomes: |
| 01:55:47 | <procyon112> | | tyX `occursIn` tyS = fmap (++[(tyX,tyS)]) $ unify rest |
| 01:56:57 | <sjanssen> | dcoutts: the report says (referring to Int, Integer, Float, Double) "For all four of these Prelude numeric types, all of the enumFrom family of functions are strict in all their arguments" |
| 01:57:14 | <sjanssen> | ACTION goes to report the bug and write a patch |
| 01:58:22 | <dcoutts> | sjanssen: ah ok, good |
| 01:58:33 | <sjanssen> | yikes, Hugs is even worse. null [undefined :: Int ..] == False in Hugs |
| 01:59:27 | <procyon112> | sorear: yes.. this appears to be testing ok for everything I throw at it. |
| 02:18:31 | <sorear> | ACTION 's brain fills with swirling pointers... |
| 02:22:31 | <procyon112> | night all |
| 02:32:51 | <sorear> | @spell Patterson |
| 02:32:52 | <lambdabot> | Patterson |
| 02:32:54 | <sorear> | @spell Paterson |
| 02:32:54 | <lambdabot> | Paterson |
| 02:33:00 | <sorear> | aww, both valid |
| 02:33:44 | <sorear> | ok, Paterson it is |
| 02:35:02 | <sjanssen> | @spell Janssen |
| 02:35:02 | <lambdabot> | Jansen Jensen Kansan Hansen Jasen |
| 02:35:11 | <sjanssen> | I hate you |
| 02:35:47 | <dons> | hehe. |
| 02:35:58 | <dons> | @spell Spencer |
| 02:35:58 | <lambdabot> | Spencer |
| 02:36:06 | <dons> | Spencer Kansan, I like it! |
| 02:37:03 | <sorear> | The algorithm intro I'm writing for my uber-fast pretty printing lib is start to sound a bit ... papery |
| 02:38:23 | <dons> | sorear: you'll be pleased to know, one of the nobench programs now runs 159x faster with a new ghc optimisation rl added. |
| 02:38:33 | <sorear> | nice. |
| 02:38:33 | <kc5tja_> | @spell supercalifragilisticexpialidocious. |
| 02:38:56 | <sorear> | dons: is this like a RULESy thing? |
| 02:39:08 | <dons> | nah, static argument transformation |
| 02:39:16 | <sorear> | nice |
| 02:39:23 | <sorear> | will that fold 2^26? |
| 02:39:26 | <dons> | fixes the 'atom' benchmark, which ghc has been losing to hugs on for 20 years. |
| 02:39:53 | <sorear> | so when was this added? |
| 02:40:02 | <sorear> | ACTION intents to read the patch long description |
| 02:40:05 | <dons> | its in our stream fusion branch of ghc. |
| 02:40:13 | <dons> | so you'll have to wait to till the paper is done |
| 02:40:24 | <sorear> | ouch, you branched all of gcc? not just base? |
| 02:40:36 | <sorear> | what did you have to change in the compiler proper? |
| 02:41:12 | <dons> | too busy right now. deadline in 30 hours. read the paper to find out ;-) |
| 02:41:25 | <dons> | new optimisations, in ghc. stuff in base too. |
| 02:42:38 | <sorear> | so how badly do we want general computation and guards in rules? |
| 02:43:31 | <dons> | not really. lightweight term rewriting does 98% of things we need. |
| 02:44:25 | <sjanssen> | hmm, where might Hugs define instance Enum Integer? |
| 02:44:26 | <sorear> | ACTION also has an idea for adding inline C to GHC |
| 02:44:33 | <sorear> | Hugs.Prelude maybe? |
| 02:44:59 | <sjanssen> | oh, that isn't in base? |
| 02:45:08 | <kc5tja_> | You creates of sheep! Everyone knows, BCPL is the language of the future! :D |
| 02:45:15 | <sorear> | grep saves the day! |
| 02:45:17 | <sorear> | /usr/src/hugs98-98.200503.08/libraries/hugsbase/Hugs/Prelude.hs:instance Enum Integer where |
| 02:45:24 | <sjanssen> | please don't tell me I have to use CVS to fix this simple patch... |
| 02:45:29 | <sjanssen> | s/patch/bug |
| 02:45:58 | <sjanssen> | @google hugsbase |
| 02:46:01 | <lambdabot> | http://osdir.com/ml/lang.haskell.cvs.hugs/2005-04/msg00000.html |
| 02:46:01 | <lambdabot> | Title: cvs commit: hugs98/libraries/hugsbase/Hugs Array.hs Int.hs Prelude.hs Wo |
| 02:46:14 | <dons> | yeah, there's a few other strictness bugs floating around |
| 02:46:25 | <dons> | since StrictCheck et al are fairly new. |
| 02:47:30 | <dons> | sjanssen: i note that tiled xmonad is perfect for doing 3 or 4 simultaneous ghc builds , and observing their progress |
| 02:48:02 | <sjanssen> | :) |
| 02:48:04 | <c> | or compiling dwm with different settings (say if youre on an 8086 and it takes 12 seconds) |
| 02:48:49 | <sjanssen> | pssh, why would you build dwm if you've already got xmonad? |
| 02:49:05 | <c> | does xmonad run on win32 yet? |
| 02:49:12 | <c> | im longing for a proper wm on win32 |
| 02:49:25 | <c> | bblean can mostly be hacked to work like dwm, but its tiling plugin sucks |
| 02:49:35 | <dons> | are there any unixy window managers that run on windows? |
| 02:49:36 | <sjanssen> | xmonad would probably run on one of those X servers in windows |
| 02:50:30 | <glguy> | you can use nearly any window manager with cygwin/X |
| 02:50:37 | <glguy> | or do you mean directly instead of explorer? |
| 02:51:03 | <c> | i mean i dont know the goals of xmonad |
| 02:51:11 | <sorear> | I'd like to hack yi to use the xmonad/dwm layout algorith. |
| 02:51:13 | <c> | but , it seems it could be abstracted away from X to work on win32, plan9, osux |
| 02:51:21 | <sorear> | would anyone hurt me if I did that? |
| 02:51:22 | <dons> | its a dwm clone, with some nicer features, in 400 lines. |
| 02:51:23 | <c> | that is, if theres alternate WMs on osux and an API for them , besids X. (afaik there isnt, is there?) |
| 02:52:15 | <sjanssen> | xmonad (and every X window manager, I imagine) is very X specific |
| 02:54:51 | <sjanssen> | ARG talked about an API for window management on the dwm list recently |
| 02:56:34 | <sorear> | does anyone but arg know the suckless wiki password? |
| 02:57:00 | <kc5tja_> | sorear: only those with 80x25 screens. ;D |
| 02:57:25 | <sorear> | Oh well. my screen is 128x48 |
| 02:57:44 | <sjanssen> | sorear: 128 still seems a bit tight |
| 02:57:57 | <kc5tja_> | With the font that I use at work, it's 262x100 something. |
| 02:58:06 | <kc5tja_> | People are always like, "How can you SEE anything?!" |
| 02:58:25 | <sorear> | No, I care about legibility. |
| 02:58:36 | <sorear> | I use the linux console because fixed is ugly |
| 02:58:41 | <kc5tja_> | So do I. I just happen to be able to read a 5x8 pixel font just fine. |
| 02:58:54 | <sorear> | much, much, much, uglier than the intel rom font |
| 02:59:18 | <kc5tja_> | I disagree, personally, but then, I did come up with my own 8x8 font that I'd like to expand to 8x16 for the PC. |
| 02:59:46 | <kc5tja_> | I could write it in Forth in about a couple hours, but I am thinking of using Haskell for the practice. |
| 03:00:12 | <sorear> | oh this sounds really easy |
| 03:00:14 | <sorear> | ACTION man psf |
| 03:00:22 | <sorear> | just a little list processing |
| 03:00:37 | <kc5tja_> | Nothing appropriate. |
| 03:00:56 | <kc5tja_> | Is that the console font format, or the X11 format? |
| 03:01:03 | <sorear> | fst |
| 03:01:19 | <sorear> | and it turns out there's no manpage |
| 03:01:21 | <allbery_b> | grah. still can't solve my gtk2hs right-justified label issue |
| 03:01:27 | <sorear> | ACTION mg /usr/share/doc/console-tools/file-formats/psf |
| 03:01:31 | <kc5tja_> | sorear: OK, I'm confused. What are you talking about? |
| 03:01:36 | <kc5tja_> | Oh, OK, just making sure. :) |
| 03:01:41 | <kc5tja_> | There is no list processing involved. |
| 03:02:04 | <allbery_b> | okay, maybe I could use a textview... after figuring out how to make it match everything else in that part of the window instead of a white background, and using the rather more complicated API (since it's really an embedded editor window) |
| 03:02:05 | <kc5tja_> | ACTION was going to read in the full 2KB binary image that I used for the Kestrel, scan double it, then spit it back out in the PSF. |
| 03:02:33 | <sorear> | exactly - list processing. |
| 03:02:43 | <dcoutts> | allbery_b: single line or multi-line ? have you used the x-align? |
| 03:02:44 | <kc5tja_> | Well, more like ByteString processing. :) |
| 03:02:50 | <sorear> | [Word8] |
| 03:02:51 | <allbery_b> | single line |
| 03:03:05 | <sorear> | bytestrings are slower than strings when you have that little data |
| 03:03:08 | <allbery_b> | I'm putting a number in it, I want it to be up to 5 digits and right-justified |
| 03:03:32 | <allbery_b> | the justification attribute doesn't apply to single line labels; the x-align aligns the *widget* but not the text within it |
| 03:03:33 | <dcoutts> | allbery_b: and have you set the packing so your label expand to fill the space available? |
| 03:03:39 | <jcreigh> | dons: what, you don't like red? :) |
| 03:03:42 | <kc5tja_> | sorear: Believe me, speed isn't the issue. |
| 03:03:55 | <allbery_b> | as far as I can tell, gtk2 formats the next into a "box" whose alignment you cannot specify |
| 03:03:57 | <kc5tja_> | For that little data, on an 800MHz box, even if it takes one second, I'm still happy. |
| 03:04:08 | <dcoutts> | allbery_b: you can see how it all works interactively if you try it in glade |
| 03:04:29 | <allbery_b> | I can right-justify the text that gets formatted into that "box", or the label that holds the "box", but the "box" itself is always left-justified |
| 03:04:29 | <sorear> | ok, I just finished writing the scan doubler |
| 03:04:31 | <sorear> | in haskell |
| 03:04:41 | <allbery_b> | and yes, I've been mucking with this in glade as well |
| 03:04:44 | <kc5tja_> | You don't even know the source image format. |
| 03:04:52 | <sorear> | it's not psf? |
| 03:05:00 | <kc5tja_> | No, it's for my Kestrel. |
| 03:05:02 | <sorear> | if it was: interact (\(0x36:0x04:0:8:xs) -> 0x36:0x04:0:16:(xs >>= replicate 2)) |
| 03:05:04 | <kc5tja_> | I want to move it to the PC though. |
| 03:05:09 | <sorear> | standard formats ftw! |
| 03:05:34 | <sorear> | given that you're copying the nicities off it, does this mean the kestrel project is dead? |
| 03:05:36 | <kc5tja_> | sorear: The format is pretty standard -- it's a raw, uncompressed, 2048x8 bitmap. |
| 03:05:39 | <sorear> | would be a shame |
| 03:06:11 | <sorear> | kc5tja_: so <data for char0> <data for char1> ... |
| 03:06:12 | <kc5tja_> | sorear: No. Quite the contrary; the whole reason I'm moving the font onto the PC is precisely because I *LIKE* the Kestrel's font. |
| 03:06:32 | <sorear> | kc5tja_: where the data for each character is stored left-right top-bottom? |
| 03:06:54 | <dcoutts> | allbery_b: your label needs to "expand" and "fill" and the x-align needs to be 1. |
| 03:06:55 | <kc5tja_> | Right; take a 2048 pixel wide bitmap, draw all 256 characters onto it, then save it as a bitmap. |
| 03:07:11 | <kc5tja_> | PSF is column-major in comparison. |
| 03:07:48 | <dons> | woot,! http://www.haskell.org/pipermail/haskell-cafe/2007-April/024127.html |
| 03:07:49 | <lambdabot> | Title: [Haskell-cafe] 25k lines of ASP to 4.2k lines of Haskell, with considerably more ..., http://tinyurl.com/yporof |
| 03:07:49 | <kc5tja_> | (the reason this layout was chosen is because it's the most natural font format for use in the ROM's drawing routines) |
| 03:07:51 | <sorear> | no, psf is left-right-top-bottom too |
| 03:07:56 | <dons> | jcreigh: i guess red is ok. |
| 03:08:15 | <jcreigh> | dons: I gues circular dependencies are out, huh? |
| 03:08:16 | <sorear> | so kestrel is column major? |
| 03:08:27 | <sorear> | psf is row major |
| 03:08:33 | <sjanssen> | jcreigh: yes |
| 03:08:41 | <allbery_b> | dcoutts: did all three. the 0 that is specified as the default text in glade is still left-justified; likewise the actual data when I run the program |
| 03:09:16 | <sjanssen> | jcreigh: I've noticed this annoying issue too. It seems like every config (even the static stuff) has to be in XState, which is getting bloated |
| 03:09:54 | <dons> | mod up! http://programming.reddit.com/info/1fh3p/comments |
| 03:09:55 | <lambdabot> | Title: Haskell in the real world: building a commercial website in Haskell with WASH (r ... |
| 03:10:03 | <dons> | (its been a while, and reddit has been sucking muchly) |
| 03:10:18 | <dcoutts> | allbery_b: so you can see in glade that the label widget is wider than the text, the x-align is 1.0 and it still is at the far left of it's allocated space ? |
| 03:10:36 | <dcoutts> | allbery_b: I tried what I said a second ago in glade and it works, I'm not sure what's going on |
| 03:11:00 | <kc5tja_> | sorear: Dude, I just cannot find the format for this file. |
| 03:11:12 | <sorear> | @remember AdamPeacock Once I looked at the source code, 25000 lines of ASP, I reckoned it would be easier to rewrite it in a real language. |
| 03:11:12 | <lambdabot> | Done. |
| 03:11:22 | <allbery_b> | the label widget has a width of 5 characters, this is visible in the window view. the content "0" i left-justified |
| 03:12:35 | <jcreigh> | sjanssen: is defaultLayoutDesc the only one like that, or are there others? |
| 03:12:41 | <sorear> | kc5tja_: if it is 2048 bytes and uncompressed, there isn't much flexibility. try my script (assumes char/row/col ordering, but that's the natural thing for a row-major display), if your screen is garbled then try the 5 other permutations |
| 03:12:42 | <allbery_b> | http://www.ece.cmu.edu/~allbery/hwatchmwp.glade (hwatchmwp.gladep is also there if it matters) |
| 03:13:01 | <allbery_b> | it's possible the installed version of glade here is buggy, but it looks like I set everything right |
| 03:13:30 | <sorear> | display tubes have are row-major in a tradition that goes back to the days of Farnsworth |
| 03:14:23 | <sjanssen> | jcreigh: defaultLayoutDesc is the only one right now |
| 03:14:36 | <jcreigh> | everything was fine when you just had to edit Main.hs :) |
| 03:14:39 | <jcreigh> | hmm. |
| 03:15:01 | <sjanssen> | jcreigh: but grabkeys is stuck in Main.hs because it uses several variables in Config.hs |
| 03:15:03 | <dcoutts> | allbery_b: I can't reproduce that, if the packing were wrong you'd end up with centered text, not left aligned |
| 03:15:26 | <kc5tja_> | sorear: The Kestrel's ROM expects a 2Kx8 bitmap -- period. It doesn't need flexibility, because it's not a full-on GUI. |
| 03:15:27 | <allbery_b> | hrm. I have another one that actually works correctly, but it's in a table instead of an hbox |
| 03:15:28 | <sorear> | Hmm. Haskell-mode seems to think ^#define .... is a comment |
| 03:15:53 | <allbery_b> | yeh, I noticed it thinks # is a comment character |
| 03:15:57 | <sorear> | kc5tja_: Does the kestral have a row-major display bitmap? |
| 03:16:04 | <allbery_b> | {-# ... #-} also messes it up |
| 03:16:06 | <kc5tja_> | sorear: that's all it has. |
| 03:16:27 | <kc5tja_> | no text mode; the ROM blits text to the screen in using a row-major blitting operation. |
| 03:16:41 | <sorear> | kc5tja_: ok, then, unless your display rom programmer was intoxicated you are storing fonts row-major |
| 03:17:03 | <sorear> | kc5tja_: the only question is whether the data for characters is interleaved |
| 03:17:10 | <allbery_b> | guess I'll try reimplementing that part of the gui as a table instead and see if it obeys the x-align |
| 03:17:46 | <sorear> | a0 a1 a2 a3 a4 a5 a6 a7 b0 b1 b2 b3 b4 b5 b6 b7 ... vs |
| 03:18:06 | <sorear> | ... a0 b0 ... a1 b1 ... a2 b2 ... a3 b3 ... a4 b4 ...a5 b5 |
| 03:18:23 | <sorear> | both would be about equally easy to implement I think |
| 03:18:54 | <sorear> | I think the former would be ever-so-slightly simpler still, which would be Very Good since that's the psf order |
| 03:19:06 | <sorear> | in which case you can just double bytes |
| 03:20:03 | <sorear> | @botsnack |
| 03:20:03 | <lambdabot> | :) |
| 03:20:10 | <kc5tja_> | sorear: It's stored like a normal 2048 pixel by 8 pixel bitmap would be (like a graphic image). So it's a0 b0 c0 ... a1 b1 c1 ... etc. |
| 03:20:18 | <kc5tja_> | 256 bytes of raster 0 data, 256 bytes of raster 1 data, etc. |
| 03:20:46 | <sorear> | ah, ok |
| 03:20:54 | <sorear> | sounds like a job for transpose! |
| 03:21:03 | <sorear> | @type transpose |
| 03:21:05 | <lambdabot> | forall a. [[a]] -> [[a]] |
| 03:21:35 | <kc5tja_> | ACTION was just going to read in 8 chunks of 256 bytes, then zip8 them. :) |
| 03:22:36 | <sorear> | @ty let chunk k [] = [] ; chunk k l = take k l : chunk k (drop k l) in interact $ (++) [0x36,4,0,16] . concat . transpose . chunk 256 |
| 03:22:38 | <lambdabot> | No instance for (Num Char) |
| 03:22:39 | <lambdabot> | arising from the literal `54' at <interactive>:1:84-87 |
| 03:23:08 | <sorear> | @ty let chunk k [] = [] ; chunk k l = take k l : chunk k (drop k l) in interact $ (++) "\54\4\0\16" . concat . transpose . chunk 256 |
| 03:23:10 | <lambdabot> | IO () |
| 03:23:17 | <sorear> | is that a forth oneliner too? :) |
| 03:24:43 | <allbery_b> | okay, that's got it, if not quite the way I want the spacing |
| 03:25:23 | <allbery_b> | I killed the internal boxes and the field widths (which caused centering instead of right justification!) |
| 03:26:06 | <allbery_b> | the result is somewhat wider than I wanted but I don't think there's a way to tell it to shrink the middle column of the table |
| 03:28:34 | <kc5tja_> | sorear: If I had concat, transpose, and all its dependent words in the dictionary already, YES. |
| 03:28:52 | <kc5tja_> | ACTION doesn't like these chest-beating contests with respect to languages. >:( |
| 03:29:12 | <kc5tja_> | ACTION sighs -- I'm just so not intended for this world. |
| 03:29:35 | <jcreigh> | kc5tja_: which world? |
| 03:29:42 | <kc5tja_> | _This_ world. |
| 03:29:50 | <kc5tja_> | Sol 3 |
| 03:29:53 | <jcreigh> | oh. |
| 03:29:56 | <kc5tja_> | circa 21st century. |
| 03:30:17 | <jcreigh> | sorry, but there's not another one within convienent range WRT time or space. |
| 03:30:31 | <kc5tja_> | I doubt it'd help. |
| 03:30:36 | <SamB_XP> | kc5tja: that is not particularly precise |
| 03:30:53 | <kc5tja_> | SamB_XP: What isn't? |
| 03:30:59 | <SamB_XP> | that designation |
| 03:31:03 | <chessguy> | hi haskellers |
| 03:31:04 | <kc5tja_> | So what? |
| 03:31:06 | <kc5tja_> | Who cares? |
| 03:31:08 | <SamB_XP> | "circa 21st century"? |
| 03:31:10 | <kc5tja_> | It's precise enough. |
| 03:31:23 | <jcreigh> | chessguy: hey |
| 03:31:34 | <SamB_XP> | that includes the world after 90 years have passed! |
| 03:31:47 | <chessguy> | jcreigh: how goes the chess engine? |
| 03:32:03 | <jcreigh> | chessguy: oh, haven't touched it. :) |
| 03:32:08 | <chessguy> | awww |
| 03:32:09 | <kc5tja_> | Do you REALLY think that we're going to be that different in 90 years, given 6000 years of known, recorded history of war and unnecessary competition? |
| 03:32:18 | <jcreigh> | been busy with other stuff. |
| 03:32:28 | <chessguy> | i know how that goes |
| 03:33:07 | <SamB_XP> | kc5tja: what about computers! |
| 03:33:10 | <chessguy> | i haven't done much with mine lately either |
| 03:33:27 | <SamB_XP> | computers are very upsetting |
| 03:33:38 | <chessguy> | @quote computer |
| 03:33:39 | <lambdabot> | dons says: Welcome to computer science, we count from 0. |
| 03:34:46 | <SamB_XP> | @quote computer |
| 03:34:46 | <lambdabot> | conal says: monochrom: maybe that's only because computers program people as much as vice versa. |
| 03:34:52 | <SamB_XP> | @quote computer |
| 03:34:53 | <lambdabot> | Makali says: Whenever a programmer thinks, "Hey, skins, what a cool idea", their computer's speakers should create some sort of cock-shaped soundwave and plunge it repeatedly through their skulls. |
| 03:34:59 | <sorear> | Oh, I think in 20 years computers will replace us, and in 90 years nobody will be able to tell the difference |
| 03:35:06 | <sorear> | they'll be as dumb as we are |
| 03:35:22 | <chessguy> | sorear: like animal farm? |
| 03:35:34 | <kc5tja_> | @hoogle transpose |
| 03:35:35 | <lambdabot> | List.transpose :: [[a]] -> [[a]] |
| 03:35:35 | <jcreigh> | sorear: you're saying that someday computers will program in PHP? :) |
| 03:35:35 | <lambdabot> | Data.Graph.transposeG :: Graph -> Graph |
| 03:35:39 | <sorear> | more intellectual maybe, but if 5 billion years of evolotion are any guide.. |
| 03:35:44 | <SamB_XP> | gotta love the retirement plan! |
| 03:35:48 | <sorear> | chessguy: yeah! |
| 03:36:30 | <sorear> | chessguy: thank you for pointing out the reference! |
| 03:36:39 | <sorear> | Computers are the new proletariat. |
| 03:36:49 | <chessguy> | i haven't actually read the book, but i'd like to |
| 03:37:19 | <jcreigh> | chessguy: you should. it's a fun read |
| 03:37:25 | <chessguy> | i bet |
| 03:39:03 | <dmwit> | Could somebody take a look at http://hpaste.org/1248? I don't understand the meaning of the error message. |
| 03:39:32 | <kc5tja_> | sorear: Your code won't compile in GHCI. |
| 03:39:40 | <sorear> | why has hpaste stopped answeing? |
| 03:39:55 | <kc5tja_> | "Main.hs:xx:0: parse error on input `module'" |
| 03:40:03 | <kc5tja_> | And I'm not sure why. No useful error is given. |
| 03:40:07 | <sjanssen> | sorear: works for me |
| 03:40:12 | <sorear> | kc5tja_: Nice one! |
| 03:40:31 | <kc5tja_> | xx is 12 on my box, but that's only because I put in a comment. |
| 03:40:36 | <sorear> | kc5tja_: it errored on the auto-inserted module t |
| 03:40:39 | <sorear> | oh, sorry |
| 03:40:47 | <sorear> | I thought it said xx :) |
| 03:41:01 | <chessguy> | ?hoogle printName |
| 03:41:02 | <lambdabot> | No matches found |
| 03:41:10 | <kc5tja_> | module FontVerter where chunk k [] = [] chunk k l = take k l : chunk k (drop k l) |
| 03:41:15 | <kc5tja_> | doggone irssi. |
| 03:41:17 | <sorear> | ?hoogle pprName |
| 03:41:17 | <lambdabot> | Language.Haskell.TH.PprLib.pprName :: Name -> Doc |
| 03:41:25 | <dmwit> | printName x = widgetGetName x >>= putStrLn |
| 03:41:32 | <MarcWeber> | dmwit: GHC doesn't know which type to choose for a (line 6) |
| 03:41:41 | <sorear> | kc5tja_: there needs to be a ; between the defns for chunk |
| 03:41:53 | <kc5tja_> | sorear: Not when they're on separate lines. |
| 03:41:54 | <MarcWeber> | So using (printName :: WidgetClass -> IO ()) should fix it |
| 03:41:54 | <sorear> | no idea why it said module |
| 03:42:02 | <dmwit> | MarcWeber: But, why should it need to? The widgetGetName function is defined for all WidgetClass instances... |
| 03:42:07 | <kc5tja_> | module FontVerter where |
| 03:42:09 | <kc5tja_> | chunk k [] = [] |
| 03:42:11 | <kc5tja_> | chunk k l = take k l : chunk k (drop k l) |
| 03:42:13 | <kc5tja_> | ...etc... |
| 03:42:18 | <dmwit> | Wait, really? |
| 03:42:21 | <kc5tja_> | (hand pasted) |
| 03:42:29 | <chessguy> | kc5tja: that's what hpaste is for |
| 03:42:43 | <kc5tja_> | C'mon it's only 3 lines here. But I digress. |
| 03:43:31 | <sorear> | kc5tja_: and module is the first line in the file? |
| 03:43:40 | <kc5tja_> | MUST it be the first line? |
| 03:43:48 | <kc5tja_> | If so, that's contradictory with my CUT sources. |
| 03:43:55 | <sorear> | no, whitespace can precede it |
| 03:44:04 | <sorear> | incl. comments and pragmas |
| 03:44:07 | <dmwit> | MarcWeber: No, WidgetClass is a class, not a type... |
| 03:44:24 | <kc5tja_> | OK, I figured it out. |
| 03:44:30 | <kc5tja_> | I had the import preceding the module. |
| 03:44:32 | <kc5tja_> | Thanks. |
| 03:44:39 | <chessguy> | it would have to be (printName :: WidgetClass w => w -> IO ()) |
| 03:44:49 | <chessguy> | which is a mouthful |
| 03:44:54 | <MarcWeber> | dmwit: Have look at the definition of windowNew. See which type it returns |
| 03:45:10 | <chessguy> | ?hoogle windowNew |
| 03:45:11 | <lambdabot> | No matches found |
| 03:45:47 | <dmwit> | MarcWeber: IO Window, but why? I want this to work on the button/entry widgets, too. |
| 03:46:14 | <MarcWeber> | dmwit: Then you need existantial types. |
| 03:46:53 | <dmwit> | chessguy: Sorry, I missed the meaning of your comment that starts, "it would have to be..." |
| 03:47:02 | <dmwit> | ?go existential types |
| 03:47:03 | <lambdabot> | http://www.cs.washington.edu/homes/djg/papers/exists_imp.pdf |
| 03:47:39 | <chessguy> | dmwit: oh, instead of MarcWeber's (printName :: WidgetClass -> IO ()) |
| 03:48:00 | <dmwit> | chessguy: Oh, okay, yeah. What you wrote is what I currently have. |
| 03:48:28 | <MarcWeber> | dmwit: Sorry, can't test it now. Have to reinstall gtk2hs |
| 03:48:30 | <chessguy> | oh, so it is |
| 03:49:10 | <kc5tja_> | ACTION sighs |
| 03:49:15 | <kc5tja_> | Now ghc isn't building the code. |
| 03:49:15 | <dmwit> | ?go existential types haskell |
| 03:49:17 | <lambdabot> | http://en.wikibooks.org/wiki/Haskell/Existentially_quantified_types |
| 03:49:17 | <lambdabot> | Title: Haskell/Existentially quantified types - Wikibooks, collection of open-content t ... |
| 03:49:53 | <kc5tja_> | ghc -o fontverter --make Main.hs ==> no Main module. |
| 03:49:58 | <sjanssen> | dmwit: I think you want to give the parameter the type Widget |
| 03:50:05 | <kc5tja_> | ghc --make Main.hs ==> compiles to .o file, but no binary. |
| 03:50:05 | <MarcWeber> | dmwit: Just try (printName :: Window -> IO ()) |
| 03:50:16 | <kc5tja_> | module has been adjusted to expose main. |
| 03:50:24 | <dmwit> | MarcWeber: That crashes if you ever focus on any widget other than the main window. ;-) |
| 03:50:26 | <kc5tja_> | Haskell just isn't being my friend. :( |
| 03:50:30 | <dmwit> | sjanssen: Okay, I'll try it. |
| 03:50:54 | <chessguy> | dmead_: oh weird. i get a different error from that code than what you did |
| 03:51:04 | <MarcWeber> | dmwit: No because you register that handler only for the window widget: onSetFocus window (printName) |
| 03:51:34 | <dmwit> | MarcWeber: Empirically false. =P |
| 03:51:35 | <kc5tja_> | n/m -- I got it. |
| 03:51:44 | <sjanssen> | dmwit: gtk2hs simulates OO subtyping with typeclasses and data types. 'Widget' is the supertype of everything in WidgetClass |
| 03:53:27 | <dmwit> | Okay, I'm still getting crashes, but it is less often, and I think I can take it from here. |
| 03:53:28 | <dmwit> | Thanks all. |
| 03:54:15 | <chessguy> | well, i got up to the point where it says (GtkTest:4268): libglade-WARNING **: could not find glade file 'hworld.glade' |
| 03:54:20 | <chessguy> | which is expected |
| 03:54:25 | <dmwit> | Heh, oh. |
| 03:54:30 | <dmwit> | Should I hpaste it for you? |
| 03:54:35 | <chessguy> | if you want |
| 03:55:05 | <chessguy> | i'd kind of like to learn some gtk2hs stuff too, so i wouldn't mind playing around with it |
| 03:55:22 | <hpaste> | dmwit annotated "gtk2hs: ambiguous type error" with "hworld.glade" at http://hpaste.org/1248#a1 |
| 03:56:42 | <dmwit> | On a side note, urxvt doesn't really play nice with xmonad. =( |
| 03:56:44 | <chessguy> | ok, the program runs for me now, but at the end, when i quit, it gives me a segmentation fault |
| 03:57:14 | <dmwit> | ! |
| 03:57:27 | <dino-> | So I wrote a small command-line utility in Haskell recently: http://ui3.info/d/proj/storylen.html |
| 03:57:29 | <lambdabot> | Title: storylen - story word count and categorization |
| 03:57:57 | <dino-> | And I followed the How to write a Haskell program page closely, doing as many of the steps as seem relevant for this project. |
| 03:58:04 | <hpaste> | chessguy annotated "gtk2hs: ambiguous type error" with "my session" at http://hpaste.org/1248#a2 |
| 03:58:18 | <dino-> | It's under darcs, it's got the right file structure, it's cabalized. It has a web page. |
| 03:58:19 | <chessguy> | all i did was click ok and then click close |
| 03:58:44 | <dons> | dino-: nice. does it use bytestrings? |
| 03:58:46 | <dons> | dmwit: oh? |
| 03:59:06 | <dino-> | dons: It does indeed. |
| 03:59:08 | <dmwit> | dons: It doesn't repaint properly when it resizes. |
| 03:59:23 | <dino-> | And before it used ByteString, it was ridiculous slow. |
| 04:00:12 | <dino-> | So, my question is: is this suitable for announcing on haskell@haskell.org? |
| 04:00:33 | <dons> | dino-: sure, or drop it by haskell-cafe@ |
| 04:00:41 | <dino-> | It's so.. not a Functional Graph Library, or an XML toolkit or even a Gameboy emulator. |
| 04:00:50 | <sorear> | dons: nice. I just discovered another advantage of my mutable-byte-array state-block approach (cf builder hacking results, parameter reduction) - it allows highly efficient mon{a,oi}d transformers |
| 04:01:19 | <chessguy> | dino-: it seems pretty specific to your needs. i don't know that many people will use it, but it could be an interesting project for people to read through |
| 04:01:35 | <dmwit> | chessguy: Yes, I can replicate the problem. (It's 100% reproducible -- pressing close always triggers a problem, no matter what happened before.) |
| 04:01:46 | <dmwit> | I just don't know enough to say waht it is yet... =) |
| 04:01:52 | <kc5tja_> | @type replicate |
| 04:01:54 | <lambdabot> | forall a. Int -> a -> [a] |
| 04:01:58 | <dino-> | chessguy: Perhaps gathering an all plain text library of books will become a trend. :) |
| 04:02:45 | <chessguy> | dmwit: hmm, i wonder if the onClick event for close is getting captured before the printName |
| 04:03:05 | <chessguy> | dmwit: and then by the time the printName executes, the window is already destroyed |
| 04:03:30 | <dmwit> | Well, commenting out the printName line seems to fix it, so... |
| 04:03:42 | <chessguy> | hmm. all thought just tabbing around the window is enough to give a segmentation fault too |
| 04:03:48 | <chessguy> | s/thought/though/ |
| 04:03:55 | <dmwit> | Yeah, that's why I was saying it still crashed sometimes. |
| 04:04:09 | <dmwit> | Specifically, tabbing past "OK" or shift-tabbing before the entry box will do it. |
| 04:04:15 | <chessguy> | yes |
| 04:04:18 | <chessguy> | interesting |
| 04:04:32 | <jcreigh> | type ReadS a = String -> [(a,String)] -- okay, but what should I return as the String? |
| 04:04:37 | <chessguy> | you might want to chat with dcoutts about this, he knows a lot about gtk2hs |
| 04:04:41 | <dcoutts> | @arr |
| 04:04:42 | <lambdabot> | Drink up, me 'earties |
| 04:04:43 | <abz> | yeah dons, thanks for the reddit. |
| 04:04:55 | <chessguy> | speaking of which :) |
| 04:04:56 | <dmwit> | Hi dcoutts! |
| 04:04:57 | <jcreigh> | I'm trying to declare an instace of Read... |
| 04:05:04 | <dcoutts> | abz: looks good! |
| 04:05:12 | <abz> | thanks. |
| 04:05:18 | <dcoutts> | abz: nice success story too :-) |
| 04:05:36 | <dcoutts> | abz: I wonder if this'll motivate anyone to clean up wash |
| 04:05:54 | <abz> | it needs it, i've got a good list of TODOs for wash |
| 04:05:58 | <chessguy> | ?src Read |
| 04:05:59 | <lambdabot> | class Read a where |
| 04:05:59 | <lambdabot> | readsPrec :: Int -> ReadS a |
| 04:05:59 | <lambdabot> | readList :: ReadS [a] |
| 04:05:59 | <lambdabot> | readPrec :: ReadPrec a |
| 04:05:59 | <lambdabot> | readListPrec :: ReadPrec [a] |
| 04:06:07 | <dons> | abz, yeah, great work. |
| 04:06:15 | <dcoutts> | abz: heh, I bet. #1 make it build in less than 2 hours ;-) |
| 04:07:03 | <abz> | fast_cgi is the most needed thing. atm it just exitWith Success after each page... |
| 04:07:09 | <dmwit> | dcoutts: I'm playing with gtk2hs, and found a few ways to make it send makeNewGObject a NULL pointer. Do you want/have time to educate me? =) |
| 04:07:51 | <abz> | which means that you have to do everything (like send emails) before rendering the page. |
| 04:07:52 | <dcoutts> | abz: I suggest you post your wash todo list as a follow up to the mailing list and/or your general development experiences with it, I'm sure people will be interested |
| 04:08:40 | <dcoutts> | abz: you've seen Happs to of course, I wonder if any integration between the two would help since wash is mainly about front end and happs mainly about back end |
| 04:08:50 | <abz> | dcoutts: good thought, will do. |
| 04:09:01 | <dino-> | Speaking of ByteString, I was having performance problems until someone told me to use -O2. And dcoutts emailed about the inlining: http://www.haskell.org/pipermail/haskell-cafe/2007-March/024027.html |
| 04:09:03 | <lambdabot> | Title: [Haskell-cafe] Data.ByteStream.Char8.words performance, http://tinyurl.com/yrqe5b |
| 04:09:09 | <dino-> | But the discussion never went further. |
| 04:09:28 | <MarcWeber> | abz Where can I read about the success story? Do you want to join web-devel@haskell.org ? |
| 04:09:37 | <dcoutts> | dmwit: if you can give makeNewGObject a NULL pointer it's a bug and we'd apreciate if you report how to reproduce it |
| 04:09:39 | <abz> | dcoutts: yes, I've seen Happs. |
| 04:09:50 | <kc5tja_> | Awesome. I now have 8x8 and 8x16 console fonts. |
| 04:10:00 | <abz> | http://article.gmane.org/gmane.comp.lang.haskell.cafe/21159 |
| 04:10:03 | <lambdabot> | Title: Gmane -- Mail To News And Back Again |
| 04:10:10 | <kc5tja_> | sorear: However, you ruined it for me; I was going to try and do this myself for practice. But, still, it's done. |
| 04:10:38 | <abz> | I was thinking of posting it to web-devel, but I didn't know that it had enough readers... |
| 04:11:08 | <chessguy> | probably most web-devel readers are on -cafe and/or haskell@ |
| 04:11:13 | <dcoutts> | abz: yeah, -cafe has a bigger readership, web-devel is probably a good place for more detailed discussions |
| 04:11:20 | <dmwit> | dcoutts: Okay, I'll make a minimal test-case and forward it to the dev list. |
| 04:11:28 | <glguy> | Is there a reasonably efficient method of altering a doubly linked list without assignments? |
| 04:11:47 | <sorear> | use finger trees instead? :) |
| 04:11:48 | <dcoutts> | dmwit: cheers |
| 04:11:49 | <glguy> | I'm assuming you'd have to rebuild the whole skeleton |
| 04:12:12 | <glguy> | sorear, that's not a bad suggestion... maybe I'm just looking at this wrong.. |
| 04:12:24 | <dcoutts> | glguy: aye, modifying a circularly linked data structure requires rebuilding the whole thing |
| 04:13:42 | <glguy> | I'm trying to figure out how I might store the equivalent of a doubly-connected edge list in a functional data structure |
| 04:13:59 | <glguy> | and have a more efficient way to update it than copy it |
| 04:14:49 | <dcoutts> | dmwit: btw did you figure out the type error? |
| 04:15:01 | <dcoutts> | dmwit: it's just like the error you get if you (read . show) |
| 04:15:12 | <dmwit> | dcoutts: Yes, sjanssen told me I could use Widget instead of (Widgetclass a) => a. |
| 04:15:36 | <dcoutts> | right otherwise it has no idea which Widgetclass instance you mean |
| 04:15:55 | <dmwit> | I'm still not sure I understand why it should care, though. |
| 04:15:57 | <SamB_XP> | little does it know that it *doesn't really matter*... |
| 04:16:05 | <MarcWeber> | abz: We have 9 subscribers |
| 04:16:11 | <SamB_XP> | dmwit: it doesn't know any better than to care! |
| 04:16:16 | <dmwit> | I mean, widgetGetName is part of the API for WidgetClass... |
| 04:16:36 | <dmwit> | All instances of it have to provide that function! |
| 04:16:47 | <dcoutts> | dmwit: but the result might depend on which instance it is |
| 04:17:07 | <dmwit> | ACTION blinks |
| 04:17:19 | <abz> | marcweber: will join |
| 04:17:22 | <abz> | brb |
| 04:17:40 | <dcoutts> | dmwit: suppose the WidgetClass provided a method to get the type name as a string and widgetGetName used it? then it'd depend on which instance it is |
| 04:17:54 | <MarcWeber> | abz: I'm interested in adding fastcgi support, too. But I'm still struggling with WASH source code... |
| 04:18:06 | <dcoutts> | dmwit: of course in this case it doesn't actually matter, but there's no way from the outside to see that |
| 04:18:14 | <dcoutts> | ACTION -> lunch |
| 04:18:16 | <dmwit> | So there's no dynamic dispatch? |
| 04:18:21 | <dmwit> | Okay, I guess that makes sense. |
| 04:24:26 | <jcreigh> | lol...I was wondering why my chess engine was so slow after a minor refactor. turns out that runghc isn't as fast as ghc -O... :) |
| 04:24:29 | <MarcWeber> | ACTION has to checkfs his disks |
| 04:27:21 | <SamB_XP> | dmwit: well there is dynamic dispatch |
| 04:27:29 | <SamB_XP> | but the Haskell compiler knows nothing of that |
| 04:27:48 | <dmwit> | hm? |
| 04:27:54 | <SamB_XP> | it happens in C land |
| 04:28:05 | <dmwit> | Now I'm confused again. |
| 04:28:35 | <dmwit> | If there's definitely enough information at runtime to do the calculation, and it is easy to see at compile time that there will be enough information at runtime, why should the compile fail? |
| 04:28:45 | <SamB_XP> | you aren't familiar with the object oriented language known as C? |
| 04:29:11 | <dmwit> | Wait, are you saying that there is dynamic dispatch, but only in other languages? |
| 04:29:29 | <SamB_XP> | well, the Haskell compiler doesn't know anything about this, and anyway the typesystem doesn't allow it ;-) |
| 04:30:10 | <dmwit> | So, let me say more specifically that I was asking whether *Haskell* has dynamic dispatch. |
| 04:30:34 | <SamB_XP> | nope |
| 04:30:43 | <dmwit> | ok |
| 04:30:49 | <SamB_XP> | C doesn't have it either, though ;-P |
| 04:31:15 | <dmwit> | Well, it does if it's hand-crafted. ;-) |
| 04:31:23 | <SamB_XP> | yeah ;-) |
| 04:32:32 | <dmwit> | This reminds me of that joke where in the punchline, the guy says, |
| 04:32:46 | <dmwit> | You must be a mathematician, because what you said is completely true, but totally unhelpful. |
| 04:32:50 | <dmwit> | =P |
| 04:33:52 | <SamB_XP> | well, I can't help it if I see a false statement and try and correct it! |
| 04:34:01 | <dmwit> | =D |
| 04:41:51 | <SamB_XP> | at least I try to restrain myself during classes |
| 04:42:18 | <dmwit> | Heh, yeah, the person who corrects the prof always looks sort of silly. |
| 04:42:47 | <dmwit> | Plus, if you hold back, you get to feel superior! "I figured that mistake out ages before the other guy." =) |
| 04:43:21 | <SamB_XP> | I find that my lab group doesn't like it so much ;-) |
| 04:43:53 | <SamB_XP> | (when I refrain from pointing out mistakes) |
| 04:45:10 | <sorear> | There's a certain satisfaction to be had from correctly correcting the prof. |
| 04:45:21 | <SamB_XP> | yeah. |
| 04:45:24 | <SamB_XP> | but it can take time |
| 04:45:35 | <SamB_XP> | so it is best to save it for break and/or after class ;-) |
| 04:46:13 | <dmwit> | It is a delicate art. |
| 04:46:28 | <dmwit> | I find that "clarifying" certain points is an effective form of correction. |
| 04:46:51 | <SamB_XP> | another way to do it is to say that you are confused |
| 04:54:00 | <glguy> | ACTION wonders what blue cheese would smell like when it goes bad |
| 04:56:08 | <glguy> | and furthermore :) who looked at this stuff and said "hmm, that dairy product looks moldy enough, I'll try eating it |
| 04:56:49 | <abz> | MarcWeber: joined |
| 04:57:33 | <dino-> | glguy: Oh come on now, the bleu-class cheeses are delicious. |
| 04:57:55 | <glguy> | dino-, I'm not sitting here eating blue cheese on crackers because I like punishment :) |
| 04:57:56 | <dino-> | I actually did have a Roquefort go bad recently. It smelled foul in a way that was unlike its normal foulness. |
| 04:58:25 | <dino-> | There was mucus. |
| 04:58:55 | <glguy> | maybe you'd just discovered a new cheese? |
| 04:59:26 | <dino-> | heh, that could be |
| 05:01:02 | <sorear> | Muahahaha. |
| 05:02:06 | <allbery_b> | remember that italian cheese with worm in it |
| 05:02:09 | <allbery_b> | worms |
| 05:02:15 | <dino-> | Ok, that's nasty. |
| 05:02:29 | <sorear> | Why isn't : newtype Foo# = Foo# Int# : legal? |
| 05:02:30 | <allbery_b> | you can't eat it until you've left it in a paper bag for a day so the worms come out |
| 05:02:45 | <glguy> | why does the paper bag drive them out? |
| 05:02:48 | <sorear> | it seems like it should define a new type of kind # |
| 05:03:05 | <allbery_b> | ACTION doesn't know and isn't sure he wants to know |
| 05:04:05 | <allbery_b> | as to cheese in general (and I say this as a cheese lover) I have to assume someone got desperate when their milk went bad |
| 05:07:38 | <glguy> | ugh, I hate when minesweeper forces you to guess |
| 05:07:48 | <glguy> | especially at the end of a large level |
| 05:09:35 | <jcreigh> | well, it is NP-complete and all |
| 05:10:08 | <glguy> | sure, but so is sudoku |
| 05:10:33 | <jcreigh> | you have to backtrack to solve sudoku as well. |
| 05:10:37 | <glguy> | and sudoku puzzles that require guessing are considered malformed by many :) |
| 05:10:53 | <glguy> | well... I consider minesweeper in the same sense |
| 05:11:26 | <sorear> | glguy: look at simon tatham's puzzle collection |
| 05:11:47 | <sorear> | glguy: portable, has minesweeper and a couple gazillon others |
| 05:12:05 | <sorear> | and an option to never generate puzzles that require backtracking |
| 05:12:15 | <sorear> | unfortunately it requires a GUI :( |
| 05:12:18 | <glguy> | heh |
| 05:12:35 | <jcreigh> | whoa, author of putty? |
| 05:12:36 | <sorear> | http://chiark.greenend.ac.uk/~sgtatham/ |
| 05:12:41 | <jcreigh> | SimonTatham++ |
| 05:12:43 | <sorear> | jcreigh: yerp |
| 05:13:26 | <sorear> | @flush |
| 05:14:45 | <sorear> | ACTION obsessivly checks #haskell stats |
| 05:15:47 | <Heffalump> | ACTION used to work with him |
| 05:16:12 | <sorear> | cool! |
| 05:16:18 | <sorear> | dons: ping |
| 05:26:52 | <dons> | this is rather nice, http://programming.reddit.com/info/1fh8x/comments |
| 05:26:53 | <lambdabot> | Title: What Haskell teaches us about writing Enterprise-scale software (reddit.com) |
| 05:43:02 | <sorear> | nice. on my system unaligned word writes are twice as slow as aligned (256k buffer). still 25% faster than byte writes though |
| 05:44:07 | <dcoutts> | that makes sense |
| 05:44:23 | <dcoutts> | since for an unaligned access the cpu needs to do two aligned accesses |
| 05:44:50 | <sorear> | OTOH, the breaking is done in hardware (i386). |
| 05:45:18 | <dcoutts> | so an unaligned word read/write primitive would want to use byte reads/writes on hardware that doesn't allow unaligned access and on system that do allow it then use it |
| 05:45:29 | <dcoutts> | sorear: breaking? |
| 05:45:39 | <sorear> | into two accesses |
| 05:45:45 | <dcoutts> | right |
| 05:45:48 | <narain> | if i want to make a data type which combines two things of different types, is it more idiomatic to make it a record type or simply a pair? |
| 05:45:56 | <dcoutts> | hence faster than doing it manually in code |
| 05:46:15 | <dcoutts> | narain: depends how much you need to use it |
| 05:46:32 | <dcoutts> | for a one off I'd use a pair |
| 05:46:32 | <sorear> | also, the p4 manual says that unaligned writes are much slower on the p4 than previous processors, since they eliminated the integer-unit shifter |
| 05:46:52 | <narain> | dcoutts: almost all the time |
| 05:47:09 | <sorear> | so that 33% faster includes the time needed to send the data all the way from the store unit to the mmx unit and back |
| 05:47:17 | <dcoutts> | sorear: still it's probably quicker than using multiple instructions to do byte reads/writes |
| 05:47:25 | <dcoutts> | right |
| 05:47:42 | <sorear> | dcoutts: plus, I'm benchmarking on a p4 :) |
| 05:47:46 | <narain> | dcoutts: i take it you would recommend a record in that case? |
| 05:47:53 | <sorear> | unfortunately we still don't have unaligned access primops |
| 05:47:57 | <dcoutts> | narain: aye |
| 05:48:03 | <dcoutts> | sorear: you could write one for us ;-) |
| 05:48:11 | <narain> | dcoutts: ok, thanks! |
| 05:48:36 | <sorear> | perhaps even in big/little/host endian versions! |
| 05:48:44 | <dcoutts> | go for it |
| 05:48:51 | <sorear> | i386 has an opcode for byteswapping |
| 05:48:54 | <dcoutts> | narain: there's no hard and fast rule, just go for whichever feels nicer |
| 05:49:15 | <dcoutts> | sorear: I'm not suggesting going that low level |
| 05:49:33 | <dcoutts> | unless you want to pattern match that case in the x86 NCG |
| 05:49:56 | <narain> | dcoutts: i usually do, but my gut instinct wasn't telling me anything in this case |
| 05:50:14 | <narain> | so i thought i'd rely on received wisdom and do whatever was more idiomatic |
| 05:50:39 | <sorear> | heh. no, i don't think I'm going to mess with the NCG as my first hacking in ghc, sounds way too easy to introduce subtle codegen bugs |
| 05:51:14 | <sorear> | ACTION is probably over-estimating the fragility of ghc |
| 06:01:09 | <narain> | hmm... when defining a monad, i can't just supply my own definition of join directly, can i? |
| 06:01:17 | <narain> | :t join |
| 06:01:19 | <lambdabot> | forall (m :: * -> *) a. (Monad m) => m (m a) -> m a |
| 06:02:57 | <sorear> | right. |
| 06:03:09 | <sorear> | join is outside the class, stupidly. |
| 06:03:31 | <narain> | that's a little annoying, because i'm thinking of my monad in terms of fmap and join |
| 06:04:43 | <narain> | so i'll define >>= using join but have to tuck join into a where clause |
| 06:05:04 | <sjanssen> | the default definitions would probably be cyclic if join was in the class |
| 06:05:52 | <narain> | but that would be ok because of laziness, right? :) |
| 06:05:59 | <sorear> | sjanssen: Eq |
| 06:06:05 | <sorear> | @src (==) |
| 06:06:06 | <lambdabot> | x == y = not (x /= y) |
| 06:06:09 | <sorear> | @src (/=) |
| 06:06:09 | <lambdabot> | x /= y = not (x == y) |
| 06:06:36 | <sorear> | @src compare |
| 06:06:36 | <lambdabot> | compare x y | x == y = EQ |
| 06:06:37 | <lambdabot> | | x <= y = LT |
| 06:06:37 | <lambdabot> | | otherwise = GT |
| 06:06:39 | <sorear> | @src <= |
| 06:06:39 | <lambdabot> | x <= y = case compare x y of { GT -> False; _other -> True } |
| 06:06:42 | <sjanssen> | yeah, Ord too |
| 06:06:48 | <narain> | i remembered seeing that and convincing myself it was ok |
| 06:06:48 | <sorear> | @src show |
| 06:06:49 | <lambdabot> | show x = shows x "" |
| 06:06:51 | <sjanssen> | I guess it's not a big deal then :) |
| 06:06:51 | <sorear> | @src shows |
| 06:06:52 | <lambdabot> | Source not found. Just what do you think you're doing Dave? |
| 06:06:55 | <sorear> | @src showsPrec |
| 06:06:56 | <lambdabot> | showsPrec _ x s = show x ++ s |
| 06:07:03 | <sorear> | show is cyclic too! |
| 06:07:30 | <sorear> | @src (-) |
| 06:07:30 | <lambdabot> | x - y = x + negate y |
| 06:07:33 | <sorear> | @src negate |
| 06:07:34 | <lambdabot> | negate x = 0 - x |
| 06:07:39 | <sorear> | ACTION stops |
| 06:07:44 | <sjanssen> | bah, those are all over |
| 06:08:06 | <sjanssen> | and they're really not very nice -- there's no way for the compiler to give good warnings |
| 06:08:26 | <sorear> | there is imo |
| 06:08:27 | <sjanssen> | and the runtime behavior is just an infinite loop |
| 06:08:34 | <narain> | ah, #haskell, when you can ask a simple question and be taken on a whirlwind tour of the unique corners of the language |
| 06:08:46 | <sorear> | we can for each class infer a default method dependency graph |
| 06:08:49 | <sjanssen> | yeah, we could add a pragma or something that shows the dependency graph |
| 06:09:07 | <sjanssen> | sorear: but sometimes the cycles are intentional |
| 06:09:11 | <sorear> | then for each instance we create a pruned graph with the overriden methods removed |
| 06:09:29 | <sorear> | then for each instance, we warn if the pruned graph is still cyclic |
| 06:09:35 | <sjanssen> | or could theoretically be cyclic |
| 06:09:49 | <sorear> | sjanssen: emphasis on 'sometimes' |
| 06:10:06 | <sorear> | sjanssen: ndm is the only one here who complains about -Wincomplete-cases |
| 06:10:21 | <sjanssen> | why does ndm complain? |
| 06:10:23 | <sorear> | and he probably wouldn't if his thesis wasn't about improving it |
| 06:10:26 | <sorear> | catch? |
| 06:10:54 | <sjanssen> | oh right, the incomplete case checker is broken |
| 06:11:24 | <sjanssen> | I've seen those bugs before |
| 06:15:58 | <narain> | :t join . fmap |
| 06:16:00 | <lambdabot> | Occurs check: cannot construct the infinite type: f = (->) (f a) |
| 06:16:00 | <lambdabot> | Probable cause: `fmap' is applied to too many arguments |
| 06:16:04 | <dmwit> | ?remember narain ah, #haskell, where you can ask a simple question and be taken on a whirlwind tour of the unique corners of the language |
| 06:16:04 | <lambdabot> | Done. |
| 06:16:07 | <narain> | :t (join .) . fmap |
| 06:16:09 | <lambdabot> | forall a a1 (f :: * -> *). (Monad f, Functor f) => (a1 -> f a) -> f a1 -> f a |
| 06:16:19 | <narain> | :t (>>=) |
| 06:16:21 | <lambdabot> | forall (m :: * -> *) a b. (Monad m) => m a -> (a -> m b) -> m b |
| 06:16:39 | <narain> | hmm, almost |
| 06:16:58 | <dmwit> | ?unpl (join .) . fmap |
| 06:16:58 | <lambdabot> | (\ d i -> (fmap d i) >>= \ e -> e) |
| 06:17:07 | <Cheery> | today I get radical and remove all the unnecessary declaration choices and ugly possible values from cascading style sheets |
| 06:17:08 | <dmwit> | ... =( |
| 06:17:09 | <narain> | ?pl \m f -> join (fmap f m) |
| 06:17:09 | <lambdabot> | (join .) . flip fmap |
| 06:17:34 | <Cheery> | die shitty w3c! |
| 06:17:50 | <dmwit> | Yay! |
| 06:19:28 | <narain> | Cheery: what's this about, if i may ask? |
| 06:19:50 | <Cheery> | yesterday's thing |
| 06:20:06 | <narain> | Cheery: i wasn't there |
| 06:21:57 | <Cheery> | I'm writing a css generator, yesterday decided to remove all which shouldn't be there |
| 06:23:21 | <narain> | css... *generator*? i wouldn't think generating output in a language would send you into ugly corners of the spec |
| 06:23:35 | <narain> | that usually happens when trying to parse, not generate... ? |
| 06:23:54 | <Cheery> | so, this looks even more uglier when parsed? |
| 06:24:11 | <narain> | i wouldn't know |
| 06:24:25 | <narain> | i was speaking about languages in general |
| 06:25:21 | <narain> | i never found css ugly myself though |
| 06:25:39 | <narain> | offtopic: what's the syntax for single-line comments in haskell? |
| 06:25:51 | <Cheery> | well, css problems are that there are some things which are obsolete / horrible in practise / available by other means |
| 06:26:07 | <Cheery> | this is your code -- this is your comment |
| 06:26:14 | <narain> | oh right thanks |
| 06:26:23 | <Cheery> | multiline is {- COMMENT -} |
| 06:27:09 | <Cheery> | another css problem is that each property accepts totally different set of values, while they accept same sort of values |
| 06:28:01 | <narain> | hmm |
| 06:28:33 | <Cheery> | good example is font-size |
| 06:29:03 | <narain> | ..., x-small, small, um medium?, large, x-large, ... |
| 06:29:10 | <Cheery> | there's xx-small, x-small, small, medium, large, x-large xx-large smaller, and in the end you can just give a length or percents |
| 06:29:29 | <narain> | are there smaller and larger also? |
| 06:29:36 | <narain> | i forget |
| 06:30:48 | <Cheery> | font-weight is ridiculous in kind |
| 06:31:05 | <Cheery> | normal bold bolder light 100 200 300 400 500 600 700 800 900 |
| 06:31:32 | <narain> | well, they tried to keep to what the standard was in typography |
| 06:31:57 | <narain> | it's not like they could make up their own measures of typeface weight |
| 06:32:16 | <narain> | (i assume this is what they did, at least) |
| 06:33:04 | <narain> | to what purpose are you generating css? are you trying to cover all possible cases? |
| 06:33:05 | <Cheery> | I think they could have, anyways, the cascading style sheet is used mainly with web sites, not for typography |
| 06:33:48 | <narain> | well (1) it can be used in print (2) it's supposed to be used also by designers |
| 06:33:57 | <Cheery> | nah, I tried that, I think I'm fed up with it, now things are going to fly and bend |
| 06:34:59 | <Cheery> | what's the role of visibility and display -properties? |
| 06:35:03 | <narain> | what's your css generator for, anyway? if you don't need to cover all cases you could just pick the bits which are easy to generate and/or of interest to you and generate those |
| 06:35:20 | <narain> | visibility hides an element but it still takes up space |
| 06:35:25 | <narain> | display eliminates it completely |
| 06:35:38 | <sorear> | @users |
| 06:35:38 | <lambdabot> | Maximum users seen in #haskell: 336, currently: 273 (81.3%), active: 20 (7.3%) |
| 06:35:51 | <Cheery> | my target is just web sites |
| 06:35:58 | <narain> | display also changes other things |
| 06:36:09 | <Cheery> | narain: so, what's the point of an element if it is hidden? |
| 06:36:09 | <narain> | like you could make a list display inline instead of as a block |
| 06:36:26 | <Cheery> | ie. why leave the blank space in those cases? |
| 06:37:01 | <narain> | hmm i feel like i used it once but i can't recall why |
| 06:37:36 | <narain> | doesn't the spec suggest any uses? |
| 06:38:09 | <hyrax42> | aren't there non-standard selectors for hover &c.? |
| 06:38:19 | <allbery_b> | ACTION curses sorear --- he's wasted the last 45 minutes playing those puzzles instead of going to bed :) |
| 06:38:23 | <hyrax42> | they can change the visibility |
| 06:38:34 | <sorear> | muahahaha! |
| 06:38:39 | <hyrax42> | also I believe I've seen hidden used for browser warnings |
| 06:38:49 | <narain> | hyrax42: hover is a standard selector |
| 06:38:52 | <hyrax42> | serve the same HTML, different CSS based on user agen |
| 06:39:08 | <hyrax42> | so a warning can be hidden and shown for NS 4.7 or something |
| 06:39:32 | <sorear> | a lot of hidden's raison d'etre is that it can be set from javascript |
| 06:39:54 | <hyrax42> | and you get all these "pure CSS dropdowns" that I assume use hover and hidden for magic |
| 06:40:34 | <narain> | i think the question is why would you ever want to use hidden instead of display:none |
| 06:40:57 | <narain> | if you have something hidden, it still takes up space if it's in the normal flow |
| 06:41:18 | <hyrax42> | oh I confused them |
| 06:41:25 | <narain> | which you would often not want because it would look like empty space sitting there for no reason |
| 06:41:26 | <hyrax42> | narain: for really annoying link-stles |
| 06:41:29 | <hyrax42> | tyles |
| 06:41:32 | <hyrax42> | styles |
| 06:42:25 | <narain> | yeah, the only reason i can think of is if you have something in the normal flow which is revealed interactively and you don't want the following text to jump around |
| 06:42:43 | <Cheery> | I've already removed all related to background images because they indicate immature style and are the easiest thing to make web pages hyper unreadable |
| 06:42:45 | <narain> | like awful link styles, as you said |
| 06:42:57 | <hyrax42> | speakng of jumping around, I hate it when people change the typeface on hover for links |
| 06:43:07 | <narain> | background images ... indicate immature style? :) |
| 06:43:12 | <hyrax42> | I'd actually rather the text disappear than have it get reflowed |
| 06:43:19 | <narain> | often... but not necessarily always |
| 06:43:27 | <Cheery> | hmm, I wonder whether I could find some sites with background images |
| 06:43:36 | <Cheery> | I think they all suck |
| 06:43:42 | <hyrax42> | myspace.com |
| 06:43:55 | <narain> | http://www.squidfingers.com/patterns/ |
| 06:43:56 | <lambdabot> | Title: Squidfingers / Patterns |
| 06:44:17 | <narain> | stylish tileable pixel backgrounds |
| 06:44:20 | <Cheery> | http://users.nac.net/falken/annoying/backgrounds.html |
| 06:44:21 | <lambdabot> | Title: Background Images |
| 06:44:24 | <narain> | never used them personally |
| 06:44:38 | <narain> | Cheery: hahaha |
| 06:45:18 | <narain> | Cheery: good use of background images: http://www.cameronmoll.com/ |
| 06:45:19 | <lambdabot> | Title: Authentic Boredom ~ Delivered weekly by Cameron Moll |
| 06:48:04 | <narain> | what's -fallow-undecidable-instances all about? |
| 06:48:13 | <Cheery> | narain: those squidfinger patterns would be mostly annoying, I think |
| 06:48:52 | <sorear> | narain: it allows instances even when the compiler can't decide them |
| 06:48:53 | <narain> | well i certainly don't advocate putting them everywhere |
| 06:49:28 | <sorear> | narain: so that option gives GHC licence to crash when a type error occurs |
| 06:49:41 | <narain> | hmm, i don't like the sound of that |
| 06:49:46 | <sorear> | narain: but if you don't mess up, it gives you much more oleg-power |
| 06:49:53 | <sorear> | crash means infinite loop here |
| 06:50:32 | <sorear> | you can set it on a module-by-module basis, which lets oleg write his modules without making your code unsafe |
| 06:50:33 | <narain> | i'm having trouble with a mptc and ghc's telling me i might want to use that flag |
| 06:50:45 | <sorear> | coverage condition? |
| 06:50:45 | <Cheery> | lol |
| 06:50:47 | <Cheery> | http://users.nac.net/falken/annoying/main.html |
| 06:50:48 | <lambdabot> | Title: How to make an Annoying Web Page |
| 06:51:15 | <sorear> | class Foo a b | a -> b instance Foo a b => Foo (X a) (X b) -- something like this, narain? |
| 06:51:47 | <sorear> | the compiler doesn't normally look at fundeps on the left of => to decide if the fundeps on the right are satisfied |
| 06:51:56 | <narain> | no, i'm not using fundeps |
| 06:51:59 | <sorear> | so that example, while perfectly safe, needs -fallow-undecidable-instances |
| 06:52:09 | <narain> | class Foo a b where bar :: a -> a -> a |
| 06:52:23 | <sorear> | er, why don't you use b? |
| 06:52:31 | <sorear> | in the type |
| 06:52:40 | <narain> | well, there are other things in the class |
| 06:52:46 | <narain> | one of them happens to not need b |
| 06:53:00 | <sorear> | what kind of class is this? |
| 06:53:05 | <sorear> | the affine space? |
| 06:53:21 | <sorear> | the usual fix for this is fundeps |
| 06:53:22 | <narain> | a class of linear transformations, yes |
| 06:53:40 | <narain> | hmm, ok i'll try that |
| 06:53:47 | <sorear> | class LinTrans pointy veccy | pointy -> veccy, veccy -> pointy |
| 06:54:06 | <narain> | no that's not exactly what it is but i'll try it out anyway |
| 06:54:12 | <narain> | fundeps i mean |
| 06:55:44 | <narain> | it worked! |
| 06:55:45 | <Cin> | hello. I'm trying to understand the following code: parseNumber :: Parser LispVal (new line) parseNumber = liftM (Number . read) $ many1 digit |
| 06:55:49 | <Cin> | is this defining a monad? |
| 06:55:59 | <sorear> | no |
| 06:56:03 | <sorear> | that's a value |
| 06:56:08 | <sorear> | all monads are types |
| 06:56:21 | <Cin> | oh I see |
| 06:56:24 | <sorear> | monad operations are used to construct the value however |
| 06:56:35 | <Cin> | so its -type- is a monad, but it is a value |
| 06:56:45 | <sorear> | like foo = 2 + 2 doesn't define a kind of number even though it uses + |
| 06:56:46 | <sorear> | yes |
| 06:56:47 | <Cin> | okay |
| 06:57:00 | <narain> | sorear: i'm not sure i wanted the fundep though |
| 06:57:29 | <narain> | i can think of theoretical cases where a may not imply b |
| 06:57:32 | <sorear> | narain: what are the types a and b? |
| 06:58:19 | <narain> | a is a spatial transformation type and b is a vector type |
| 06:58:43 | <sorear> | like rotate, reflect, etc? |
| 06:59:00 | <narain> | class Transform t v where |
| 06:59:03 | <narain> | ident :: t |
| 06:59:10 | <narain> | apply :: t -> v -> v |
| 06:59:17 | <narain> | compose :: t -> t -> t |
| 06:59:21 | <narain> | inv :: t -> t |
| 06:59:23 | <narain> | you get the idea |
| 06:59:31 | <sorear> | you can think of t as being analagous to (v -> v) |
| 06:59:33 | <Cin> | so, (many1 digit) collects digits into a string? |
| 06:59:37 | <narain> | sorear, yes |
| 06:59:48 | <sorear> | so any given monotype t only works on one kind of v |
| 07:00:01 | <sorear> | just as any monotyped function only works on one type of argument |
| 07:00:06 | <narain> | well i could have different implementations of vectors... no? |
| 07:00:15 | <narain> | (in theory) |
| 07:00:34 | <sorear> | In that case, the usual approach is to split the class. |
| 07:01:13 | <sorear> | class Transform t where |
| 07:01:13 | <sorear> | ident :: t |
| 07:01:13 | <sorear> | compose :: t -> t -> t |
| 07:01:13 | <sorear> | inv :: t -> t |
| 07:01:16 | <sorear> | class Transform t => Transform2 t v where |
| 07:01:19 | <sorear> | apply :: t -> v -> v |
| 07:01:40 | <sorear> | but it will be harder to use that way - more ambiguity errors |
| 07:01:54 | <narain> | hmm, i see, that seems to make sense |
| 07:03:13 | <narain> | i probably could have avoided all this trouble by not using MPTCs in the first place, but i felt like making all my types instances of type classes :) |
| 07:03:28 | <sorear> | hehe. |
| 07:04:31 | <narain> | well, thanks a lot sorear, you really helped clarify all this |
| 07:04:36 | <sorear> | yw |
| 07:05:41 | <narain> | at this rate i'll keep playing with the type system and never get anything implemented :( |
| 07:06:11 | <narain> | *at the rate i'm going |
| 07:13:46 | <Cin> | parseNumber = liftM (Number . read) $ many1 digit -- so this grabs all the digits from a string to a string, then converts that string to an number, and that is then converted to a Number...? |
| 07:14:02 | <sorear> | precisely. |
| 07:14:33 | <Cin> | bare with me |
| 07:15:23 | <Cheery> | what is the point of separated-borders? |
| 07:15:24 | <Cin> | so far, that's defining operations. but we want parseNumber to return an actual Number so liftM is used? : |
| 07:15:49 | <Cheery> | or wait, border-collaps |
| 07:16:24 | <sorear> | yes. |
| 07:16:37 | <sjanssen> | Cin: liftM "lifts" a normal function into a monadic one |
| 07:16:41 | <sjanssen> | @hoogle liftM |
| 07:16:42 | <lambdabot> | Monad.liftM :: Monad a => (b -> c) -> a b -> a c |
| 07:16:42 | <lambdabot> | Control.Monad.liftM :: Monad m => (a1 -> r) -> m a1 -> m r |
| 07:16:42 | <lambdabot> | Monad.liftM2 :: Monad a => (b -> c -> d) -> a b -> a c -> a d |
| 07:17:03 | <Cin> | yes, I found that confusig |
| 07:18:29 | <sjanssen> | "liftM :: Monad m => (a -> b) -> (m a -> m b)" takes a function from a's to b's, and creates a new function that takes a's wrapped in m and yields b's wrapped in m |
| 07:20:09 | <Cin> | sjanssen, takes a function from a's what? |
| 07:20:46 | <sjanssen> | I mean that it is a function that takes an a and yields a b |
| 07:20:52 | <sorear> | ooh, nice, first reply to my perf post was direct SPJ |
| 07:21:22 | <sjanssen> | I mean that it is a function that takes an 'a' and yields a 'b' (single quotes around type variables to eliminate ambiguity) |
| 07:23:29 | <sjanssen> | let's look at your example. "Number . read" has type "String -> Expr" |
| 07:23:59 | <sjanssen> | and "many1 digit" has type "Parser String" |
| 07:24:17 | <Cin> | okay |
| 07:24:34 | <sjanssen> | so in the example code, liftM has the type "(String -> Expr) -> (Parser String -> Parser Expr)" |
| 07:25:02 | <Cin> | I see |
| 07:25:47 | <sjanssen> | in English, you could say: given a transformation from Strings to Exprs, we have a transformation from Parsers of Strings to Parsers of Exprs |
| 07:26:23 | <Bourbaki> | does anyone know this book? "The Haskell Road to Logic, Maths and Programming" and if so is it any good? |
| 07:27:52 | <Cin> | sjanssen: hm, right. I think I'll read more of A Gentle Introduction to Haskell |
| 07:28:04 | <osfameron> | "Gentle"! |
| 07:28:43 | <Cin> | hehe |
| 07:29:27 | <osfameron> | all the online haskell tutorials terrify me. I am just about able to read the School of Expression without turning into a gibbering Cthulhu cultist however |
| 07:29:38 | <Cin> | haha |
| 07:29:53 | <sjanssen> | osfameron: I've heard that YAHT isn't so bad |
| 07:30:14 | <osfameron> | the writing and explanations of YAHT are poor, but the exercises are well chosen. |
| 07:30:30 | <sorear> | night all. |
| 07:30:39 | <sjanssen> | goodnight sorear |
| 07:30:44 | <osfameron> | nighty |
| 07:31:27 | <Cin> | I'm doing Write Yourself a Scheme in 4 Hours, but I find myself wondering about most of it. |
| 07:31:56 | <sjanssen> | Write Yourself a Scheme isn't exactly introductory material |
| 07:32:09 | <osfameron> | it seems to help if you already know both scheme and haskell |
| 07:32:39 | <Cin> | oh, okay |
| 07:32:46 | <Cin> | A Gentle Introduction is the way to go? |
| 07:33:05 | <sjanssen> | Cin: have you done some sort of functional programming before? |
| 07:33:35 | <Cin> | sjanssen, not really. I have been doing Common Lisp for a few months, but it's been mainly in an imperative style |
| 07:34:13 | <osfameron> | the "Gentle" of the Gentle introduction is a little bit like the sadistic mocking "We're not going to hurt you" of Mafia hitmen |
| 07:35:25 | <Cin> | it seemed quite acceptable of what I'd read of it. but I got a little bored because I wasn't taking anything in. if I sat writing up my own examples with a compiler, I think that would work quite well as a way of learning |
| 07:35:43 | <sjanssen> | Cin: try the gentle intro, just remember that there are other resources that are actually gentle, if you get stuck |
| 07:36:18 | <sjanssen> | YAHT and the wiki book are what most people recommend |
| 07:36:37 | <Cin> | sjanssen, okay, thanks. I'll have a look at those too |
| 07:38:46 | <osfameron> | actually... the gentle introduction looks less scary now after having read some other stuff - looks like a good reference |
| 07:41:12 | <Cheery> | ok, now I removed the most useless css -stuff |
| 07:41:27 | <Cheery> | now there's about 35 or 40 |
| 07:42:28 | <_roconnor> | @docs System |
| 07:42:29 | <lambdabot> | System not available |
| 07:42:34 | <_roconnor> | @hoogle system |
| 07:42:35 | <lambdabot> | System.system :: String -> IO ExitCode |
| 07:42:35 | <lambdabot> | System :: module |
| 07:42:35 | <lambdabot> | System.Win32.Info.SystemColor :: type SystemColor |
| 07:42:46 | <_roconnor> | @docs System.system |
| 07:42:47 | <lambdabot> | System.system not available |
| 07:48:48 | <_roconnor> | @type when |
| 07:48:50 | <lambdabot> | forall (m :: * -> *). (Monad m) => Bool -> m () -> m () |
| 07:49:18 | <_roconnor> | @type (\x -> x >> return ()) |
| 07:49:20 | <lambdabot> | forall (m :: * -> *) a. (Monad m) => m a -> m () |
| 07:53:57 | <andersca> | hello |
| 07:54:30 | <bdash> | andersca: hi! |
| 07:54:48 | <andersca> | bdash: hah! didn't expect to see you here! :) |
| 07:59:21 | <joelr1> | good morning! |
| 08:00:25 | <dons> | hey joelr1 , you might be interested in today's good news story, http://article.gmane.org/gmane.comp.lang.haskell.cafe/21159 |
| 08:00:26 | <lambdabot> | Title: Gmane -- Mail To News And Back Again |
| 08:01:09 | <joelr1> | dons: the one you posted? i have reddit in my google reader ;) |
| 08:01:21 | <dons> | hah |
| 08:01:32 | <joelr1> | dons: thanks for the tip, though. it was a welcome email when i received it on haskell this morning |
| 08:02:02 | <joelr1> | dons: i plan to do something similar but using happs. no particular reason apart from alex's goals towards amazon ec2 and s3 |
| 08:02:29 | <joelr1> | dons: plan = do it before the end of the month |
| 08:03:55 | <joelr1> | dons: mine is supposed to be a money-making haskell web app of my own |
| 08:06:41 | <dons> | wonderful. i wish you success and may your lambdas always beta reduce. |
| 08:07:06 | <joelr1> | dons: i'm translating from a custom trading systems programming language to c# and plan to sell translations on the web |
| 08:08:05 | <olliej> | dons: my thesis is finally had its final submission (about 10 months after the original submission -- grumble) |
| 08:08:26 | <joelr1> | dons: i wrote a working translator in ocaml in about a month, learning ocaml included. then had conceptual trouble deploying it on the web. i thought about ruby + ocaml, etc. then remembered the warn and fuzzy haskell feelings and started a rewrite. |
| 08:09:05 | <dons> | congratulations olliej ! |
| 08:09:25 | <olliej> | dons: soon the pain than is my writing will be recorded forever |
| 08:09:40 | <dons> | ACTION disappears to finish two icfp papers in two days... WRITE WRITE WRITE! |
| 08:09:48 | <Syzygy-> | Go dons, go! |
| 08:10:01 | <Syzygy-> | olliej: http://www.phdcomics.com/comics/archive.php?comicid=844 |
| 08:10:02 | <lambdabot> | Title: Piled Higher and Deeper |
| 08:12:11 | <olliej> | Syzygy-: mines's a real degree dammit -- it's only being mailed to me because i'm no longer in the country :D |
| 08:12:21 | <olliej> | ... it better be a real degree anyway |
| 08:12:24 | <olliej> | :D |
| 08:12:30 | <alec> | :) |
| 08:12:38 | <Syzygy-> | olliej: You're not familiar with phdcomics, I hear.... |
| 08:12:39 | <robreim> | I'm trying to build/install the network library with hugs using cabal. I did 'runhugs -98 ./Setup.hs configure', but 'runhugs -98 ./Setup.hs build' returns the error message 'ERROR "dist/build/Network/BSD.hs" - Can't find imported module "GHC.IOBase"'. I'm guessing there's some sort of hsc2hs or cpphs trick I need? |
| 08:13:01 | <Syzygy-> | ACTION linked it for the empty feeling of "Well, that was the thesis .. now what...", not for the mailing quip. |
| 08:13:52 | <olliej> | Syzygy-: hehe |
| 08:14:11 | <olliej> | Syzygy-: i slept for a whlie |
| 08:14:16 | <olliej> | *while even |
| 08:14:28 | <Syzygy-> | In the storyline, the guy handing in there really is a PhD at a regular university; not a customer at some scheme... |
| 08:14:40 | <Syzygy-> | Go read the comic though - if you're doing academia it's well worth it. |
| 08:14:50 | <olliej> | Syzygy-: i've escaped academia |
| 08:14:55 | <alec> | robreim: did you get it from darcs, hackage, or somewhere else? |
| 08:14:59 | <olliej> | i dno't feel like inflicting a phd upon myself |
| 08:15:02 | <Syzygy-> | Ok, then for the nostalgia of it. ;) |
| 08:15:04 | <robreim> | alec: darcs |
| 08:15:06 | <olliej> | heh |
| 08:15:25 | <alec> | robreim: sounds like either the preprocessor isn't picking up hugs or the darcs version is broken for hugs because of a ghc dependency |
| 08:16:32 | <alec> | robreim: did you try passing the '--hugs' flag to the configure step? |
| 08:16:42 | <robreim> | Nope, I'll try that. |
| 08:17:24 | <robreim> | No difference. |
| 08:18:16 | <alec> | robreim: that's the extent of my hugs experience, I won't be of much further help |
| 08:18:44 | <robreim> | thanks for your help |
| 08:21:23 | <robreim> | I do have cpphs-hugs and hsc2hs-hugs installed on my computer (running debian with the 'experimental' distribution of hugs) but using the --with-(cpphs|hsc2hs) flags set to those different commands doesn't seem to make any difference. |
| 08:23:06 | <sjanssen> | @remember dons i wish you success and may your lambdas always beta reduce. |
| 08:23:06 | <lambdabot> | Done. |
| 08:58:35 | <joelr1> | sjanssen: yes, that's a very good one! :D |
| 09:01:36 | <tsp> | > let swedishSwearWords = [("anteckningar","shit"),("tillbehör","penis")] in first (first swedishSwearWords)) |
| 09:01:36 | <lambdabot> | Parse error |
| 09:01:54 | <tsp> | damn, my new swedish swear words gave a parse error |
| 09:02:13 | <tsp> | > "anteckningar tillbehör" |
| 09:02:14 | <lambdabot> | "anteckningar tillbeh\246r" |
| 09:02:23 | <opqdonut> | > first (1,2) |
| 09:02:24 | <lambdabot> | add an instance declaration for (Arrow (,)) |
| 09:02:24 | <lambdabot> | In the expression: first (1... |
| 09:02:32 | <opqdonut> | see, firsting doesn't work for tuples |
| 09:02:44 | <opqdonut> | :t first |
| 09:02:46 | <lambdabot> | forall (a :: * -> * -> *) b c d. (Arrow a) => a b c -> a (b, d) (c, d) |
| 09:02:52 | <tsp> | thought I was giving it a tuple? |
| 09:02:53 | <opqdonut> | and besides i think that is not the first you want? |
| 09:02:55 | <joelr1> | @where FilePath |
| 09:02:55 | <lambdabot> | http://www-users.cs.york.ac.uk/~ndm/projects/libraries.php#filepath |
| 09:02:56 | <lypanov> | huh... |
| 09:03:01 | <lypanov> | isn't it just because you did )) ? |
| 09:03:19 | <tsp> | maybe it doesn't like the swedish swear words |
| 09:03:33 | <lypanov> | your ()s don't match up |
| 09:03:35 | <tsp> | oh |
| 09:03:38 | <lypanov> | ACTION is a haskell beginner so ignore him if need be |
| 09:03:40 | <Syzygy-> | tsp: You call that a swearword? |
| 09:03:46 | <lypanov> | where beginner == hasn't used it :P |
| 09:03:54 | <tsp> | I'm assuming they're swear words |
| 09:04:05 | <opqdonut> | > let swedishSwearWords = [("anteckningar","shit"),("tillbehör","penis")] in head $ fst swedishSwearWords |
| 09:04:06 | <lambdabot> | lexical error in string/character literal (UTF-8 decoding error) |
| 09:04:11 | <lypanov> | lol |
| 09:04:12 | <tsp> | anteckningar and tillbehör are strange enough |
| 09:04:16 | <opqdonut> | that is probably what you meant |
| 09:04:21 | <Syzygy-> | ... |
| 09:04:28 | <joelr1> | is there a haskell colorizer for various languages? |
| 09:04:33 | <Syzygy-> | anteckningar = notes. tillbehör = accessories. |
| 09:04:35 | <opqdonut> | > let swedishSwearWords = [(1,2),(3,4)] in head $ fst swedishSwearWords |
| 09:04:36 | <tsp> | ah, fst |
| 09:04:36 | <lambdabot> | Couldn't match expected type `(a, b)' |
| 09:04:45 | <tsp> | Syzygy-: lol |
| 09:04:47 | <opqdonut> | > let swedishSwearWords = [(1,2),(3,4)] in fst $ head swedishSwearWords |
| 09:04:48 | <lambdabot> | 1 |
| 09:04:49 | <opqdonut> | gah i suck |
| 09:04:57 | <Syzygy-> | opqdonut: You want to map somewhere.... |
| 09:05:05 | <tsp> | ah, head, that'll work |
| 09:05:19 | <opqdonut> | Syzygy-: i was just trying to guess what tsp wanted to do |
| 09:05:30 | <Syzygy-> | Oh, yeah, I kinda missed the start.... |
| 09:05:47 | <opqdonut> | nvm |
| 09:06:03 | <tsp> | I tried to learn swedish, but gave up after hej and sverige :) |
| 09:06:34 | <lypanov> | i gave up after jag |
| 09:06:42 | <tsp> | lypanov: heh |
| 09:06:42 | <lypanov> | "Jag ... (tried to etc)" |
| 09:07:13 | <tsp> | I happen to like the verbs - similar to esperanto, jag har/du har/etc - never got to adjectives and adverbs though |
| 09:07:39 | <Syzygy-> | lypanov: "Jag försökte" :) |
| 09:07:52 | <tsp> | but I accidentally used the wrong windows iso on my test vm, too lazy to reinstall - so its all in swedish |
| 09:08:00 | <lypanov> | Syzygy-: gave up i guess? |
| 09:08:15 | <Syzygy-> | Hehe |
| 09:09:08 | <tsp> | oh, that's why I couldn't telnet to the box - I kept typing administrator, it wants administratör |
| 09:09:14 | <lypanov> | lol |
| 09:09:16 | <Syzygy-> | Hehe |
| 09:10:37 | <tsp> | weird, even among windows oses the o isn't showing up right - my screen reader says quote, when its an o when I type it in here |
| 09:10:52 | <Syzygy-> | oO |
| 09:11:01 | <Syzygy-> | You use a screen reader? |
| 09:11:18 | <opqdonut> | tsp: probably charset issue |
| 09:11:35 | <tsp> | yup, screen reader |
| 09:12:01 | <tsp> | luckally enough I have a synth that speaks swedish, even if jag talär svenska inte :) |
| 09:12:07 | <Syzygy-> | Hehe |
| 09:12:17 | <lypanov> | inte -> ? |
| 09:12:20 | <tsp> | not |
| 09:12:24 | <lypanov> | ACTION nods |
| 09:12:25 | <Syzygy-> | You blind^Wvisually impaired^W^Wwhatever the right term should be? |
| 09:12:29 | <lypanov> | wanted to confirm :) |
| 09:12:29 | <tsp> | Syzygy-: yup |
| 09:12:34 | <Syzygy-> | lypanov: inte == not |
| 09:12:39 | <Syzygy-> | tsp: Right-o. |
| 09:12:53 | <Syzygy-> | First time I encounter someone I know works that way. |
| 09:13:02 | <tsp> | I have the choice of learning english, spanish, swedish, finnish, german, etc - but I finally found a synth that speaks esperanto |
| 09:13:08 | <Syzygy-> | Wow! |
| 09:13:11 | <Syzygy-> | Cooooooool. |
| 09:13:18 | <tsp> | I even ahve one that speaks chinese, but that is weird |
| 09:13:24 | <lypanov> | i suppose i really should take up esperanto again lol |
| 09:13:36 | <Syzygy-> | How do the readers handle mixed languages? |
| 09:13:37 | <tsp> | I need a terminal that supports utf-8 :) |
| 09:13:51 | <tsp> | ah, haskell always breaks on this |
| 09:13:52 | <Syzygy-> | Say, if I should start talking swedish or german in here, what'd happen at your end? |
| 09:13:55 | <Syzygy-> | Hehe |
| 09:14:07 | <tsp> | > 'Ã¥' |
| 09:14:07 | <lambdabot> | Improperly terminated character constant |
| 09:14:20 | <tsp> | Syzygy-: it'd try to read it currently using an english synth |
| 09:14:29 | <Syzygy-> | For that matter, what does haskell code sound like? |
| 09:14:41 | <tsp> | why does it break? I hit the apostrophe, alt-0229, and another ' - it shows up on my end just fine |
| 09:14:56 | <lypanov> | shows up here fine also |
| 09:14:57 | <tsp> | code, and most everything else is just a really fast stream of words |
| 09:14:59 | <Syzygy-> | tsp: So it'd be a combination of horribly mispronounced fragments and reading letters? |
| 09:15:09 | <tsp> | http://silenceisdefeat.org/~tspivey/eloquence.flac |
| 09:15:14 | <tsp> | Syzygy-: yup |
| 09:15:29 | <tsp> | reading anteckningar in english and swedish sounds totally different |
| 09:15:36 | <Syzygy-> | Oh bugger. The computer I got here doesn't do sound. :( |
| 09:15:48 | <tsp> | wow |
| 09:15:50 | <tsp> | a computer without sound |
| 09:16:10 | <Syzygy-> | Well, it has a soudncard, but I don't seem to have access rights to it, and I don' |
| 09:16:25 | <Syzygy-> | I don't want to bug the sysadmin about it, since I'm only here for two weeks. |
| 09:16:30 | <tsp> | running windows? you should be able to access it - volume control :) |
| 09:16:36 | <Syzygy-> | Nonono, linux |
| 09:16:39 | <tsp> | ah |
| 09:16:41 | <Syzygy-> | And I'm very much not root. |
| 09:16:45 | <Syzygy-> | So I can't poke at it. |
| 09:16:48 | <tsp> | damn |
| 09:17:00 | <Syzygy-> | I'm getting through by bringing my mp3player to work :) |
| 09:17:16 | <tsp> | your at work, on a linux box? neat |
| 09:17:22 | <Syzygy-> | Suer |
| 09:17:34 | <Syzygy-> | The only ppl here who don't use linux are the MacOS 9 diehards.... |
| 09:17:39 | <tsp> | ouch OS 9 |
| 09:17:47 | <tsp> | would that even run haskell? |
| 09:17:54 | <Syzygy-> | Dunno, wouldn't bet on it. |
| 09:17:57 | <lypanov> | os9?? lol |
| 09:18:05 | <Syzygy-> | Though haskell isn't very much around here either. |
| 09:18:15 | <dmead> | lol |
| 09:18:20 | <Syzygy-> | Which is funky, since I'm at on of the departments housing Per Martin-Löf. |
| 09:19:05 | <tsp> | lol, my screen reader said martin-loaf due to the ö - how do you type those anyways? I have to hit ^a^v"o |
| 09:19:15 | <Syzygy-> | http://en.wikipedia.org/wiki/Per_Martin-L%C3%B6f |
| 09:19:16 | <lambdabot> | Title: Per Martin-Löf - Wikipedia, the free encyclopedia |
| 09:19:28 | <Syzygy-> | tsp: I'm in sweden, using a swedish keyboard. |
| 09:19:35 | <Syzygy-> | I haev a key dedicated to each of å ä ö |
| 09:19:45 | <tsp> | ah |
| 09:20:19 | <tsp> | I hate this windows terminal emulator - it works great for normal text, but it only supports unicode in japanese |
| 09:20:44 | <tsp> | utf-8 tera term pro |
| 09:21:36 | <joelr1> | folks, can you please try building the deriver library? |
| 09:21:38 | <joelr1> | @where derive |
| 09:21:38 | <lambdabot> | http://www.cs.york.ac.uk/fp/darcs/derive |
| 09:21:47 | <joelr1> | it doesn't build on my end with ghc 6.6 |
| 09:21:51 | <joelr1> | and i wonder why |
| 09:21:58 | <dmead> | 6.6 isn't on portage yet |
| 09:22:09 | <Syzygy-> | It's in the layover though. |
| 09:23:32 | <joelr1> | ugh... portage? layover? what brand of ... ah, gentoo! |
| 09:23:32 | <eivuokko> | Yo. |
| 09:23:54 | <joelr1> | eivuokko: long time |
| 09:24:35 | <eivuokko> | Yeah :) |
| 09:24:57 | <joelr1> | ACTION high-fives eivuokko |
| 09:26:01 | <eivuokko> | Anyone got experiences if newStablePtr/freeStablePtr are expensive to call (calling atleast hundreds of times per second) or for gc? (with ghc rts) |
| 09:26:05 | <fuzan> | is there some function for lifting multiple items? ie. i want to compare three IO Integers, a b c |
| 09:26:13 | <eivuokko> | ACTION returns high-five |
| 09:26:31 | <joelr1> | fuzan: lifM3? |
| 09:26:39 | <Syzygy-> | liftM3 you mean |
| 09:26:42 | <fuzan> | :t liftM3 |
| 09:26:45 | <lambdabot> | forall a1 a2 a3 r (m :: * -> *). (Monad m) => (a1 -> a2 -> a3 -> r) -> m a1 -> m a2 -> m a3 -> m r |
| 09:26:45 | <joelr1> | yes |
| 09:27:10 | <fuzan> | oh snap, that'll do it. thanks :) |
| 09:27:18 | <joelr1> | i used to wonder what liftMx do until i looked at the implementation |
| 09:27:33 | <Syzygy-> | ?src liftM3 |
| 09:27:33 | <lambdabot> | liftM3 f m1 m2 m3 = do { x1 <- m1; x2 <- m2; x3 <- m3; return (f x1 x2 x3) } |
| 09:27:38 | <Syzygy-> | Hehe |
| 09:27:46 | <joelr1> | Syzygy-: that's neat! i didn't know you could do that |
| 09:29:41 | <dmead> | >liftM2 (,) [1..5] [6..10] |
| 09:29:41 | <joelr1> | @paste |
| 09:29:42 | <lambdabot> | Haskell pastebin: http://hpaste.org/new |
| 09:29:49 | <dmead> | > liftM2 (,) [1..5] [6..10] |
| 09:29:51 | <lambdabot> | [(1,6),(1,7),(1,8),(1,9),(1,10),(2,6),(2,7),(2,8),(2,9),(2,10),(3,6),(3,7),(... |
| 09:29:58 | <dmead> | cartesian product |
| 09:30:29 | <Syzygy-> | ?t guard |
| 09:30:29 | <lambdabot> | Maybe you meant: tell temp time tiny-url todo todo-add todo-delete topic-cons topic-init topic-null topic-snoc topic-tail topic-tell type . ft v |
| 09:30:34 | <Syzygy-> | ?type guard |
| 09:30:37 | <lambdabot> | forall (m :: * -> *). (MonadPlus m) => Bool -> m () |
| 09:30:44 | <hpaste> | joelr1 pasted "Error compiling derive" at http://hpaste.org/1261 |
| 09:30:54 | <joelr1> | any suggestions on how to fix this? |
| 09:31:17 | <Syzygy-> | > liftM2 (,) [1..5] [6..10] >>= guard . (\(x,y) -> x+y==10) |
| 09:31:18 | <lambdabot> | [(),(),(),()] |
| 09:31:27 | <Syzygy-> | > liftM2 (,) [1..5] [6..10] >> guard . (\(x,y) -> x+y==10) |
| 09:31:28 | <lambdabot> | Couldn't match expected type `[]' |
| 09:32:11 | <Syzygy-> | Hmmmm... Some way to do the do x <- a; y <- b; guard (a+b==10); return (a+b) with binds? |
| 09:33:21 | <dmead> | with a comprehension i think |
| 09:33:35 | <dmead> | peoples |
| 09:33:42 | <dmead> | is there a constructor for upper case chars? |
| 09:34:01 | <dmead> | outside of the parsec class |
| 09:34:02 | <tsp> | what does this come out as? Ĉ |
| 09:34:04 | <tsp> | damn |
| 09:34:11 | <tsp> | utf-8 doesn't work on any modern os |
| 09:34:16 | <dmead> | :s |
| 09:34:19 | <tsp> | at least not mine :) |
| 09:34:27 | <dmead> | maybe it's just irc |
| 09:34:31 | <tsp> | maybe |
| 09:34:44 | <dmead> | what did you mean to type? |
| 09:35:00 | <Syzygy-> | dmead: Constructor? |
| 09:35:03 | <tsp> | esperanto C |
| 09:35:04 | <dmead> | ah |
| 09:35:06 | <tsp> | it comes up in the shell as a normal C |
| 09:35:13 | <tsp> | but appears in files as an esperanto one |
| 09:35:24 | <fuzan> | tsc, i saw a "? ??" :) |
| 09:35:25 | <Syzygy-> | I saw Ä and a white blob |
| 09:35:34 | <tsp> | hmm, I'll put it in a text file |
| 09:35:34 | <xpika> | what does * -> * translate to? |
| 09:35:41 | <tsp> | Ŝ |
| 09:35:49 | <tsp> | I heard a ? |
| 09:35:54 | <Syzygy-> | xpika: It's a kind that takes another type in order to generate a basic type. |
| 09:35:57 | <xpika> | as in return :: forall a (m :: * -> *). (Monad m) => a -> m a |
| 09:36:17 | <tsp> | what did taht come out as for you? |
| 09:36:27 | <Syzygy-> | xpika: It means that m needs to be a type of kind * -> *, i.e. something that maps types to new types. |
| 09:36:41 | <Syzygy-> | tsp: ^S |
| 09:36:46 | <tsp> | ah, there should be a unicode help channel around here somewhere |
| 09:37:02 | <tsp> | damn, it just shows up as either (1) a normal S in the shell, or 2) a ? in irc |
| 09:37:03 | <Syzygy-> | #unicode for instance |
| 09:37:14 | <Syzygy-> | Topic for #unicode: UNICODE users channel |
| 09:37:19 | <Syzygy-> | (and it goes on to more) |
| 09:37:25 | <tsp> | hmm, good idea |
| 09:38:01 | <xpika> | so M needs to be of type data Something x = ... |
| 09:38:48 | <ski> | @kind Maybe |
| 09:38:49 | <Syzygy-> | M needs to be something like IO, or [], or Maybe |
| 09:38:51 | <lambdabot> | * -> * |
| 09:38:54 | <ski> | @src Maybe |
| 09:38:54 | <lambdabot> | data Maybe a = Nothing | Just a |
| 09:39:14 | <xpika> | ok |
| 09:39:15 | <ski> | @kind Either |
| 09:39:18 | <lambdabot> | * -> * -> * |
| 09:39:20 | <ski> | @kind Either String |
| 09:39:23 | <lambdabot> | * -> * |
| 09:39:30 | <xpika> | :) |
| 09:40:21 | <Syzygy-> | :k Either undefined |
| 09:40:24 | <lambdabot> | Not in scope: type variable `undefined' |
| 09:40:27 | <Syzygy-> | Doh. |
| 09:40:32 | <Syzygy-> | Is there smoe null type? |
| 09:40:42 | <ski> | Void |
| 09:40:43 | <ski> | ? |
| 09:40:53 | <Syzygy-> | :k Either Void |
| 09:40:56 | <lambdabot> | Not in scope: type constructor or class `Void' |
| 09:41:09 | <eivuokko> | There isn't for context like that. Usually such type is designated by type variable. |
| 09:41:41 | <ski> | @djinn Either a b -> (a -> Void,b -> Void) -> Void |
| 09:41:41 | <lambdabot> | f a = |
| 09:41:41 | <lambdabot> | case a of |
| 09:41:41 | <lambdabot> | Left b -> \ (c, _) -> c b |
| 09:41:41 | <lambdabot> | Right d -> \ (_, e) -> e d |
| 09:41:42 | <xerox> | ?kind Control.Monad.RWS.RWST |
| 09:41:44 | <lambdabot> | * -> * -> * -> (* -> *) -> * -> * |
| 09:41:48 | <dmead> | tsp: hows your haskell coming? |
| 09:42:01 | <ski> | @djinn-env |
| 09:42:01 | <lambdabot> | Plugin `djinn' failed with: IRCRaised Data.ByteString.last: empty ByteString |
| 09:42:07 | <ski> | bah |
| 09:42:51 | <dmead> | ski: you doing some logic too? |
| 09:44:29 | <joelr1> | any template haskell experts in the house? |
| 09:46:54 | <dcoutts> | joelr1: the Q functor instance is almost certainly defined in one of the TH modules |
| 09:47:13 | <joelr1> | dcoutts: oh... |
| 09:47:14 | <dcoutts> | otherwise if it's only in a later version or something then just define it locally |
| 09:47:18 | <dcoutts> | @type fmap |
| 09:47:20 | <lambdabot> | forall a b (f :: * -> *). (Functor f) => (a -> b) -> f a -> f b |
| 09:47:22 | <dcoutts> | @type liftM1 |
| 09:47:24 | <lambdabot> | Not in scope: `liftM1' |
| 09:47:28 | <joelr1> | M |
| 09:47:30 | <dcoutts> | @type Control.Monad.liftM1 |
| 09:47:33 | <lambdabot> | Not in scope: `Control.Monad.liftM1' |
| 09:47:36 | <joelr1> | @type liftM |
| 09:47:38 | <lambdabot> | forall a1 r (m :: * -> *). (Monad m) => (a1 -> r) -> m a1 -> m r |
| 09:47:41 | <dcoutts> | ah yes :-) |
| 09:48:11 | <joelr1> | dcoutts: i guess i have to go fishing through the TH modules. i wouldn't know how to define a local instance |
| 09:48:19 | <joelr1> | meaning what to put in the "methods" |
| 09:48:51 | <dcoutts> | joelr1: notice how the type of fmap and liftM have a remarkable similarity? :-) |
| 09:49:19 | <joelr1> | dcoutts: they look almost the same, yes, except functor is used instead of monad |
| 09:49:21 | <joelr1> | why? |
| 09:51:59 | <ski> | mathematically, every monad is a functor, and its 'fmap' is just 'liftM' as defined by |
| 09:52:03 | <ski> | @src liftM |
| 09:52:03 | <lambdabot> | liftM f m1 = do { x1 <- m1; return (f x1) } |
| 09:52:19 | <Syzygy-> | ?src IO fmap |
| 09:52:20 | <lambdabot> | fmap f x = x >>= (return . f) |
| 09:52:35 | <ski> | @undo do { x1 <- m1; return (f x1) } |
| 09:52:35 | <lambdabot> | m1 >>= \ x1 -> return (f x1) |
| 09:52:36 | <Syzygy-> | See any similarity? |
| 09:53:08 | <ski> | (every instance of 'Monad' ought to be an instance of 'Functor' with 'fmap' being same as 'liftM') |
| 09:53:18 | <joelr1> | yes, yes, just trying to figure out what this has to do with the TH error in http://hpaste.org/1261 |
| 09:54:26 | <ski> | (it would be better if 'Functor' was a superclass of 'Monad' .. but we'd like the class system to be more expressive so we can e.g. state defaults nicer ..) |
| 09:56:45 | <joelr1> | dcoutts: it's TH.Syntax |
| 09:58:11 | <joelr1> | dcoutts: but it's imported. weird... |
| 09:58:24 | <dcoutts> | weird indeed |
| 10:00:43 | <joelr1> | dcoutts: if I add a local instance declaration then i get "Data/Derive/TH.hs:23:10: Not in scope: data constructor `Q`" any tips on this? |
| 10:00:49 | <mnislaih> | I can't find the Functor instance fo Q either |
| 10:00:55 | <joelr1> | i guess i have to go fishing for Q |
| 10:01:05 | <joelr1> | mnislaih: TH.Syntax, it's there |
| 10:01:13 | <joelr1> | mnislaih: in ghc 6.6 at leat |
| 10:01:16 | <joelr1> | least |
| 10:02:25 | <mnislaih> | if you were in 6.7, you can say "derive instance Functor Q" inside your module |
| 10:02:27 | <joelr1> | cause Q is exported from TH.Syntax as well |
| 10:02:47 | <joelr1> | is the list of new features in 6.7 posted someplace? |
| 10:02:54 | <mnislaih> | but you can always write the instance manually |
| 10:03:02 | <mnislaih> | or just use liftM |
| 10:03:03 | <joelr1> | mnislaih: i did. see the next error above |
| 10:05:39 | <joelr1> | i'm starting to get this weird feeling... that the sources in darcs that i'm looking at do not match the compiler that i'm using |
| 10:06:14 | <mnislaih> | you are looking at the ghc sources? |
| 10:06:28 | <joelr1> | mnislaih: yes, but i pulled them from darcs so they must be 6.7 |
| 10:06:39 | <joelr1> | i'll have to go find my darwinports sources for 6.6 |
| 10:07:01 | <mnislaih> | those would be at http://darcs.haskell.org/ghc-6.6/ |
| 10:07:02 | <lambdabot> | Title: Index of /ghc-6.6 |
| 10:07:24 | <joelr1> | mnislaih: well, that makes my job easier. so hunting for Functor Q I go |
| 10:07:43 | <mnislaih> | but if you are looking for the Language.Syntax.TH, that is at http://darcs.haskell.org/libraries/template-haskell/ |
| 10:07:44 | <lambdabot> | Title: Index of /libraries/template-haskell |
| 10:08:29 | <joelr1> | mnislaih: thanks, looking... |
| 10:09:13 | <joelr1> | mnislaih: same thing. look at the instance Functior Q |
| 10:09:15 | <joelr1> | it's there |
| 10:09:18 | <joelr1> | and Q is exported |
| 10:09:26 | <mnislaih> | Thu Mar 22 18:01:42 CET 2007 Ian Lynagh <igloo@earth.li> |
| 10:09:26 | <mnislaih> | * Add instance Functor Q and make Quasi require Functor |
| 10:09:38 | <joelr1> | @#%@# |
| 10:09:38 | <lambdabot> | Unknown command, try @list |
| 10:09:39 | <mnislaih> | that was added a few days ago |
| 10:09:44 | <mnislaih> | :S |
| 10:10:00 | <joelr1> | mnislaih: added to 6.6 a few days ago? |
| 10:10:17 | <mnislaih> | it's not there in 6.6, |
| 10:10:29 | <joelr1> | much grief |
| 10:10:46 | <mnislaih> | yep, |
| 10:10:48 | <joelr1> | still, why does it complain about missing Q? |
| 10:10:57 | <joelr1> | not in scope data constructor Q |
| 10:11:31 | <mnislaih> | because it is not exported |
| 10:12:02 | <mnislaih> | you cannot really define the Functor instance locally, |
| 10:12:13 | <mnislaih> | (I believe) |
| 10:12:45 | <joelr1> | mnislaih: are you on 6.7? how close is that to production? |
| 10:13:07 | <joelr1> | mnislaih: where can i read up on the features? /googling/ |
| 10:13:13 | <mnislaih> | it flukes sometimes, but not very often |
| 10:13:32 | <mnislaih> | dont know. |
| 10:14:05 | <mnislaih> | well there is the debugger, stand-alone deriving, a lot of improvements to the type system |
| 10:14:28 | <mnislaih> | and now that Gregory Wright added a ghc-devel MacPort |
| 10:14:32 | <joelr1> | mnislaih: what's standalone deriving? |
| 10:14:52 | <joelr1> | mnislaih: don't see that (macport) |
| 10:14:59 | <joelr1> | i only see 6.5 and 6.7 |
| 10:15:02 | <joelr1> | err, 6.6 |
| 10:15:04 | <mnislaih> | ask ghc to derive Q for you, even if you didn't define Q yourself |
| 10:15:27 | <mnislaih> | G Wright posted to the glasgow-haskell-users list 2 or 3 days ago |
| 10:16:54 | <joelr1> | mnislaih: looking... |
| 10:18:02 | <joelr1> | found the post |
| 10:18:22 | <joelr1> | i guess i'll just build the darcs version ... but then i'll have to rebuild my libs... argh... |
| 10:18:41 | <mnislaih> | oh, and cabal-install comes there too |
| 10:18:47 | <joelr1> | i guess i gotta look at it as my contribution to ghc development |
| 10:18:53 | <joelr1> | mnislaih: what's cabal-install? |
| 10:19:16 | <mnislaih> | havent seen hackage ? |
| 10:19:20 | <mnislaih> | hackage.haskell.org |
| 10:20:42 | <joelr1> | mnislaih: yes, seen hackage |
| 10:20:47 | <joelr1> | so it pulls from there automatically? |
| 10:21:22 | <mnislaih> | yep |
| 10:23:04 | <joelr1> | mnislaih: does it automatically get invoked when you do runhaskell Setup.hs configure? |
| 10:23:46 | <mnislaih> | no, I don't think so. But you better ask one of the cabal developers, they might have plans for that |
| 10:25:15 | <joelr1> | mnislaih: ok, i'll try to run it |
| 10:26:14 | <joelr1> | mnislaih = pepe? |
| 10:26:22 | <mnislaih> | yes |
| 10:26:30 | <joelr1> | ah, i'll have to remember that |
| 10:26:35 | <joelr1> | @localtime mnislaih |
| 10:26:37 | <lambdabot> | Local time for mnislaih is 2007-04-05 12:26:25 +0200 |
| 10:26:45 | <mnislaih> | hm? |
| 10:26:47 | <joelr1> | mnislaih: madrid? |
| 10:26:51 | <mnislaih> | valencia |
| 10:27:03 | <joelr1> | mnislaih: cool! what do you dedicate yourself to? |
| 10:27:13 | <mnislaih> | PhD procrastinating |
| 10:27:21 | <mnislaih> | you are in Mallorca right ? |
| 10:27:28 | <joelr1> | mnislaih: nah, tenerife |
| 10:27:32 | <mnislaih> | ahh |
| 10:28:04 | <joelr1> | but i can't afford to procrastinate :-) |
| 10:30:05 | <joelr1> | mnislaih: i gave meself 1 week to rewrite my translator from ocaml |
| 10:51:30 | <joelr1> | mnislaih: what did you do about System.Console.Readline? |
| 10:52:00 | <mnislaih> | follow the instructions at http://mult.ifario.us/articles/tag/macosx |
| 10:52:49 | <joelr1> | ok |
| 10:52:53 | <mnislaih> | if that's dead, http://hackage.haskell.org/trac/ghc/wiki/X86OSXGhc |
| 10:52:54 | <lambdabot> | Title: X86OSXGhc - GHC - Trac |
| 10:53:48 | <joelr1> | i have readline in macports |
| 10:53:53 | <joelr1> | i guess i should just point to it |
| 10:53:59 | <mnislaih> | yep |
| 11:07:24 | <mnislaih> | joelr1: I forgot to mention, after configure, check that ghc/libraries/readline says READLINE_BUILD_PACKAGE=yes |
| 11:07:44 | <joelr1> | shoot, i didn't |
| 11:07:55 | <mnislaih> | that's ghc/libraries/readline/config.mg sry |
| 11:08:00 | <mnislaih> | config.mk |
| 11:08:20 | <joelr1> | yes, it does |
| 11:08:28 | <joelr1> | maybe i need to do a global distclean |
| 11:08:33 | <mnislaih> | it's all fine then |
| 11:08:54 | <joelr1> | cause just reconfiguring and letting make take over from where it left does not work |
| 11:09:14 | <mnislaih> | you don't need to reconfigure if it was already allright |
| 11:09:29 | <joelr1> | i did need to reconfigure to add readline paths |
| 11:09:43 | <joelr1> | i still get the error from ghci |
| 11:09:46 | <mnislaih> | oh, |
| 11:09:56 | <joelr1> | should i clean in the compiler/ghci and deriver/ghci directories? |
| 11:09:59 | <mnislaih> | yeah, make distclean is the safest way |
| 11:10:09 | <mnislaih> | but you can try to touch InteractiveUI.hs |
| 11:10:20 | <mnislaih> | and make clean only the readline package |
| 11:10:28 | <mnislaih> | not sure if that will do the trick though |
| 11:10:36 | <joelr1> | doing |
| 11:17:07 | <tsp> | I guess if teckningar are notes, I can shorten that to teckning = one of them? |
| 11:17:22 | <tsp> | that's an odd sounding word if I ever heard one :) |
| 11:21:14 | <joelr1> | tsp: are you learning swedish? |
| 11:21:21 | <Syzygy-> | teckningar = drawings |
| 11:21:32 | <Syzygy-> | And yes, -ar is one of the plural suffices. |
| 11:26:35 | <joelr1> | folks, is there a syntax colorizer in haskell? |
| 11:26:42 | <joelr1> | that can be used for other languages |
| 11:30:24 | <dons> | yeah, HsColour |
| 11:30:27 | <dons> | ?where HsColour |
| 11:30:28 | <lambdabot> | http://www.cs.york.ac.uk/fp/darcs/hscolour/ |
| 11:30:36 | <dons> | its the one we use on hpaste.org |
| 11:32:15 | <joelr1> | dons: that only colorizes haskell code, though. can be be extended? |
| 11:32:48 | <SamB_XP> | it doesn't even support .txt! |
| 11:32:50 | <joelr1> | i guess i'll need to loook into it |
| 11:36:42 | <joelr1> | what do you make of this error? |
| 11:36:43 | <joelr1> | unknown symbol `_derivezm0zi1_DataziDeriveziPeephole_zdf7_closure' |
| 11:36:50 | <joelr1> | this happens when i run ghci -package derive |
| 11:37:06 | <joelr1> | where derive is something that i just built and installed |
| 11:37:41 | <phoniq> | running that through c++filt might help |
| 11:38:00 | <joelr1> | phoniq: but derive is a haskell package, cabalized too |
| 11:38:45 | <phoniq> | oops. haskell's mangled symbol names look c++ish. |
| 11:47:16 | <joelr1> | mnislaih: have you see this? |
| 11:47:17 | <joelr1> | basicTypes/OccName.lhs-boot:1:0: |
| 11:47:17 | <joelr1> | Failed to load interface for `Prelude': |
| 11:47:42 | <mnislaih> | that would be for stage2 ? |
| 11:48:11 | <joelr1> | mnislaih: i guess. this is when trying to pick up after reconfigure. let me clean, configure and rebuild everything again |
| 11:48:16 | <mnislaih> | oh, yes, that's probably due to split-objs causing trouble |
| 11:48:27 | <mnislaih> | create a custom build.mk with split-objs disabled |
| 11:48:42 | <mnislaih> | you know, in ghc/mk/ |
| 11:49:14 | <mnislaih> | and then make clean ; make -j2 |
| 11:49:54 | <mnislaih> | joelr1: make sure you disable split objs |
| 11:50:16 | <joelr1> | mnislaih: why would i? |
| 11:50:28 | <joelr1> | mnislaih: doesn't it help? or it fails on mac osx |
| 11:50:52 | <joelr1> | or does it help pick up after a failure? |
| 11:51:03 | <mnislaih> | split objs seem to cause trouble on mac osx. I disabled it a while ago and never looked back |
| 11:52:18 | <mnislaih> | spying the macports settings for ghc-6.6 might give some further info on why |
| 11:53:52 | <joelr1> | mnislaih: fishing for those settings ... |
| 11:55:15 | <mnislaih> | note that everything works finely without split objs, it's just that link times will be a bit longer |
| 11:56:17 | <joelr1> | mnislaih: where should i look for settings in macports? i have 3 files under distfiles/ghc |
| 11:56:23 | <joelr1> | untar the bootstrap files? |
| 11:56:46 | <mnislaih> | I guess in the port file mainly |
| 11:56:52 | <joelr1> | i wish i could just look at the patches |
| 11:57:08 | <joelr1> | mnislaih: silly question, i know, but where's the port file? |
| 11:57:26 | <mnislaih> | dont know in the filesystem. it's all available in the macports subversion repo |
| 11:57:46 | <mnislaih> | let me find the link |
| 11:58:21 | <mnislaih> | http://trac.macosforge.org/projects/macports/browser/trunk/dports/lang/ghc |
| 11:58:23 | <lambdabot> | Title: /trunk/dports/lang/ghc - MacPorts - Trac, http://tinyurl.com/yuucya |
| 11:58:30 | <mnislaih> | there you go, the portfile and the patches are there |
| 11:59:13 | <joelr1> | thanks pepe |
| 11:59:15 | <joelr1> | Plan to throw one away. You will do that, anyway. Your only choice is whether to try to sell the throwaway to customers. - Frederick Brooks. |
| 11:59:29 | <joelr1> | boom! right in the spot |
| 11:59:47 | <mnislaih> | heh :) |
| 11:59:49 | <joelr1> | this has happened to me a couple of times already, except i never sold the throwaway to customers :D |
| 12:00:03 | <joelr1> | mnislaih: happening to me right now, throwing away the ocaml version |
| 12:00:38 | <mnislaih> | viva haskell ;) |
| 12:01:22 | <joelr1> | viva le haskell |
| 12:01:29 | <joelr1> | sounds cooler i think |
| 12:01:45 | <joelr1> | how do the french say viva? |
| 12:02:04 | <joelr1> | pepe, didn't find anything re: split-objs |
| 12:02:41 | <joelr1> | mnislaih: not in the port file anyway. loooking at patches |
| 12:04:18 | <joelr1> | patch files have nothing on that either. but ghc is still building so i'll wait until it fails |
| 12:04:32 | <joelr1> | make -j 2 screams |
| 12:04:42 | <joelr1> | make -j fails with out of memory errors eventually |
| 12:05:34 | <mnislaih> | º |
| 12:06:03 | <joelr1> | mnislaih: i think you are right. editing config |
| 12:07:03 | <mnislaih> | hopefully I'm right. Otherwise you won't be building ghc today :) |
| 12:07:19 | <joelr1> | mnislaih: still, it's strange that there's no patch for it in macports |
| 12:08:10 | <mnislaih> | prolly the port does something that solves the problem with split objs |
| 12:08:17 | <robreim> | Are there any plans to implement a String typeclass to clean up all the different String and ByteString.* functions? |
| 12:08:40 | <mnislaih> | gotta go for lunch |
| 12:08:42 | <mnislaih> | gl joelr1 |
| 12:08:44 | <dcoutts> | robreim: not any immediate plans |
| 12:08:53 | <joelr1> | thanks mnislaih |
| 12:09:52 | <robreim> | dcoutts: got any idea what sort of problems such an endeavor might encounter? |
| 12:10:18 | <dcoutts> | type, performance and dependencies |
| 12:10:58 | <joelr1> | dcoutts: duncan, any idea on why a module of an installed library could not be found? |
| 12:11:46 | <robreim> | dcoutts: sorry if it should be obvious to me, but can you elaborate? Perhaps with examples? |
| 12:13:04 | <quicksilver> | joelr1: you got mail :) |
| 12:13:14 | <joelr1> | quicksilver: checking... |
| 12:13:26 | <quicksilver> | let constructorName = takeWhile (/=' ') . show in constructorName (Just undefined) |
| 12:13:29 | <quicksilver> | > let constructorName = takeWhile (/=' ') . show in constructorName (Just undefined) |
| 12:13:30 | <lambdabot> | "Just" |
| 12:13:34 | <quicksilver> | that's pretty interesting |
| 12:13:37 | <dcoutts> | joelr1: the module might not be exposed, the package might not be exposed |
| 12:13:39 | <quicksilver> | not that that: |
| 12:13:41 | <joelr1> | quicksilver: is that you, jules? |
| 12:13:44 | <quicksilver> | > show (just undefined) |
| 12:13:44 | <lambdabot> | Not in scope: `just' |
| 12:13:50 | <quicksilver> | > show (Just undefined) |
| 12:13:51 | <lambdabot> | Undefined |
| 12:13:53 | <dcoutts> | robreim: sorry, no time. I've got to finish a paper. |
| 12:13:57 | <quicksilver> | joelr1: right |
| 12:14:04 | <robreim> | no problem. Thanks anyhow. |
| 12:14:17 | <dcoutts> | robreim: ask me in two days time :-) |
| 12:14:18 | <quicksilver> | joelr1: metaprogramming is great but you don't always needs it; or not as much as many people imagine, anyhow |
| 12:14:36 | <joelr1> | quicksilver: appreciate your email. reading... |
| 12:15:03 | <Syzygy-> | > let constructorName = takeWhile (/=' ') . show; test = (23 :: Integer) in constructorName test |
| 12:15:05 | <lambdabot> | "23" |
| 12:15:09 | <Syzygy-> | Doh. |
| 12:15:17 | <robreim> | dcoutts: I'll try to remember. In the meantime, best wishes with your paper. Gambatte! |
| 12:15:30 | <Syzygy-> | > let constructorName = takeWhile (/=' ') . show; test = (return 15)::Maybe Integer in constructorName test |
| 12:15:31 | <mux> | dcoutts: out of curiosity, what's your paper about? |
| 12:15:32 | <lambdabot> | "Just" |
| 12:15:35 | <Syzygy-> | Ah! |
| 12:16:01 | <quicksilver> | and I bet there is a typeclass hack to tell how many parameters a constructor has, too |
| 12:16:04 | <quicksilver> | although not a h98 one |
| 12:16:13 | <dcoutts> | mux: fusion |
| 12:16:38 | <mux> | dcoutts: again! ;-) |
| 12:16:52 | <dcoutts> | mux: up, more fusion |
| 12:17:19 | <joelr1> | quicksilver: awesome suggestions, thanks. i do want to automatically choose the # and type of args, though |
| 12:17:25 | <mux> | great, I wish I could attend your talk |
| 12:17:59 | <joelr1> | i have a suspicion that dons and dcoutts are writing a paper together. maybe two! |
| 12:18:01 | <joelr1> | hehe |
| 12:18:10 | <dcoutts> | @arr |
| 12:18:10 | <lambdabot> | Aye |
| 12:18:16 | <joelr1> | two days... two papers... two days |
| 12:18:44 | <dcoutts> | mux: the conference isn't for months yet but the deadline... |
| 12:19:05 | <mux> | oh |
| 12:20:16 | <joelr1> | @arr |
| 12:20:16 | <lambdabot> | Aye |
| 12:20:21 | <joelr1> | hmm |
| 12:21:48 | <quicksilver> | joelr1: yeah, I know you do |
| 12:21:55 | <quicksilver> | joelr1: I thought the point was interesting to make nonetheless |
| 12:22:54 | <joelr1> | quicksilver: i concur |
| 12:25:34 | <araujo> | morning! |
| 12:26:13 | <joelr1> | @where derive |
| 12:26:14 | <lambdabot> | http://www.cs.york.ac.uk/fp/darcs/derive |
| 12:32:14 | <hpaste> | quicksilver pasted "Counting arguments using a type class" at http://hpaste.org/1262 |
| 12:32:22 | <quicksilver> | joelr1: check that paste |
| 12:34:05 | <joelr1> | quicksilver: hmm... |
| 12:34:24 | <joelr1> | quicksilver: i gotta save that pasted! |
| 12:36:34 | <joelr1> | quicksilver: totally love it but i have something like 100 parsers to derive, very tedious by hand |
| 12:36:53 | <hpaste> | quicksilver annotated "Counting arguments using a type class" with "Now it describes the arguments too" at http://hpaste.org/1262#a1 |
| 12:36:54 | <joelr1> | quicksilver: i'm saving your tips for other future uses, though |
| 12:37:16 | <quicksilver> | joelr1: now it counts the types and lists them in a concrete data type |
| 12:37:28 | <joelr1> | hmm... |
| 12:37:42 | <joelr1> | let me take that a bit further |
| 12:38:21 | <joelr1> | quicksilver: if you can grab the arg types as a list you can probably generate a list of parsers from it, right? if you were able to attach a type to each parser |
| 12:38:46 | <tsp> | Program som körs! |
| 12:38:54 | <joelr1> | although you would need to collect all your types into the separate ArgTypes ADT. |
| 12:38:59 | <joelr1> | quicksilver: right ^^^ |
| 12:39:00 | <joelr1> | ? |
| 12:39:04 | <tsp> | just realized I had to reinstall inux :) |
| 12:39:07 | <tsp> | linux |
| 12:39:29 | <tsp> | accidentally deleted the disk |
| 12:39:35 | <Cale> | whoa |
| 12:39:49 | <Cale> | Which linux will you be installing? :) |
| 12:39:54 | <joelr1> | ubuntu! |
| 12:40:04 | <Cale> | Yeah, Ubuntu rocks :) |
| 12:40:43 | <tsp> | arch |
| 12:40:45 | <Cale> | Too bad about your files, eh? Or do you keep a separate home partition? |
| 12:40:53 | <tsp> | Cale: heh, it was my haskell test box |
| 12:40:54 | <tsp> | vmware |
| 12:40:56 | <Cale> | ah |
| 12:41:02 | <tsp> | everything important is on this sparc box |
| 12:41:05 | <quicksilver> | joelr1: sorry, I was tidying it up |
| 12:41:06 | <joelr1> | Cale: big time! that's what i'm using as my amazon EC2 image |
| 12:41:15 | <joelr1> | quicksilver: another paste? :D |
| 12:41:26 | <hpaste> | quicksilver annotated "Counting arguments using a type class" with "tidier, no more quadratic instance explosion" at http://hpaste.org/1262#a2 |
| 12:41:26 | <joelr1> | quicksilver: more type class magic? |
| 12:42:02 | <hpaste> | quicksilver annotated "Counting arguments using a type class" with "try again (was tidier, no more quadratic instance explosion" at http://hpaste.org/1262#a3 |
| 12:42:23 | <quicksilver> | joelr1: better, don't need instances for each possible a->b->TestType |
| 12:42:35 | <quicksilver> | Cale: care to take a look? |
| 12:42:46 | <Cale> | okay |
| 12:42:48 | <quicksilver> | Cale: objective was to produce a type class framework which can 'introspect' an ADT |
| 12:42:58 | <quicksilver> | Cale: and find out how many parameters, and what types, each constructor has |
| 12:43:07 | <joelr1> | quicksilver: i still see them there, instances for each possible a -> b -> TestType |
| 12:43:15 | <Cale> | Have you looked at SYB? |
| 12:43:19 | <quicksilver> | joelr1: not in the last paste. just one instance for it |
| 12:43:33 | <quicksilver> | Cale: yes; this is just H98 + MPTCs, though |
| 12:43:37 | <joelr1> | quicksilver: instance DescribeArgs (String->TestType) TestType where descArgs _ = [JStr] |
| 12:43:38 | <Cale> | ah, okay |
| 12:43:42 | <joelr1> | this is the last paste |
| 12:43:45 | <joelr1> | the tidier one |
| 12:44:05 | <quicksilver> | joelr1: no, that's the one I pasted in error :) |
| 12:44:07 | <quicksilver> | joelr1: scroll down |
| 12:44:11 | <quicksilver> | joelr1: or, reload |
| 12:44:23 | <joelr1> | quicksilver: got it |
| 12:45:12 | <joelr1> | quicksilver: can you paste the output from a couple of examples? like you did with CountArgs? |
| 12:45:24 | <joelr1> | quicksilver: well, never mind that, a better question |
| 12:46:07 | <joelr1> | quicksilver: does this solution require me to gather all my types under ArgTypes? |
| 12:47:14 | <hpaste> | quicksilver annotated "Counting arguments using a type class" with "with examples + one more constructor" at http://hpaste.org/1262#a4 |
| 12:47:21 | <quicksilver> | joelr1: yes |
| 12:47:33 | <quicksilver> | joelr1: or you could use 'Typeable' |
| 12:47:41 | <quicksilver> | joelr1: which is a built-in haskell way to concretize most types |
| 12:47:58 | <quicksilver> | joelr1: I don't think it's a bad thing to collect all your types into a constructor, personally |
| 12:48:20 | <joelr1> | quicksilver: well, i have a huge AST, right |
| 12:48:37 | <quicksilver> | joelr1: but presumably no more than a dozen parameter types... |
| 12:48:44 | <quicksilver> | joelr1: most DSLs just have 4 or 5 |
| 12:48:54 | <joelr1> | quicksilver: with types for numerical expressions, string expressions, etc. |
| 12:49:30 | <quicksilver> | joelr1: ok, go on |
| 12:49:47 | <joelr1> | quicksilver: to tell you the truth, arguments can only be of type string, bool or num expression, although just NumExpr can look like this |
| 12:49:50 | <joelr1> | @paste |
| 12:49:50 | <lambdabot> | Haskell pastebin: http://hpaste.org/new |
| 12:50:06 | <hpaste> | joelr1 pasted "NumExpr" at http://hpaste.org/1263 |
| 12:50:30 | <joelr1> | string expression and bool expr look similar |
| 12:50:44 | <joelr1> | so what i'm trying to derive is a parser for this type |
| 12:51:03 | <quicksilver> | ACTION nods |
| 12:51:33 | <joelr1> | the reason that have separate NumGroup, BoolGroup, StrGroup or same for NumArrayVar, etc. is because i would like to use quickcheck to generate a random AST |
| 12:51:57 | <joelr1> | and with just Expr (like i had before) i would be generating invalid ASTs at times |
| 12:52:15 | <joelr1> | the way i have it now everything is nicely typed |
| 12:52:19 | <quicksilver> | ACTION nods |
| 12:52:26 | <quicksilver> | well, perhaps you want data.typeable then |
| 12:52:33 | <joelr1> | the kicker is that NumCall has about... let me count them |
| 12:52:33 | <quicksilver> | http://haskell.org/ghc/docs/latest/html/libraries/base/Data-Typeable.html |
| 12:52:35 | <lambdabot> | http://tinyurl.com/ylclo5 |
| 12:52:51 | <quicksilver> | or, to be honest, perhaps you want to do with with Data.Derive just as stefan suggested |
| 12:53:03 | <quicksilver> | but it's interesting to see how far you can get within the language |
| 12:53:21 | <joelr1> | quicksilver: 170+ constructors describing the various numerical functions |
| 12:53:50 | <quicksilver> | joelr1: funky :) |
| 12:53:56 | <joelr1> | quicksilver: so i want to 1) derive parsers for these and 2) derive pretty-printers |
| 12:54:02 | <joelr1> | based on the types |
| 12:54:24 | <quicksilver> | there was a paper on writing pretty printers and parsers simultaneously |
| 12:54:31 | <joelr1> | it's easy because the format is always the same: function name '(' arg1 , arg2 , ... ')' |
| 12:54:43 | <joelr1> | quicksilver: don't have that paper |
| 12:54:55 | <quicksilver> | joelr1: http://www.cs.ru.nl/A.vanWeelden/bi-arrows/ |
| 12:54:57 | <lambdabot> | Title: There and Back Again: Arrows for Invertible Programming |
| 12:55:12 | <joelr1> | quicksilver: but i can imagine that you can somehow derive an unparser from a parser |
| 12:55:32 | <joelr1> | this sucks. full paper at ACM :( |
| 12:55:34 | <joelr1> | why?! |
| 12:55:40 | <quicksilver> | joelr1: http://citeseer.ist.psu.edu/alimarine05there.html |
| 12:55:41 | <lambdabot> | Title: There and Back Again: Arrows for Invertible Programming (ResearchIndex) |
| 12:55:47 | <quicksilver> | joelr1: downloadable from that link |
| 12:55:50 | <quicksilver> | joelr1: good old citeseer |
| 12:56:11 | <Botje> | damn you, quicksilver. I was just about to start working for school :) |
| 12:56:21 | <quicksilver> | Botje: damn me for which part? |
| 12:56:34 | <Botje> | linking to a paper about stuff that interests me ;) |
| 12:56:36 | <quicksilver> | joelr1: I think I'll email my snippet to the -cafe. No doubt there are even cleverer generalisations |
| 12:57:22 | <joelr1> | quicksilver: definitely but hang on a bit |
| 12:57:36 | <joelr1> | quicksilver: maybe you'll want to add more? |
| 12:57:58 | <joelr1> | anyway, i downloaded the paper. eventually, i'll set friday aside (saturday maybe?) just to reading papers :D |
| 12:58:07 | <quicksilver> | :) |
| 12:58:25 | <quicksilver> | Cale: seen this trick before? |
| 12:58:28 | <joelr1> | but back to the grind, would you say that it's impossible to derive the parser and pprinter with your approach? |
| 12:58:45 | <quicksilver> | > let constructorName = takeWhile (/=' ') . show in constructorName (Just undefined) |
| 12:58:46 | <lambdabot> | "Just" |
| 12:59:10 | <quicksilver> | joelr1: no, not at all. But in the design I outlines, you would need to 'shadow' all your argument types with a concrete ADT |
| 12:59:20 | <quicksilver> | joelr1: at least that's work that only needs doing once, but it's still work |
| 12:59:40 | <quicksilver> | joelr1: on the other hand, you could use Data.Typeable and its TypeRep, which gives you that for most types for free |
| 13:00:03 | <joelr1> | quicksilver: another paste would be _very_ welcome ;-) |
| 13:00:08 | <quicksilver> | joelr1: for which part? |
| 13:00:14 | <joelr1> | keeping the old paste, of course, for pasting both |
| 13:00:16 | <joelr1> | for typeable |
| 13:03:59 | <hpaste> | quicksilver annotated "Counting arguments using a type class" with "Data.Typeable version" at http://hpaste.org/1262#a5 |
| 13:04:33 | <quicksilver> | joelr1: it's almost identical. My 'ConcreteType' class was just a cut-down special case of Data.Typeable. |
| 13:04:59 | <joelr1> | quicksilver: looking |
| 13:05:31 | <quicksilver> | joelr1: oh, I removed the multiple parameters on the typeclasses. They were not necessary, I was just confused. |
| 13:05:37 | <joelr1> | quicksilver: so no unification of types is needed |
| 13:06:39 | <joelr1> | quicksilver: would you say that what's needed now is to make each type an instance of a Parseable class and an instance of a PrettyPrintable one? |
| 13:07:08 | <joelr1> | cause then, it would seem, you could just fmap parse to the list of returned types to parse |
| 13:07:19 | <joelr1> | and fmap pprint to pretty-print |
| 13:07:29 | <quicksilver> | joelr1: more or less, yes |
| 13:07:39 | <joelr1> | quicksilver: why more or less? |
| 13:07:41 | <quicksilver> | joelr1: there's some work to do in terms of adding commas in between productions |
| 13:07:47 | <quicksilver> | joelr1: and brackets around them |
| 13:07:52 | <quicksilver> | joelr1: it's not hard though |
| 13:08:27 | <quicksilver> | joelr1: anyhow I have to grab some lunch and do some work. I'll paste the last version to the -cafe in case anyone has a comment |
| 13:08:46 | <joelr1> | quicksilver: thanks a lot! can i ask one more q? |
| 13:09:16 | <joelr1> | should this support liftMx to return the type? |
| 13:09:24 | <joelr1> | with the parsed arguments |
| 13:10:20 | <joelr1> | anyhow, this is extra-fine type class hackery! :D thanks again |
| 13:10:44 | <Saizan> | couldn't this be handled once with some typeclass recursiveness hackery? |
| 13:10:56 | <joelr1> | Saizan: don't ask me |
| 13:11:02 | <joelr1> | i wouldn't know |
| 13:11:07 | <Saizan> | enumering types i mean |
| 13:13:18 | <joelr1> | i also wonder if there's a hack to get the constructors... hmm... after this i figure (almost) anything is possible! |
| 13:13:43 | <quicksilver> | joelr1: 'get the constructors'? what do you mean? |
| 13:13:53 | <Saizan> | given the type? |
| 13:13:57 | <joelr1> | quicksilver: i thought you were out to lunch :D |
| 13:14:00 | <desp> | odd |
| 13:14:12 | <quicksilver> | joelr1: no, I went to buy a sandwich |
| 13:14:25 | <quicksilver> | joelr1: ah, well that's a tricky one because they don't have the same type as each other |
| 13:14:38 | <quicksilver> | joelr1: so without existentials, you can't make that a well-typed program |
| 13:14:41 | <joelr1> | well, if my NumCall has 170 constructors metaprogramming lets me derive the whole type |
| 13:16:33 | <quicksilver> | joelr1: yes, I see what you mean |
| 13:16:45 | <quicksilver> | joelr1: I can't think of a way of doing that :) |
| 13:17:01 | <Saizan> | something with TH maybe? |
| 13:17:15 | <quicksilver> | Saizan: the point of this tangent it to avoid true metaprogramming |
| 13:17:26 | <quicksilver> | Saizan: he already knows how to do it with Data.Derive, more or less |
| 13:17:33 | <Saizan> | oh ok |
| 13:17:39 | <quicksilver> | Saizan: I was showing how close he could get with a couple of typeclasses |
| 13:23:28 | <joelr1> | very close indeed |
| 13:32:58 | <hpaste> | Saizan annotated "Counting arguments using a type class" with "one recursive case to rule them all" at http://hpaste.org/1262#a6 |
| 13:34:50 | <joelr1> | Saizan: looking |
| 13:35:19 | <joelr1> | Saizan: what does that give you? you didn't post example output |
| 13:36:30 | <Saizan> | it gives you the same functionality with only 2 instances, regardless of the number and arity of the contructors |
| 13:37:02 | <joelr1> | i see |
| 13:37:45 | <joelr1> | haskell is awesome precisely because of folks like you (Saizan, quicksilver, dons, dcoutts, Cale and others too) |
| 13:37:54 | <joelr1> | i think this is strongly underappreciated |
| 13:38:57 | <quicksilver> | Saizan: ah, I wondered if that was possibly |
| 13:39:07 | <quicksilver> | Saizan: I ran out of time to try it :) |
| 13:39:19 | <joelr1> | i had a chance to compare during my ocaml month |
| 13:40:08 | <joelr1> | how in the world do i debug: baz.hs:30:3: Not in scope: `a1' |
| 13:40:28 | <joelr1> | (posted to cafe) |
| 13:41:01 | <Philippa> | joelr1: well, in one sense it's trivial - the identifier's not bound there |
| 13:41:16 | <Philippa> | do you know where the binding site you're expecting to cover that use is? |
| 13:41:20 | <joelr1> | Philippa: right. the question where does it come from. it's not obvious by looking at the code |
| 13:41:39 | <Philippa> | you mean there's no a1 at 30:3? Oh |
| 13:41:43 | <Philippa> | which implementation're you using? |
| 13:41:46 | <joelr1> | Philippa: beats me! it's template haskell generating a1, i think |
| 13:41:52 | <Philippa> | ah |
| 13:41:54 | <joelr1> | Philippa: take a look at the last message in cafe |
| 13:41:59 | <Philippa> | in that case, have fun |
| 13:42:13 | <joelr1> | Philippa: under Automatic Derivation of .. |
| 13:42:23 | <Philippa> | something that might help: you can use debug.trace from inside the quotation monad |
| 13:42:32 | <joelr1> | hmm |
| 13:42:44 | <Philippa> | another something that /might/ help: IIRC there're show instances on the quoted code types |
| 13:42:51 | <Saizan> | you can dump the code generated by TH with a flag |
| 13:42:56 | <Igloo> | joelr1: Pretty-print the stuff you're splicing in |
| 13:43:19 | <Igloo> | and if that doesn't help, print the actual datatype |
| 13:43:27 | <joelr1> | Saizan: which flag |
| 13:43:32 | <Saizan> | -ddump-splices iirc |
| 13:43:46 | <joelr1> | Igloo: i'm not splicing much in, just a name |
| 13:44:19 | <joelr1> | that helped |
| 13:44:54 | <hpaste> | joelr1 annotated "NumExpr" with "result from -ddump-splices" at http://hpaste.org/1263#a1 |
| 13:46:26 | <hpaste> | sebell pasted "Parsec string matching issues" at http://hpaste.org/1264 |
| 13:48:44 | <joelr1> | sebell: if you are matching a portion you need to use try, i think |
| 13:48:50 | <hpaste> | sebell annotated "Parsec string matching issues" with "Possible fix" at http://hpaste.org/1264#a1 |
| 13:48:53 | <joelr1> | sebell: so that the parser fakes reading no input |
| 13:49:07 | <joelr1> | sebell: yup |
| 13:51:12 | <sebell> | bleh. That was a blonde moment |
| 13:53:38 | <joelr1> | happens to me all the time |
| 13:56:13 | <joelr1> | :t fromInteger |
| 13:56:15 | <lambdabot> | forall a. (Num a) => Integer -> a |
| 14:03:03 | <sorear> | hello! |
| 14:03:12 | <glguy> | oh dear :) |
| 14:03:24 | <joelr1> | hi stefan |
| 14:03:30 | <joelr1> | glguy: what;s your name? |
| 14:03:46 | <glguy> | *** glguy is n=eric@ |
| 14:03:53 | <glguy> | :) |
| 14:03:59 | <sorear> | and with a little googling... |
| 14:04:07 | <sorear> | you may have heard of 'hpaste' |
| 14:04:15 | <sorear> | read the who-wrote-it |
| 14:04:16 | <glguy> | it doesn't take much to tie me to a real name and email |
| 14:04:22 | <sorear> | then read the darcs inventor |
| 14:04:42 | <glguy> | joelr1: why do you ask? |
| 14:04:49 | <sorear> | Eric Mertens <emertens@gmail.com>**20070118045641] |
| 14:04:55 | <glguy> | yeah, that one |
| 14:05:52 | <joelr1> | glguy: for no reason. just like to associate nicks with name of people posting to cafe |
| 14:06:05 | <glguy> | ah |
| 14:06:10 | <glguy> | I don't post *too* much |
| 14:06:18 | <joelr1> | yeah, unlike me |
| 14:06:30 | <beelsebob> | is there a properly proscribed way of renaming files/folers in darcs? |
| 14:06:33 | <sebell> | My inbox is full of Joel ;) |
| 14:07:25 | <sorear> | beelsebob: darcs mv |
| 14:07:28 | <glguy> | I remembered that I am off tomorrow when I got to work today |
| 14:07:41 | <glguy> | that really brightened up my morning :) |
| 14:15:09 | <sorear> | ACTION has a LOT of mails to reply to... |
| 14:19:02 | <joelr1> | sorear: not really :D |
| 14:19:46 | <joelr1> | sorear: you should get one last mail that clears it up. well, almost since the parser doesn't work right but that doesn't seem to have anything to do with derivation |
| 14:26:38 | <matthew-_> | @src Monad ((,) a) |
| 14:26:39 | <lambdabot> | Source not found. You type like i drive. |
| 14:26:42 | <matthew-_> | @src Monad (,) |
| 14:26:42 | <lambdabot> | Source not found. Your mind just hasn't been the same since the electro-shock, has it? |
| 14:27:32 | <matthew-_> | @type \f -> f >>= (,) |
| 14:27:34 | <lambdabot> | forall a b. (b -> a) -> b -> (a, b) |
| 14:27:44 | <sorear> | @src (,) >>= |
| 14:27:45 | <lambdabot> | Source not found. Maybe if you used more than just two fingers... |
| 14:27:49 | <sorear> | @src , >>= |
| 14:27:49 | <lambdabot> | Source not found. BOB says: You seem to have forgotten your passwd, enter another! |
| 14:27:53 | <glguy> | (,) isn't a monad? |
| 14:28:02 | <sorear> | glguy: import |
| 14:28:05 | <mauke> | (,) takes two args, it can't be a monad |
| 14:28:08 | <sorear> | @instances Monad |
| 14:28:09 | <lambdabot> | ((->) r), ArrowMonad a, Cont r, ContT r m, Either e, ErrorT e m, IO, Maybe, RWS r w s, RWST r w s m, Reader r, ReaderT r m, ST s, State s, StateT s m, Writer w, WriterT w m, [] |
| 14:28:13 | <matthew-_> | it's the -> r |
| 14:28:19 | <sorear> | @instances-importing Control.Monad.Instances Monad |
| 14:28:20 | <lambdabot> | ((->) r), ArrowMonad a, Cont r, ContT r m, Either e, ErrorT e m, IO, Maybe, RWS r w s, RWST r w s m, Reader r, ReaderT r m, ST s, State s, StateT s m, Writer w, WriterT w m, [] |
| 14:28:25 | <matthew-_> | must be. (,) is a function |
| 14:28:27 | <sorear> | @instances-importing Control.Monad.Writer Monad |
| 14:28:28 | <lambdabot> | ((->) r), ArrowMonad a, Cont r, ContT r m, Either e, ErrorT e m, IO, Maybe, RWS r w s, RWST r w s m, Reader r, ReaderT r m, ST s, State s, StateT s m, Writer w, WriterT w m, [] |
| 14:28:38 | <matthew-_> | so it must be the ((->) r) which is the monad |
| 14:28:41 | <glguy> | ah, I see what you were trying to do |
| 14:28:47 | <sorear> | @tell dons @instances-importing Control.Monad.Writer Monad doesn't find (,) a |
| 14:28:47 | <lambdabot> | Consider it noted. |
| 14:28:55 | <sorear> | @flush |
| 14:29:03 | <glguy> | @wipe |
| 14:29:03 | <lambdabot> | Maybe you meant: dice time type wiki |
| 14:29:41 | <dhart> | I have a quick question: in using HOpenGL, GHC seems to be bafled by "Vertex2 100 100", saying "ambigouis type variable". Is there a way of "casting" he values to GLfloat? |
| 14:29:56 | <matthew-_> | 100::GLfloat |
| 14:30:05 | <mauke> | (100 :: GLfloat) |
| 14:30:05 | <dhart> | thx a bunch |
| 14:30:24 | <mauke> | @bot |
| 14:30:24 | <lambdabot> | :) |
| 14:30:30 | <mauke> | bah, lag |
| 14:34:33 | <sorear> | joelr1: so how long did it take you to figure out I was S. O'Rear? A *lot* of people seem to think I'm a random guy with a pseudonym and an earache... |
| 14:35:05 | <joelr1> | sorear: ugh... a minute yesterday. you said something and then sent me an email |
| 14:37:57 | <joelr1> | mnislaih: i wonder if my prelude error is caused by my using make -j 2 |
| 14:38:27 | <mnislaih> | the prelude error means that base did not build |
| 14:38:48 | <sebell> | joelr1: Which snapshot are you building? |
| 14:38:49 | <mnislaih> | make -j2 works fine here |
| 14:39:02 | <mnislaih> | did split-objs help ? |
| 14:39:08 | <dhart> | And how would I do a conversion from Integer to GLfloat? |
| 14:39:13 | <sorear> | fromInteger |
| 14:39:44 | <joelr1> | sebell: ugh... 0404 |
| 14:39:57 | <joelr1> | mnislaih: no, it did not |
| 14:40:18 | <joelr1> | mnislaih: i put NO in my build.mk |
| 14:40:29 | <mnislaih> | can you check what happens when you call make from libraries/base ? |
| 14:40:33 | <joelr1> | mnislaih: oh, i did manage to build the "quickest" configuration, btw |
| 14:40:38 | <joelr1> | i can't build the perf one |
| 14:40:46 | <joelr1> | with or without split objects |
| 14:41:57 | <sebell> | joelr1: Hmm. I build 0404 yesterday without any issues... (default build) |
| 14:42:21 | <joelr1> | sebell: platform? |
| 14:42:30 | <sebell> | Linux x86. Are you using Mac OS X? |
| 14:42:36 | <joelr1> | sebell: 100% |
| 14:42:40 | <sebell> | ACTION doesn't think he has 0404 on his Mac |
| 14:44:11 | <mnislaih> | joelr1: it's certainly possible to build "perf" in Os X, although I don't have 0404 either. I have Mar 29 here |
| 14:44:27 | <sorear> | I'm using 0402, built from darcs with my TH patches |
| 14:44:40 | <joelr1> | sorear: mac osx ? intel? |
| 14:44:45 | <sorear> | wanted to spare myself the embarrasement of sending patches that broke the build |
| 14:44:46 | <mnislaih> | ACTION builds from darcs too |
| 14:44:53 | <sorear> | sorry, PC intel |
| 14:45:04 | <joelr1> | i'm building from darcs myself |
| 14:45:11 | <sorear> | having mukke, why not make install? |
| 14:45:35 | <joelr1> | mukke? |
| 14:45:39 | <joelr1> | who is mukke |
| 14:46:06 | <sorear> | past tense of make? |
| 14:48:15 | <sebell> | joelr1: I don't know if it would make a different, but I would try the source snapshot distribution |
| 14:48:18 | <sebell> | *difference |
| 14:52:29 | <joelr1> | my example parser doesn't work :-( |
| 14:52:59 | <joelr1> | can anyone suggest a reason? |
| 14:53:44 | <sorear> | have you looked at the generated code? |
| 14:54:12 | <joelr1> | i should, let me just make a paste |
| 14:54:15 | <joelr1> | @paste |
| 14:54:15 | <lambdabot> | Haskell pastebin: http://hpaste.org/new |
| 14:54:45 | <hpaste> | joelr1 pasted "simple parser not working" at http://hpaste.org/1265 |
| 14:55:42 | <hpaste> | joelr1 annotated "simple parser not working" with "generated code" at http://hpaste.org/1265#a1 |
| 14:56:21 | <joelr1> | sorear: i don't think it's the code |
| 14:58:10 | <joelr1> | meaning i don't think it's the derivation that's causing the problem, it's the structure of the parser |
| 15:00:44 | <sorear> | joelr1: found it |
| 15:00:52 | <joelr1> | sorear: found what? |
| 15:02:33 | <hpaste> | joelr1 annotated "simple parser not working" with "still no dice" at http://hpaste.org/1265#a2 |
| 15:02:56 | <joelr1> | this is what i'm trying: run fooParser "Bar (10, 10.345)" |
| 15:03:43 | <beelsebob> | joelr1: why not just use Happy? |
| 15:04:12 | <joelr1> | beelsebob: too late. plus, my original ocaml parser was written yacc-style and i much rather use the combinators |
| 15:04:39 | <joelr1> | beelsebob: i can test things in chunks |
| 15:06:23 | <joelr1> | sorear: what did you find? |
| 15:09:38 | <sorear> | joelr1: mail sent |
| 15:11:03 | <joelr1> | sorear: thanks stefan. can you tell me why the choice version with try doesn't work? |
| 15:11:07 | <joelr1> | in the paste |
| 15:11:27 | <joelr1> | i.e. one that has choice [ try $ float, integer ] |
| 15:11:45 | <joelr1> | it would seem to me that it would try float first and then integer |
| 15:12:08 | <joelr1> | this works on Foo, for example |
| 15:12:26 | <joelr1> | *Main> run fooParser "Foo (10.345)" |
| 15:12:27 | <joelr1> | Foo (Num 10.345) |
| 15:12:47 | <joelr1> | it doesn't work for run fooParser "Bar (10, 10.345), though. let me try your version anyhow |
| 15:16:00 | <quicksilver> | joelr1: surely the float will match an int though |
| 15:16:12 | <quicksilver> | joelr1: and just return a floatified version of the int? |
| 15:16:31 | <yaarg> | why did the designer of sed decide to use / as a delimter? |
| 15:16:36 | <yaarg> | i mean really WTF was he thinking? |
| 15:16:53 | <joelr1> | quicksilver: right but there's on constructor that starts with a double |
| 15:17:18 | <joelr1> | i mean bar wants int first ... oh, i see... i think i see |
| 15:17:32 | <joelr1> | it eats up the int as a float and bombs out |
| 15:17:34 | <ibid> | yaarg: i believe it predates sed :) |
| 15:17:49 | <ibid> | yaarg: probably comes from the original ed |
| 15:18:09 | <yaarg> | probably right |
| 15:18:10 | <ibid> | (you do know that grep comes from the ed command g/re/p :) |
| 15:18:24 | <ibid> | (the name, that is) |
| 15:18:39 | <yaarg> | yup |
| 15:19:06 | <robreim> | I can't make changes to my trac proposal. Has this something to do with that big spamming I heard mention of? |
| 15:19:21 | <yaarg> | i just realised i spammed the wrong channel too :) |
| 15:19:39 | <sorear> | robreim: log in as guest, pw guest |
| 15:20:01 | <sorear> | anons can still make changes, it's just registered users that are locked out |
| 15:20:06 | <robreim> | sorear: I'm logged in as myself. Is guest preferable? |
| 15:20:10 | <robreim> | oh ok, thanks :) |
| 15:20:21 | <ibid> | why does that make sense? |
| 15:20:37 | <sorear> | robreim: and nag Igloo (?) to add you to the 'developers' group |
| 15:20:43 | <joelr1> | sorear: what is your sign parser? |
| 15:20:58 | <sorear> | joelr1: it's from the Token source |
| 15:21:08 | <sorear> | ACTION makes sure it is exported |
| 15:21:16 | <joelr1> | sorear: it's not exported... i think |
| 15:21:25 | <Igloo> | robreim: What's your username? |
| 15:21:41 | <sorear> | sign = (char '-' >> return negate) |
| 15:21:41 | <sorear> | <|> (char '+' >> return id) |
| 15:21:41 | <sorear> | <|> return id |
| 15:21:50 | <robreim> | Igloo: robreim |
| 15:21:50 | <sorear> | that shouldn't be too much to copy :) |
| 15:22:20 | <Igloo> | robreim: OK, should work now |
| 15:22:21 | <joelr1> | sorear: right, thanks |
| 15:22:35 | <robreim> | Igloo: yup! Thanks plenty :) |
| 15:24:11 | <joelr1> | sign doesn't work with doubles :( |
| 15:24:15 | <joelr1> | working on it |
| 15:26:19 | <sorear> | joelr1: it does too |
| 15:26:37 | <sorear> | joelr1: you're just running into the fact lambda bound variables are monomorphic :( |
| 15:26:42 | <joelr1> | ugh |
| 15:27:20 | <joelr1> | sorear: so what am i to do? |
| 15:27:26 | <sorear> | you could work around it by having sign return +1 or -1 |
| 15:28:00 | <sorear> | then replace each sg with (fromIntegral sg *) |
| 15:28:08 | <sorear> | fromInteger even |
| 15:28:45 | <joelr1> | right |
| 15:28:55 | <joelr1> | trying |
| 15:32:10 | <joelr1> | :t either |
| 15:32:14 | <lambdabot> | forall a c b. (a -> c) -> (b -> c) -> Either a b -> c |
| 15:32:33 | <sorear> | @djinn (a -> c) -> (b -> c) -> Either a b -> c |
| 15:32:34 | <lambdabot> | f a b c = |
| 15:32:34 | <lambdabot> | case c of |
| 15:32:34 | <lambdabot> | Left d -> a d |
| 15:32:34 | <lambdabot> | Right e -> b e |
| 15:32:52 | <sorear> | @src either |
| 15:32:52 | <lambdabot> | either f _ (Left x) = f x |
| 15:32:52 | <lambdabot> | either _ g (Right y) = g y |
| 15:33:09 | <sorear> | probably that last is the clearest :)j |
| 15:34:13 | <joelr1> | sorear: i like the 1st one, actually. so either takes two functions and one either a b returned by naturalOrFloat |
| 15:34:27 | <joelr1> | :t naturalOrFloat |
| 15:34:29 | <lambdabot> | Not in scope: `naturalOrFloat' |
| 15:34:37 | <joelr1> | @where naturalOrFloat |
| 15:34:37 | <lambdabot> | I know nothing about naturalorfloat. |
| 15:34:47 | <sorear> | :t Text.ParserCombinators.Parsec.Token.natOrFloat |
| 15:34:49 | <lambdabot> | Not in scope: `Text.ParserCombinators.Parsec.Token.natOrFloat' |
| 15:34:53 | <sorear> | :t Text.ParserCombinators.Parsec.Token.naturalOrFloat |
| 15:34:55 | <lambdabot> | forall st. Text.ParserCombinators.Parsec.Token.TokenParser st -> Text.ParserCombinators.Parsec.Char.CharParser st (Either Integer Double) |
| 15:35:45 | <quicksilver> | sorear: would be a small-but-cute enhancement to djinn |
| 15:35:56 | <quicksilver> | sorear: for it to rewrite top-level cases as multiple definitions |
| 15:36:34 | <sjanssen> | top level cases on a simple variable |
| 15:36:53 | <sorear> | hehe. djinn was given to us by a hacker of far greature stature than I |
| 15:37:09 | <sjanssen> | sorear: but it'd be great fun to send a patch to lennart! |
| 15:37:09 | <sorear> | or were you not implying I wrote it? |
| 15:38:50 | <tsp> | > "haskell är en Program som körs" |
| 15:38:52 | <lambdabot> | "haskell \228r en Program som k\246rs" |
| 15:39:20 | <earthy> | haskell is a program with what now? |
| 15:39:24 | <tsp> | no idea |
| 15:39:34 | <mauke> | > "€" |
| 15:39:35 | <lambdabot> | "\8364" |
| 15:39:41 | <sorear> | > "?" |
| 15:39:43 | <lambdabot> | "?" |
| 15:39:50 | <tsp> | same thing here - a ? |
| 15:39:55 | <quicksilver> | sorear: I didn't intend to imply you wrote it. Just you had contrasted the two definitions |
| 15:40:03 | <earthy> | > "\8364" |
| 15:40:04 | <lambdabot> | "\8364" |
| 15:40:08 | <earthy> | pity. :) |
| 15:40:11 | <mauke> | U+20AC (0xe2 0x82 0xac): EURO SIGN [€] |
| 15:40:26 | <tsp> | what is a program som körs? |
| 15:40:27 | <sjanssen> | since when did lambdabot learn UTF-8? |
| 15:40:47 | <sorear> | sjanssen: it doesn' |
| 15:41:04 | <sorear> | sjanssen: it passes an uninterpreted string of bytes to GHCI 6.6 |
| 15:41:12 | <sorear> | sjanssen: which does know utf8 |
| 15:41:32 | <sorear> | well, GHC actually |
| 15:41:46 | <chessguy> | ?version |
| 15:41:47 | <lambdabot> | lambdabot 4p517, GHC 6.6 (OpenBSD i386) |
| 15:41:47 | <lambdabot> | darcs get http://www.cse.unsw.edu.au/~dons/lambdabot |
| 15:42:23 | <tsp> | OpenBSD? yuck |
| 15:42:26 | <tsp> | freebsd ftw |
| 15:42:26 | <glguy> | lol |
| 15:42:30 | <hpaste> | joelr1 annotated "simple parser not working" with "a generated parser issue?" at http://hpaste.org/1265#a3 |
| 15:42:54 | <joelr1> | sorear: still not working but i'm fighting it |
| 15:44:44 | <sjanssen> | wow, 77 Haskell mailing list posts since I sent to sleep |
| 15:45:31 | <chessguy> | ?vera ftw |
| 15:45:33 | <lambdabot> | No match for "ftw". |
| 15:45:40 | <sorear> | it's all joelr1's fault ;) |
| 15:45:46 | <sorear> | chessguy: for teh win |
| 15:46:01 | <joelr1> | ACTION hides in his cave |
| 15:46:06 | <chessguy> | win? |
| 15:46:21 | <syntaxfree> | I'm trying out common lisp. |
| 15:46:22 | <sorear> | hello syntaxfree! |
| 15:46:26 | <sorear> | yay? |
| 15:46:30 | <tsp> | common lisp? heh |
| 15:46:32 | <syntaxfree> | I feel like I'm cheating on my wife or something. |
| 15:46:39 | <opqdonut> | :DD |
| 15:46:42 | <hpaste> | joelr1 annotated "simple parser not working" with "close but no cigar" at http://hpaste.org/1265#a4 |
| 15:46:48 | <syntaxfree> | I became curious because of the CLOS article that came out on reddit. |
| 15:46:53 | <glguy> | syntaxfree: well, she's probably cheating with lambdabot |
| 15:46:59 | <glguy> | syntaxfree: so it all works out |
| 15:47:02 | <chessguy> | @quote cheat |
| 15:47:02 | <lambdabot> | sebazzz says: * sebazzz sabe que emonk es un negro sucio y cheatero |
| 15:47:16 | <syntaxfree> | the "generic dispatch" thing is actually very very close to what I'm used to in gnu r. |
| 15:47:16 | <chessguy> | uh |
| 15:47:21 | <syntaxfree> | but no lazy evaluation is weird! |
| 15:47:21 | <joelr1> | syntaxfree: set your feelings free |
| 15:47:36 | <greenrd> | Is it my overactive imagination, or did someone once announce a library to make it easier to use the GHC API? |
| 15:47:38 | <syntaxfree> | anyway, hello everyone! |
| 15:47:40 | <chessguy> | i don't think that quote should be in there |
| 15:47:43 | <joelr1> | syntaxfree: i'm a language whore although haskell is becoming my favorite |
| 15:47:46 | <sorear> | greenrd: hs-plugins |
| 15:47:51 | <int-e> | joelr1: you're using the result of sign with two different types I think; use fromIntegral sign in the Num case? |
| 15:47:53 | <syntaxfree> | hey sorear :) |
| 15:48:04 | <sorear> | @remember syntaxfree I'm trying out common lisp. ... I feel like I'm cheating on my wife or something. |
| 15:48:04 | <lambdabot> | Done. |
| 15:48:09 | <greenrd> | sorear: I don't think it's that I'm thinking of. hs-plugins is for loading code dynamically, right? |
| 15:48:19 | <syntaxfree> | (defun g (x y) (setq x (+ y x))) |
| 15:48:23 | <sjanssen> | greenrd: I vaguely remember that announcement |
| 15:48:25 | <joelr1> | int-e: let me try |
| 15:48:25 | <greenrd> | I've found lemmih's ghc-api library, but that's not what I'm thinking of - I think it's just a snapshot of the ghc api |
| 15:48:29 | <sorear> | greenrd: yeah, but it uses ghc-plugis |
| 15:48:39 | <sorear> | er, ghc-api |
| 15:48:39 | <syntaxfree> | (defun dup (f i) (funcall f i i)) |
| 15:48:40 | <greenrd> | ok |
| 15:48:44 | <mauke> | syntaxfree: (defun g (x y) (incf x y)) |
| 15:48:59 | <chessguy> | is this #haskell or #lisp? |
| 15:48:59 | <greenrd> | but I'm actually interested in parsing and pretty printing Haskell code with all ghc extensions enabled |
| 15:49:03 | <greenrd> | hehe |
| 15:49:11 | <sorear> | syntaxfree: don't you just love languages where funcall is explicit? :) |
| 15:49:17 | <syntaxfree> | guess what (dup #'+ (g x x)) yields? |
| 15:49:26 | <syntaxfree> | sorear: I prefer the scheme way, I think. |
| 15:49:35 | <emu> | (setq x (+ y x)) is pointless (sorry) |
| 15:49:45 | <glguy> | incf? |
| 15:49:46 | <emu> | take it to #lisp |
| 15:49:54 | <chessguy> | yeah, anyway |
| 15:49:56 | <joelr1> | int-e: doesn't work |
| 15:50:02 | <syntaxfree> | I'm just bitching about "no lazy evaluation". |
| 15:50:12 | <syntaxfree> | It's meant for haskellers to feel all smug and superior. |
| 15:50:27 | <emu> | bitch about the lack in scheme |
| 15:50:39 | <emu> | CLers love their side-effects |
| 15:51:03 | <sjanssen> | most Haskell features are meant to make us feel smug and superior |
| 15:51:17 | <joelr1> | int-e: Expected type: Either Integer Integer |
| 15:51:18 | <greenrd> | hehe |
| 15:51:20 | <chessguy> | that's gotta be even weirder than african tribal people who stick bones through their noses |
| 15:51:36 | <glguy> | ACTION wouldn't mind learning a lisp dialect that didn't have crappy naming conventions |
| 15:51:53 | <sebell> | glguy: What's wrong with Scheme? |
| 15:51:58 | <emu> | hey dontcha love nthcdr or rplacd |
| 15:52:20 | <syntaxfree> | I thought CLers loved their macros. |
| 15:52:27 | <emu> | that too |
| 15:52:27 | <sorear> | emu: "If I were to do it again? I would spell creat with an e." |
| 15:52:32 | <emu> | hehe |
| 15:52:38 | <syntaxfree> | In the end, it strikes me that I'm losing out by knowing just one programming language. |
| 15:52:40 | <joelr1> | i feel particularly stupid today |
| 15:52:55 | <mauke> | yeah, a common function like rplacd needs a short name, while a seldom used low-level function like multiple-value-bind ... er |
| 15:53:01 | <syntaxfree> | If I'm going for a mistress, I might as well choose Lisp instead of Cobol or something. |
| 15:53:02 | <sorear> | ACTION has done memcpy(dst, dst, sizeof src) in haskell |
| 15:53:16 | <joelr1> | mind you, i wrote a fully-functioning translator from one language to another in a month, while learning ocaml |
| 15:53:23 | <joelr1> | what does that say about haskell?! |
| 15:53:27 | <int-e> | joelr1: what is the type of naturalOrFloat? |
| 15:53:32 | <emu> | mauke: except no one uses rplacd anymore =) |
| 15:53:37 | <sorear> | :t Text.ParserCombinators.Parsec.Token.naturalOrFloat |
| 15:53:38 | <joelr1> | int-e: one sec |
| 15:53:39 | <hpaste> | DIARULE annotated "(no title)" with "NEW YORK" at http://hpaste.org/1109#a4 |
| 15:53:39 | <lambdabot> | forall st. Text.ParserCombinators.Parsec.Token.TokenParser st -> Text.ParserCombinators.Parsec.Char.CharParser st (Either Integer Double) |
| 15:53:44 | <emu> | it mostly exists for the purposes of dredging up funny names |
| 15:53:45 | <sorear> | int-e: ^^ |
| 15:53:48 | <syntaxfree> | learning a ML would really be cheating. |
| 15:54:03 | <mauke> | rplaca, cdadr, ... |
| 15:54:09 | <syntaxfree> | why would I learn another functional typeful language? What's wrong with the one I'd have? That'd be vicious thrill-seeking. |
| 15:54:14 | <syntaxfree> | Oh, c'mon. I love cadaddr. |
| 15:54:18 | <hpaste> | joelr1 annotated "simple parser not working" with "not working" at http://hpaste.org/1265#a5 |
| 15:54:26 | <joelr1> | there |
| 15:54:35 | <sorear> | syntaxfree: they only go up to 4, for no good reason :) |
| 15:54:37 | <emu> | syntaxfree: actually, only up to 3 (a|d) are defined |
| 15:54:49 | <emu> | in scheme it is special cased for any number |
| 15:55:07 | <mauke> | what? |
| 15:55:10 | <sorear> | Oh. ISTR getting that 4 from r5rs. |
| 15:55:12 | <mauke> | I don't want to learn scheme anymore :( |
| 15:55:19 | <sjanssen> | joelr1: what is the type of naturalOrFloat? |
| 15:55:23 | <sorear> | :t Text.ParserCombinators.Parsec.Token.naturalOrFloat |
| 15:55:24 | <cjeris> | emu: scheme has an infinite number of symbols bound in the initial environment??? |
| 15:55:26 | <lambdabot> | forall st. Text.ParserCombinators.Parsec.Token.TokenParser st -> Text.ParserCombinators.Parsec.Char.CharParser st (Either Integer Double) |
| 15:55:27 | <sorear> | sjanssen: ^^ |
| 15:55:40 | <joelr1> | i dig a fair bit of lisp programming, on two projects. ran away from lisp afterwards. |
| 15:55:45 | <joelr1> | no, make that 3 projects |
| 15:55:46 | <int-e> | joelr1: I think you want either (Int . (sign *)) (Num . (fromIntegral sign *)) num |
| 15:55:46 | <emu> | i was pretty sure it was scheme. maybe just mzscheme. but scheme doesnt have symbols like lisp. |
| 15:55:46 | <cjeris> | sorear: i think it's 4 in Common Lisp. |
| 15:55:54 | <joelr1> | that's excluding those of my own |
| 15:55:59 | <sjanssen> | @hoogle lexeme |
| 15:55:59 | <lambdabot> | Text.ParserCombinators.Parsec.Token.lexeme :: TokenParser st -> CharParser st a -> CharParser st a |
| 15:55:59 | <lambdabot> | Text.Read.Lexeme :: data Lexeme |
| 15:56:06 | <sjanssen> | @hoogle sign |
| 15:56:06 | <lambdabot> | Prelude.significand :: RealFloat a => a -> a |
| 15:56:07 | <lambdabot> | Prelude.signum :: Num a => a -> a |
| 15:56:07 | <lambdabot> | Control.Concurrent.QSem.signalQSem :: QSem -> IO () |
| 15:56:18 | <joelr1> | int-e: trying |
| 15:56:18 | <int-e> | sjanssen: sign is defined in the paste |
| 15:56:21 | <sorear> | int-e: i had intended for him to use (fromInteger sign *) in the Num case |
| 15:56:23 | <emu> | my job is work in CL. but anyway, i have to run. |
| 15:56:24 | <mauke> | hmm, I should write an Acme::CADR |
| 15:56:25 | <sebell> | emu: What do you mean, doesn't have symbols like CL? |
| 15:56:28 | <quicksilver> | XLISP had cadadrs up to 7, if I'm not mistaken |
| 15:56:28 | <int-e> | sorear: me too |
| 15:56:33 | <hpaste> | sjanssen annotated "simple parser not working" with "should be fixed" at http://hpaste.org/1265#a6 |
| 15:56:39 | <joelr1> | finally! the magic moment |
| 15:56:41 | <int-e> | sorear: well, fromIntegral but that's a small difference :) |
| 15:56:48 | <joelr1> | int-e and sjanssen win |
| 15:56:52 | <dhart> | How does one combine Monads togther? |
| 15:57:08 | <joelr1> | that's one ugly little parser, though |
| 15:57:10 | <joelr1> | dammit |
| 15:57:15 | <joelr1> | lets see if it works now |
| 15:57:17 | <sorear> | dhart: you can't |
| 15:57:22 | <syntaxfree> | with monad combinators! \o/ |
| 15:57:40 | <sorear> | dhart: values of a monad can be combined with >>=, and functions derived from it like replicateM |
| 15:57:41 | <dhart> | sorear: I thought it was possible? Monad combinators? |
| 15:58:08 | <Cale> | dhart: By combining monads, you do mean the types, right? |
| 15:58:35 | <Cale> | There's a technique which is commonly used called monad transformers to build up monads from pieces. |
| 15:58:39 | <hpaste> | joelr1 annotated "simple parser not working" with "yay! i can move on now!" at http://hpaste.org/1265#a7 |
| 15:58:42 | <syntaxfree> | I'm joking about monad combinators. |
| 15:58:44 | <syntaxfree> | sorry. |
| 15:58:45 | <joelr1> | thank you folks |
| 15:58:47 | <dhart> | Let me introduce the simple example I am trying to implement, to make things clearer |
| 15:58:51 | <dhart> | <dhart> class Renderable a where |
| 15:58:51 | <dhart> | <dhart> render :: a -> IO () |
| 15:59:02 | <Cale> | (But you can't automatically construct a monad transformer from a monad.) |
| 15:59:22 | <dhart> | I want to add a "Reader RenderParams" and a "State GameState" |
| 15:59:28 | <Cale> | (I might not be around long, I haven't had any sleep) |
| 16:00:02 | <sorear> | Oh, then just use ReaderT RenderParams (State GameState) as your monad |
| 16:00:16 | <sorear> | the order matters in general, but not in this case |
| 16:00:17 | <Cale> | dhart: You can: newtype MyMonad a = MM (ReaderT RenderParams (StateT GameState IO) a) |
| 16:00:27 | <Cale> | deriving (Functor, Monad) |
| 16:00:49 | <dhart> | Why "ReaderT" and not "Reader", same question for "State" and "StateT" |
| 16:00:57 | <joelr1> | as for why i ran away from lisp... one project was a new one and i totally hated my boss's technique since he wrote lisp c-style. the next one was ITA and just grokking the code was difficult. you couldn't just run a bunch of unit tests to see what breaks. HUGE codebase. then, i joined another project and had to refactor. again, impossible to change and see what breaks, no unit tests and dynamic typing. argh! |
| 16:01:01 | <sorear> | @kind ReaderT |
| 16:01:04 | <lambdabot> | * -> (* -> *) -> * -> * |
| 16:01:04 | <sorear> | @kind Reader |
| 16:01:05 | <Cale> | Because they're transforming the underlying monad. |
| 16:01:06 | <lambdabot> | * -> * -> * |
| 16:01:10 | <hpaste> | sjanssen annotated "simple parser not working" with "reduce the ugly" at http://hpaste.org/1265#a8 |
| 16:01:13 | <joelr1> | ocaml is fast and cool but haskell is wicked! |
| 16:01:21 | <Cale> | ReaderT is a thing which adds reader-like functionality to another monad. |
| 16:01:23 | <sorear> | as you se ReaderT takes an extra parameter |
| 16:01:35 | <dhart> | right |
| 16:01:39 | <sorear> | (kinds are to types as types are to values) |
| 16:01:49 | <joelr1> | sjanssen: you cheated! you just removed the parens :D |
| 16:02:13 | <joelr1> | err, the curly braces, etc |
| 16:02:32 | <joelr1> | how do i remember my own quote? |
| 16:02:33 | <sjanssen> | yes, cuz they're ugly ;) |
| 16:02:46 | <joelr1> | sjanssen: nah, i'm taking to that style nowadays |
| 16:02:47 | <Cale> | Newtype-deriving is a simple Haskell extension which lets you derive any class for your newtype that the old type supports. |
| 16:02:50 | <glguy> | {; } is good for one thing: one-liners for lambdabot |
| 16:02:53 | <Cale> | (well, almost any class) |
| 16:03:03 | <sjanssen> | joelr1: how come? |
| 16:03:05 | <sorear> | @remember joelr1 foo bar? |
| 16:03:06 | <glguy> | (also for machine generated code) |
| 16:03:12 | <joelr1> | oh, c'mon you guys |
| 16:03:12 | <glguy> | but not for normal human use! |
| 16:03:27 | <joelr1> | @remember joelr1 ocaml is fast and cool but haskell is wicked! |
| 16:03:27 | <lambdabot> | Done. |
| 16:03:30 | <joelr1> | there |
| 16:03:52 | <glguy> | ACTION reminds the insane masses not to quote themselves :-p |
| 16:04:14 | <Cale> | dhart: Because the types get so long and complicated, and it's easy to abuse/get confused about the layering which occurs when building up monads with monad transformers, it's a good idea to newtype the whole stack and then write your own run function. |
| 16:04:21 | <joelr1> | sjanssen: dunno why. i'll try reformatting some code to see. i have also taken to putting commas and semis before the statements |
| 16:04:46 | <sebell> | joelr1: I agree, that's gross :) |
| 16:04:56 | <joelr1> | Cale: what about putting the whole stack into a monad of its own? |
| 16:05:09 | <joelr1> | sebell: do { ... } is gross? or putting commas in front? |
| 16:05:09 | <dhart> | Cale and sorear: Thx for the help. I'll try to play around with the newtype you provided le Cale. |
| 16:05:13 | <sjanssen> | joelr1: yes, that style is common in Haskell. It makes it easier to rearrange the order of lines |
| 16:05:24 | <Cale> | joelr1: Well, it is a monad already -- you could smash it out by hand, but there's not a whole lot of point to that usually. |
| 16:05:56 | <joelr1> | Cale: i remember having a lot of trouble with lifting when i had ErrorT in the middle of the stack |
| 16:06:05 | <joelr1> | Cale: so i guess my point would be to simplify the lifting? |
| 16:06:31 | <Cale> | Well, what you should normally do is what I was about to describe... |
| 16:06:52 | <Cale> | You can either derive MonadState GameState and MonadReader RenderParams, or you can provide your own interfaces to their functionality. |
| 16:07:35 | <sebell> | joelr1: do { ; } -- especially if you're not binding |
| 16:07:48 | <joelr1> | Cale: is there an example of doing so? could you add a tiny bit to the wiki maybe? |
| 16:07:55 | <Cale> | Providing your own names for get and put, etc in your monad is generally a good idea, because the names can help the library user understand what it is that's being manipulated, and if the monad representation ever has to change, you only need to change the operations in one place. |
| 16:08:11 | <joelr1> | sebell: well, yes, i agree that one-liners are cooler with >>= |
| 16:08:34 | <dhart> | Cale: I got this error with the newtype you gave me: Can't make a derived instance of `Monad GameMonad' |
| 16:08:34 | <dhart> | (`Monad' is not a derivable class |
| 16:08:50 | <Cale> | dhart: Which Haskell implementation are you using? |
| 16:09:02 | <Cale> | -fglasgow-exts ? |
| 16:09:08 | <dhart> | I'm compiling with GHC 6.6 |
| 16:09:29 | <Cale> | okay, then add {-# OPTIONS_GHC -fglasgow-exts #-} to the top of your file |
| 16:09:37 | <kmg> | @pl \x -> (rem x 3 == 0) || (rem x 5 == 0) |
| 16:09:38 | <lambdabot> | ap ((||) . (0 ==) . flip rem 3) ((0 ==) . flip rem 5) |
| 16:09:41 | <Cale> | which should turn on newtype deriving |
| 16:09:58 | <dhart> | It worked with that compile option, thx a lot. |
| 16:10:20 | <dhart> | I like to understand what I did. What is the difference brought by that compile option? |
| 16:10:39 | <Cale> | It turns on all the features of GHC beyond the Haskell 98 standard. |
| 16:10:40 | <kmg> | @pl \x -> x <= 10^6 |
| 16:10:40 | <lambdabot> | (<= 10 ^ 6) |
| 16:10:51 | <Cale> | Well, all the non-dangerous ones, anyway. |
| 16:11:08 | <dhart> | I didn't know that newtype could not derive in H98 |
| 16:11:24 | <Cale> | It can derive only the usual things like Read and Show in H98. |
| 16:11:31 | <kmg> | @pl \x -> rem x 2 == 0 |
| 16:11:32 | <lambdabot> | (0 ==) . flip rem 2 |
| 16:11:39 | <joelr1> | sorear: my next step would be to derive a pretty printer |
| 16:12:07 | <joelr1> | sorear: the inverse of the derived parser. just so that i could print the constructor name and then the args within parens and separated by comma |
| 16:12:25 | <kmg> | @pl \y -> y*y |
| 16:12:26 | <lambdabot> | join (*) |
| 16:13:54 | <dhart> | Cale: I'm still confused as to what the run function is used for in Monads: is it for executing the list of actions? In that case, how would one define and implement a run method for my new monad? |
| 16:14:13 | <joelr1> | sorear: i'm sure it's gona be a struggle but i really like working with derive so far. as opposed to dealing with naked TH |
| 16:15:24 | <Cale> | dhart: Well, when you're working with monadic values, you think of those as abstract computations waiting to be run. They often need something extra -- like an initial state or environment, in order to begin working. |
| 16:16:12 | <Cale> | (We know that internally, State and Reader are implemented as ordinary functions, but that's an implementation detail) |
| 16:17:06 | <Cale> | runState, given a State computation, and an initial state, runs the computation using the initial state, and gives a resulting value together with the final state. |
| 16:17:44 | <dhart> | I know I have to give my custom monad the starting ReaderT state (RenderParams) and a starting StateT state (GameState), and that it should return the finale GameState + some extra computation values |
| 16:18:30 | <dhart> | runGameMonad :: r -> s -> (s',a) -- but I think I'm missing something there |
| 16:18:38 | <sorear> | @tell ndm [Apr 5 16:14 UTC 2007] <joelr1> sorear: I'm sure it's gonna be a struggle but I really like working with derive so far; esp. as opposed to dealing with naked TH. |
| 16:18:38 | <lambdabot> | Consider it noted. |
| 16:18:50 | <Cale> | Right, so runMyMonad :: MyMonad a -> RenderParms -> GameState -> IO (a, GameState) |
| 16:18:52 | <Cale> | perhaps |
| 16:18:53 | <sorear> | he'll be pleased when he gets back :) |
| 16:19:06 | <joelr1> | dhart: did you look at the code for ... what was that game's name? implemented using Yampa |
| 16:19:10 | <Cale> | Note that we can't escape IO here. |
| 16:19:13 | <sorear> | joelr1: frag |
| 16:19:22 | <joelr1> | yep, frag |
| 16:19:27 | <joelr1> | dhart: did you look at frag? |
| 16:19:27 | <Cale> | So we'll actually be producing an IO action. |
| 16:19:50 | <joelr1> | sorear: i'm about to save hundreds of lines of code |
| 16:19:53 | <joelr1> | parsing code |
| 16:19:57 | <dhart> | joelr1: I did look at frag but failed to compile it succesfully |
| 16:20:18 | <joelr1> | dhart: ok, then you can see how they are doing state, etc |
| 16:20:33 | <dhart> | Cale: it's a good thing we cant escape IO here as the function calling my render function expects an IO back |
| 16:21:43 | <Cale> | (of course, you probably want to find a more creative name than MyMonad :) |
| 16:22:02 | <Cale> | If you want more detail about how I think monad transformers are best used... |
| 16:22:20 | <Cale> | http://cale.yi.org/index.php/How_To_Use_Monad_Transformers |
| 16:22:23 | <lambdabot> | Title: How To Use Monad Transformers - CaleWiki |
| 16:23:03 | <Cale> | I'm afraid I aimed that article a little higher than at the complete monad transformer beginner, but hopefully it's not too bad. |
| 16:23:30 | <chessguy> | i was wondering when Cale would trot out that link :) |
| 16:23:41 | <sorear> | is yi.org named after tuomov/dons/jyp's yi? |
| 16:24:01 | <Cale> | chessguy: I've finally begun to package up the things where I've repeated myself over and over in the past :) |
| 16:24:08 | <chessguy> | hehe |
| 16:24:19 | <dmwit> | Does anybody have some suggested reading for Typeable? |
| 16:24:32 | <sorear> | @google Typing dynamic typing |
| 16:24:36 | <lambdabot> | http://www.cs.uu.nl/~arthurb/dynamic.html |
| 16:24:37 | <lambdabot> | Title: Typing Dynamic Typing |
| 16:24:37 | <sorear> | iirc |
| 16:25:02 | <dmwit> | Awesome, thanks! |
| 16:25:59 | <dhart> | Cale: I have pmed you if you dont mind. |
| 16:26:03 | <sorear> | dmwit: no, that's not the write paper |
| 16:26:09 | <Cale> | dhart: no you haven't |
| 16:26:15 | <sorear> | dhart: check your sever messages |
| 16:26:19 | <Cale> | dhart: You're probably not registered. |
| 16:26:20 | <sorear> | <*dhart> Cale: I have pmed you if you dont mind. |
| 16:26:29 | <dhart> | Cale: right |
| 16:26:30 | <sorear> | the * means not-identified |
| 16:26:36 | <dhart> | Cale: must not be registered |
| 16:26:43 | <dmwit> | Hmmm... this paper seems to be about Dynamic, not Typeable. |
| 16:26:49 | <sorear> | only registered users are allowed to send pms |
| 16:26:57 | <Cale> | sorear: My client doesn't show that marker. |
| 16:27:04 | <glguy> | what marker? |
| 16:27:08 | <sorear> | dmwit: actually it's about neither, I pulled the wrong name |
| 16:27:11 | <dhart> | ok, so this is my current runGameMonad implementation: |
| 16:27:12 | <dhart> | runGameMonad :: GameMonad a -> RenderParms -> GameState -> IO (a, GameState) |
| 16:27:12 | <dhart> | runGameMonad gm rp gs = |
| 16:27:12 | <dhart> | let (a, newState) = runStateT gs in |
| 16:27:12 | <dhart> | let b = runReaderT rp in |
| 16:27:26 | <dhart> | but I'm really stuck due to all this Monadic layering. |
| 16:27:35 | <sorear> | Cale: apparantly your client has to explicitly request it |
| 16:27:44 | <sorear> | Cale: i'm using erc-capab-module |
| 16:28:28 | <Cale> | runGameMonad (GM x) rp gs = runStateT gs (runReaderT rp x) |
| 16:28:32 | <sorear> | dmwit: the first "Scrap your boilerplate" paper has a section on typeable |
| 16:28:35 | <Cale> | I think that'll do it. |
| 16:28:53 | <sorear> | @google scrap your bolrod |
| 16:28:55 | <lambdabot> | http://tunes.org/~nef/logs/haskell/06.03.15 |
| 16:28:57 | <Cale> | er |
| 16:28:58 | <sorear> | @google scrap your boilerplate |
| 16:29:00 | <lambdabot> | http://www.cs.vu.nl/boilerplate/ |
| 16:29:00 | <lambdabot> | Title: Scrap your boilerplate ... in Haskell |
| 16:29:04 | <Cale> | modulo bad parameter order |
| 16:29:13 | <Cale> | er... |
| 16:29:21 | <Cale> | never mind, I'm tired :) |
| 16:29:27 | <dhart> | Cale: I'm baffled. I still have so much to learn about Haskell |
| 16:29:32 | <chessguy> | bolrod? |
| 16:29:36 | <dhart> | Cale: thx, I'll try it out |
| 16:29:39 | <Cale> | runGameMonad (GM x) rp gs = runStateT (runReaderT x rp) gs |
| 16:29:42 | <Cale> | I think |
| 16:29:46 | <Cale> | @type runReaderT |
| 16:29:48 | <lambdabot> | forall r (m :: * -> *) a. ReaderT r m a -> r -> m a |
| 16:29:52 | <Cale> | yes. |
| 16:30:04 | <Cale> | GM is the newtype constructor for your newtype. |
| 16:30:14 | <dmwit> | sorear: Thanks again. |
| 16:30:29 | <Cale> | If you keep it within the module where you're defining your monad, and don' |
| 16:30:33 | <Cale> | and don't export it |
| 16:30:34 | <sorear> | Thu Mar 8 08:32:47 PST 2007 Ian Lynagh <igloo@earth.li> |
| 16:30:34 | <sorear> | * Build the libraries with cabal |
| 16:30:41 | <sorear> | Igloo++ Igloo++ Igloo++ |
| 16:30:46 | <sorear> | fptools must die. |
| 16:30:49 | <sorear> | @flush |
| 16:30:57 | <Cale> | then you'll have control over which parts of the application can define new operations for your GameMonad |
| 16:31:11 | <Cale> | You might want to derive MonadIO though |
| 16:31:29 | <Cale> | Which will let you liftIO any IO action, making it a GameMonad action. |
| 16:33:50 | <dhart> | why do you provide 2 arguments to runStateT? |
| 16:33:52 | <chessguy> | ?hoogle lift |
| 16:33:52 | <lambdabot> | Control.Monad.Trans.lift :: (MonadTrans t, Monad m) => m a -> t m a |
| 16:33:53 | <lambdabot> | Text.ParserCombinators.ReadPrec.lift :: ReadP a -> ReadPrec a |
| 16:33:53 | <lambdabot> | Language.Haskell.TH.Syntax.lift :: Lift t => t -> Q Exp |
| 16:33:59 | <Cale> | dhart: because it takes two |
| 16:34:03 | <chessguy> | @hoogle+ |
| 16:34:03 | <lambdabot> | Language.Haskell.TH.Syntax.Lift :: class Lift t |
| 16:34:04 | <lambdabot> | Monad.liftM :: Monad a => (b -> c) -> a b -> a c |
| 16:34:04 | <lambdabot> | Monad.liftM2 :: Monad a => (b -> c -> d) -> a b -> a c -> a d |
| 16:34:07 | <Cale> | dhart: the computation, and the initial state |
| 16:34:08 | <glguy> | cool, now dhart has a ~ in front of his name for not being identified |
| 16:34:15 | <Cale> | @type runStateT |
| 16:34:17 | <lambdabot> | forall s (m :: * -> *) a. StateT s m a -> s -> m (a, s) |
| 16:34:22 | <dhart> | @hoogle runStateT |
| 16:34:22 | <lambdabot> | Control.Monad.State.runStateT :: StateT s m a -> (s -> m (a, s)) |
| 16:34:24 | <chessguy> | ?where hoogle |
| 16:34:25 | <lambdabot> | http://www.haskell.org/hoogle |
| 16:34:37 | <dmead_> | > "hello channel" |
| 16:34:38 | <lambdabot> | "hello channel" |
| 16:35:25 | <dhart> | It's weird, my hoogle says "runStateT :: (s -> m (a, s))", but the extra parameter probably comes from the fact the function is inside the monad definition |
| 16:35:37 | <sorear> | yeah |
| 16:35:43 | <sorear> | it's a record selector |
| 16:35:47 | <sorear> | @src StateT |
| 16:35:47 | <lambdabot> | Source not found. Maybe if you used more than just two fingers... |
| 16:35:50 | <sorear> | @src State |
| 16:35:50 | <lambdabot> | Source not found. Just what do you think you're doing Dave? |
| 16:36:01 | <sorear> | ACTION beats up @src |
| 16:36:08 | <dmwit> | lambdabot, his name isn't Dave. |
| 16:36:12 | <chessguy> | why is there a liftIO, but not, say, a liftState |
| 16:36:23 | <dmwit> | :t liftIO |
| 16:36:25 | <lambdabot> | forall a (m :: * -> *). (MonadIO m) => IO a -> m a |
| 16:36:30 | <Cale> | dhart: hoogle suffers from some irritating bugs |
| 16:36:36 | <sorear> | chessguy: you don't need it |
| 16:36:41 | <sorear> | :t put |
| 16:36:43 | <lambdabot> | forall s (m :: * -> *). (MonadState s m) => s -> m () |
| 16:37:11 | <Cale> | dhart: You probably shouldn't trust it to give accurate type signatures for everything, use :t instead. |
| 16:37:12 | <chessguy> | ... |
| 16:37:13 | <sorear> | chessguy: mtl uses undecidable instances to automatically lift get/put into any monad stack containing a State or StateT |
| 16:37:40 | <sorear> | it's IO that is broken, for needing liftIO |
| 16:37:50 | <sorear> | we should have all IO ops auto-lift |
| 16:38:03 | <Cale> | It doesn't *need* liftIO, it's just extremely convenient to have it. |
| 16:38:17 | <Cale> | You could also do lift . lift . lift |
| 16:38:27 | <Cale> | With as many lifts as needed. |
| 16:38:50 | <sorear> | "io theMRSucks = liftIO theMRSucks" ftw |
| 16:38:55 | <dhart> | Cale: thx for everything, I understand it much better now. I'm curious, now that the StateT monad is inside the ReaderT, how conveluted it will be to do "put" and "get" functions? |
| 16:39:07 | <sorear> | value restriction probably sucks more though |
| 16:39:22 | <Cale> | dhart: not at all, because there's an instance which makes them available "through" the ReaderT |
| 16:40:05 | <Cale> | dhart: Though that's a good observation. |
| 16:40:29 | <Cale> | Ordinarily, without that instance, you'd have to use lift in order to get at them. |
| 16:40:50 | <Cale> | lift takes an action in an underlying monad, and gives the equivalent in the transformed monad. |
| 16:40:59 | <Cale> | @type lift |
| 16:41:01 | <lambdabot> | forall (m :: * -> *) a (t :: (* -> *) -> * -> *). (MonadTrans t, Monad m) => m a -> t m a |
| 16:41:22 | <Cale> | m a -> t m a |
| 16:41:44 | <Cale> | @type lift get |
| 16:41:46 | <lambdabot> | forall (m :: * -> *) a (t :: (* -> *) -> * -> *). (MonadTrans t, MonadState a m) => t m a |
| 16:42:03 | <Cale> | @type \x -> lift (put x) |
| 16:42:05 | <lambdabot> | forall s (m :: * -> *) (t :: (* -> *) -> * -> *). (MonadTrans t, MonadState s m) => s -> t m () |
| 16:42:47 | <sorear> | No wonder darcs stopped... |
| 16:43:00 | <sorear> | ACTION wishes flow control was harder to accidentally use |
| 16:43:01 | <Cale> | But that's unnecessary in this case, because there's a "lifting instance" for MonadState for any ReaderT over another MonadState instance. |
| 16:43:20 | <sorear> | @src ReaderT get |
| 16:43:20 | <lambdabot> | Source not found. That's something I cannot allow to happen. |
| 16:43:23 | <sorear> | :( |
| 16:43:46 | <sioraiocht> | @src fix |
| 16:43:47 | <lambdabot> | fix f = let x = f x in x |
| 16:43:56 | <chessguy> | Cale: i definitely agree with your comments in your article about the arbitrary existence of some lifting operations and not others being *very* confusing |
| 16:46:07 | <tsp> | argh, everything's being a Meddelandefältet today |
| 16:46:52 | <Cale> | chessguy: Yes, that was one thing which made monad transformers a lot harder for me to figure out at first. Once I fully realised that was what was going on, things started making a whole lot more sense. |
| 16:47:25 | <Cale> | Part of me would sort of prefer that those instances were left out. |
| 16:47:29 | <dhart> | Cale: nice, I'll check it out first thing tomorow. Now I need to leave work ^^ thx for all the help |
| 16:47:37 | <sorear> | @babel fr en Meddelandefält |
| 16:47:38 | <lambdabot> | Plugin `babel' failed with: IRCRaised Data.ByteString.last: empty ByteString |
| 16:47:42 | <sorear> | :( |
| 16:47:53 | <Cale> | dhart: quite welcome :) |
| 16:47:56 | <Cale> | oop |
| 16:47:58 | <chessguy> | Cale: or maybe require an extra import at least |
| 16:47:59 | <tsp> | it's swedish for something |
| 16:48:05 | <chessguy> | sorear: that's not french |
| 16:48:06 | <Cale> | chessguy: yes! |
| 16:48:15 | <Cale> | import Control.Monad.LiftingInstances |
| 16:48:21 | <tsp> | there's a Meddelandefältet, and a program som körs |
| 16:48:53 | <sorear> | tsp: Oh. So I went through all the trouble of /whois, host, and whois(1) only to discover "Calgary, CA" isn't relevant? |
| 16:48:55 | <sorear> | :) |
| 16:49:14 | <monochrom> | They don't use French in Calgary. |
| 16:49:25 | <sorear> | ACTION has no clue where calgary is, but knows that the only non-english language spoken in any large part of CA is french |
| 16:49:26 | <tsp> | sorear: heh, british columbia, but close enough |
| 16:49:35 | <tsp> | I hate french |
| 16:49:38 | <tsp> | mi parolas Esperanton |
| 16:49:59 | <monochrom> | Western Canada is very English and also pro-US actually. |
| 16:50:14 | <chessguy> | what a bunch of weirdos |
| 16:50:21 | <Cale> | Boooo US. |
| 16:50:35 | <sorear> | tsp: Calgary is from the snail-mail contact address for your ISP, it isn't specifically tagged as referring to you. |
| 16:50:40 | <tsp> | heh, french can manges la merde francaise for all I'm concerned |
| 16:50:42 | <Cale> | (I live in Ontario) |
| 16:50:42 | <joelr1> | :t >>= |
| 16:50:43 | <[malte]> | tsp, ankaÅ vi? tamen mi dezirus pli bone paroli Haskelon |
| 16:50:44 | <lambdabot> | parse error on input `>>=' |
| 16:51:03 | <chessguy> | :t (>>=) |
| 16:51:05 | <lambdabot> | forall (m :: * -> *) a b. (Monad m) => m a -> (a -> m b) -> m b |
| 16:51:08 | <sebell> | :t (>>=) |
| 16:51:10 | <lambdabot> | forall (m :: * -> *) a b. (Monad m) => m a -> (a -> m b) -> m b |
| 16:51:26 | <joelr1> | aha |
| 16:51:28 | <joelr1> | :t (>>) |
| 16:51:31 | <lambdabot> | forall (m :: * -> *) a b. (Monad m) => m a -> m b -> m b |
| 16:51:41 | <tsp> | [malte]: mi lernas esperanton :) Cxu vi parolas bona haskelon? |
| 16:51:43 | <chessguy> | @hoogle m a -> (a -> b) |
| 16:51:44 | <lambdabot> | No matches, try a more general search |
| 16:51:58 | <tsp> | I seriously need an utf-8 client |
| 16:52:07 | <[malte]> | tsp, ni daÅrigu en #esperanto |
| 16:52:18 | <chessguy> | @hoogle (a -> b) -> m a |
| 16:52:18 | <lambdabot> | Prelude.($!) :: (a -> b) -> a -> b |
| 16:52:19 | <lambdabot> | Prelude.($) :: (a -> b) -> a -> b |
| 16:52:19 | <lambdabot> | Control.Monad.Error.catchError :: MonadError e m => m a -> (e -> m a) -> m a |
| 16:52:28 | <chessguy> | @hoogle (a -> b) -> m a -> m b |
| 16:52:29 | <lambdabot> | Prelude.($!) :: (a -> b) -> a -> b |
| 16:52:29 | <lambdabot> | Prelude.($) :: (a -> b) -> a -> b |
| 16:52:29 | <lambdabot> | Prelude.maybe :: b -> (a -> b) -> Maybe a -> b |
| 16:55:21 | <chessguy> | ?where ndm |
| 16:55:22 | <lambdabot> | http://www.cs.york.ac.uk/~ndm/ |
| 16:55:59 | <sorear> | neil d. mitchell, phd student, yhc hacker, #haskeller |
| 16:56:25 | <daniel_larsson> | "meddelandefält" = message field |
| 16:56:26 | <daniel_larsson> | "Programmet som körs" = The program that's running |
| 16:56:31 | <sorear> | gone for a week |
| 16:56:54 | <chessguy> | yeah, i was actually looking for his homepage |
| 16:57:12 | <daniel_larsson> | tsp: above, if you're still confused :) |
| 16:57:30 | <ozzilee> | Can someone help me with some really newbie questions? First, how do I get out of GHCi? |
| 16:57:47 | <Igloo> | :q |
| 16:59:18 | <Cale> | Or ctrl-c |
| 16:59:20 | <chessguy> | ozzilee: many commands to ghci involve a colon, followed by a letter, as just demonstrated |
| 16:59:21 | <sorear> | Igloo: when I pulled your most recent 4 patches, darcs died with an internal consistency error |
| 16:59:28 | <sorear> | Cale: only on windows |
| 16:59:33 | <sorear> | :? |
| 16:59:34 | <Cale> | sorear: huh? |
| 16:59:47 | <Cale> | sorear: ctrl-c works for me on ubuntu |
| 17:00:03 | <Cale> | er... |
| 17:00:05 | <daniel_larsson> | Cale: not on my ubuntu. Ctrl-D works fine though |
| 17:00:07 | <Cale> | sorry, ctrl-d |
| 17:00:12 | <Cale> | yeah |
| 17:00:12 | <ozzilee> | ctrl-c doesn't work on os x either |
| 17:00:14 | <sorear> | Cale: on windows ghci dies messily on ctrl-c, on (Debian sid) it stops computation, displays "Interrupted", adn returns to the repl |
| 17:00:28 | <ozzilee> | crtl-d does though, thanks |
| 17:00:32 | <glguy> | : reloads the current file |
| 17:00:47 | <sorear> | ozzilee: type :? <ENTER> |
| 17:00:50 | <Cale> | ACTION wonders how long he's been "awake" |
| 17:00:54 | <sorear> | read what it says |
| 17:01:55 | <ozzilee> | sorear: thanks |
| 17:01:56 | <Cale> | Hmm, 23 hours. |
| 17:02:04 | <joelr1> | can this be made shorter? |
| 17:02:05 | <joelr1> | do { x <- foo; y <- bar; return $ N x y } |
| 17:02:17 | <joelr1> | :t liftM2 |
| 17:02:19 | <lambdabot> | forall a1 a2 r (m :: * -> *). (Monad m) => (a1 -> a2 -> r) -> m a1 -> m a2 -> m r |
| 17:02:20 | <Cale> | joelr1: liftM2 N foo bar |
| 17:02:26 | <joelr1> | argh |
| 17:02:32 | <joelr1> | i'm an idiot. thanks cale |
| 17:03:57 | <narain> | :t do {x <- foo; y <- bar; return $ N x y} |
| 17:04:00 | <lambdabot> | Not in scope: `foo' |
| 17:04:00 | <lambdabot> | |
| 17:04:00 | <lambdabot> | <interactive>:1:19: Not in scope: `bar' |
| 17:04:18 | <narain> | ignore that |
| 17:05:38 | <ozzilee> | Ok, I can declare a function "foo" with "let foo x = x + 1", is this correct? |
| 17:05:45 | <chessguy> | :t do {?x <- ?foo; ?y <- ?bar; return $ ?N ?x ?y} |
| 17:05:47 | <lambdabot> | Parse error in pattern |
| 17:06:01 | <narain> | ozzilee: yes |
| 17:06:18 | <chessguy> | > let foo x = x + 1 in foo 1 |
| 17:06:19 | <lambdabot> | 2 |
| 17:06:22 | <joelr1> | how abou tthis one? |
| 17:06:25 | <joelr1> | foo = option Nothing $ do { reserved "xx"; strExpr >>= return . Just } |
| 17:06:26 | <narain> | :t do {x <- ?foo; y <- ?bar; return $ ?N x y} |
| 17:06:28 | <lambdabot> | parse error on input `?' |
| 17:06:34 | <joelr1> | even shorter without sacrificing clarity? |
| 17:07:35 | <narain> | :t ?x + ?y |
| 17:07:37 | <lambdabot> | forall a. (?x::a, ?y::a, Num a) => a |
| 17:08:08 | <chessguy> | :t do {return ?x} |
| 17:08:10 | <lambdabot> | forall t (t1 :: * -> *). (?x::t, Monad t1) => t1 t |
| 17:08:57 | <sorear> | foo = option Nothing $ reserved "xx" >> fmap Just strExpr |
| 17:09:24 | <narain> | ?t option |
| 17:09:24 | <lambdabot> | Maybe you meant: tell temp time tiny-url todo todo-add todo-delete topic-cons topic-init topic-null topic-snoc topic-tail topic-tell type . ft v |
| 17:09:30 | <narain> | :t option |
| 17:09:31 | <lambdabot> | Not in scope: `option' |
| 17:09:59 | <sorear> | it is in parsec |
| 17:10:09 | <narain> | oh, ok |
| 17:10:57 | <sorear> | OR, if you are willing to use a recent addition (maybe just in trac, but I think it's in darcs too) |
| 17:11:08 | <sorear> | foo = optionMaybe $ reserved "xx" >> strExpr |
| 17:15:24 | <joelr1> | sorear: hmm |
| 17:15:51 | <joelr1> | sorear: @#@#% of course! we are in the maybe monad!!! |
| 17:16:00 | <sorear> | no we are in parsec |
| 17:16:17 | <joelr1> | sorear: in that particular parser |
| 17:16:24 | <sorear> | haskell-mode doesn't properly highlight CPP :( |
| 17:17:50 | <dmwit> | sorear: :set ft=cpp ;-) |
| 17:18:00 | <joelr1> | sorear: what about this one? shares = do { e <- numExpr; reserved "xx"; return $ Qty e } |
| 17:18:10 | <sorear> | dmwit: but if I do that it doesn't properly hl Haskell :) |
| 17:18:38 | <sorear> | shares = fmap Qty numExpr *> reserved "xx" |
| 17:18:53 | <sorear> | where x *> y = do { a <- x ; y ; return x} |
| 17:18:59 | <dmwit> | sorear: I think the real problem here is that you have C++ and Haskell in the same buffer. =P |
| 17:19:04 | <joelr1> | hmm |
| 17:19:11 | <sorear> | dmwit: CPP, not C++ |
| 17:19:33 | <dmwit> | ?where cpp |
| 17:19:33 | <lambdabot> | I know nothing about cpp. |
| 17:19:34 | <sorear> | dmwit: I don't want to write all this unboxed primop mangling by hand :) |
| 17:19:41 | <sorear> | dmwit: C PreProcessor |
| 17:19:44 | <sorear> | ghc -cpp |
| 17:19:44 | <dmwit> | o |
| 17:19:56 | <sorear> | ?where hscpp |
| 17:19:56 | <lambdabot> | I know nothing about hscpp. |
| 17:20:06 | <sorear> | ?google hscpp |
| 17:20:07 | <lambdabot> | http://urchin.earth.li/pipermail/debian-haskell/2004-September/000012.html |
| 17:20:08 | <lambdabot> | Title: [Debian-haskell] Re: [Haskell] hscpp |
| 17:20:16 | <thedward> | is it likely that there will be some sort of String type class in the future? To generalize over [Char] and Bytestring (and whatever else)? |
| 17:20:19 | <joelr1> | sorear: thanks |
| 17:20:48 | <sorear> | ... who here has heard of heat sink corner press pins? |
| 17:21:07 | <sorear> | thedward: not likely at all |
| 17:21:11 | <sorear> | ACTION cynicises |
| 17:21:25 | <sorear> | look at ross' collections |
| 17:21:29 | <sorear> | @where collections |
| 17:21:30 | <lambdabot> | I know nothing about collections. |
| 17:22:06 | <glguy> | Foldable and Traversable generalize of some of that stuff though |
| 17:22:29 | <thedward> | well, I am mainly wondering if there is some reason it would be a bad idea. |
| 17:22:45 | <glguy> | thedward: what function would it have? |
| 17:23:32 | <sorear> | @where+ hscpp It's called cpphs, dimwit! |
| 17:23:32 | <lambdabot> | Done. |
| 17:23:39 | <sorear> | @where cpphs |
| 17:23:39 | <lambdabot> | http://www.cs.york.ac.uk/fp/cpphs/ |
| 17:23:43 | <thedward> | glguy: to make it easier to write code that operates on any type of String? |
| 17:23:52 | <dmwit> | haha |
| 17:23:57 | <dmwit> | ?where hscpp |
| 17:23:57 | <lambdabot> | It's called cpphs, dimwit! |
| 17:24:00 | <dmwit> | =) |
| 17:24:44 | <glguy> | thedward: ok... so I'll clarify: what function*s* would it have? |
| 17:24:51 | <sorear> | also http://haskell.org/cpphs/ |
| 17:24:52 | <lambdabot> | Title: cpphs |
| 17:25:50 | <thedward> | glguy: hmm. well all the ones that come immediately to mind are list functions; maybe it should be a List type class :) |
| 17:26:03 | <glguy> | thedward: like Traversable? |
| 17:26:07 | <sjanssen> | thedward: I've hacked up a little 'Stringable' class before |
| 17:26:24 | <thedward> | ooh. Traversable, I've been meaning to read up on that. |
| 17:26:24 | <sorear> | thedward: ross paterson's collections probably do what you want |
| 17:26:37 | <Debolaz> | Which book is the defacto guide to Haskell? |
| 17:26:43 | <Debolaz> | ACTION is on a shopping spree. |
| 17:26:58 | <sjanssen> | it'd be a nice thing to have, especially with the half a dozen string representations around |
| 17:27:01 | <thedward> | well, what I actually want is to write some code that with either Strings or Bytestrings |
| 17:27:08 | <sjanssen> | sorear: I think you mean jyp's collections package |
| 17:27:17 | <sjanssen> | there is also Edison |
| 17:27:42 | <sebell> | thedward: You could always just write your code with a qualified module name |
| 17:27:58 | <sjanssen> | thedward: feel free to yank http://darcs.haskell.org/SoC/fps-soc/Data/Stringable.hs |
| 17:28:37 | <sorear> | we NED properly overloaded basic ops, like ++ |
| 17:29:07 | <sjanssen> | emphasized typos are especially humorous |
| 17:29:19 | <glguy> | (.) = fmap (++) = mappend ? :) |
| 17:29:21 | <thedward> | sjanssen: that is exactly the sort of thing I want. Thanks. :) |
| 17:29:29 | <sorear> | glguy: yah |
| 17:29:53 | <narain> | :t mappend |
| 17:29:55 | <lambdabot> | forall a. (Monoid a) => a -> a -> a |
| 17:30:11 | <narain> | Mon*oid*? eep |
| 17:30:24 | <sjanssen> | narain: they're much simpler than Monads |
| 17:30:27 | <sorear> | it's elementary abstract algebra! |
| 17:30:28 | <sorear> | !! |
| 17:30:32 | <opqdonut> | yeh |
| 17:30:33 | <sorear> | @src Monoid |
| 17:30:34 | <lambdabot> | class Monoid a where |
| 17:30:34 | <lambdabot> | mempty :: a |
| 17:30:34 | <lambdabot> | mappend :: a -> a -> a |
| 17:30:34 | <lambdabot> | mconcat :: [a] -> a |
| 17:30:39 | <opqdonut> | ?instances Monoid |
| 17:30:40 | <glguy> | monoids have an associative operation and an identity |
| 17:30:40 | <lambdabot> | (), (a -> b), (a, b), (a, b, c), All, Any, Dual a, Endo a, Ordering, Product a, Sum a, [a] |
| 17:31:05 | <opqdonut> | (++) for [] i'd guess? |
| 17:31:06 | <narain> | that's it? haha, that's cute |
| 17:31:14 | <sorear> | opqdonut: yes |
| 17:31:15 | <opqdonut> | and . for (a->b)? |
| 17:31:24 | <sorear> | opqdonut: no, liftM2 mappend |
| 17:31:28 | <narain> | > mappend (1,2) (3,4) |
| 17:31:29 | <sorear> | @src (->) mappend |
| 17:31:29 | <lambdabot> | Add a type signature |
| 17:31:29 | <lambdabot> | Source not found. That's something I cannot allow to happen. |
| 17:31:36 | <glguy> | > Sum 1 `mappend` Sum 10 |
| 17:31:37 | <lambdabot> | Sum {getSum = 11} |
| 17:32:00 | <sorear> | > mappend ("foo"++) ("bar"++) "baz" -- silently different in 6.6 and 6.4! |
| 17:32:01 | <lambdabot> | "foobazbarbaz" |
| 17:32:05 | <sjanssen> | > mappend (Sum 1, Product 2) (Sum 3, Product 4) |
| 17:32:06 | <lambdabot> | (Sum {getSum = 4},Product {getProduct = 8}) |
| 17:32:07 | <narain> | :t mappend (?a,?b) (?c,?d) |
| 17:32:09 | <lambdabot> | forall t t1. (?a::t, ?b::t1, ?c::t, ?d::t1, Monoid (t, t1)) => (t, t1) |
| 17:32:13 | <sorear> | 6.4 saz: "foobarbaz" |
| 17:32:17 | <glguy> | > (Product 3, Sum 3) `mappend` (Product 2, Sum 2) |
| 17:32:19 | <lambdabot> | (Product {getProduct = 6},Sum {getSum = 5}) |
| 17:32:27 | <glguy> | oh, someone already did that :) |
| 17:32:38 | <narain> | ohh |
| 17:32:56 | <narain> | fun |
| 17:33:06 | <glguy> | > mempty :: Sum Int |
| 17:33:07 | <lambdabot> | Sum {getSum = 0} |
| 17:33:13 | <sjanssen> | narain: there are several monoids with the numbers, so we have newtype wrappers to choose them |
| 17:33:44 | <narain> | sjanssen: that makes sense |
| 17:33:59 | <narain> | > mempty :: All |
| 17:34:00 | <lambdabot> | All {getAll = True} |
| 17:34:17 | <narain> | just checking |
| 17:34:41 | <sjanssen> | to see monoids used to great effect, see the fingertree paper by Hinze and Paterson |
| 17:34:52 | <narain> | wait, () is a monoid? what's mappend for ()? const? |
| 17:35:11 | <narain> | > mappend () () |
| 17:35:11 | <lambdabot> | () |
| 17:35:14 | <glguy> | > () `mappend` undefined |
| 17:35:14 | <lambdabot> | () |
| 17:35:25 | <glguy> | > undefined `mappend` () |
| 17:35:26 | <lambdabot> | () |
| 17:35:34 | <glguy> | mappend _ _ = () |
| 17:36:06 | <narain> | now that's just silly |
| 17:36:17 | <sorear> | The Ordering instance is very useful. |
| 17:36:36 | <opqdonut> | > [()..] |
| 17:36:38 | <lambdabot> | [()] |
| 17:36:41 | <opqdonut> | :D |
| 17:36:44 | <opqdonut> | excellent |
| 17:36:49 | <glguy> | > liftM2 mappend [LT ..] [LT ..] |
| 17:36:49 | <sjanssen> | yes, compare on lists is something like: |
| 17:36:51 | <lambdabot> | [LT,LT,LT,LT,EQ,GT,GT,GT,GT] |
| 17:37:18 | <sjanssen> | > mconcat (zipWith compare "stuff" "stuff2") |
| 17:37:19 | <lambdabot> | EQ |
| 17:37:23 | <sjanssen> | bah |
| 17:37:29 | <sorear> | @scheck \a b c d -> ((a `compare` b) `mappend` (c `compare` d)) == ((a,b) `compare` (c,d)) :: Bool -> Bool -> Bool -> Bool -> Bool |
| 17:37:31 | <lambdabot> | Couldn't match expected type `Bool -> Bool -> Bool -> Bool -> Bool' |
| 17:37:33 | <sjanssen> | close |
| 17:37:36 | <glguy> | > liftM2 (\x y -> (x,y,mappend x y)) [LT ..] [LT ..] |
| 17:37:37 | <lambdabot> | [(LT,LT,LT),(LT,EQ,LT),(LT,GT,LT),(EQ,LT,LT),(EQ,EQ,EQ),(EQ,GT,GT),(GT,LT,GT... |
| 17:37:40 | <sorear> | @scheck (\a b c d -> ((a `compare` b) `mappend` (c `compare` d)) == ((a,b) `compare` (c,d))) :: Bool -> Bool -> Bool -> Bool -> Bool |
| 17:37:42 | <lambdabot> | Failed test no. 3. Test values follow.: True, True, False, True |
| 17:37:48 | <narain> | ?src Ordering |
| 17:37:48 | <lambdabot> | data Ordering = LT | EQ | GT |
| 17:37:58 | <sorear> | @scheck (\a b c d -> ((a `compare` b) `mappend` (c `compare` d)) == ((a,c) `compare` (b,d))) :: Bool -> Bool -> Bool -> Bool -> Bool |
| 17:37:59 | <lambdabot> | Completed 16 test(s) without failure. |
| 17:38:17 | <sorear> | @src (,) compare |
| 17:38:17 | <lambdabot> | Source not found. Have you considered trying to match wits with a rutabaga? |
| 17:38:29 | <narain> | ?src Ordering compare |
| 17:38:29 | <lambdabot> | Source not found. Do you think like you type? |
| 17:38:33 | <narain> | oh wait |
| 17:38:38 | <narain> | ?src Ordering mappend |
| 17:38:39 | <lambdabot> | Source not found. You type like i drive. |
| 17:38:50 | <narain> | i didn't come here to be insulted |
| 17:38:56 | <glguy> | |
| 17:39:26 | <dmwit> | How does lambdabot drive? |
| 17:39:31 | <sorear> | glguy: how dare you! |
| 17:39:38 | <sorear> | :) |
| 17:39:47 | <glguy> | umm... lol? ;) |
| 17:40:03 | <sorear> | narain: patch it |
| 17:40:28 | <glguy> | and if you think "sure, that's a good idea", then fix ?vixen and ?let first |
| 17:40:36 | <glguy> | and then bother changing the error messages |
| 17:40:50 | <dmwit> | What's wrong with ?vixen? |
| 17:40:51 | <glguy> | afterward |
| 17:40:59 | <glguy> | ?vixen didn't used to say this: |
| 17:41:00 | <lambdabot> | <undefined> |
| 17:41:12 | <narain> | ?vixen |
| 17:41:12 | <lambdabot> | <undefined> |
| 17:41:13 | <dmwit> | I thought dons just disabled it for a while. |
| 17:41:14 | <narain> | ?vixen |
| 17:41:14 | <sjanssen> | I think we're waiting on a fix for the regex libraries for let |
| 17:41:14 | <lambdabot> | <undefined> |
| 17:41:18 | <narain> | ?vixen |
| 17:41:18 | <lambdabot> | <undefined> |
| 17:41:25 | <narain> | take that you foul mouthed bot |
| 17:41:27 | <glguy> | or fix this: |
| 17:41:34 | <glguy> | ?type mapM |
| 17:41:36 | <lambdabot> | Ambiguous occurrence `mapM' |
| 17:41:37 | <lambdabot> | It could refer to either `mapM', imported from Control.Monad.Writer |
| 17:41:42 | <sorear> | glguy: don't discourage fixes |
| 17:41:51 | <sjanssen> | glguy: you know how to use darcs send |
| 17:42:16 | <glguy> | sjanssen: I'm only arguing that there are much more important things to worry about than sudo insults |
| 17:42:17 | <narain> | not like i know nearly enough to fix any bugs in lambdabot |
| 17:42:29 | <sorear> | glguy: also, if narain removes the insults, @let will be AUTOMATICALLY fixed |
| 17:42:39 | <glguy> | ? |
| 17:42:47 | <sorear> | since @let is fixed in darcs but dons hasn't bothered to recompile |
| 17:42:51 | <glguy> | ahh |
| 17:43:07 | <glguy> | but if the insults are fixed, dons will be hot to recompile? |
| 17:43:11 | <sjanssen> | narain: you'd be surprised. lambdabot is a good project to cut your teeth on |
| 17:43:12 | <glguy> | "fixed" |
| 17:43:27 | <sjanssen> | bug dons to recompile in 24 hours or so, after the ICFP deadline |
| 17:43:41 | <narain> | im already cutting my teeth on something else |
| 17:43:44 | <sorear> | when are ICFP papers published? |
| 17:43:58 | <sorear> | ACTION is still managing to avoid hacking ghc |
| 17:44:08 | <sjanssen> | probably at ICFP, August or so? |
| 17:46:15 | <narain> | anyway i wouldn't want to lobotomize lambdabot's testy personality |
| 17:51:40 | <sebell> | Hmmm. I can't seem to use runInteractiveCommand from GHCI |
| 18:09:34 | <sjanssen> | @yow |
| 18:09:34 | <lambdabot> | ... the MYSTERIANS are in here with my CORDUROY SOAP DISH!! |
| 18:10:28 | <opqdonut> | @. elite yow |
| 18:10:29 | <lambdabot> | WhY iS ev3Ry+hinG mAd3 of lY(r4 zp4nd3x? |
| 18:12:02 | <sjanssen> | sebell: what happens? |
| 18:13:48 | <sebell> | sjanssen: I get EOF exception when trying to read from the output handle |
| 18:15:57 | <hpaste> | sjanssen pasted "ghci session with runInteractiveCommand" at http://hpaste.org/1266 |
| 18:18:45 | <sjanssen> | sebell: do simple things like "ls" work? |
| 18:20:01 | <sebell> | sjanssen: that seems to work for me |
| 18:21:17 | <sebell> | sjanssen: the issue seems to be with my own function (which in turn uses unsafeInterleaveIO) |
| 18:24:26 | <sioraiocht> | anyone know how to get haskell mode in emacs to use ghci instead of hugs? i fixed it once but i can't remember how and this is a new install |
| 18:24:51 | <sebell> | (add-hook 'haskell-mode-hook 'turn-on-haskell-ghci) |
| 18:25:23 | <sorear> | aslo look at the haskell interpreter command in customize |
| 18:25:31 | <freshhawk> | is there a current best practice for avoiding thread starvation / memory issues when you have multiple producers and one consumer communication over a tchan? |
| 18:25:54 | <freshhawk> | *communicating |
| 18:25:55 | <sioraiocht> | sebell, sorear : tahnks |
| 18:26:32 | <sorear> | yay, I finally got a chance to use Any |
| 18:26:55 | <sjanssen> | @pl ssfold p f a0 xs = foldr (\x xs a -> if p a then a else xs (f a x)) id xs a0 |
| 18:26:56 | <lambdabot> | ssfold = (flip .) . flip flip id . ((foldr . (const .)) .) . (. ((ap id .) . flip)) . (.) . ap . (if' =<<) |
| 18:27:20 | <mauke> | @. pl . unpl pl ssfold p f a0 xs = foldr (\x xs a -> if p a then a else xs (f a x)) id xs a0 |
| 18:27:23 | <lambdabot> | ssfold = (flip .) . flip flip id . ((foldr . (const .)) .) . (. flip (flip . (((.) . (>>=) . (id >>=)) .) . flip flip (return .) . (((.) . (>>=)) .) . flip) (return .)) . (.) . (>>=) . (if' =<<) |
| 18:28:27 | <narain> | @. unpl pl ssfold p f a0 xs = foldr (\x xs a -> if p a then a else xs (f a x)) id xs a0 |
| 18:28:28 | <lambdabot> | ssfold aa bl b c = foldr (\ br _ -> (aa >>= if') >>= \ be -> ((\ o -> o) >>= \ ay -> (\ s -> bl s br) >>= \ ax -> return (ay ax)) >>= \ bd -> return (be bd)) (\ k -> k) c b |
| 18:28:45 | <sjanssen> | hmm, lambdabot seems to be generating bad results here |
| 18:29:14 | <sjanssen> | @pl ssfold p f a0 xs0 = foldr (\x xs a -> if p a then a else xs (f a x)) id xs0 a0 |
| 18:29:15 | <lambdabot> | ssfold = (flip .) . flip flip id . (foldr .) . (. ((flip (.) .) . flip)) . (.) . (.) . ap . (if' =<<) |
| 18:29:21 | <kpreid> | pl or unpl? |
| 18:29:26 | <mauke> | unpl is almost pure obfuscation |
| 18:29:28 | <sjanssen> | kpreid: pl |
| 18:29:44 | <kpreid> | both have flaws in their parsing, but I wrote unpl :) |
| 18:31:10 | <narain> | does pl work by eliminating one argument at a time? |
| 18:31:11 | <sorear> | kpreid: unfortunately you aren't doing enough type checking :) I want eta expansion and monad op inlining |
| 18:31:27 | <sioraiocht> | is it possible to do parallel list comprehension? |
| 18:31:39 | <kpreid> | sorear: so write it :-) |
| 18:31:48 | <narain> | @pl \xs0 -> foldr (\x xs a -> if p a then a else xs (f a x)) id xs0 a0 |
| 18:31:48 | <lambdabot> | flip (foldr ((ap (if' =<< p) .) . flip (.) . flip f) id) a0 |
| 18:31:54 | <sioraiocht> | and by that i mean do the evaluation on each member of the list in parallel |
| 18:31:55 | <mauke> | I think ghc supports it as an extension |
| 18:32:01 | <narain> | @pl \a0 xs0 = foldr (\x xs a -> if p a then a else xs (f a x)) id xs0 a0 |
| 18:32:01 | <lambdabot> | (line 1, column 9): |
| 18:32:02 | <lambdabot> | unexpected "=" |
| 18:32:02 | <lambdabot> | expecting pattern or "->" |
| 18:32:09 | <LeCamarade> | Stuck with a problem. I need to iterate over data that I collect from an IO monad (hence can't get rid of the blemish). But I can't return that, because it would ruin my returns ... I don't know if I should just think harder, or it isnormal to meet such quirks (and think of a work-around). Do I just need some sleep? :D |
| 18:32:09 | <narain> | @pl \a0 xs0 -> foldr (\x xs a -> if p a then a else xs (f a x)) id xs0 a0 |
| 18:32:10 | <lambdabot> | flip (foldr ((ap (if' =<< p) .) . flip (.) . flip f) id) |
| 18:32:14 | <sorear> | narain: it is ordinary abstraction elimination, followed by the most sophisticated optimizer in the Haskell world. |
| 18:32:27 | <sorear> | sioraiocht: Control.Parallel.Strategies.parList |
| 18:32:37 | <mauke> | LeCamarade: I don't understand |
| 18:32:49 | <mauke> | sorear: wtf? what's wrong with zipWith? |
| 18:32:59 | <LeCamarade> | I am trying to get the filesystem as a tree. |
| 18:33:01 | <mnislaih> | help. where is the combinator for onclick in the xhtml package ? |
| 18:33:02 | <sorear> | <sioraiocht> and by that i mean do the evaluation on each member of the list in |
| 18:33:02 | <sorear> | parallel |
| 18:33:16 | <sorear> | mauke: does zipWith do parallelism? |
| 18:33:17 | <sebell> | LeCamarade: You stay in the IO Monad |
| 18:33:27 | <sioraiocht> | sorear: thanks |
| 18:33:38 | <LeCamarade> | mauke: Yeah. |
| 18:33:45 | <mauke> | sorear: oh, right :( my interpretation was faulty |
| 18:34:10 | <LeCamarade> | The problem is that I think I should be returning the next file/dir, yet that would require ... wait a second ... |
| 18:34:29 | <LeCamarade> | mauke: Lemme first try what I thought of ... |
| 18:34:48 | <sorear> | sioraiocht: confusingly, parallel list comprehension is also the name of an unrelated ghc extension |
| 18:35:04 | <sioraiocht> | sorear: ah |
| 18:36:24 | <sebell> | sjanssen: I added EOF checking to my hGetContentsTimeout, but now it doesn't behave as it does in the compiled program -- in fact, the compiled program never reached EOF before at all |
| 18:38:44 | <sjanssen> | sebell: may I see the code? |
| 18:41:37 | <hpaste> | sebell pasted "hGetContentsTimeout" at http://hpaste.org/1267 |
| 18:42:57 | <stepcut> | haddock barfs on this line: deriving (Monad, MonadIO, MonadFix, Functor, MonadReader TaskIOStyle), because of the MonadReader TaskIOStyle |
| 18:43:04 | <stepcut> | what is the best work-around ? |
| 18:43:22 | <sjanssen> | stepcut #ifndef HADDOCK |
| 18:44:05 | <procyon112> | fix haddock? *nudge-nudge-wink-wink* |
| 18:44:29 | <sjanssen> | or perhaps a dummy instance guarded by #ifdef HADDOCK |
| 18:45:08 | <sjanssen> | sebell: you should probably use waitProcess p to make sure you don't leave zombies |
| 18:45:24 | <stepcut> | sjanssen: spiffy, thanks |
| 18:45:37 | <sioraiocht> | sorear: so if Strategy is a -> Done, then how does parList get a value from a Strategy to build the list |
| 18:45:51 | <sorear> | sioraiocht: ? |
| 18:45:56 | <sorear> | type Done = () |
| 18:46:06 | <sorear> | it just represents a side effect |
| 18:46:14 | <sjanssen> | @hoogle using |
| 18:46:14 | <lambdabot> | Control.Parallel.Strategies.using :: a -> Strategy a -> a |
| 18:46:17 | <sorear> | par :: Done -> Done -> Done -- do it in parallel! |
| 18:46:39 | <sioraiocht> | @t Control.Parallel.Strategies.parList |
| 18:46:39 | <lambdabot> | Maybe you meant: tell temp time tiny-url todo todo-add todo-delete topic-cons topic-init topic-null topic-snoc topic-tail topic-tell type . ft v |
| 18:46:47 | <sioraiocht> | @type Control.Parallel.Strategies.parList |
| 18:46:49 | <lambdabot> | forall a. Strategy a -> [a] -> Done |
| 18:47:06 | <sioraiocht> | ohh, n/m i get it, now |
| 18:49:30 | <stepcut> | hrm, now haddocks dies on: #ifndef HADDOCK |
| 18:49:32 | <stepcut> | :-l |
| 18:51:25 | <sorear> | stepcut: is your program a program? |
| 18:52:02 | <stepcut> | ?? it's a module in a library... |
| 18:52:09 | <sorear> | http://hackage.haskell.org/trac/hackage/ticket/102 |
| 18:52:10 | <lambdabot> | Title: #102 (building haddock for executables wont CPP) - Hackage - Trac |
| 18:52:28 | <sorear> | haddock doesn't cpp, you need to have something else (like cabal) do it first |
| 18:52:36 | <sorear> | or just upgrade to haddock.ghc |
| 18:53:00 | <stepcut> | ok, I was justing running 'haddock ABIO.hs', but in practice, it is run by cabal |
| 18:53:30 | <sorear> | haddock ABIO.hs wouldn't work anyway |
| 18:53:39 | <Saizan> | ?where derive |
| 18:53:44 | <sorear> | you need to specify an output format and a output directory |
| 18:53:45 | <Saizan> | ?where derive |
| 18:53:45 | <lambdabot> | http://www.cs.york.ac.uk/fp/darcs/derive |
| 18:53:53 | <sjanssen> | sebell: does the program work differently if you compile with -threaded? |
| 18:53:57 | <sorear> | derive is pouplar? what have I done :( |
| 18:55:11 | <Saizan> | well i have to do that "derive a 'untyped' representation for my state type", and thought derive could help |
| 19:00:28 | <glguy> | @temp |
| 19:00:30 | <lambdabot> | now 17.6°, min 17.5°, max 24.2°, rain 0.0mm, wind 31km/h SW |
| 19:02:05 | <glguy> | !say hello |
| 19:02:05 | <hpaste> | hello |
| 19:02:15 | <sorear> | !say hi? |
| 19:02:38 | <glguy> | ?admin+ :n=user@ip68-7-248-101.sd.sd.cox.net |
| 19:02:38 | <lambdabot> | Not enough privileges |
| 19:02:45 | <glguy> | err |
| 19:02:48 | <glguy> | !admin+ :n=user@ip68-7-248-101.sd.sd.cox.net |
| 19:02:56 | <emu> | @temp where? |
| 19:02:56 | <lambdabot> | now 17.6°, min 17.5°, max 24.2°, rain 0.0mm, wind 31km/h SW |
| 19:03:01 | <glguy> | I forget if the leading : needed to be there |
| 19:03:01 | <Saizan> | derive's cabal misses mtl in build-depends |
| 19:03:35 | <sorear> | it shouldn't depend on mtl |
| 19:04:00 | <sorear> | @admin + hpaste |
| 19:04:11 | <sorear> | !msg lambdabot @msg #haskell Muahahaha? |
| 19:04:25 | <sorear> | !say testing? |
| 19:05:36 | <Saizan> | Data/Derive/Play.hs:import Control.Monad.State |
| 19:05:48 | <glguy> | !admin+ n=user@ip68-7-248-101.sd.sd.cox.net |
| 19:05:54 | <sorear> | !msg lambdabot @msg #haskell Muahahaha? |
| 19:05:58 | <Saizan> | well maybe i've pulled a very recent patch |
| 19:06:04 | <sorear> | !say hi? |
| 19:06:07 | <glguy> | ACTION goes to check the code |
| 19:06:11 | <glguy> | did I remove !admin+ ?? |
| 19:06:30 | <sjanssen> | speaking of derive, aren't these blank lines in derive.cabal illegal? |
| 19:06:41 | <Igloo> | Hmm, what's the flag to show which RULES are being applied? |
| 19:06:49 | <sorear> | -ddump-simpl-stats |
| 19:07:12 | <sorear> | sjanssen: I only wrote the .cabal, never tested it :) |
| 19:07:17 | <Igloo> | Thanks |
| 19:07:23 | <Igloo> | ACTION wonders what -ddump-rules is meant to do |
| 19:07:26 | <glguy> | !admin+ :sorear!n=user@ip68-7-248-101.sd.sd.cox.net |
| 19:07:32 | <glguy> | sorear: that should do it |
| 19:07:35 | <glguy> | i was forgetting the nick |
| 19:07:39 | <sorear> | !msg lambdabot @msg #haskell Muahahaha? |
| 19:07:44 | <sorear> | !say hi? |
| 19:07:44 | <hpaste> | hi? |
| 19:07:52 | <glguy> | > PRIVMSG lambdabot @msg #haskell Muahahaha? |
| 19:07:52 | <glguy> | :sorear!n=user@ip68-7-248-101.sd.sd.cox.net PRIVMSG #haskell :!say hi? |
| 19:07:52 | <lambdabot> | Parse error |
| 19:08:09 | <glguy> | lambdabot rejected the command, but hpaste tried making it |
| 19:08:25 | <glguy> | !say @admin + glguy |
| 19:08:26 | <hpaste> | @admin + glguy |
| 19:08:26 | <sorear> | @admin + hpaste |
| 19:08:41 | <sorear> | !say @msg #haskell Bwahaha! |
| 19:08:42 | <hpaste> | @msg #haskell Bwahaha! |
| 19:08:42 | <lambdabot> | Bwahaha! |
| 19:08:47 | <Igloo> | sorear: Do you happen to know if you can print the source before and after the rules are applied? |
| 19:08:49 | <lambdabot> | ? |
| 19:09:26 | <sorear> | -ddump-simpl-iterations, -dverbose-core2core |
| 19:09:31 | <sorear> | or so the wiki said |
| 19:09:44 | <sorear> | @wiki GHC/Using_rules |
| 19:09:45 | <lambdabot> | http://www.haskell.org/haskellwiki/GHC/Using_rules |
| 19:10:51 | <sjanssen> | boo, derive isn't darcs sendable |
| 19:11:17 | <sorear> | I've directly bood ndm twice. |
| 19:11:33 | <sorear> | Just for that I'll say his email in cleartext: |
| 19:11:38 | <sorear> | ndm@cs.york.ac.uk |
| 19:11:49 | <sorear> | FTR he is out of communication for a week |
| 19:11:57 | <sorear> | with intent to continue hacking derive |
| 19:12:34 | <sjanssen> | he must be paranoid, he doesn't even include his email in darcs record messages |
| 19:14:01 | <sjanssen> | anyway, I just fixed the .cabal and replaced fmap with liftM for 6.6 compatibility |
| 19:14:33 | <sjanssen> | sorear: do you want the patches too? |
| 19:14:37 | <sorear> | darcs send --to=ndm@cs.york.ac.uk |
| 19:14:55 | <sorear> | sjanssen: no, I'm respecting his absense as if it were a lock |
| 19:14:59 | <sjanssen> | yep, already sent to ndm |
| 19:15:42 | <Igloo> | sorear: Thanks! |
| 19:16:00 | <joelr1> | sorear: he took a week off to hack derive? |
| 19:17:06 | <sorear> | he's also attending some kind of conference |
| 19:17:18 | <sorear> | but the details are a bit fuzzy |
| 19:19:27 | <joelr1> | sorear: what do you do for a living? |
| 19:19:42 | <joelr1> | sjanssen: and what about you? |
| 19:20:15 | <pejo> | sorear, wasn't it BTCS or something like that? |
| 19:21:05 | <sjanssen> | joelr1: I'm a computer science undergrad |
| 19:21:21 | <joelr1> | sjanssen: cool |
| 19:22:15 | <sebell> | sjanssen: Yes, I see the same behavior with GHCI as -threaded |
| 19:22:27 | <sebell> | (sorry for the late response) |
| 19:22:30 | <sorear> | joelr1: I stay at home and act like a 16-year-old dependant? |
| 19:22:53 | <joelr1> | sorear: :D |
| 19:23:25 | <joelr1> | sorear: i stay at home too |
| 19:23:40 | <sjanssen> | sebell: ghci uses the threaded RTS, so I think you've officially narrowed your problem |
| 19:23:55 | <sjanssen> | sorear: just "act like"? |
| 19:24:00 | <sorear> | yeah, but you probably don't get infinite free food, shelter, internet, and education :) |
| 19:24:08 | <sjanssen> | infinite? |
| 19:24:12 | <sebell> | sjanssen: Indeed. |
| 19:24:12 | <joelr1> | sorear: infinite? |
| 19:24:15 | <glguy> | sjanssen: in the verizon sense |
| 19:24:23 | <joelr1> | 5Gb of food |
| 19:24:24 | <joelr1> | hehe |
| 19:24:24 | <glguy> | infinite for 5gig, which ever comes first |
| 19:24:25 | <sorear> | in the Haskell sense |
| 19:24:27 | <sjanssen> | glguy: 5 gigabytes of steak/month |
| 19:24:45 | <sorear> | no, my supplies are infinite but lazily evaluated |
| 19:25:01 | <sorear> | I only consume a finite amount in any given timeslot |
| 19:25:19 | <qwr> | so you're just not forced a strict evaluation? |
| 19:25:57 | <sjanssen> | sorear's refrigerator is just one big thunk |
| 19:26:03 | <joelr1> | hehe |
| 19:26:34 | <glguy> | /set format_unidentified_nick Loser:$0 |
| 19:26:36 | <glguy> | oops |
| 19:26:44 | <glguy> | /set format_unidentified_nick Loser:$0 |
| 19:26:46 | <glguy> | err! |
| 19:26:52 | <glguy> | damn spaces |
| 19:27:03 | <glguy> | you'd think that when you press UP |
| 19:27:06 | <glguy> | that it would fix that |
| 19:27:30 | <sjanssen> | now we all know that glguy hates unregistered nicks |
| 19:28:23 | <joelr1> | i think i just eliminated over 1000 lines of code thanks to sorear, maybe 1500 |
| 19:28:26 | <glguy> | hate is such a strong word |
| 19:28:31 | <joelr1> | and that's not the end |
| 19:28:45 | <joelr1> | i should be able to blow away 500-1000 more |
| 19:28:45 | <joelr1> | he |
| 19:28:46 | <sorear> | apr1 : 09:56:07 <ndm> oh, btw - i'm off to Oxford all next week, out of email contact |
| 19:29:07 | <sorear> | BCTCS |
| 19:29:18 | <mauke> | 0xf0rd |
| 19:29:19 | <sorear> | http://www.cs.swan.ac.uk/BCTCS2006/ |
| 19:29:21 | <lambdabot> | Title: BCTCS 2006 |
| 19:29:35 | <qwr> | ACTION is too lazy to register the damn nick. and there seem to be plenty of free nicks for backup. ;) |
| 19:30:03 | <glguy> | 14:29 (*) Loser:qwr is too lazy to register the ... ;-) |
| 19:30:18 | <qwr> | yeah :) |
| 19:31:08 | <sorear> | * *qwr is too lazy to register the damn nick. and there seem to be plenty of free |
| 19:31:09 | <sorear> | nicks for backup. ;) |
| 19:31:22 | <sorear> | you realize it would have taken less typing just to register? |
| 19:31:34 | <sorear> | /msg nickserv register <password> |
| 19:32:52 | <qwr> | now, that i didn't have to figure out how, it was easy ;) |
| 19:33:26 | <glguy> | ACTION wonders where Loser:qwr went |
| 19:33:51 | <glguy> | /set format_unidentified_nick ~$0 |
| 19:33:55 | <glguy> | :( |
| 19:34:04 | <sjanssen> | @slap glguy |
| 19:34:04 | <lambdabot> | why on earth would I slap glguy |
| 19:34:14 | <glguy> | ACTION isn't actually intentionally sending this stuff to channel |
| 19:34:32 | <glguy> | Time to drink some more coffee I guess |
| 19:36:27 | <glguy> | ACTION wonders why people would want to turn garbage pop music *up* when we aren't supposed to have audible music at all |
| 19:36:47 | <glguy> | I wonder if he just doesn't know what he's listening to |
| 19:36:59 | <monochrom> | in an office? |
| 19:37:02 | <glguy> | yeah |
| 19:37:27 | <glguy> | ACTION is completely ready to start his new job in Portland |
| 19:37:29 | <monochrom> | someone should let him know he should stop |
| 19:38:01 | <glguy> | monochrom: I Don't want to open myself up to the possibility of a conversation with him |
| 19:38:29 | <procyon112> | lol |
| 19:39:03 | <glguy> | which I would consider worse than having to block out christina aguilara(?) |
| 19:39:25 | <astrolabe> | You could try singing along |
| 19:39:55 | <monochrom> | You could compete and play Carmina Burana. |
| 19:40:28 | <stepcut> | hrm, for some reason ./Setup haddock, is running cphhs on ABIO.hs, but it never actually runs haddock on it :( |
| 19:40:57 | <stepcut> | oh wait, hahaha |
| 19:41:09 | <stepcut> | I'm just looking in the wrong output directory :) |
| 19:42:46 | <sjanssen> | glguy: throw things at him |
| 19:42:49 | <sorear> | glguy: How about explosives? |
| 19:43:25 | <monochrom> | A power outage. |
| 19:44:11 | <sjanssen> | one of those EMPs the movies always use? |
| 19:44:12 | <sorear> | ghc -e 'cycle "\0\255"' > /dev/audio |
| 19:44:19 | <glguy> | poweroutage woudl be great |
| 19:44:24 | <glguy> | because I would get to go home earlier |
| 19:44:26 | <sjanssen> | sorear++ |
| 19:44:43 | <glguy> | sorear: does that do more than static? |
| 19:44:45 | <sjanssen> | ACTION waits for @flush |
| 19:44:54 | <sorear> | glguy: eeeeeeeeeeeeeee! |
| 19:45:01 | <glguy> | ?help flush |
| 19:45:02 | <lambdabot> | flush. flush state to disk |
| 19:45:14 | <sorear> | glguy: maximum amplitude, maximum frequency square wave |
| 19:45:22 | <monochrom> | ACTION tries |
| 19:45:24 | <glguy> | Oh, does sorear @flush after karma to ensure that it is saved? |
| 19:45:26 | <sorear> | glguy: or silence if you aren't using 8-bit unsigned :) |
| 19:45:29 | <sorear> | @flush |
| 19:45:30 | <sorear> | yeah |
| 19:45:32 | <glguy> | ?karma |
| 19:45:33 | <lambdabot> | You have a karma of 50 |
| 19:45:38 | <glguy> | mine hasn't changed for a while |
| 19:45:59 | <sorear> | My karma high watered at fourty after shapr rewarded my for yi synhl. |
| 19:46:02 | <glguy> | everyone was excited about hpaste, but then I faded away |
| 19:46:09 | <sorear> | Then I lost it all to the bot crash. |
| 19:46:14 | <glguy> | ahh |
| 19:46:17 | <glguy> | ?karma sorear |
| 19:46:17 | <lambdabot> | sorear has a karma of 30 |
| 19:46:23 | <sorear> | glguy: hpaste is boring, you need new features |
| 19:46:24 | <glguy> | The great bot crash of 2007? |
| 19:46:29 | <glguy> | sorear: I agree |
| 19:46:56 | <SamB_XP> | like nethack? |
| 19:47:15 | <glguy> | ACTION has been playing with Factor recently |
| 19:47:24 | <sorear> | glguy: like ... type checking ... more channels ... multisyntax with autodetection ... an announce api ... |
| 19:47:35 | <glguy> | announce api? |
| 19:47:41 | <sorear> | you realize they are still using lisppaste over in #darcs? |
| 19:47:52 | <sorear> | like connect and recieve a lazy [Announce] |
| 19:48:02 | <SamB_XP> | interesting idea, sorear |
| 19:48:20 | <SamB_XP> | why don't you go write an RFC for one of those? |
| 19:48:39 | <sorear> | ACTION would write hpaste-mode for emacs if it had a decent api |
| 19:48:43 | <sjanssen> | type checking would be really nice IMO |
| 19:48:56 | <sorear> | SamB_XP: you really think the ietf cares about pastebots?! |
| 19:48:58 | <glguy> | I've never used the GHC API, what would that take? |
| 19:49:07 | <sjanssen> | newbies always forget to state the error messages |
| 19:49:10 | <glguy> | sorear: nope, they care about cash money |
| 19:49:19 | <SamB_XP> | sorear: I was thinking something more along the lines of "lazy RSS" |
| 19:49:40 | <sorear> | glguy: GHC API will probably segfault in no time, it's too big. Do as lambdabot does and fork a subprocess. |
| 19:50:04 | <hpaste> | boowax annotated "Monad Issue" with "(no title)" at http://hpaste.org/1268#a1 |
| 19:50:12 | <sorear> | > array (0,maxBound::Int) [(1000000,'a')] -- look, segfault is caught! |
| 19:50:14 | <lambdabot> | Terminated |
| 19:50:15 | <boowax> | Since you guys were helpful to me last time, I've come back for more! I'm having a problem getting something to compile, can any of you guys give me some insight as to how to fix it? http://hpaste.org/1268#a1 |
| 19:50:18 | <sjanssen> | GHC API also won't protect you from evil TH hacks |
| 19:50:36 | <sorear> | filtering out {-# OPTION will :) |
| 19:50:39 | <monochrom> | ghc -e 'map Char.chr $ Random.randomRs (0,255) $ Random.mkStdGen 42' > /dev/audio |
| 19:51:05 | <sorear> | monochrom: THAT is static. |
| 19:51:08 | <SamB_XP> | GHC API doesn't support -fno-th? |
| 19:51:26 | <monochrom> | Yeah |
| 19:51:30 | <sorear> | SamB_XP: -fno-th can be overriden with {-# OPTIONS -fth #-} |
| 19:51:38 | <sorear> | so we need to filter as well |
| 19:51:51 | <SamB_XP> | okay, so -fno-th-I-really-mean-it |
| 19:51:54 | <monochrom> | Ha, I don't need map. |
| 19:52:03 | <sjanssen> | or just run ghc with dropped privileges |
| 19:52:14 | <sorear> | tuukkah: ping? |
| 19:52:24 | <sorear> | I meant twanvl :) |
| 19:52:29 | <sorear> | twanvl: ping? |
| 19:52:35 | <twanvl> | sorear: pong |
| 19:52:39 | <monochrom> | ghc -e 'Random.randomRs ('\0','\255') $ Random.mkStdGen 42' > /dev/audio |
| 19:54:09 | <xerox> | ghc -e 'Control.Monad.Fix.fix (\xs -> 1 : 1 : zipWith (+) xs (tail xs))' > /dev/audio |
| 19:54:30 | <joelr1> | requiring TH seems to make ghc want functions to be defined before use |
| 19:54:31 | <sorear> | twanvl: My original plan had been to concentrate all the prettying logic in Peephole ... any particular reason you are doing the prettying in lK ? |
| 19:54:50 | <twanvl> | No, I just didn't know that |
| 19:54:54 | <sorear> | joelr1: yeah, that's why we have the derivations in separate modules |
| 19:55:05 | <joelr1> | aha! |
| 19:55:19 | <twanvl> | But why not do it immediatly? |
| 19:55:51 | <sorear> | separation of concerns? |
| 19:56:08 | <sorear> | smart constructors can only act as bottom-up traversals |
| 19:56:18 | <xerox> | ghc -e 'cycle (replicate 1000 '\0' ++ replicate 1000 '\255')' > /dev/audio |
| 19:56:19 | <xerox> | alarm |
| 19:56:20 | <sorear> | the peephole can be context sensitive |
| 19:57:39 | <sorear> | also, we really don't want special case code in derivations, because the goal of derive is to make adding derivations easier, and it helps if there is less code involved. |
| 19:58:11 | <boowax> | any takers for helping me with this: http://hpaste.org/1268#a1 ? I'm new to Haskell so it may be a fairly simple fix |
| 19:58:56 | <Saizan> | boowax: you are trying to compare two values which have different types, it seems |
| 19:59:03 | <chessguy> | procyon112: ping? |
| 20:00:20 | <Saizan> | your valTy is not a simple Type, but a Either ErrorTy Type, so you have to unpack it |
| 20:00:35 | <SamB_XP> | Saizan: well, the paste is called "Monad issues" |
| 20:00:36 | <boowax> | saizan: yes, that seems to be the case...the type in question is an Either ErrorTy Type and I just want to compare on Type....is there an easy way to extract it from the Either? |
| 20:00:44 | <SamB_XP> | so remember to use monads in the solution ;-P |
| 20:00:47 | <joelr1> | is there a way to resolve recursive module dependencies in haskell? |
| 20:00:53 | <joelr1> | or with ghc to be precise |
| 20:01:09 | <SamB_XP> | joelr1: you mean {-# SOURCE #-}? |
| 20:01:10 | <sjanssen> | with pain, yes |
| 20:01:18 | <SamB_XP> | and .hs-boot |
| 20:01:26 | <joelr1> | i put my derived parsers in a separate module but just realized (ghc told me) that mod1 (non-derived) needs funs from mod2 |
| 20:01:38 | <sjanssen> | joelr1: it's better to eliminate the cyclic dependencies |
| 20:01:46 | <SamB_XP> | yes, quite |
| 20:01:54 | <procyon112> | sorear: I just had a weird realization |
| 20:01:55 | <joelr1> | sjanssen: that would require moving huge chunks of code from one module to another |
| 20:02:01 | <joelr1> | i can try ... i guess |
| 20:02:07 | <SamB_XP> | well, you can try the other way first |
| 20:02:15 | <sorear> | procyon112: cool, type theory ocks |
| 20:02:24 | <SamB_XP> | but don't blame *me* if you somehow manage to crash GHC (if this somehow involves TH) |
| 20:02:29 | <joelr1> | SamB_XP: which way? moving chunks is not gonna work, chunks are too huge |
| 20:02:36 | <procyon112> | sorear: You know Turing's halting problem thought experiment proof? Imagine a list of all possible computer programs..... |
| 20:02:36 | <joelr1> | this does involve TH |
| 20:02:51 | <sorear> | procyon112: yeah, semi-decidable |
| 20:02:51 | <procyon112> | sorear: My code generates that list :) |
| 20:02:51 | <sjanssen> | joelr1: there is no real limit to cut and paste |
| 20:03:08 | <sorear> | joelr1: stop using _derive_print_instance |
| 20:03:12 | <SamB_XP> | sorear: well, don't worry about it until GHC crashes okay? |
| 20:03:18 | <Saizan> | boowax: you can pattern match on it, like how you do with ty, or use the function either, or reqrite that function in a MonadError fashion maybe |
| 20:03:19 | <joelr1> | sorear: i'm not! |
| 20:03:27 | <joelr1> | sorear: i just ghc --make |
| 20:03:38 | <sorear> | joelr1: you just have a bunch of $( derive ... )'s rgith? |
| 20:03:43 | <joelr1> | $( derive makeFunParser ''StrCall ) |
| 20:03:44 | <joelr1> | yea |
| 20:03:51 | <joelr1> | in mod2 that's required by mod1 |
| 20:03:51 | <xerox> | ghc -e 'let tune = 42 in Monad.join $ map (\(t,s) -> replicate (t*10) s) (zip (Random.randomRs (0::Int,10) (Random.mkStdGen tune)) (Random.randomRs ('\0','\255') (Random.mkStdGen tune)))' > /dev/audio |
| 20:03:54 | <xerox> | mewzik |
| 20:04:08 | <joelr1> | i think i'm an idiot |
| 20:04:19 | <SamB_XP> | procyon112: do you have a decision procedure that determines whether or not a given file is a computer program? |
| 20:04:20 | <sorear> | so mod2 is a bunch of derives,and mod1 is the data types, and they are mutually recursive? |
| 20:04:23 | <procyon112> | sorear: All my future career programming asignments are in that list... Now I just need an algorithm to map specifications to an index, and my entire career is lazily evaluated :) |
| 20:04:34 | <xerox> | (t*100) is more audible, maybe. |
| 20:04:39 | <sorear> | procyon112: beware of the exponential. |
| 20:04:58 | <sorear> | what, people are having fun writing haskell audio programs? |
| 20:05:00 | <SamB_XP> | procyon112: you'd be surprise how long it takes to generate some of those entries |
| 20:05:13 | <SamB_XP> | also, isn't the index essentially the same thing as the entry? |
| 20:05:30 | <joelr1> | i just need to reverse the require order |
| 20:05:32 | <sorear> | my BF theorem prover can't prove S, it's just too complicated |
| 20:05:39 | <joelr1> | and move a few little bits to the TH module |
| 20:06:03 | <stepcut> | argh. So, I tried to use #ifdef __HADDOCK__, but now cpp is stripping out most of my file because it has the strings /* and later */ |
| 20:06:26 | <procyon112> | SamB_XP: Yeah... mapping spec to index is a substantially more difficult problem than generating a lazy list of all possible well-typed programs :) |
| 20:06:46 | <joelr1> | no, does not work |
| 20:07:46 | <joelr1> | sorear: as soon as i derive it wants stuff from ModuleA but it I keep everything in one module then i need to order the functions |
| 20:08:13 | <procyon112> | SamB_XP: It works by generating all possible well-typed abstract syntax trees in a turing complete language. |
| 20:08:20 | <joelr1> | and in at least one case i don't think i can order the functions |
| 20:08:41 | <sorear> | joelr1: can you hpaste a small thing? |
| 20:08:57 | <joelr1> | sorear: sure |
| 20:09:00 | <joelr1> | what do you need? |
| 20:09:11 | <Gwern> | has anyone else been using dmenu? am I just imagining it or does dmenu convert apostrophes to backticks? |
| 20:09:14 | <joelr1> | @paste |
| 20:09:15 | <lambdabot> | Haskell pastebin: http://hpaste.org/new |
| 20:09:29 | <procyon112> | SamB_XP: Using S,K and Y (ap, const and fix) as primitives, all possible programs can be expressed, so this just filters out the poorly typed ones and returns the valid ones in a lazy list. |
| 20:10:07 | <mauke> | Gwern: you're imagining it :-) |
| 20:10:16 | <SamB_XP> | procyon112: dude, that won't be able to fulful the requirements of most specs! |
| 20:10:21 | <chessguy> | procyon112: so when are we going to be ready to integrate? i wanna play! |
| 20:10:31 | <hpaste> | joelr1 pasted "small thing" at http://hpaste.org/1269 |
| 20:10:39 | <joelr1> | sorear: there ^^^ |
| 20:10:47 | <SamB_XP> | people are always wanting the programs to read files or do something on the intarweb or something like that |
| 20:10:51 | <sjanssen> | joelr1: if you're trying to build ghc 6.7 just because you want to use derive, I have patches for you |
| 20:11:15 | <sorear> | procyon112: just pull the occurs check out, and eliminite Y, and optimize the type checker to (const True) |
| 20:11:25 | <joelr1> | sjanssen: i thought i could use derive with 6.6. i'm doing that at the moment, actually. do you have patches for ghc 6.7? because it doesn't build for me :D |
| 20:11:33 | <joelr1> | sjanssen: thanks, of course |
| 20:11:36 | <procyon112> | SamB_XP: I can just add primitives to access those things. |
| 20:11:48 | <mauke> | Gwern: it seems to use LEFT SINGLE QUOTATION MARK [‘] and RIGHT SINGLE QUOTATION MARK [’], though |
| 20:11:54 | <SamB_XP> | what about EGA graphics! |
| 20:12:04 | <joelr1> | sjanssen: i patched derive this morning (someone complained about my 70 messages :D) |
| 20:12:22 | <sjanssen> | joelr1: ah, did you send your fix to ndm? |
| 20:12:24 | <Gwern> | mauke: weird. I wonder how I should fix that? |
| 20:12:24 | <procyon112> | chessguy: Next step is a well-typed crossover operator.... then I should be able to integrate. |
| 20:12:30 | <joelr1> | sjanssen: nope |
| 20:12:40 | <SamB_XP> | hoarder of patches! |
| 20:12:41 | <joelr1> | sjanssen: i thought it was a glitch in the matrix |
| 20:12:41 | <chessguy> | procyon112: cool! any guess on an ETA? |
| 20:12:59 | <SamB_XP> | procyon112: so you can use it to make pretty pictures? |
| 20:13:00 | <sjanssen> | joelr1: was it the *same* cat? |
| 20:13:04 | <joelr1> | i thought that surely sorear and ndm would have the stuff that builds |
| 20:13:08 | <sorear> | joelr1: what's BoolCall, what is boolCall |
| 20:13:08 | <sorear> | ? |
| 20:13:17 | <sjanssen> | joelr1: anyway, I sent ndm a fix for that |
| 20:13:22 | <procyon112> | chessguy: no firm estimate.. couple of days. I might get lots of time to work on it this weekend, or I might get none at all. |
| 20:13:30 | <joelr1> | sorear: BoolCall has a bunch of constructors for funs that return Bool (conceptually) |
| 20:13:32 | <joelr1> | same for others |
| 20:13:33 | <sorear> | joelr1: I don't usually bother to test except on <1 week old ghc:) |
| 20:13:33 | <chessguy> | lol, ok |
| 20:13:38 | <procyon112> | SamB_XP: I could... it's turing complete :) |
| 20:13:40 | <sorear> | joelr1: boolCall? |
| 20:13:44 | <joelr1> | boolCall is the parser, same for numCall, etc. |
| 20:13:56 | <SamB_XP> | procyon112: I was referring to the term "integrate" |
| 20:13:59 | <sorear> | hmm. what do you use boolCall for? |
| 20:13:59 | <joelr1> | sorear: boolCall is supposed to parse all the boolean functions |
| 20:14:12 | <SamB_XP> | it seems that the chief use of integration is making pretty pictures ;-P |
| 20:14:14 | <joelr1> | sorear: i use it in the condExpr, hang on |
| 20:14:29 | <chessguy> | procyon112: i'm thinking about just outputing stats to text files, and then basically using gnuplot for the interface |
| 20:14:52 | <hpaste> | joelr1 annotated "small thing" with "boolCall used here" at http://hpaste.org/1269#a1 |
| 20:14:55 | <sorear> | joelr1: I think you might be able to avoid dependencies by using the class method directly. |
| 20:14:56 | <joelr1> | sorear: check |
| 20:14:57 | <procyon112> | chessguy: That sounds cool for now. |
| 20:14:57 | <boowax> | saizan: thanks for the help! I tried doing it with the pattern matching as you suggested and I'm having other issues with that...can you show me how to use the Either function you mentioned? my attempt at google-ing it was less than helpful |
| 20:15:54 | <chessguy> | procyon112: we can extend your REPL to seed and evolve populations, i think |
| 20:15:55 | <SamB_XP> | ACTION wishes you could change resolution in DOS by dragging on the corner of the window |
| 20:15:56 | <sorear> | joelr1: replace boolCall with parse globally, the type checker will know what you mean |
| 20:16:14 | <sorear> | @docs Data.Either |
| 20:16:14 | <lambdabot> | http://haskell.org/ghc/docs/latest/html/libraries/base/Data-Either.html |
| 20:16:37 | <joelr1> | sorear: interesting... that would surely be hard to grok, though. parse? where does it come from? that type of thing |
| 20:16:46 | <SamB_XP> | well, more to the point, in near-DOS |
| 20:16:54 | <joelr1> | let me try with boolCall |
| 20:16:57 | <procyon112> | chessguy: I can't get my unifier to do infinite types... I've got a thread on haskell-cafe about it... but it's no hold up. I can still do turing-complete trees by adding fix as a primitive. |
| 20:17:31 | <joelr1> | sorear: you rock! |
| 20:17:34 | <sorear> | if you want to be explicit you could use (parse :: CharParser s BoolExpr) |
| 20:17:35 | <sorear> | :) |
| 20:17:36 | <chessguy> | procyon112: i can't even grok infinite types |
| 20:18:04 | <procyon112> | chessguy: Yea.. I can see the REPL and/or GUI doing cool things.. like selecting primitives, kicking off runs, selecting type schemes, all sorts of stuff. |
| 20:19:19 | <procyon112> | chessguy: infinite types are easy. It's the type of general recursion... (a->a)->a. It just means that this might not halt (all non-infinite types halt) because the type refers back to itself. |
| 20:19:51 | <chessguy> | hmm |
| 20:20:13 | <joelr1> | sorear: no, not needed |
| 20:20:24 | <procyon112> | chessguy: so, my unifier's problem is that it can' |
| 20:20:33 | <procyon112> | t figure out infinite types |
| 20:21:18 | <procyon112> | chessguy: but that's ok, because you can cheat, by giving fix as a primitive, and telling the unifier it's type (a->a)->a, so it doesn't have to figure it out. |
| 20:21:46 | <procyon112> | chessguy: and anything that can be resolved to an infinite type, can also be resolved in terms of fix. |
| 20:21:47 | <chessguy> | o |
| 20:21:49 | <chessguy> | ok |
| 20:22:00 | <Saizan> | boowax: another thing, you are using throwError in one of the case branches, so the typesign for checker doesn't seem quite right, unless you have some custom throwError |
| 20:23:26 | <procyon112> | chessguy: Haskell uses that method... It doesn't do infinite type either, but it gives fix as part of the language, and recursive lets and functions in terms of fix, so that any program can be written. |
| 20:23:39 | <chessguy> | i see |
| 20:25:38 | <kscaldef> | anyone around who might be able to help me with a HaXml problem (hopefully a simple one)? |
| 20:26:51 | <Heffalump> | depends on how simple |
| 20:27:32 | <kscaldef> | I'm just trying to parse and validate a document. The parsing goes fine, and the DTD and content look sane |
| 20:27:43 | <procyon112> | chessguy: It might be nice to have an infinite type checker as a kind of GP system because more funky recursive programs can be expressed easier if infinite types are allowed, but using fix works in a pinch, even if it makes it more difficult on the evolution, which I'm not confident it does. |
| 20:27:49 | <kscaldef> | but when I call validate, it compains about everything |
| 20:28:01 | <kscaldef> | every element is unknown, every atttribute is unknown |
| 20:28:09 | <chessguy> | procyon112: we can always add it later, too |
| 20:28:20 | <chessguy> | (assuming we can get it working |
| 20:29:22 | <chessguy> | ACTION finally breaks down and orders TAPL from amazon |
| 20:29:22 | <procyon112> | chessguy: I've never seen a paper on infinite types vs. fix in GP... actually, I've only seen about 3 papers on recursive GP ever :) |
| 20:29:50 | <joelr1> | yay, my stuff compiles! with derive and everything! kudos to sorear |
| 20:29:55 | <kscaldef> | so, it seems like what I want to do is very simple, but I must be doing something dumb |
| 20:30:13 | <joelr1> | ACTION high-fives sorear and goes to watch a movie. hacking to continue a little later |
| 20:30:15 | <procyon112> | PolyGP is the only deep work I've seen on this stuff. |
| 20:30:43 | <chessguy> | that was an interesting paper, if a little out-dated |
| 20:30:52 | <joelr1> | quality time with one's wife is very important! wives can only handle so many all-day haskell hacking sessions |
| 20:31:13 | <Heffalump> | kscaldef: nothing obvious comes to mind |
| 20:31:27 | <chessguy> | joelr1: that's why i need to solve all the world's problems in haskell in the next 10.5 months, before i get married! |
| 20:31:29 | <Heffalump> | but HaXml is a little temperamental in my experience |
| 20:31:39 | <joelr1> | ehehe |
| 20:31:48 | <kscaldef> | hmm... is there another validating XML parser I should consider? |
| 20:32:02 | <joelr1> | ACTION looks forward to using HaXml /will know who to go to/ |
| 20:32:08 | <chessguy> | (man, i'm going to have to start pulling all-nighters soon) |
| 20:32:11 | <joelr1> | ACTION is away |
| 20:32:25 | <monochrom> | I swear by HXT |
| 20:32:35 | <joelr1> | oh, year, here's a going-away present: http://flickr.com/photos/joelr1 |
| 20:32:36 | <lambdabot> | Title: Flickr: Photos from joelr1 |
| 20:32:38 | <procyon112> | chessguy: I've already solved all the worlds problems in Haskell... All the solutions are somewhere in that lazy list of mine :) |
| 20:34:18 | <kscaldef> | monochrom: well, the existence of a cookbook for hxt is encouraging |
| 20:34:40 | <procyon112> | map eval allPossiblePrograms -- is fun to watch |
| 20:35:27 | <Saizan> | hxt is very nice the moment you realize it's like a list monad with arrow combinators |
| 20:36:04 | <emu> | count halting allPrograms / count (not . halting) allPrograms |
| 20:36:19 | <procyon112> | hehe |
| 20:36:23 | <chessguy> | lol |
| 20:36:51 | <Saizan> | you can calculate the chaitin constant! :D |
| 20:36:52 | <monochrom> | I have a code optimiser that simplifies that code to 0. |
| 20:37:00 | <kscaldef> | Saizan: I'm not sure I have enough haskell experience for that realization to apply |
| 20:37:15 | <procyon112> | Saizan: Isn't it effectively 1? |
| 20:37:24 | <Heffalump> | so, arrows are supposed to be good at tracking static properties of functions, right? |
| 20:37:26 | <emu> | count halting allPrograms / length allPrograms, sorry |
| 20:37:44 | <chessguy> | it's oo/oo |
| 20:38:03 | <emu> | counting and length are probably not good ways to put this |
| 20:38:21 | <Saizan> | procyon112, isn't it the probability that a random program halts? and is quite impossible to know its value |
| 20:38:56 | <monochrom> | lim n->oo (count programs under n that halts / n) |
| 20:39:22 | <emu> | that sounds better |
| 20:39:31 | <monochrom> | ACTION was a mathematician |
| 20:39:40 | <emu> | ACTION is not |
| 20:39:47 | <procyon112> | Saizan: The research I've seen estimates the value as infintessimally close to 1. |
| 20:39:48 | <monochrom> | ACTION is not |
| 20:39:52 | <Syzygy-> | monochrom: Was? What happened? |
| 20:40:04 | <monochrom> | I entered computer science. |
| 20:40:43 | <dylan> | ACTION wants a new job |
| 20:40:48 | <procyon112> | Saizan: So, all programs are non-halting, except for the infinitely small subset we are actually interested in ;) |
| 20:41:52 | <emu> | that speaks about decidable languages then too? |
| 20:44:30 | <procyon112> | Essentually, if you introduce general recursion into the language, then the the ratio of non-halting programs to halting approaches 1 as the allowable size of the programs approach infinity. |
| 20:44:58 | <Heffalump> | dylan: what's wrong with your current one? |
| 20:45:15 | <emu> | approaches 1? that would be 1:1 |
| 20:45:44 | <Heffalump> | procyon112: is the argument easy to summarise? |
| 20:45:44 | <procyon112> | emu: er... sorry. ratio of non-halting to total count.. my bad |
| 20:46:28 | <procyon112> | Heffalump: I've only seen it done experimentally.. I'm not sure if it's proven. |
| 20:47:37 | <emu> | whacha do, pick a compact representation of a turing machine and simulate n-bit ones ? |
| 20:47:44 | <emu> | *all n-bit ones |
| 20:47:49 | <emu> | for each n |
| 20:48:10 | <emu> | they still need some kind of bound |
| 20:48:26 | <Saizan> | yeah, you generate all bit-strings of lenght < n and see if they halt |
| 20:48:36 | <procyon112> | emu: I haven't personally, but I've seen a few papers that did more or less that for different representations. |
| 20:49:14 | <emu> | then you're testing linear bounded automatons |
| 20:50:17 | <procyon112> | emu: I'm more interested in the practical applications than the proof, so whether the limit is assymptotically 1 or not is less important to me than "it looks like it, so it's pretty close to something like that", since I'm developing large GP trees. |
| 20:52:08 | <sorear> | What order are the fields in a foreign import decl? |
| 20:53:03 | <sjanssen> | foreign import ccall "foo.h funcname" funcname :: stuff |
| 20:54:15 | <sorear> | where does unsafe go? |
| 20:54:45 | <mauke> | foreign import ccall safe "unistd.h execv" c_execv :: Ptr CChar -> Ptr (Ptr CChar) -> IO CInt |
| 20:54:46 | <sorear> | ok, random search found it |
| 20:54:59 | <sorear> | the ffi could use more DWIM |
| 20:55:07 | <sorear> | permutation parsers aren't that hard |
| 20:55:29 | <sjanssen> | or you could just learn the syntax |
| 20:55:54 | <sorear> | that's like expecting me to memorize the options to fsck |
| 20:56:05 | <sorear> | i just about never use it |
| 20:56:07 | <monochrom> | I heard some programmers bitched about the freedom in Java of allowing "public static final" to be permuted. |
| 20:56:17 | <sorear> | but when I need it, I need documentation or dwim |
| 20:56:28 | <monochrom> | dwim = ? |
| 20:56:39 | <sorear> | do what I (obviously) mean |
| 20:56:59 | <mauke> | int const unsigned typedef long foo; |
| 20:57:58 | <sorear> | bwahaha! |
| 20:58:31 | <sorear> | I've thoughroughly read that section of the spec, and implemented it |
| 20:58:36 | <sorear> | and I still did a double take |
| 20:58:57 | <sorear> | Oh, right, typedef *is* a storage class qualifier... |
| 21:01:02 | <sorear> | *Main Control.Concurrent System.Posix.Unistd> ghc-6.7.20070402: internal error: interpretBCO: unknown or unimplemented opcode 36080 |
| 21:01:05 | <sorear> | nice. |
| 21:01:19 | <monochrom> | haha |
| 21:07:19 | <matthew-_> | ok, I need a map with fold in a monad |
| 21:07:30 | <sjanssen> | @type foldM |
| 21:07:32 | <lambdabot> | forall a b (m :: * -> *). (Monad m) => (a -> b -> m a) -> a -> [b] -> m a |
| 21:07:45 | <matthew-_> | fantastic. cheers. |
| 21:08:51 | <syntaxfree> | oh, boy. |
| 21:09:51 | <monochrom> | hey, girl. |
| 21:12:00 | <Saizan> | syntaxfree, it has been a long time, isn't it? |
| 21:12:08 | <syntaxfree> | yeah. |
| 21:12:18 | <syntaxfree> | I haven't been hacking much. |
| 21:12:20 | <syntaxfree> | or at all. |
| 21:13:22 | <Saizan> | we could have used a rant or two :) |
| 21:14:00 | <matthew-_> | @pl foldM (\(ls,acc) l -> f (l,acc) >>= \(l',acc') -> return (l:ls, acc')) |
| 21:14:02 | <lambdabot> | foldM (uncurry (flip (ap . (((>>=) . f) .) . flip (,)) . flip flip snd . (ap .) . flip flip fst . (((.) . const . (return .) . (,)) .) . flip (:))) |
| 21:14:13 | <matthew-_> | ew. Um, any clues?! |
| 21:14:27 | <syntaxfree> | Lispers have no zygomorphisms. |
| 21:14:35 | <syntaxfree> | I think I'm gonna rant about that. |
| 21:14:50 | <syntaxfree> | I saw an article on CLOS on reddit. |
| 21:15:01 | <Saizan> | zygomorphism? |
| 21:15:07 | <syntaxfree> | the generic dispatch mechanism is precisely what I'm used to in GNU R. |
| 21:15:13 | <syntaxfree> | So I decided to hack on it for a while. |
| 21:15:20 | <syntaxfree> | In the beginning, it's great fun. |
| 21:15:40 | <syntaxfree> | You feel somewhat like cheating on your wife, but, c'mon. Variadic functions, named arguments, macros... |
| 21:15:55 | <syntaxfree> | It's like going to this bordello where the girls will do all that kinky stuff your wife won't. |
| 21:16:19 | <syntaxfree> | but then, after a while, the fun wears off and you start feeling dirty for messing with the impure stuff. |
| 21:16:36 | <syntaxfree> | Finally, you return home thinking of how much you really love your wife and shouldn't have done that. |
| 21:18:16 | <Saizan> | ?type \f -> foldM (\(ls,acc) l -> first (l:) `fmap` f (l,acc) ) |
| 21:18:16 | <syntaxfree> | @help define |
| 21:18:18 | <lambdabot> | forall b d (m :: * -> *). (Functor m, Monad m) => ((b, d) -> m ([b], d)) -> ([b], d) -> [b] -> m ([b], d) |
| 21:18:19 | <lambdabot> | help <command>. Ask for help for <command>. Try 'list' for all commands |
| 21:19:37 | <matthew-_> | ahh, I thought I didn't know "first" - in Arrows. |
| 21:19:58 | <matthew-_> | Saizan: cheers, a solution that forces me to learn Arrows ;) |
| 21:21:09 | <Saizan> | well first f (x,y) = (f x,y) for the (->) arrow |
| 21:21:10 | <sorear> | Is there a recommended style for implementation documentation? |
| 21:21:16 | <sorear> | @src (->) first |
| 21:21:17 | <lambdabot> | first f = f *** id |
| 21:21:23 | <sorear> | @src (->) *** |
| 21:21:23 | <lambdabot> | Source not found. Sorry about this, I know it's a bit silly. |
| 21:21:28 | <sorear> | @src (->) (***) |
| 21:21:29 | <lambdabot> | (f *** g) ~(x,y) = (f x, g y) |
| 21:21:49 | <syntaxfree> | > ((+1) *** (-1)) (3,5) |
| 21:21:49 | <lambdabot> | add an instance declaration for (Num (b' -> c')) |
| 21:21:59 | <sorear> | :t (-1) |
| 21:22:01 | <lambdabot> | forall a. (Num a) => a |
| 21:22:02 | <sorear> | nice try. |
| 21:22:11 | <sorear> | sections aren't that simple :( |
| 21:22:26 | <Saizan> | screw unary minus! |
| 21:22:34 | <luis> | ACTION eyes syntaxfree |
| 21:22:41 | <sorear> | ACTION almost never uses negative numbers in Haskell |
| 21:22:41 | <syntaxfree> | > ((+1) *** (subtract 1)) (3,5) |
| 21:22:43 | <lambdabot> | (4,4) |
| 21:23:01 | <sorear> | ACTION wouldn't mind if unary minus was spelled N E G A T E |
| 21:23:06 | <sjanssen> | I think - should be made part of the lexical syntax for num literals |
| 21:23:42 | <sorear> | sjanssen: but then foo-1 woudn't work |
| 21:23:47 | <sjanssen> | such that (-1) parses as negative one, and (- 1) parses as subtract 1 |
| 21:24:02 | <sorear> | foo-1 lexes now as foo - 1 |
| 21:24:09 | <sorear> | foo-1 would lex as foo -1 |
| 21:24:11 | <sjanssen> | sorear: correct. I think it's less hackish than what we have now |
| 21:24:25 | <syntaxfree> | maybe we should undefine (-) altogether. |
| 21:24:28 | <syntaxfree> | leave it for custom operators. |
| 21:24:29 | <sorear> | I think it should be eliminated altogether |
| 21:24:39 | <matthew-_> | Saizan: sorry, my original was wrong! |
| 21:24:43 | <syntaxfree> | it'd be great fun to make (-) = (*) |
| 21:24:49 | <matthew-_> | foldM (\(ls,acc) l -> f (l,acc) >>= \(l',acc') -> return (l':ls, acc')) |
| 21:25:00 | <matthew-_> | i.e. the l' in the return |
| 21:26:34 | <Saizan> | ?type \f -> foldM (\(ls,acc) l -> first (:ls) `fmap` f (l,acc)) --? |
| 21:26:36 | <lambdabot> | forall b b1 d (m :: * -> *). (Functor m, Monad m) => ((b, d) -> m (b1, d)) -> ([b1], d) -> [b] -> m ([b1], d) |
| 21:27:16 | <matthew-_> | Saizan: yep, quite why I couldn't work that out, I'm not sure ;) |
| 21:27:22 | <procyon112> | ?hoogle (***) |
| 21:27:22 | <lambdabot> | Did you mean: (***) |
| 21:27:23 | <lambdabot> | Prelude.undefined :: a |
| 21:27:23 | <lambdabot> | Control.Monad.Reader.ask :: MonadReader r m => m r |
| 21:28:29 | <sjanssen> | ACTION wonders how import Prelude hiding ((-)) would work |
| 21:29:06 | <mauke> | ?hoogle *** |
| 21:29:07 | <lambdabot> | Control.Arrow.(***) :: Arrow a => a b c -> a b' c' -> a (b, b') (c, c') |
| 21:30:25 | <procyon112> | that's handy. I have no idea what arrows are, but *** will be very useful :) |
| 21:30:56 | <mauke> | ?hoogle &&& |
| 21:30:56 | <lambdabot> | Control.Arrow.(&&&) :: Arrow a => a b c -> a b c' -> a b (c, c') |
| 21:31:12 | <Saizan> | ?hoogle +++ |
| 21:31:12 | <lambdabot> | Control.Arrow.(+++) :: ArrowChoice a => a b c -> a b' c' -> a (Either b b') (Either c c') |
| 21:31:13 | <lambdabot> | Text.Html.(+++) :: (HTML a, HTML b) => a -> b -> Html |
| 21:31:13 | <lambdabot> | Text.ParserCombinators.ReadP.(+++) :: ReadP a -> ReadP a -> ReadP a |
| 21:41:39 | <sorear> | ACTION uses Haskell to write the world's first unlambda compiler! |
| 21:45:26 | <matthew-_> | grrr. Do I take it that readFile and writeFile close the file handles as lazily as possible? |
| 21:45:43 | <bos> | yes. |
| 21:45:52 | <matthew-_> | so if you've read it with readFile and then later on you try to write to it you have major issues? |
| 21:45:59 | <bos> | yes. |
| 21:46:09 | <bos> | at least if you're using windows. |
| 21:46:14 | <matthew-_> | linux here! |
| 21:46:15 | <matthew-_> | so the way round is to actually get the handles properly and hClose them? |
| 21:46:18 | <bos> | but probably on other OSes too. |
| 21:46:32 | <bos> | yes, you need to do the I/O quite imperatively. |
| 21:46:37 | <matthew-_> | ok, thanks |
| 21:49:27 | <qwr> | matthew-_: use strict bytestring's readFile |
| 21:49:41 | <qwr> | matthew-_: if you want the whole file |
| 21:49:58 | <qwr> | and want to close it right away |
| 21:52:20 | <matthew-_> | qwr: ahh, thanks. I'll look into that. |
| 21:53:37 | <qwr> | and Data.ByteString.Char8 unpacks as [Char] |
| 21:54:12 | <desp> | @whereis tackling |
| 21:54:13 | <lambdabot> | Maybe you meant: where where+ |
| 21:54:17 | <desp> | @where tackling |
| 21:54:18 | <lambdabot> | I know nothing about tackling. |
| 21:54:24 | <desp> | @where awkward squad |
| 21:54:25 | <lambdabot> | I know nothing about awkward. |
| 21:54:48 | <mauke> | @google awkward squad |
| 21:54:51 | <lambdabot> | http://en.wikipedia.org/wiki/Awkward_Squad |
| 21:54:51 | <lambdabot> | Title: Awkward Squad - Wikipedia, the free encyclopedia |
| 21:54:57 | <desp> | @remember tackling http://research.microsoft.com/~simonpj/papers/marktoberdorf/ |
| 21:54:58 | <lambdabot> | Done. |
| 21:56:29 | <sorear> | @flush |
| 21:56:47 | <mauke> | @unflush |
| 21:56:47 | <lambdabot> | Not enough privileges |
| 21:56:52 | <sorear> | just ... so many opportunities for side tracking |
| 21:56:53 | <mauke> | :( |
| 21:57:32 | <desp> | @remember awkward http://research.microsoft.com/~simonpj/papers/marktoberdorf/ |
| 21:57:32 | <lambdabot> | Done. |
| 21:57:36 | <desp> | @where awkward |
| 21:57:36 | <lambdabot> | I know nothing about awkward. |
| 21:57:39 | <desp> | argh! |
| 21:57:49 | <desp> | I lose. |
| 21:58:14 | <sorear> | @quote awkward |
| 21:58:14 | <lambdabot> | awkward says: http://research.microsoft.com/~simonpj/papers/marktoberdorf/ |
| 21:58:25 | <sorear> | @quote tackling |
| 21:58:26 | <lambdabot> | tackling says: http://research.microsoft.com/~simonpj/papers/marktoberdorf/ |
| 21:59:00 | <sorear> | @where+ tackling http://research.microsoft.com/~simonpj/papers/marktoberdorf/ |
| 21:59:00 | <lambdabot> | Done. |
| 21:59:08 | <sorear> | @where+ awkward http://research.microsoft.com/~simonpj/papers/marktoberdorf/ |
| 21:59:09 | <lambdabot> | Done. |
| 21:59:18 | <mauke> | @where tackling the awkward squad |
| 21:59:19 | <lambdabot> | http://research.microsoft.com/~simonpj/papers/marktoberdorf/ |
| 22:00:09 | <desp> | sorear: thanks |
| 22:00:16 | <sorear> | yw |
| 22:01:07 | <desp> | sorear: care to get sidetracked again? how do you like http://desp.night.pl/juwenalia3.png as a t-shirt design? ;) |
| 22:02:15 | <sorear> | Heh. cute. (Just personally not big on patterned clothing) |
| 22:02:23 | <desp> | ah. |
| 22:02:46 | <sorear> | I can't read the text but something screams "promoting a technical conference and/or video game" |
| 22:03:08 | <desp> | it's actually for a student's festival. for the students of the departament of mathematics and computer science. |
| 22:03:18 | <desp> | s/'s/s'/ |
| 22:03:59 | <dmwit> | Just a guess, but is matematyki "mathematics"? ;-) |
| 22:04:05 | <desp> | yup |
| 22:04:16 | <desp> | and you could probably say "informatics" |
| 22:04:21 | <dmwit> | And "wydziat" is clearly "Wuzdat?" |
| 22:04:29 | <desp> | ;) |
| 22:04:45 | <sorear> | What's the best haskell compiler for recursive-newtype-intensive programs? (hbc?) |
| 22:05:07 | <mauke> | haha, "what's the the best compiler for X" |
| 22:05:35 | <sorear> | there is a undisputed best compiler for general haskell, GHC |
| 22:05:38 | <mauke> | .oO( metahc -fghc ) |
| 22:05:54 | <sorear> | but a long-standing bug renders it useless for my purpose |
| 22:05:58 | <mauke> | just change the build options, etc |
| 22:06:07 | <sorear> | we actually have that |
| 22:06:10 | <sorear> | @where hmake |
| 22:06:10 | <lambdabot> | http://haskell.org/hmake |
| 22:21:51 | <dmead> | hey channel |
| 22:22:00 | <dmead> | i'm in type constructor hell =/ |
| 22:24:22 | <stepcut> | heh |
| 22:24:32 | <monochrom> | type constructor is easy |
| 22:25:46 | <monochrom> | actually everything is easy. then humans cook up messy code. so there is some chance you are in complexity hell and not type constructor per se, and the complexity is self-inflicted. |
| 22:26:41 | <wmind> | why don't imperative languages have nice simple sum types |
| 22:26:52 | <wmind> | most even have some sort of case for the pattern matching... |
| 22:27:10 | <mauke> | OCaml does :-) |
| 22:27:30 | <dmead> | i have a dependant typing problem actually |
| 22:27:30 | <wmind> | does OCalm count as imperative? |
| 22:27:38 | <dmead> | it's in between |
| 22:27:57 | <stepcut> | what about javascript ? |
| 22:27:58 | <dmead> | ?seen sorear |
| 22:27:59 | <lambdabot> | sorear is in #darcs, #haskell-overflow, #ghc and #haskell. I last heard sorear speak 21m 48s ago. |
| 22:28:05 | <dmead> | ?seen dons |
| 22:28:06 | <lambdabot> | dons is in #haskell-soc, #haskell-overflow and #haskell. I last heard dons speak 10h 57m 30s ago. |
| 22:28:19 | <dmead> | yeaaa |
| 22:28:20 | <wmind> | stepcut: I don't know :) |
| 22:28:27 | <monochrom> | Pascal has nice simple sum type. |
| 22:28:30 | <dmead> | i'm trying to figure out a dependant type structure for FOL |
| 22:28:34 | <dmead> | i mean |
| 22:28:36 | <dcoutts> | dmead: he's probably still a sleep |
| 22:28:40 | <dmead> | ah |
| 22:28:45 | <wmind> | monochrom: you mean variant records? |
| 22:28:52 | <monochrom> | Yes. |
| 22:29:07 | <dmead> | you know whats nice |
| 22:29:27 | <dmead> | if you include a type synonym in an data declaration |
| 22:29:31 | <dmead> | it can't be shown |
| 22:29:34 | <dmead> | =/ |
| 22:29:35 | <lightstep> | i think Fortress is functional, since you have some kind of tagging for side-effects of functions, like monadic operations in askell |
| 22:29:36 | <wmind> | monochrom: well that's not even safe ... |
| 22:29:37 | <dmead> | apparently... |
| 22:29:52 | <monochrom> | I recall that it's safe. C's union not safe. |
| 22:30:39 | <wmind> | monochrom: how can it be safe? |
| 22:30:47 | <monochrom> | OK, I forgot. |
| 22:31:02 | <wmind> | monochrom: it's a record, and some fields are enabled depending on others |
| 22:31:18 | <wmind> | monochrom: you check the flag field and decide what to access |
| 22:31:24 | <wmind> | monochrom: it can't be safe |
| 22:32:17 | <dmead> | ?paste |
| 22:32:18 | <lambdabot> | Haskell pastebin: http://hpaste.org/new |
| 22:32:26 | <wmind> | monochrom: plus I don't even find that simple :D an haskell like sum and then pattern matching in case would do the job |
| 22:33:18 | <augustss> | yo |
| 22:34:46 | <sorear> | hello! |
| 22:35:22 | <hpaste> | dmead pasted "dependant types for FOL?" at http://hpaste.org/1270 |
| 22:35:23 | <sorear> | dmead: you called? |
| 22:35:47 | <hpaste> | dmead annotated "dependant types for FOL?" with "(no title)" at http://hpaste.org/1270#a1 |
| 22:35:55 | <dmead> | allo |
| 22:36:14 | <dmead> | augustss: you probably know the answer to this |
| 22:36:44 | <dmead> | i'm trying to figure out a nice way to represent FOL as type constructors |
| 22:37:47 | <sorear> | augustss: can hbc survive newtype V = V (U V -> U V) |
| 22:37:48 | <augustss> | dmead: you need dependent types to represent quantifiers |
| 22:38:14 | <sorear> | I non-negotiably have that type in my program. I could use hugs, but I'd like an optimizing compiler. |
| 22:38:20 | <augustss> | sorear: probably, but i don't know |
| 22:39:03 | <dmead> | augustss: indeed, did you have seperate types for propositions, predicates and variables in dijnn? |
| 22:39:11 | <sorear> | About how much of a fight do you think hbc install will put up? |
| 22:39:15 | <Igloo> | sorear: Is this the known GHC-loops bug? If so, and you have a real use for it, then you should let SPJ know |
| 22:39:36 | <augustss> | what djinn does have little relevance, since djinn only does propositional logic |
| 22:40:03 | <dmead> | oh |
| 22:40:11 | <dmead> | i thought yours did FOL |
| 22:40:24 | <sorear> | Igloo: does it matter if the program is not one anybody will actually use? |
| 22:40:43 | <sorear> | Igloo: or rather, the functionality is pretty useless |
| 22:40:58 | <sorear> | Igloo: it's an Unlambda compiler |
| 22:41:06 | <augustss> | dmead: no quantifiers. quantifiers makes it undecidable |
| 22:41:10 | <sorear> | should I still tell him? |
| 22:41:36 | <augustss> | sorear: installing hbc is most likely a major pain |
| 22:41:38 | <dmead> | augustss: what method for proving did you use? |
| 22:42:15 | <sorear> | some kind of CJK calculus |
| 22:42:18 | <Igloo> | sorear: If you want to run the program, then yes |
| 22:42:27 | <augustss> | dmead: a variation of Gentzen's LJ |
| 22:42:29 | <Igloo> | sorear: The description on http://www.haskell.org/ghc/docs/6.6/html/users_guide/bugs.html#bugs-ghc says: |
| 22:42:31 | <sorear> | there's a well commented source file in the djinn repo |
| 22:42:32 | <lambdabot> | Title: 12.2. Known bugs or infelicities, http://tinyurl.com/22mp2v |
| 22:42:33 | <dmead> | hmm |
| 22:42:35 | <Igloo> | We have never found another class of programs, other than this contrived one, that makes GHC diverge, and fixing the problem would impose an extra overhead on every compilation. So the bug remains un-fixed. |
| 22:42:46 | <dmead> | i'm trying to do resolution |
| 22:42:55 | <sorear> | @where spj |
| 22:42:56 | <lambdabot> | I know nothing about spj. |
| 22:43:12 | <Igloo> | assuming that is the same problem you're running into |
| 22:43:34 | <Igloo> | glasgow-haskell-users@haskell.org is probably the best place to bring it up |
| 22:43:36 | <augustss> | dmead: does resolution give you a constructive proof? |
| 22:43:55 | <dmead> | augustss: yes |
| 22:44:16 | <dmead> | err |
| 22:44:19 | <dmead> | it's gives you a proof |
| 22:44:24 | <dmead> | what do you mean by constructive proof? |
| 22:44:57 | <monochrom> | Whenever "a or b" is proved, the proof tells you "it's actually a" or "it's actually b". |
| 22:45:01 | <augustss> | you can't use the law of excluded middle (or double negation, or something equivalent) |
| 22:45:15 | <dmead> | ah |
| 22:45:21 | <monochrom> | Whenever "exist x. Px" is proved, the proof tells you a lot about the x chosen. |
| 22:45:29 | <augustss> | I wanted a proof method that gave me programs |
| 22:45:35 | <dmead> | ah |
| 22:45:59 | <dmead> | resolution just says when you have a v b and ~a v c |
| 22:46:05 | <dmead> | you can write b v c |
| 22:46:27 | <monochrom> | yeah, usually, when proving "exist x. Px", an algorithm (maybe slow but still) is produced for crunching out an x. |
| 22:47:06 | <dmead> | hmm, resolution in FOL is still much simpler than that |
| 22:47:07 | <monochrom> | Yes that one kills constructivity. |
| 22:47:08 | <augustss> | dmead: but that uses excluded middle |
| 22:47:32 | <dmead> | whats wrong with that? |
| 22:47:41 | <dmead> | @hoogle halp |
| 22:47:42 | <lambdabot> | No matches found |
| 22:47:45 | <dmead> | hmm |
| 22:47:46 | <augustss> | proving for propositional calculus in classical logic is easy |
| 22:47:58 | <dmead> | right |
| 22:48:06 | <dmead> | in FOL you skolemize each predicate |
| 22:48:08 | <augustss> | it's the constructive part that is the challange |
| 22:48:21 | <dmead> | and it gives you an interpretation ( i think ) that is valid to do resolution with |
| 22:48:22 | <monochrom> | It doesn't tell you "it's b" or "it's c". More detailedly, it says "a or ~a, but I don't care which, it's probably impossible to find out anyway" |
| 22:48:38 | <dmead> | oh well |
| 22:48:40 | <stepcut> | Igloo: are you subscribed to debian-haskell@lists.urchin.earth.li ? |
| 22:48:49 | <dmead> | thats just the resolution rulew |
| 22:49:02 | <dmead> | the entire proof method requires you to break a sentence into a list of conjuncts |
| 22:49:14 | <Igloo> | stepcut: Yes |
| 22:49:20 | <augustss> | dmead: what method you use depends on what your purpose is (and your philosophy :) ) |
| 22:49:26 | <dmead> | hmm |
| 22:49:33 | <monochrom> | Yes, one could refrain from that rule. Fewer things can be proved, but all proofs provide more information. |
| 22:50:00 | <dmead> | resolution is an inference rule |
| 22:50:05 | <monochrom> | It is okay because constructivity does mean fewer things provable anyway. |
| 22:50:57 | <sorear> | @all-dicts skolemize |
| 22:50:58 | <lambdabot> | No match for "skolemize". |
| 22:51:18 | <dmead> | http://planetmath.org/encyclopedia/Skolemization.html |
| 22:51:20 | <lambdabot> | Title: PlanetMath: Skolemization |
| 22:52:13 | <augustss> | Thoralf Skolem |
| 22:52:16 | <sorear> | brilliant |
| 22:52:45 | <sorear> | computing a supertype of an arbitrary type with no existentail quantification... |
| 22:52:57 | <monochrom> | it sounds like resolution is in-grained in classical non-constructive logic then. |
| 22:53:11 | <luis> | Does anybody have an HTML version of YAHT? |
| 22:53:26 | <dmead> | ni |
| 22:53:28 | <dmead> | no |
| 22:53:32 | <Saizan> | ?google wikibook YAHT |
| 22:53:34 | <lambdabot> | http://en.wikibooks.org/wiki/Haskell/YAHT |
| 22:53:43 | <sorear> | how embarrasing |
| 22:53:46 | <dmead> | you can use resolution in FOL if you have a correct skolem function |
| 22:53:50 | <sorear> | @hoogle Char -> IO () |
| 22:53:50 | <lambdabot> | Prelude.putChar :: Char -> IO () |
| 22:53:51 | <lambdabot> | System.Console.Readline.completeInternal :: Char -> IO () |
| 22:53:51 | <lambdabot> | System.Console.Readline.setPendingInput :: Char -> IO () |
| 22:53:57 | <luis> | Saizan: thanks |
| 22:54:00 | <monochrom> | <a href="yaht.pdf">yaht</a> |
| 22:54:01 | <sorear> | I can't believe putChar slipped :) |
| 22:54:11 | <dmead> | hey |
| 22:54:20 | <dmead> | i'm making a type synonym |
| 22:54:21 | <sorear> | ndm++ |
| 22:54:24 | <sorear> | @flush |
| 22:54:24 | <dmead> | which is Char -> [Char] |
| 22:54:30 | <dmead> | can i add a show instance for that? |
| 22:54:36 | <sorear> | no |
| 22:54:38 | <dmead> | =/ |
| 22:54:41 | <sorear> | you can if it's a newtype |
| 22:54:47 | <dmead> | HMM |
| 22:54:48 | <dmead> | alright |
| 22:54:54 | <sorear> | but type synonyms are basically macros |
| 22:55:14 | <dmead> | right |
| 22:55:23 | <dmead> | it's sorta tricky |
| 22:55:46 | <dmead> | i've got a few paragraphs what what a well formed formula for FOL is |
| 22:56:03 | <dmead> | and i'm trying to cram that into not as many lines of data types |
| 22:56:14 | <dmead> | *of what |
| 22:57:52 | <monochrom> | should be just a few lines |
| 22:58:00 | <dmead> | yea |
| 22:58:18 | <dmead> | i keep running into problems with show =/ |
| 22:58:27 | <augustss> | why |
| 22:58:35 | <augustss> | can't you just derive it? |
| 23:01:48 | <Igloo> | stepcut: Why did you ask? Did I miss a message there? |
| 23:02:11 | <hpaste> | dmead pasted "layout manager hates me" at http://hpaste.org/1271 |
| 23:03:38 | <stepcut> | Igloo: no, but I am going to send one, and I wanted to make sure you had the opportunity to see it |
| 23:03:47 | <Igloo> | OK :-) |
| 23:03:53 | <stepcut> | Igloo: debian packages (everyones, not just yours) install the documentation in /usr/share/doc |
| 23:04:05 | <hpaste> | sorear pasted "Is this a GHCi bug?" at http://hpaste.org/1272 |
| 23:04:17 | <Igloo> | stepcut: I have plans to do something else, probably for 6.6.1 |
| 23:04:29 | <stepcut> | Igloo: but ghc-pkg describe pkgName, says that it is stored in /usr/share/pkgName-1.0/doc/html |
| 23:04:51 | <stepcut> | Igloo: which is, (a) wrong (b) unfortunate, because haddock uses that information to try to create links between packages |
| 23:04:51 | <Igloo> | stepcut: I'm not quite sure what something else is, but I do now know how to merge docs into one big haddock index |
| 23:04:58 | <Igloo> | stepcut: Right, there's a bug about that too |
| 23:05:02 | <Igloo> | I'll look into it at the same time |
| 23:05:23 | <sorear> | Can someone take a quick look at that paste and tell me if I should file a ticket? |
| 23:05:23 | <stepcut> | Igloo: spiffy, I wasn't sure if anyone had noticed yet |
| 23:05:40 | <Igloo> | I think Ross Patterson filed it a week or so ago |
| 23:06:04 | <augustss> | dmead: what's the 'a in 'Formula a' anyway? |
| 23:06:16 | <monochrom> | s/Not Formula/Not (Formula a)/ |
| 23:06:54 | <augustss> | I mean, i can see it in quantifier, but it makes little sense to me |
| 23:06:58 | <chessguy> | ?uptime |
| 23:06:58 | <lambdabot> | uptime: 1d 7h 56m 42s, longest uptime: 5d 3h 40m 45s |
| 23:07:09 | <stepcut> | Igloo: spiffy, I want to make sure I update my cabalDebianTemplate when we figure out what the right thing to do is |
| 23:08:09 | <augustss> | sorear: seems a bit fishy |
| 23:08:12 | <stepcut> | Igloo: also, since I am bugging you, what is the policy for updating things like libghc6-time-dev ? Version 1.1 has bringerts time parsing functions, and fixes a bug in formatDate |
| 23:08:14 | <Igloo> | *nod* |
| 23:08:46 | <sorear> | Igloo: I can't find the ticket |
| 23:09:02 | <Igloo> | stepcut: Prod me, I guess :-) |
| 23:09:08 | <stepcut> | ok |
| 23:09:33 | <stepcut> | I wasn't sure if it was frozen due to etch, policy, or just no one complaining :) |
| 23:09:48 | <Igloo> | sorear: It's not in tract AFAIK, just documented at the URL I gave above (in the GHC user guide) |
| 23:10:52 | <stepcut> | you can't make rfc2822 date's using time-1.0 because there is no (clean) way to show seconds rounded to whole numbers ;) |
| 23:11:06 | <Igloo> | stepcut: Oh, true, technically I probably shouldn't upload it to unstable |
| 23:11:11 | <Cin> | hi. the wikibook about Haskell is really good, so far. I especially like the summaries and excersizes, they help a lot to re-enforce what I has just been read. thanks for the link :) |
| 23:11:12 | <stepcut> | s/date's/dates/ |
| 23:11:39 | <stepcut> | Igloo: are you involved with the GSoC project to try to auto debianize and autobuild hackagedb ? |
| 23:12:24 | <sorear> | Igloo: Oh, this paste has nothing whatsoever with recursive newtypes. |
| 23:12:25 | <sorear> | isearch-backward doesn't take regexes?! |
| 23:12:33 | <sorear> | found it (isearch-backward-regexp, C-M-r) |
| 23:12:37 | <stepcut> | :p |
| 23:13:20 | <Igloo> | sorear: Oh, well, if it's an unknown bug then file it! |
| 23:13:26 | <sorear> | stepcut: does that mean I should take aptage off my TODO list? |
| 23:13:50 | <stepcut> | sorear: dunno, I am not even sure if the project was approved |
| 23:14:32 | <sorear> | Igloo: Is a registered wiki account 'sorear' usable for bug reporting on GHC-trac? ISTR some of the others were locked. |
| 23:14:47 | <sorear> | s/wiki// |
| 23:16:35 | <Igloo> | sorear: It is now |
| 23:17:17 | <sorear> | ty |
| 23:25:06 | <dmead> | augustss: dont i need it there for the other types? |
| 23:25:23 | <dmead> | wait |
| 23:25:24 | <dmead> | nm |
| 23:25:25 | <dmead> | XD |
| 23:25:37 | <augustss> | dmead: what other types would you need? |
| 23:26:04 | <dmead> | for efficency reasons in resolution i need a seperate type for propositional classes |
| 23:26:25 | <augustss> | hmmmm |
| 23:26:32 | <dmead> | and predicates, and the quantifiers |
| 23:26:42 | <dmead> | i know it could all be in the same recursive type |
| 23:26:50 | <augustss> | go for the simple solution first, then optimize |
| 23:26:51 | <dmead> | but that complicates things at many stages |
| 23:27:06 | <sorear> | #1268 |
| 23:27:08 | <dmead> | this is my attempt at optimization |
| 23:27:10 | <dmead> | hehe |
| 23:27:22 | <dmead> | my prover thats up on the wiki already does propositional stuff |
| 23:28:11 | <sorear> | There's efficiency and then there is efficiency. |
| 23:28:21 | <sorear> | Avoiding nub is the first kind. |
| 23:28:35 | <dmead> | is nub short for something? |
| 23:28:39 | <sorear> | no |
| 23:28:41 | <dmead> | cause thats e-slang for newb |
| 23:28:47 | <augustss> | dmead: no |
| 23:28:53 | <augustss> | it's an english word |
| 23:28:54 | <sorear> | Writing pretty printing combinator libraries that do not allocate using GHC primops is the second kind. |
| 23:29:24 | <dmead> | ah |
| 23:29:37 | <sorear> | Wow, it's even in my dictionary! |
| 23:29:43 | <augustss> | dmead: meaning "the essence of" |
| 23:30:15 | <narain> | is there a good type class for things that behave like invertible functions? |
| 23:30:26 | <augustss> | narain: no |
| 23:30:41 | <augustss> | narain: but you can make one |
| 23:30:42 | <sorear> | Transform isn't good enough for you? |
| 23:30:49 | <narain> | yes, i guess i will |
| 23:30:54 | <sorear> | augustss: he did, yesterday |
| 23:31:19 | <augustss> | i guess it all depends on what ones means by "is" |
| 23:31:20 | <narain> | sorear: yes, then i saw a discussion of Monoid today |
| 23:31:43 | <sorear> | Invertable functions form a group, which is a ... duh! |
| 23:31:44 | <narain> | then i thought, hey, a transform is a monoid |
| 23:31:52 | <sorear> | a transform is a group. |
| 23:31:57 | <sorear> | a group is a transform. |
| 23:32:01 | <sorear> | they are the same. |
| 23:32:09 | <narain> | not really...? |
| 23:32:12 | <sorear> | transform2 is still needed however |
| 23:32:35 | <sorear> | class Transform t where |
| 23:32:35 | <sorear> | comp :: t -> t -> t |
| 23:32:35 | <sorear> | nill :: t |
| 23:32:35 | <sorear> | invr :: t -> t |
| 23:32:36 | <narain> | a transform is something that transforms other things |
| 23:33:00 | <sorear> | class Group g where |
| 23:33:00 | <sorear> | operate :: t -> t -> t |
| 23:33:00 | <sorear> | identity :: t |
| 23:33:00 | <sorear> | inverse :: t -> t |
| 23:33:18 | <narain> | ah well |
| 23:33:27 | <narain> | guess it's more of a different perspective |
| 23:33:35 | <sorear> | class Group t => Transform t e where |
| 23:33:36 | <sorear> | transform :: t -> e -> e |
| 23:33:46 | <narain> | the reals form a group but not really a transform |
| 23:33:52 | <narain> | sorear: right |
| 23:34:03 | <narain> | so is Group built-in? |
| 23:34:07 | <sorear> | no :( |
| 23:34:15 | <sorear> | further we require a 'homomorphism' law |
| 23:34:28 | <sorear> | transform a . transform b == transform (a `comp` b) |
| 23:34:40 | <sorear> | id == transform nill |
| 23:34:47 | <sek> | what module are abstract algebra notions in? |
| 23:34:53 | <dmead> | can you guys see an error here? |
| 23:34:54 | <dmead> | http://hpaste.org/1271 |
| 23:34:58 | <sorear> | sek: none |
| 23:35:04 | <dmead> | ghci says i have a bad layout |
| 23:35:04 | <sorear> | @where numericprelude |
| 23:35:04 | <lambdabot> | darcs.haskell.org/numericprelude/ |
| 23:35:06 | <dmead> | but i don't see how |
| 23:35:14 | <sorear> | ^^^ non standard lib |
| 23:35:32 | <sjanssen> | dmead: "newtype Proposition = Char" probably doesn't mean what you think it does |
| 23:35:44 | <sek> | nice, thanks. Will check it out |
| 23:35:51 | <dmead> | ah |
| 23:36:11 | <sjanssen> | dmead: but the compiler is complaining about layout? |
| 23:36:16 | <dmead> | yes |
| 23:36:20 | <dmead> | i changed newtype to data |
| 23:36:22 | <dmead> | and it's alright |
| 23:36:25 | <narain> | yeah, i really feel the numeric prelude should be part of the standard |
| 23:36:30 | <narain> | ACTION <3 math |
| 23:36:33 | <sorear> | yay for type systems |
| 23:36:35 | <sorear> | curr = unsafePerformIO (newIORef Nothing) |
| 23:36:43 | <dmead> | narain: theres a discussion on the mailing list to that effect |
| 23:36:48 | <sorear> | thankfully I caught that before trying to use it :) |
| 23:37:00 | <narain> | dmead: what was the consensus? |
| 23:37:11 | <dmead> | i forget |
| 23:37:17 | <dmead> | sorear started the topic |
| 23:37:24 | <narain> | ok, i'll look |
| 23:37:29 | <sorear> | "Why the Prelude must die" ? |
| 23:37:53 | <sorear> | IMO it was fairly OT by the time the numeric prelude was mentioned. |
| 23:38:28 | <sjanssen> | dmead: I'm not getting layout errors |
| 23:38:52 | <dmead> | weird |
| 23:38:59 | <hpaste> | sjanssen annotated "layout manager hates me" with "fixed" at http://hpaste.org/1271#a1 |
| 23:39:11 | <ptolomy> | Y'know, I used to think the Haskell Anti-Syntactical Sugar Lobby was a bit goofy, but just yesterday I was converting a ~200 line program that used strings extensively to use ByteString, and I realized how often I use limiting syntax. |
| 23:39:37 | <sjanssen> | dmead: I got a kind error " `Formula' is not applied to enough type arguments" |
| 23:39:48 | <dcoutts> | ptolomy: syntactic sugar is great :-) |
| 23:39:51 | <Botje> | HASSL? that's cool. |
| 23:40:02 | <dmead> | ah |
| 23:40:06 | <dmead> | are you on 6.6 per chance? |
| 23:40:06 | <ptolomy> | Now I think the Haskell Anti-Syntactical Sugar Lobby is slightly less goofy. :) |
| 23:40:08 | <dcoutts> | ptolomy: and yeah, I miss it when using ByteString |
| 23:40:09 | <narain> | sorear: good golly that's a long thread |
| 23:40:16 | <sjanssen> | dmead: yes. Are you using tabs? |
| 23:40:21 | <dmead> | yes |
| 23:40:22 | <dmead> | hehe |
| 23:40:27 | <dmead> | i think |
| 23:40:42 | <sjanssen> | I think we just need to make our sugar overloaded |
| 23:40:42 | <dmead> | i forget if emacs just maps tab to spaces for haskell-mode |
| 23:40:47 | <dcoutts> | ptolomy: so surely it's an argument for extending the syntactic sugar rather than eliminating it |
| 23:40:56 | <dcoutts> | sjanssen: just |
| 23:40:58 | <dcoutts> | erm |
| 23:41:01 | <dcoutts> | just/right |
| 23:41:02 | <ptolomy> | dcoutts: Yeah, ruins all my fun pattern matching. |
| 23:41:09 | <ptolomy> | dcoutts: Absolutely. |
| 23:41:12 | <dmead> | i wouldn't want to use haskell without the sugar |
| 23:41:19 | <dmead> | it's really ugly and perlish at times otherwise |
| 23:41:19 | <ptolomy> | I'm all for string-as-a-typeclass now. |
| 23:41:23 | <dmead> | from what i've seen |
| 23:41:27 | <sjanssen> | dmead: if you're using tabs, and they aren't displayed as 8 spaces in your editor, you're going to have problems |
| 23:41:30 | <sorear> | What is the best way to get IO (Maybe Char) in the style of C's getchar? |
| 23:41:33 | <dmead> | ah |
| 23:41:36 | <dcoutts> | ptolomy: heh, I'm just writing a rant against that particular idea... |
| 23:41:40 | <mauke> | no, I'm not :( |
| 23:41:51 | <sjanssen> | dcoutts: so is ICFP crunch time over? |
| 23:41:51 | <dmead> | i have another question |
| 23:41:53 | <ptolomy> | dcoutts: I should add "or any viable alternative". |
| 23:41:55 | <dmead> | if anyones up for it |
| 23:42:02 | <dcoutts> | sjanssen: oh not yet, but I was inspired to rant |
| 23:42:03 | <sorear> | I can't figure out any way better than catchJust getChar... |
| 23:42:11 | <dmead> | how do you define a type with just the upper case letters? |
| 23:42:15 | <mauke> | sorear: yeah, that's how I'd do it |
| 23:42:16 | <sorear> | dmead: newtype |
| 23:42:25 | <dmead> | newtype Upper = ? |
| 23:42:29 | <sorear> | mauke: seriously? *gulp* |
| 23:42:46 | <dmead> | i see theres UppercaseLetter in data.char |
| 23:42:50 | <mauke> | yeah, like OCaml |
| 23:42:56 | <dmead> | but theres nothing writting on/with it |
| 23:43:03 | <sorear> | ocaml uses eof exceptions :( |
| 23:43:04 | <sorear> | ? |
| 23:43:17 | <mauke> | yes |
| 23:43:27 | <monochrom> | yes |
| 23:43:32 | <sjanssen> | yes |
| 23:43:42 | <sjanssen> | ACTION is just trusting mauke and monochrom |
| 23:44:04 | <dmead> | yes |
| 23:44:07 | <dmead> | (follower) |
| 23:44:11 | <mauke> | val input_char : in_channel -> char |
| 23:44:20 | <mauke> | Raise End_of_file if there are no more characters to read. |
| 23:44:28 | <ptolomy> | "Ocaml reminds me of my ex-girlfriend: fast and ugly." |
| 23:44:33 | <sjanssen> | sorear: you could also check the return value of hGetBuf |
| 23:44:50 | <monochrom> | plus ocaml is eager. the two together causes it to easy to write naive code that stack-overflows because of layers and layers of try-catch. |
| 23:45:17 | <dmead> | ?type toupper |
| 23:45:19 | <lambdabot> | Not in scope: `toupper' |
| 23:45:24 | <sorear> | but .. try-catch can be implemented so as to have the try-bluck in tail call context! |
| 23:45:30 | <sorear> | or is ml too dumb? |
| 23:45:36 | <sorear> | ?type toUpper |
| 23:45:38 | <lambdabot> | Char -> Char |
| 23:45:46 | <narain> | something i saw in the Prelude-must-die thread: "For example: we know that all Monads are Functors, but that is not expressed in the type classes, because it would be too much a pain to do so." |
| 23:45:49 | <sorear> | much as I hate it, haskell uses CamelCase |
| 23:45:53 | <narain> | is this true? why would it be a pain? |
| 23:46:04 | <mauke> | is there a name for the match/try inversion pattern I'm thinking of? |
| 23:46:09 | <sorear> | because we don't have the class system extension proposal? |
| 23:46:24 | <sjanssen> | narain: the "pain" is exactly one line of boilerplate |
| 23:46:36 | <monochrom> | boilerplate is pain :) |
| 23:47:00 | <dmead> | > newtype Upper = toUpper Char |
| 23:47:01 | <lambdabot> | Parse error |
| 23:47:15 | <dmead> | hmm |
| 23:47:16 | <sjanssen> | sorear: please don't tie the functor/monad issue to class system extensions -- every time somebody proposes to fix the hierarchy we get sidetracked with extension proposals |
| 23:47:28 | <mauke> | let rec mangle_lines fh a = match ( try Some (words (input_line fh)) with | _ -> None ) with | None -> a | Some w -> mangle_lines fh (parse_info w :: a) |
| 23:47:37 | <narain> | sjanssen: do you mean having to also instantiate Functor every time you want to define a Monad? |
| 23:47:44 | <mauke> | ^ does that pattern have a name? |
| 23:47:58 | <dmead> | mauke: either? |
| 23:47:59 | <sjanssen> | there are certainly advantages to such extensions, but it's really not needed for Functor/Monad |
| 23:48:05 | <sjanssen> | narain: correct |
| 23:48:06 | <dmead> | ?type Either |
| 23:48:08 | <lambdabot> | Not in scope: data constructor `Either' |
| 23:48:39 | <narain> | sjanssen: i wouldn't mind that |
| 23:48:42 | <mauke> | dmead: huh? |
| 23:48:51 | <dmead> | Either type |
| 23:48:54 | <sjanssen> | instance Functor T where fmap f x = x >>= return . f -- is not very painful at all |
| 23:49:08 | <dmead> | Either a = Left a | Right a |
| 23:49:12 | <dmead> | or something like that |
| 23:49:16 | <mauke> | fmap = liftM |
| 23:49:23 | <mauke> | dmead: ... ok? |
| 23:49:55 | <sjanssen> | mauke: yes, assuming liftM isn't removed/rewritten when Functor becomes a superclass of Monad |
| 23:50:15 | <sorear> | @hoogle isJust |
| 23:50:15 | <lambdabot> | Maybe.isJust :: Maybe a -> Bool |
| 23:54:52 | <stepcut> | ACTION looks forward to that day |
| 23:55:33 | <stepcut> | anyone know why the NatTransf class from gofer never made it into haskell? Was it a bad idea? Or just underappreciated ? |
| 23:58:17 | <dan__> | blah |
| 23:58:33 | <mauke> | yes, master? |
| 23:58:40 | <dan__> | :P |
| 23:59:18 | <Cale> | stepcut: a bit of both. |
| 23:59:55 | <dmead> | wonderful |
| 23:59:55 | <narain> | ?quote |
| 23:59:56 | <lambdabot> | beelsebob says: > filter isJuggler #haskell |
| 23:59:58 | <Cale> | stepcut: It was less appreciated than it should have been, but it also makes a strange assumption that there should be some canonical natural transformation where there is one. |
Back to channel and daily index: content-negotiated html turtle
