- Haskell-Cafe - Haskell.org

List Monads and non-determinism
by Matt Ford 20 Jul '13

20 Jul '13

Hi All, I thought I'd have a go at destructing [1,2] >>= \n -> [3,4] >>= \m -> return (n,m) which results in [(1,3)(1,4),(2,3),(2,4)] I started by putting brackets in ([1,2] >>= \n -> [3,4]) >>= \m -> return (n,m) This immediately fails when evaluated: I expect it's something to do with the n value now not being seen by the final return. It seems to me that the return function is doing something more than it's definition (return x = [x]). If ignore the error introduced by the brackets I have and continue to simplify I get. [3,4,3,4] >>= \m -> return (n,m) Now this obviously won't work as there is no 'n' value. So what's happening here? Return seems to be doing way more work than lifting the result to a list, how does Haskell know to do this? Why's it not in the function definition? Are lists somehow a special case? Any pointers appreciated. Cheers, -- Matt

6 9

Default installation path of haskell platform in windows
by Ivan Perez 20 Jul '13

20 Jul '13

Hi café, I just spent the whole day fighting ghc and cabal to get my program compiled in windows (and I haven't succeeded yet :) ) realgcc is not very happy about paths with spaces in them (I think there's an open bug about the way ghc calls realgcc, which manifests if you use extra-lib-dirs in the cabal file). I'd like to ask you: Might it be wise to change the setup file of the Haskell Platform to default to an installation path with no spaces in it, and to tell users why this is a good idea (possibly during the installation process)? Other projects are doing the same, including msys. Please be aware that I know that this "solution" is far from ideal. Ideally, it should just work (with spaces and without them). This is purely a practical decision, a workaround, aiming at making haskell more accessible to windows users, and making it easier for us to develop software that works in all platforms. Regards Ivan

1 0

Re: [Haskell-cafe] Generating Haskell Code out of Haskell AST (GHC API)
by John Blackbox 19 Jul '13

19 Jul '13

I accidentally didn't send that email to haskell-cafe, so I'm pasting it here also: Alan - I do NOT want to generate Haskell code. I want only to generate AST and compile it. The question about generating the code was only to have a "debugging tool" - to see if the generated AST is good - I wanted to generate the Haskell code only to check if its correct, but normally I would not do it, because it makes no sense to generate AST -> code -> AST (by GHC) again etc :) Additional - I want to connect to GHC's type-checking also and translate the errors to be appropriate to my language syntax - so maybe the pure GHC API is the best way to go? 2013/7/19 John Blackbox <blackbox.dev.ml(a)gmail.com> > Additional - I want to connect to GHC's type-checking also and translate > the errors to be appropriate to my language syntax. > > > 2013/7/19 John Blackbox <blackbox.dev.ml(a)gmail.com> > >> Alan - I do NOT want to generate Haskell code. I want only to generate >> AST and compile it. >> The question about generating the code was only to have a "debugging >> tool" - to see if the generated AST is good - I wanted to generate the >> Haskell code only to check if its correct, but normally I would not do it, >> because it makes no sense to generate AST -> code -> AST (by GHC) again etc >> :) >> >> >> 2013/7/19 AlanKim Zimmerman <alan.zimm(a)gmail.com> >> >>> I have not used haskell-src-exts so I may be talking out of turn, but it >>> seems that if you want to generate an AST which you then turn into source >>> code and compile it makes more sense than using than GHC AST, which has a >>> number of wrinkles, including fields that are only valid at certain phases >>> of the compilation process. >>> >>> For my purposes, in the Haskell Refactorer, I need access to the >>> renaming and type-checking, which to my knowledge is not currently >>> available in haskell-src-exts, although there is work happening to bring it >>> in, e.g. https://github.com/haskell-suite/haskell-names. >>> >>> >>> On Fri, Jul 19, 2013 at 10:09 AM, John Blackbox < >>> blackbox.dev.ml(a)gmail.com> wrote: >>> >>>> Thank you! >>>> So, if I'm writing a compiler of custom language, which I want to >>>> generate Haskell AST and further compile it with GHC, you prefer >>>> something like haskell-src-exts over pure GHC API? >>>> >>>> >>>> 2013/7/19 Antoine Latter <aslatter(a)gmail.com> >>>> >>>>> The package haskell-src-exts is a lot less intimidating if all you are >>>>> trying to do is programmatically generate Haskell source: >>>>> >>>>> http://hackage.haskell.org/package/haskell-src-exts >>>>> >>>>> The base types are here: >>>>> >>>>> http://hackage.haskell.org/packages/archive/haskell-src-exts/1.13.5/doc/htm… >>>>> >>>>> This module has some helper function for generating parts of the AST: >>>>> >>>>> http://hackage.haskell.org/packages/archive/haskell-src-exts/1.13.5/doc/htm… >>>>> >>>>> >>>>> On Thu, Jul 18, 2013 at 1:11 PM, John Blackbox >>>>> <blackbox.dev.ml(a)gmail.com> wrote: >>>>> > Hi! >>>>> > I dont know GHC API very well, but I want to generate AST of a >>>>> program using >>>>> > GHC API. >>>>> > Is there any standard method to generate Haskell code out of it? >>>>> (something >>>>> > like "print_this_for_me_please" function? :D >>>>> > >>>>> > _______________________________________________ >>>>> > Haskell-Cafe mailing list >>>>> > Haskell-Cafe(a)haskell.org >>>>> > http://www.haskell.org/mailman/listinfo/haskell-cafe >>>>> > >>>>> >>>> >>>> >>>> _______________________________________________ >>>> Haskell-Cafe mailing list >>>> Haskell-Cafe(a)haskell.org >>>> http://www.haskell.org/mailman/listinfo/haskell-cafe >>>> >>>> >>> >> >

3 2

Generating Haskell Code out of Haskell AST (GHC API)
by John Blackbox 19 Jul '13

19 Jul '13

Hi! I dont know GHC API very well, but I want to generate AST of a program using GHC API. Is there any standard method to generate Haskell code out of it? (something like "print_this_for_me_please" function? :D

2 2

Monad Transformer Space Leak
by Clark Gaebel 19 Jul '13

19 Jul '13

Hi everyone! For some reason, this leaks thunks: module Main where import Control.Monad import Control.Monad.MC -- from monte-carlo import Control.Monad.ST.Strict go :: Int -> MCT (ST s) () go k = replicateM_ k (return ()) main = print $ runST $ evalMCT (go 100000000) rng where rng = mt19937 0 while this does not: module Main where import Control.Monad import Control.Monad.MC go :: Int -> MC () go k = replicateM_ k (return ()) main = print $ evalMC (go 100000000) rng where rng = mt19937 0 Can anyone help me figure out what's going on here? Thanks, - Clark

5 11

ghc and jump table generation
by Erik Rantapaa 18 Jul '13

18 Jul '13

I've been looking at the assembly code that ghc generates for simple pattern matching for situations like: foo :: Int -> Int foo 1 = 3 foo 2 = 10 foo 3 = 2 foo 4 = 42 -- ... other random assignments and I noticed that it doesn't seem to generate a jump table (or even a table lookup) like, for instance, gcc would for this switch statement: int foo(int x) { switch (c) { case 1: return 3; case 2: return 10; case 3: return 2; case 4: return 42; // ... etc. } } Under -O3 ghc seems to produce a binary-search of the cases. Questions: Would generating a jump/lookup table for pattern matches like this (i.e. with a suitable density of cases) be beneficial? And, if so, would it be difficult to add to ghc? And would it be a good first project for someone who wants to get into the ghc code base (perhaps with some mentoring from the community?)

2 1

Automatic error traces
by Alberto G. Corona 18 Jul '13

18 Jul '13

It is important to have execution traces in case of error. specially in server applications that run 24/7 such are web applications Thanks to the wonderful package monadloc by Pepe Iborra, now MFlow can generate a complete execution trace in case of error. The control-monad-exception uses monadLoc to generate stack traces, but MFlow makes use of his backtracking mechanism to generate a complete execution trace. Here I explain what and how: http://haskell-web.blogspot.com.es/2013/07/automatic-error-trace-generation… The MFlow version that implements this is in gitHub. Not in hackage yet. https://github.com/agocorona/MFlow I´m quite proud of it since it is one of the things closest to magic that I have done. Feedback? I do not want to keep MFlow as a single person development. I think that it has many unique and nice features not available in other languages and frameworks, and it can be raised to a serious exploitation level by the Haskell community. This will attract people working in Industrial Web development to Haskell thanks to the edge in expressiveness and safety necessary for creating industrial solutions that Haskell has over other languages. It uses most other Haskell web developments blaze, wai, hamlet etc and there are many other things to come. monadLoc : http://hackage.haskell.org/package/monadloc -- Alberto.

2 3

Not working examples in GHC API documentation
by John Blackbox 18 Jul '13

18 Jul '13

Hi! Please take a look here: http://www.haskell.org/haskellwiki/GHC/As_a_library The examples are not working. Even the simpelst one: import GHC import GHC.Paths ( libdir ) import DynFlags ( defaultLogAction ) main = defaultErrorHandler defaultLogAction $ do runGhc (Just libdir) $ do dflags <- getSessionDynFlags setSessionDynFlags dflags target <- guessTarget "test_main.hs" Nothing setTargets [target] load LoadAllTargets throws: $> ghc -package ghc Main.hs [1 of 1] Compiling Main ( Main.hs, Main.o ) Main.hs:6:25: Couldn't match type `DynFlags' with `[Char]' Expected type: DynFlags.FatalMessager Actual type: DynFlags.LogAction In the first argument of `defaultErrorHandler', namely `defaultLogAction' In the expression: defaultErrorHandler defaultLogAction In the expression: defaultErrorHandler defaultLogAction $ do { runGhc (Just libdir) $ do { dflags <- getSessionDynFlags; setSessionDynFlags dflags; .... } } Main.hs:7:7: Couldn't match expected type `DynFlags.FlushOut' with actual type `IO SuccessFlag' In a stmt of a 'do' block: runGhc (Just libdir) $ do { dflags <- getSessionDynFlags; setSessionDynFlags dflags; target <- guessTarget "test_main.hs" Nothing; setTargets [target]; .... } In the second argument of `($)', namely `do { runGhc (Just libdir) $ do { dflags <- getSessionDynFlags; setSessionDynFlags dflags; .... } }' In the expression: defaultErrorHandler defaultLogAction $ do { runGhc (Just libdir) $ do { dflags <- getSessionDynFlags; setSessionDynFlags dflags; .... } }

2 2

Haskell Programmers for Distributed Systems
by Amanda Schwoegler 18 Jul '13

18 Jul '13

Parallel Scientific, Inc. is a Boulder, CO based, funded startup company with a presence around the globe, working in the cutting edge area of scalable parallelization. We seek experienced Haskell programmers to build a sophisticated, ultra-scalable high availability framework to be used in conjunction with scientific applications, big data applications and I/O systems. Successful candidates will work in a modern, virtual environment, with a distributed international team. All candidates are welcome to apply no matter their location. However, for this particular position, preference may be given to someone residing in or willing to relocate to the UK. Responsibilities include design, implementation in Haskell and in C, testing, infrastructure development, and optionally formal verification. We provide training in advanced processes for software design and implementation, as well as domain specific knowledge. Required skills: - Very strong background in computer science or mathematics. - Experience with Haskell, or network and cluster programming in other functional languages (e.g. Erlang, OCaml, Clojure). - Knowledge of systems programming and operating systems. Experience in one or more of the following areas is desirable: - Experience in distributed and/or parallel programming. - Knowledge of sophisticated distributed algorithms (fault tolerance, consensus, distributed transactions, etc). - Experience with delivering quality products on time and within budget. To apply, please send a resume to jobs(a)parsci.com. About Parallel Scientific Parallel Scientific was founded by Dr Peter Braam in 2010. Peter formerly taught mathematics and computer science at Oxford and Carnegie Mellon. He then contributed file systems to Linux and invented Lustre (which provides storage to 9 of the top 10 systems in the world). He ran several successful startups, and Parallel Scientific is run by a very experienced management team and board. Our expertise is in the areas of distributed scientific parallel computing, compilers, domain specific languages, scalable computing for cloud and HPC, and the synthesis of hardware appliances.

1 0

what is wrong w my IORef Word32 ?
by Joerg Fritsch 18 Jul '13

18 Jul '13

All, what is wrong w the below code? I get an type error related to the operation shiftL import Data.Bits import Data.Word import Data.IORef data Word32s = Word32s { x :: IORef Word32 } bitfield :: Word32 bitfield = 0 mkbitfield :: Word32 -> IO Word32s mkbitfield i = do the_bf <- newIORef i return (Word32s the_bf) sLbitfield :: Integer -> Word32s -> IO () sLbitfield i (Word32s bf) = do modifyIORef bf (shiftL i) main::IO() main = do oper_bf <- mkbitfield bitfield sLbitfield 2 oper_bf bf_003.hs:15:48: Couldn't match type `Int' with `Word32' Expected type: Word32 -> Word32 Actual type: Int -> Word32 In the return type of a call of `shiftL' In the second argument of `modifyIORef', namely `(shiftL i)' In a stmt of a 'do' block: modifyIORef bf (shiftL i) bf_003.hs:15:55: Couldn't match expected type `Word32' with actual type `Integer' In the first argument of `shiftL', namely `i' In the second argument of `modifyIORef', namely `(shiftL i)' In a stmt of a 'do' block: modifyIORef bf (shiftL i) Thanks, --Joerg

3 3