August 2014 - Haskell-Cafe

Locking of threads in one OS thread
by Nikolay Amiantov 26 Aug '14

26 Aug '14

Hello Cafe, I'm using FFI to interact with a library which calls, when fail, leave the reason in some kind of "errno"-like variable which is retrived via another call. AFAIU, this is not thread-safe in Haskell even if thread-local storage is used inside the library, because Haskell uses its own thread management and the Haskell thread in the same OS thread might be switched between the actual call and the retrival of errno value. This should be somehow handled already in Haskell (errno is widely used with syscalls in Linux, for example), but the source of Foreign.C.Error suggests that this is not handled in any way at all. For example, throwErrnoIf is implemented as such: throwErrno loc = do errno <- getErrno ioError (errnoToIOError loc errno Nothing Nothing) throwErrnoIf pred loc f = do res <- f if pred res then throwErrno loc else return res So, the question is: how is it ensured that this block is "atomic" in sense that at most one Haskell thread computes this whole function at every moment in every OS thread? Thanks for any explanations! (And sorry if my English is poor) Nikolay.

4 5

Types depending on tuples.
by Mike Meyer 26 Aug '14

26 Aug '14

I'm doing some geometry in 2 and 3 dimensions. To do this, I've got Vector2d and Vector3d classes, which are basically just wrappers around 2 and 3-tuples of doubles: type Vector2d = (Double, Double) type Vector3d = (Double, Double, Double) I've also got a typeclass Vector so I can do polymorphic things on them: type Vector3d = (Double, Double, Double) (Ok, polymorphic thing). I now have a need to convert lists of Vector's to a list of lists of Doubles. I'd like that to be part of the Vector typeclass, but - well, declaring it is interesting. I can't just add a method "toLists :: a -> [Double]", because (according to ghc), there's no match between the expected type 'Double' and the actual type '[t1]'. Can I declare this typeclass?

2 1

Undesired email
by felipe zapata 25 Aug '14

25 Aug '14

Dear List, I have a problem with me new smartphone. Sorry for the inconvenient. Best, FelipeZ

1 0

Munich Haskell Meeting
by Heinrich Hördegen 24 Aug '14

24 Aug '14

Dear all, next Tuesday, 26th of August, Munich's Haskell enthusiasts will meet again at Cafe Puck around 19h30. Everyone is wellcome. Please note that we will meet at Cafe Puck. To help us reserve enough tables, please go to http://www.haskell-munich.de/dates and click the button. Heinrich

1 0

ghc-mod v5.0.0
by Kazu Yamamoto 24 Aug '14

24 Aug '14

Hi cafe, We are happy to announce that ghc-mod v5.0.0 has been released. Highlights of this version is as follows: - ghc-mod v4.x.y is suffering from the space leak of GHC 7.8.x(*1). So, ghc-modi consumes huge memory. ghc-modi v5.0.0 consumes much less memory thanks to workaround. - Alejandro Serrano merges his results of Google Summer of Code. He implements case splitting and sophisticated type hole handling. I believe that he will write an article on his results but I also updated the manual of Emacs fronted. (*2) - Daniel Gröber brushed up the internal code by bringing the GhcMod monad. So, the API of ghc-mod drastically changed. If you are users of the ghc-mod library, you should check its document. The main purpose of this release is to show the results of GSoC to everyone. This is a snapshot release and its API would change in the future. I thank two guys above since they worked hard and did good jobs. We hope that everyone enjoys v5.0.0. (*1) https://ghc.haskell.org/trac/ghc/ticket/9314 (*2) http://www.mew.org/~kazu/proj/ghc-mod/en/emacs.html --Kazu

2 2

How to write transform one set of ADTs to another with SYB?
by ducis 23 Aug '14

23 Aug '14

More specifically, I want to transform a tree of SomeExpr to a tree of (SomethingElse, SomeExpr), "prefixing" each node with a default value in the process. The result have essentially the same structure as the input. data ASTAttachment = AA {} deriving (Eq,Read,Show,Ord,Typeable,Data) type family AST a :: * type instance AST () = ASTExpr () type instance AST AA = (AA, ASTExpr AA) data ASTExpr a = ALiteral String | AApplication [AST a] (ASTOp a) | ARef String | ABind (AST a) String | AMany (ASTMany a) data ASTOp a = AOSym String | AOAlpha String | AOMany (ASTMany a) data ASTMany a = AMSimple [AST a] | AMAggeregate [AST a] deriving instance Typeable ASTMany deriving instance Typeable ASTOp deriving instance Typeable ASTExpr deriving instance Data (ASTMany ()) deriving instance Data (ASTOp ()) deriving instance Data (ASTExpr ()) Currently I have values of type (AST ()), and I need to transform them to (AST ASTAttachment) in the obvious way. Now I'm stuck at "derive instance Data (ASTExpr ASTAttachment)", where ghc reports "Multiple declarations of '$cSomeConstructor'". Even if I can get past this, I still don't know how to use SYB to transform between trees of different types.

2 1

Temporarily taking over hdevtools on Hackage.
by Schell Scivally 23 Aug '14

23 Aug '14

Hi all, it seems that Bit Connor (the author of hdevtools - https://github.com/bitc/hdevtools) has been MIA since January. There are a number of pull requests that need to be merged and uploaded to hackage for GHC 7.8. I tried contacting him last week through both emails listed on hackage - one bounced and the other is unanswered. Has anyone seen/heard from the author? I'd like to temporarily hijack hackage responsibilities until we hear from him again in order to merge and upload these changes ( https://github.com/bitc/hdevtools/issues/24#ref-commit-a996411) -- Schell Scivally | github <https://github.com/schell> | twitter <http://twitter.com/schellsan> | blog <http://efnx.com> | wiki <http://wiki.zyghost.com>

3 3

How to add Haddock comment for standalone derived instances?
by Hiromi ISHII 23 Aug '14

23 Aug '14

Hi cafe, Is there any way to add the documentation comment for the instances defined with StandaloneDeriving? I'm currently defining data-type using GADTs and its Typeable instance. Normally, this can be done only StandaloneDeriving and DeriveDataTypeable extensions: > {-# LANGUAGE StandaloneDeriving, GADTs, DeriveDataTypeable #-} > data Ordinal n where > OZ :: Ordinal (S n) > OS :: Ordinal n -> Ordinal (S n) > > deriving instance Typeable Ordinal I added `Typeable` instance for Ordinal recently, so I want to add some comments like "Since 0.2.3.0". But any of the following doesn't work or, even worse, haddock won't compile: * Just before `deriving instance` line using `-- | ` * Just after `deriving` keyword but before `instance` using `-- | ` * Just after the `Typeable Ordinal`, but no newline in-between, with `-- ^ ` * The next line of `deriving` clause with `-- ^ ` Is there any way to add documentation for instances with standalone deriving, or it's just not supported yet? -- Hiromi ISHII konn.jinro(a)gmail.com

2 1

Why this existential type could not work?
by Magicloud Magiclouds 22 Aug '14

22 Aug '14

Hi, For example, code like this: newtype Sealed = Sealed { unSealed :: forall a. SomeClass a => TVar a } mkSealed :: (SomeClass a) => a -> IO Sealed mkSealed = fmap Sealed . newTVarIO When compiling, I would get: Expected type: a -> IO (forall a1. SomeClass a1 => TVar a1) Actual type: a -> IO (TVar a) How to correctly restrict type parameter a here? -- 竹密岂妨流水过山高哪阻野云飞 And for G+, please use magiclouds#gmail.com.

5 5

Vector sort poor performance
by Alexander Pakhomov 22 Aug '14

22 Aug '14

Hi! I've got dramatically low sort performamce: about 2 us of time and 5 kB of allocation for unpacked vector of 4 (four) Double. Code: import Data.List import Control.Monad import Control.Monad.ST import qualified Data.Vector.Unboxed as V import qualified Data.Vector.Unboxed.Mutable as MV import qualified Data.Vector.Algorithms.Intro as Intro import Data.IORef arr = V.fromList ([1,2] :: [Double]) foo x = V.head q where q = runST $ do res <- V.unsafeThaw $ V.concat [ V.map (\e -> e + x) arr , V.map (\e -> e - x) arr] Intro.sort res V.unsafeFreeze res main = do ref <- newIORef 0 forM_ [0..100000] $! \i -> do modifyIORef' ref (+(foo $ fromInteger i)) -- for foo not to optimize out readIORef ref >>= print ghc -O2 sort.hs && time ./sort [1 of 1] Compiling Main ( sort.hs, sort.o ) Linking sort ... -4.999949999e9 real 0m0.189s user 0m0.184s sys 0m0.000s Does anybody know what's going on?

4 7