- Haskell-Cafe - Haskell.org

GHCI on ARM (Nvidia Jetson TK1)
by Tomas Korcak 21 Jul '14

21 Jul '14

Hi There, I would like to ask you what is the situation with ghci on arm. I bought Jetson TK1 and I am really sad that ghci does not works there. I have found mentioned somewhere that this should be fixed in 7.4.2 but still it is not. Can somebody clarify what is the situation at the moment and what are further plans? Best regards, T. -- <warning> This e-mail is intended for the named recipient(s). It may contain privileged and/or confidential information. If you are not one of the intended recipients, please notify the sender immediately and destroy this e-mail and attachment(s): you must not copy, distribute, retain or take any action in reliance upon the email or attachment(s). While all reasonable efforts are made to safeguard inbound and outbound e-mails, Tomas Korcak cannot guarantee that attachments are virus-free or are compatible with your systems, and does not accept liability in respect of viruses or computer problems experienced. Thank you. </warning> <notice> Your Skills In Reading Have Improved +1 </notice> <idea> Some days you're the dog; some days you're the hydrant. </idea>

3 2

What is the state of the art in testing code generation?
by oleg＠okmij.org 21 Jul '14

21 Jul '14

Tom Ellis wrote: > The first impediment to using QuickCheck is that my EDSL is typed. I don't > know how to generate random terms in the presence of types. First I should point out Magic Haskeller (which generates typed (Haskell) terms and then accepts those that satisfy the given input-output relationship). http://nautilus.cs.miyazaki-u.ac.jp/~skata/MagicHaskeller.html It is MagicHaskeller on Hackage. The most straightforward method is to generate random terms and filter well-typed ones. This is usually a bad method since many or most randomly generated terms will be ill-typed. One should generate well-typed terms from the beginning, without any rejections. That is actually not difficult: one merely needs to take the type checker (which one has to write anyway) and run it backwards. Perhaps it is not as simple as I made it sound, depending on the type system (for example, the type inference for simply-typed lambda-calculus without any annotations requires guessing. One has to guess correctly all the type, otherwise the process becomes slow). It has been done, in a mainstream functional language: OCaml. The code can be re-written for Haskell using the tricks for non-determinism with sharing (the Share monad). Due to many deadlines I cannot re-write myself, not until the middle of the next week. Also, I don't know your type system. If you are curious about the OCaml code, it can be found at http://okmij.org/ftp/kakuritu/type_inference.ml The main function is let rec typeof : gamma -> term -> tp = fun gamma exp -> match exp () with | I _ -> int | V name -> begin try List.assoc name gamma with Not_found -> fail () end | L (v,body) -> let targ = new_tvar() in let tbody = typeof ((v,targ) :: gamma) body in arr targ tbody | A (e1,e2) -> let tres = new_tvar() in tp_same (typeof gamma e1) (arr (typeof gamma e2) tres); tres It can infer a type of a term (or fail if it is will-typed), it can generate well-typed terms one-by-one, or it can generate all terms of a particular structure (e.g., with two lambdas at the top) and/or of a particular type or type shape (e.g., an arrow).

3 2

Problems with type family in a class
by Leza Morais Lutonda 21 Jul '14

21 Jul '14

Hi, Consider the following type classes: class (Signal sx) => Complexable sx where type ComplexSignalType .... class (Complexable sx) => FourierTransformable sx where fft :: (cx ~ ComplexSignalType sx) => cx -> cx .... and an instance: instance (RealFloat e) => Complexable [e] where type ComplexSignalType = [Complex e] instance (RealFloat e) => FourierTransformable [e] where fft = ... Why this example will give errors: *DSP.Signals.Core Prelude Data.Complex> :t ax ax :: [Complex Double] *DSP.Signals.Core Prelude Data.Complex> fft ax <interactive>:90:1: No instance for (FourierTransformable s0 [Complex Double]) arising from a use of ‘fft’ The type variable ‘s0’ is ambiguous Note: there is a potential instance available: instance [overlap ok] (RealFloat e, Math.FFT.Base.FFTWReal e) => FourierTransformable [e] [Complex e] -- Defined at src/DSP/Signals/Instances/List.hs:40:10 In the expression: fft ax In an equation for ‘it’: it = fft ax Thanks! -- Lemol-C 50 Aniversario de la Cujae. Inaugurada por Fidel el 2 de diciembre de 1964 http://cujae.edu.cu

3 8

Extensible Effects & Multiple Operations
by Justin Bailey 21 Jul '14

21 Jul '14

I am playing around with the extensible-effects framework and cannot figure out how to solve this problem. I am trying to define a set of operations that only work after a file has been opened. For example, reading a character from a file or detecting EOF. My problem is I cannot figure out how to define one "runner" for my Effect that can deal with operations with different result types. For example, I would like my `readFrom` and `atEOF` operations to have the following types: data OpenFile r v = ... -- my "OpenFile" effect -- read a character from the file readFrom :: Eff r Char -- detect EOF atEOF :: Eff r Bool And I would like to use this operations as follows (where I read a byte from the file, check EOF, and return the concatenation of the two values): main = do c <- runOpenFile "Setup.hs" (do a <- readFrom -- `a` has `Char` type eof <- atEOF -- `eof` has `Bool` type return $ [a] ++ show eof) putStrLn c However, I cannot define a `runOpenFile` that compiles. For example, this definition gives me "Overlapping Instance" errors: runOpenFile :: FilePath -> Eff (OpenFile v :> r) result -> IO result The best I've come up with is to define a `Result` type that holds all possible results: data Result = AtEOF Bool | ReadChar Char The `Result` type then lets me put a single type on the `OpenFile` effect in `runOpenFile`: runOpenFile :: FilePath -> Eff (OpenFile Result :> r) result -> IO result But that makes my `main` program pretty ugly, because I have to deconstruct those result types each time: main = do c <- runOpenFile "Setup.hs" (do (ReadChar a) <- readFrom (AtEOF eof) <- atEOF return $ [a] ++ show eof) putStrLn c Does anyone have suggestions? Justin p.s. The complete code (in a sort-of similar form) can be seen at https://github.com/m4dc4p/ext-eff-sample/blob/master/src/OpenFile.hs

2 2

Cabal question
by Jurriaan Hage 21 Jul '14

21 Jul '14

Hi there, I am busy making the Helium compiler available as a Cabal install. What I still need is a way to run some kind of postprocessing (like running a Makefile) after installation (as it happens this is for compiling the libraries that come with Helium). Does anyone know here whether that is supported and how, without having to resort to build-types like Make that for I actually want to avoid? best, Jur

2 3

Another Cabal question
by Jurriaan Hage 21 Jul '14

21 Jul '14

Dear all, I have another issue: I have a Cabal package (call it A) that runtime depends on another (say B), where B provides (only) an executable that is compiled with the Make build-type. Now, I would like to list B under the build-depends of A (since there is no way to specify having just a run-time dependency), but it seems that when I cabal install B the package is not registered (although the executable does in fact become available to use within the standard directory.) So when I then compile A with Cabal it will complain the package is not registered. Does anyone know whether the behaviour for B is normal, or whether this is a bug of some kind? If the former, what is an easy and clean way of dealing with this situation. best, Jur

3 2

applicative question
by grant 20 Jul '14

20 Jul '14

Hi guys, I am trying to simplify some code. A user provides a combination of boolean expressions and for each expression an async task is run until that expression becomes true. The boolean expressions have type (Event e => e -> Bool) eg (PC.availableFlag .&&. "asmith" .== PC.userId) Unfortunately the way I have coded this, the user needs to include the \r -> r ? for each expression, with a unique number indicating the position in the sequence (so I can keep track of the asyncs when restarting after a failure). The 'r' function just wraps the expression in a Concurrently constructor. r :: (forall e.Event e => (e -> Bool) -> Concurrently e) -> Concurrently a Here is a sample expression that someone might provide: expr1 = runit (\r -> (,) <$> r 1 (PC.availableFlag .&&. "asmith" .== PC.userId) <*> r 2 (22 .== SC.signalNumber)) So, is there a way to rewrite or simplify this to avoid having to pass the r + unique Int to each subexpression? Or, is there a different format that I can write this in and then transform to the Concurrently applicative? Here is the runit method: runit expr = runConcurrently (expr (Concurrently . queryToIOAction)) And some more sample queries: expr1 = runit (\r -> r (PC.availableFlag .&&. "asmith" .== PC.userId)) expr2 = runit (\r -> (,) <$> r 1 (PC.availableFlag .&&. "asmith" .== PC.userId) <*> r 2 (22 .== SC.signalNumber)) expr3 = runit (\r -> (\(a b -> Left (a,b)) <$> r 1 (PC.availableFlag .&&. "asmith" .== PC.userId) -- (Event e => e -> Bool) <*> r 2 (2 .== SC.signalNumber)) <|> Right <$> r 3 (1 .== SC.signalNumber)) Thanks for any help! Grant

1 0

Q about last&init
by Закиров Марат 20 Jul '14

20 Jul '14

I am teaching myself haskell. The first impression is very good. But phrase "haskell is polynomially reducible" is making me sad :(. Anyway I am trying to backport my algorithm written in C. The key to performance is to have ability to remove element from the end of a list in O(1). But the original haskell functions last and init are O(n). My questions are: 1) Is last function is something like "black box" written in C++ which perform O(1)? So I shouldn't even try to imagine some haskell O(1) equivalent. 2) Or will optimizer (llvm?) reduce init&last complexity to 1? 3) Some people suggest to use sequences package, but still how do they implement O(1) init&last sequences equivalent in haskell? -- Regards, Marat.

15 22

Memory consumption issues under heavy network throughput/concurrency loads
by Ben Bangert 20 Jul '14

20 Jul '14

I have been testing solutions in several languages for running a network daemon that accepts hundreds of thousands of websocket connections, and moves messages between them. Originally I used this websocket lib, http://jaspervdj.be/websockets/, but upon discovering a rather severe memory leak issue even when sending just a basic ping, switched to Michael Snoyman's first stab at a websocket impl for Yesod here: https://github.com/yesodweb/yesod/commit/66437453f57e6a2747ff7c2199aa7ad25d…. When under mild load (5k connections, each pinging once every 10+ seconds), memory usage remained stable and acceptable around 205 MB. Switching to higher load (pinging every second or less), memory usage spiked to 560 MB, and continued to 'leak' slowly. When I quit the server with the profile diagnostics, it indicate that hundreds of MB were "lost due to fragmentation". In addition, merely opening the connections and dropping them, repeatedly, made the base memory usage go up. Somewhere, memory is not being fully reclaimed when the connections are dropped. For a few thousand connections doing little, this wouldn't matter much. However, I'm trying to gauge whether its feasible to use Haskell to handle 150-200k connections that regularly come/go and are held open for long periods of time. Such massive memory use with what looks like leaks (or fragmentation issues) is problematic. I have created a very simple TCP based echo client/server here: https://github.com/bbangert/echo The server can be run after compiling with no additional options, and will listen on port 8080. The client can be run like so: ./dist/build/echoclient/echoclient localhost 8080 2000 0.5 The last number is the frequency to ping (every half second), the second to last is how many clients to connect. Under my local tests, when pinging every 5+ seconds, 2k clients will take about 50-75 MB of ram. Pinging every 0.5 seconds jumps to ~ 180 MB of ram, and this is for a mere 2k clients. Starting/stopping the echoclient repeatedly also causes the servers overall memory usage to continue to climb higher and higher. Releasing the connections never quite gets the memory usage down to where it started. The issue seems to occur under both GHC 7.6.3 and 7.8.2. I know there's a variety of ghc options that might be tuned, am I missing some critical option to keep memory usage under control? Is there a better way to build high TCP throughput/concurrent servers in Haskell that I'm missing? Thanks, Ben

4 11

Creating a Point
by Rafael Almeida 20 Jul '14

20 Jul '14

Hello, I was talking to friends about how could you make a Point type in haskell. It's intuitive to make a 2D point such as: type Point2D = (Double, Double) Let's define one operation on this new type: add2D (x1, y1) (x2, y2) = (x1+x2, y1+y2) Let's say now we want a 3D point. Then we'd be tempted to do: type Point3D = (Double, Double, Double) add3D (x1, y1, z1) (x2, y2, z2) = (x1+x2, y1+y2, z1+z2) Although those types work great and you could make a type class so you don't have to suffix each function with 2D and 3D. It feels like we are just repeating code when defining the add function. If we want to go 4D, 5D, etc it would be more repeated code. Using a list would be more general: type Point = [Double] now we have a nice, general add function add = zipWith (+) It's not so fun that we are able to do something like: add [2,3] [5,7,11] We have no type-safety that we can only operate on points with the same dimension. How could we address this? Can we make a general function, yet retain the type safety? I suppose maybe there's something that could be done with TH so that we automatically generate those Point2D, Point3D, etc code. I'm not sure that would be a nice path to follow, though. []'s Rafael

4 3