- Haskell-Cafe - Haskell.org

GSoC Proposal: Refactor program with HLint suggestions
by Matthew Pickering 22 Mar '15

22 Mar '15

Dear Café, I would appreciate any comments on the proposal I have written for this years GSoC. The proposal focuses on an oft-requested feature, the ability for HLint to perform the refactorings which it suggests. I believe this is now possible with the work that Alan Zimmerman has recently completed on ghc-exactprint. More details can be found here - http://mpickering.github.io/gsoc2015.html (There is also a pdf version available - http://mpickering.github.io/gsoc2015.pdf) Matt

1 0

netwire example error message
by briand＠aracnet.com 22 Mar '15

22 Mar '15

The error message: time.hs:13:8: No instance for (Show e0) arising from a use of ‘testWire’ The type variable ‘e0’ is ambiguous Note: there are several potential instances: instance Show Double -- Defined in ‘GHC.Float’ instance Show Float -- Defined in ‘GHC.Float’ instance (Integral a, Show a) => Show (GHC.Real.Ratio a) -- Defined in ‘GHC.Real’ ...plus 45 others In the expression: testWire clockSession_ pos In an equation for ‘main’: main = testWire clockSession_ pos Failed, modules loaded: none. So here's my type debugging. (netwire sure is convoluted) pos => Wire s e m a Float integral => a -> Wire s e m a a clockSession_ => Session m (Timed NominalDiffTime ()) looks to me like the problem is probably rooted in the interaction of pos with (Timed NominalDiffTime) as it does not seem to have a float instance, but that doesn't seem like it should result in an error about ambiguity. Any help appreciated. Thanks Brian ---- import Prelude hiding ((.)) -- To use (.) in the scope of Categories instead import Control.Wire import FRP.Netwire speed :: Float speed = 0.5 pos :: (HasTime t s, Monad m) => Wire s e m a Float pos = integral 0 . pure speed -- The main function main :: IO () main = testWire clockSession_ pos

2 2

Re: [Haskell-cafe] [Haskell-beginners] Multi-parameter type classes and ambiguous type variables...
by Sumit Sahrawat, Maths & Computing, IIT (BHU) 22 Mar '15

22 Mar '15

This might be better answered at the haskell-cafe. Sending to cafe. On 21 March 2015 at 03:09, Stuart Hungerford <stuart.hungerford(a)gmail.com> wrote: > Hi, > > As a learning exercise I'm modelling some algebraic structures as > Haskell typeclasses. Some of these are multi-parameter type classes. > Here's a very simplified version of the type class relationships: > > class MM a where > one :: a > > class AM a where > zero :: a > > class (AM a, MM a) => SR a > > class (AM a) => AG a where > inv :: a -> a > > class (SR a) => R a where > neg :: a -> a > > class (R r, AG g) => M r g where > sca :: r -> g -> g > > check :: (Eq g, M r g) => g -> Bool > check x = sca one x == x > > The problem I have is that GHC is finding the "r" type variable in the > "check" function ambiguous. Given my still limited Haskell knowledge > I'm not surprised this is happening. What I would like to know is how > experienced Haskellers handle this situation in practice: is there an > idiomatic way of disambiguating "r" or is it a sign of poor type class > design? > > Thanks, > > Stu > _______________________________________________ > Beginners mailing list > Beginners(a)haskell.org > http://mail.haskell.org/cgi-bin/mailman/listinfo/beginners > -- Regards Sumit Sahrawat

3 3

Continuation vs Codensity
by silvio 21 Mar '15

21 Mar '15

I just saw this http://stackoverflow.com/questions/25827227/why-cant-there-be-an-instance-o… Personally, I think the suggested type for callCC newtype ContT m a = ContT { runContT :: forall r. (a -> m r) -> m r } class MonadCont m where callCC :: (forall b. (a -> m b) -> m b) -> m a is much more intuitive and resembles the actual type of ContT much more closely then the current callCC and it can be Fixed. So I was wondering if there was an other reason than convention for not using that type. Silvio

1 0

What's the story of this weird type?
by David Feuer 21 Mar '15

21 Mar '15

I was reading an answer by Luis Casillas on Stack Overflow http://stackoverflow.com/a/23418746/1477667 and he used a somewhat peculiar type, with apparently not-so-useful constraints: newtype f :-> g = Natural { eta :: forall x. (Functor f, Functor g) => f x -> g x } There are a couple things about it that I don't understand. 1. I can construct a :-> from a function that doesn't satisfy the constraints, and I can pattern match to get access to it, but I can't use it. Why am I allowed to construct it at all? data Mb a = N | J a -- Not a Functor instance -- This is accepted (why?) pot = Natural (\case [] -> N (a:_) -> J a) -- And I can even do this: mb :: Functor Mb => Mb Int mb = case pot of Natural f -> f [3] But then of course I can't satisfy the constraint to use it. 2. I would have thought that such a weird/broken thing as this would only be allowed with -XDatatypeContexts, but in fact it is allowed without it. Why? David

2 3

apt-get install haskell-platform fails on ubuntu 14.04
by David Gladstein 21 Mar '15

21 Mar '15

http://askubuntu.com/questions/597278/haskell-platform-unmet-dependencies I'm installing on a newly-built system. I looked at the haskell.org download instructions, and I couldn't find anything that made sense. apt-get install haskell-platform worked about a month ago. Does anyone know what happened?

2 3

Strongly typed data / API specification librares
by Ignas Vyšniauskas 21 Mar '15

21 Mar '15

Hi cafe, Does anyone know of any libraries that allow one to describe data / interfaces (think IDLs), with emphasis on good support for modules, strong typing and extensibility? I am thinking in the lines of Thrift[1], Piqi[2], Swagger[3], but with the core data specification language as decoupled as possible from any kind of implementation related things (such as actual representation of data, or RPC / RESTful bindings). haskell-servant[4] and silk-rest[5] are kind of relevant, but they are very much oriented towards RESTful APIs. Sorry if this is slightly off-topic, but I couldn't think anywhere else to inquire about this. Thanks, Ignas [1]: https://thrift.apache.org/ [2]: http://piqi.org/ [3]: http://swagger.io/ [4]: http://haskell-servant.github.io/ [5]: http://silkapp.github.io/rest/

2 1

Existentials for multi-parameter type classes?
by Lyle Kopnicky 20 Mar '15

20 Mar '15

Hi folks, I know how to write existentially quantified data constructors for single-parameter type classes, but I'm having trouble doing it for multi-parameter type classes. Is it not possible, or am I just using the wrong syntax? I have this class: class Collidable a b where collideWith :: a -> b -> String And I tried to define an existentially quantified data constructor thusly: data Collision = forall a. (forall b. Collidable a b => Collision {collider1 :: a, collider2 :: b}) I get a parse error. So then I tried: data Collision = forall a. forall b. Collidable a b => Collision {collider1 :: a, collider2 :: b} I get a different parse error. Is there any way to make this work? Thanks, Lyle

3 4

can't STOP after interrupt winGHCi or ghci
by Kees Bleijenberg 19 Mar '15

19 Mar '15

Now I realize maybe I didn't make my point clear. What I want is not just interrupting the output of the compiler, but also stop the output of the compiler after interrupting. Van: Haskell-Cafe [mailto:haskell-cafe-bounces@haskell.org] Namens Kees Bleijenberg Verzonden: woensdag 18 maart 2015 11:12 Aan: haskell-cafe(a)haskell.org Onderwerp: [Haskell-cafe] can't nterrupt winGHCi or ghci With a haskell program I generated a very big haskell file. Sometimes the generated haskell is not valid. If I load the genearated file in winGHCi a flood of error messages streams over the screen. If I press ctrl-break WinGHCi stops and says interrupted. If I click Ok the program continues printing error messages. Ctrl-c or just c, doesn't help. So I have to wait until WinGHCi has shown all error messages (can take minutes) or close the WinGHCi window. Ghci has (on Windows) the same behaviour. Is there a special key combination to stop Ghc or Ghci or is this a bug? Kees

1 0

Incremental sequence sorting
by David Feuer 19 Mar '15

19 Mar '15

I had an idea, and it's kind of complicated, so I was wondering if I could get a sanity check from someone before putting *too* much time into it. Specifically, I was thinking about a sort of incremental sorting of sequences (in Data.Sequence). The idea is very closely related to the incremental sorting described by Navarro and Paredes ( http://www.dcc.uchile.cl/~gnavarro/ps/algor09.pdf ). The basic concept: sequences are represented as Hinze-Paterson 2-3 finger trees. They are fairly lazy, so building them from the top down tends to be a good thing. Suppose I have a sequence to be sorted. To start building the result sequence, I need to know which elements go in the first digit and which go in the last digit. To do this, I need two order statistics, starting with one separating the first digit from the rest, then one separating the last digit from the rest. Once I have separated these, I can produce the top of the sequence. The digit separation is the same all down the spine. Within digits, 2-3 trees split similarly. I can get the order statistics using median-of-medians. As Navarro and Paredes (and presumably others) describe, median-of-medians partially sorts the sequence in the process of finding the desired order statistic. Navarro and Paredes use a stack of indices into an array to take advantage of this, but I need things rather more incremental. What I was thinking is that I could use a finger tree of sequences (a sequence of sequences making it easy to find where a particular total length is reached) to separate the different segments from each other. Each time I need to split on a particular order statistic, I can isolate the necessary segment, use median-of-medians (splitting it into pieces), and then append the pieces onto either side as appropriate. Does anyone have a sense of 1. Whether this is sane and/or 2. What the chances are that its performance will be competitive with either of the current Data.Sequence sorting functions and/or 3. How I should actually implement median-of-medians for this particular application (as there seem to be very few Haskell implementations of the algorithm, and I've found no optimized ones)? Thanks, David Feuer

1 0