- Haskell-Cafe - Haskell.org

Re: [Haskell-cafe] typed tagless-final interpretation examples broken with recent ghc
by oleg＠okmij.org 12 Oct '14

12 Oct '14

Mathijs Kwik wrote more than two months ago, on Aug 2 2014: > I've been reading up on typed tagless-final interpretations [1] and the > accompanying examples. > > It seems the stuff about CPS transformations does no longer compile, > probably because of stricter typing rules. I tried both ghc 7.6 and ghc > 7.8, which both give a different kind of error message. GHC 7.6 has introduced the ambiguity check, which has been causing quite a bit of pain. For an example, please see http://ghc.haskell.org/trac/ghc/ticket/9587 The example in that ticket is a simplified version of the tagless final CPS transformations. >From experience, the best solution to that sort of problems is to re-write the code. Often the result is simpler. This has been the case for CPS transformations. The new version of the tagless-final CPS transformations is notably shorter http://okmij.org/ftp/tagless-final/course/CPS.hs and it should work in any GHC with type families, including GHC 7.8.3.

1 0

Current state-of-the-art in getting a dev env set up (windows 8)
by Alistair Bayley 12 Oct '14

12 Oct '14

What is the current advice for getting a haskell dev env set up on windows? There's a lot of options, and it can be hard to quickly determine which are current and which are deprecated/bit-rotted. For example, apparently cabal sandbox is preferred over cabal-dev these days. I've read: http://onoffswitch.net/started-haskell/ http://stackoverflow.com/questions/304614/haskell-on-windows-setup http://coldwa.st/e/blog/2013-08-20-Cabal-sandbox.html http://bob.ippoli.to/archives/2013/01/11/getting-started-with-haskell/ I've installed the latest haskell platform (2014.2), and am using Sublime Text 2 as my editor. SublimeHaskell is installed, but apparently it needs aeson, haskell-src-ext, and haddock. cabal install aeson indicates that it will break unordered-containers-0.2.4.0 and case-insensitive-1.1.0.3, so I must use --force-reinstalls to get it. I'm doing this in a sandbox, in case it is a terrible idea. https://github.com/SublimeHaskell/SublimeHaskell Not yet sure how I'm going to get the sandbox install of these tools to work with SublimeHaskell. Presumably some PATH magic? hdevtools looks cool and seems like it can be used from SublimeHaskell, although a windows install is from a fork. Will install ghc-mod too. What else am I missing? Thanks, Alistair

4 6

Map with incremental serialization
by Britt Mathis 11 Oct '14

11 Oct '14

Hello everyone, I'm not sure if something like this already exists so I figured I would ask here first before I reinvent the wheel. What I need is a data structure that behaves like Data.Map, but is serialized to disk incrementally so I only have the overhead of writing what has changed every time.

7 10

Pretty Printing Libraries with an Emphasis on Consistency
by Oliver Charles 11 Oct '14

11 Oct '14

Hi all, It seems that all of the pretty printing libraries on Hackage at the moment have an emphasis on being compact, but not necessarily consistent. By consistent, I mean that similar blocks of code are indented in the same fashion, even when not strictly necessary. Take for example, the start of "Pretty Printing with Lazy Dequeues". Here, the example is: if True then if True then True else True else if False then False else False I call this inconsistent, because the two sides of the top-level if have been formatted under different rules. A consistent formatting of this would be: if True then if True then True else True else if False then False else False The above can obviously be reached if you *always* format `if` in this style, but that's not what I'm after. If it's possible to format both sides on one-line, then that should be preferred. So an equally desirable output (that meets my "consistency" requirement) is: if True then if True then True else True else if True then True else True It doesn't seem that there is a library on Hackage that lets one express such a pretty printer. It seems that a pretty printer of this nature would need to have some sort of backtracking ability - I can imagine trying to print both sides of the if statement on one-line, and if this fails, try printing *both* on multiple lines. Curious to hear what people have to say on this matter, - ocharles

2 1

Chaining a map over lists with lenses
by Nicola Gigante 10 Oct '14

10 Oct '14

Hi! I’m sure the answer to this question is under my eyes but I can’t understand it only from the docs. Suppose a data type like this: data Type1 = Type1 { _myField :: [Type2] } makeLenses ‘’Type1 data Type2 = Type2 { _otherField :: String } Then with t^.myField I get access to the [Type2] list. Now, i would like to go through each element of the list, apply the otherField lens to each element, and get the list of the results. Of course this means: map (view otherField) (t^.myField) but I’m sure there’s some combinator to do the same thing by chaining lenses/prisms/traversals/whatever, right? E.g. something like t^.myField.something.otherField I’m still trying to understand the full picture about lenses, but afaik what I need is a prism, because a list is a sum type, right? Thank you for your help, Nicola

4 4

Need help testing hGetContents modification
by David Feuer 10 Oct '14

10 Oct '14

I uploaded a change to hGetContents as D327 on Phabricator to address TRAC #9236. In particular, this generates an error when attempting to read from a closed handle instead of silently returning an empty string. I would love to get some help testing this against code that uses lazy I/O heavily, *especially* with pipes, sockets, and other such weirdness, if such things exist in the wild. I don't usually read haskell-cafe, so please send responses to my address as well as the list. Thanks a lot, David Feuer

1 0

ICFP 2015 Call for Workshop and Co-located Event Proposals
by David Van Horn 09 Oct '14

09 Oct '14

CALL FOR WORKSHOP AND CO-LOCATED EVENT PROPOSALS ICFP 2015 20th ACM SIGPLAN International Conference on Functional Programming August 30 - September 5, 2015 Vancouver, Canada http://icfpconference.org/icfp2015/ The 120th ACM SIGPLAN International Conference on Functional Programming will be held in Vancouver, British Columbia, Canada on August 30-September 5, 2015. ICFP provides a forum for researchers and developers to hear about the latest work on the design, implementations, principles, and uses of functional programming. Proposals are invited for workshops (and other co-located events, such as tutorials) to be affiliated with ICFP 2015 and sponsored by SIGPLAN. These events should be less formal and more focused than ICFP itself, include sessions that enable interaction among the attendees, and foster the exchange of new ideas. The preference is for one-day events, but other schedules can also be considered. The workshops are scheduled to occur on August 30 (the day before ICFP) and September 3-5 (the three days after ICFP). ---------------------------------------------------------------------- Submission details Deadline for submission: November 16, 2014 Notification of acceptance: December 15, 2014 Prospective organizers of workshops or other co-located events are invited to submit a completed workshop proposal form in plain text format to the ICFP 2015 workshop co-chairs (Tom Schrijvers and Nicolas Wu), via email to icfp2015-workshops(a)cs.kuleuven.be by November 16, 2014. (For proposals of co-located events other than workshops, please fill in the workshop proposal form and just leave blank any sections that do not apply.) Please note that this is a firm deadline. Organizers will be notified if their event proposal is accepted by December 15, 2014, and if successful, depending on the event, they will be asked to produce a final report after the event has taken place that is suitable for publication in SIGPLAN Notices. The proposal form is available at: http://www.icfpconference.org/icfp2015/icfp15-workshops-form.txt Further information about SIGPLAN sponsorship is available at: http://www.sigplan.org/Resources/Guidelines/Workshops ---------------------------------------------------------------------- Selection committee The proposals will be evaluated by a committee comprising the following members of the ICFP 2015 organizing committee, together with the members of the SIGPLAN executive committee. Workshop Co-Chair: Tom Schrijvers (KU Leuven) Workshop Co-Chair: Nicolas Wu (University of Oxford) General Chair: Kathleen Fisher (Tufts University) Program Chair: John Reppy (University of Chicago) ---------------------------------------------------------------------- Further information Any queries should be addressed to the workshop co-chairs (Tom Schrijvers and Nicolas Wu), via email to icfp2015-workshops(a)cs.kuleuven.be.

1 0

Combine type family return value
by Dmitry Bogatov 09 Oct '14

09 Oct '14

Hello! Following code is my attempt to comprehend type families and data kinds. I want to annotate boolean formula with variables it contains, so I get function `value` checked at compile time. `value` should return Bool, if no variables are present in formula, list of variables otherwise. Of course, `value` can just return Either [Var] Bool, but I want to get Type-level programming-fu. {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE StandaloneDeriving #-} {-# LANGUAGE GADTs #-} {-# LANGUAGE UndecidableInstances #-} {-# LANGUAGE TypeFamilies #-} {-# LANGUAGE DataKinds #-} data Nat = Zero | Succ Nat type family Plus (x :: Nat) (y :: Nat) :: Nat type instance Plus Zero x = x type instance Plus (Succ x) y = Plus x (Succ y) type family Multi (x :: Nat) (y :: Nat) :: Nat type instance Multi Zero x = Zero type instance Multi (Succ x) y = Plus y (Multi x y) type One = Succ Zero type Two = Succ One type Three = Succ Two type Power2 a = Power Two a type family Power (n :: Nat) (k :: Nat) :: Nat type instance Power n Zero = One type instance Power n (Succ x) = Multi n (Power n x) data Var = Pv | Qv | Rv type family PEncode (v :: Var) :: Nat type instance PEncode Pv = Power2 Zero type instance PEncode Qv = Power2 One type instance PEncode Rv = Power2 Two -- type instance PDecode (v :: Nat) :: * data Formula (n :: Nat) where P :: Formula (PEncode Pv) Q :: Formula (PEncode Qv) R :: Formula (PEncode Rv) Const :: Bool -> Formula Zero Conj :: Formula n -> Formula m -> Formula (Plus n m) type family FormulaValue (x :: Nat) :: * type instance FormulaValue Zero = Bool type instance FormulaValue (Succ n) = [Var] value :: Formula n -> FormulaValue n value (Const v) = v value P = [Pv] value Q = [Pv] value R = [Pv] value (Conj x y) = {- I have no idea -} deriving instance Show (Formula n) main = print $ Conj P Q -- Best regards, Dmitry Bogatov <KAction(a)gnu.org>, Free Software supporter, esperantisto and netiquette guardian. GPG: 54B7F00D

1 0

ANN: FunGEn game engine 0.4.6 released, now on github
by Simon Michael 09 Oct '14

09 Oct '14

FunGEn 0.4.3 was uninstallable from hackage for a long time (and no-one noticed). Now, thanks to gelisam's bug report and some further cleanup work, I'm happy to announce that FunGEn 0.4.6 is available! I've also moved it from darcs hub to github. FunGEn (Functional Game Engine) is a BSD-licensed, cross-platform, OpenGL/GLUT-based, non-FRP game engine/framework written in Haskell. Created by Andre Furtado in 2002, it's the oldest Haskell game engine, and with very few dependencies and two example games, it's one of the easiest ways to get started with Haskell game development. It provides: - Initialization, updating, removing, rendering and grouping routines for game objects - Definition of a game background (or map), including texture-based maps and tile maps - Reading and interpretation of the player's keyboard and mouse input - Collision detection - Time-based functions and predefined game actions - Loading and displaying of 24-bit bitmap files - Some debugging and game performance evaluation facilities If you'd like to take over this package and run with it, or co-maintain, contact me. Best! -Simon

1 1

`zip` doesn't work with infinite `Seq`s
by Matteo Ferrando 09 Oct '14

09 Oct '14

Hello, We are writing a compiler[1] for a course and found that the `zip` function included in the `Data.Sequence` module, `zip :: Seq a -> Seq b -> Seq (a, b)` would hang on the following code: > -- using the `zip` from `Data.Sequence` > zip ys (fomList $ repeat x) We checked the implementation[2] in the source of `Data.Sequence` and found the following: > zip :: Seq a -> Seq b -> Seq (a, b) > zip = zipWith (,) > -- Here `zipWith` assumes *non-infinite `Seq`s* > zipWith :: (a -> b -> c) -> Seq a -> Seq b -> Seq c > zipWith f xs ys > | length xs <= length ys = zipWith' f xs ys > | otherwise = zipWith' (flip f) ys xs > -- Function not exported by `Data.Seq`, assumes `length xs <= length ys` > zipWith' :: (a -> b -> c) -> Seq a -> Seq b -> Seq c > zipWith' f xs ys = snd (mapAccumL k ys xs) > where > k kys x = case viewl kys of > (z :< zs) -> (zs, f x z) EmptyL -> error "zipWith': unexpected EmptyL" In the lazy reading of the documentation we did, we didn't find any warning of using infinite `Seq`s for zips. (Maybe there are warings that we didn't see). But looking at the code of `zip` in the `Prelude`: > zip :: [a] -> [b] -> [(a,b)] > zip (a:as) (b:bs) = (a,b) : zip as bs > zip _ _ = [] We see that we could just *pattern-match* both heads, instead of making assumptions. Maybe this should be better explained in the documentation[3] of `zip` for `Seq`s: > zip :: Seq a -> Seq b -> Seq (a, b) > O(min(n1,n2)). zip takes two sequences and returns a sequence of > corresponding pairs. If one input is short, excess elements are discarded > from the right end of the longer sequence. Or just change the implementation for it to work with infinite `Seq`s. For those of you who are curious, we ended up using the following code to fix the *infinite `Seq`s problem*: > -- using the `zip` from `Prelude` > zip (toList ys) (repeat x) [1] https://github.com/chamini2/sapphire [2] http://hackage.haskell.org/package/containers-0.5.5.1/docs/src/Data-Sequenc… [3] http://hackage.haskell.org/package/containers-0.5.5.1/docs/Data-Sequence.ht…

4 5