- Haskell-Cafe - Haskell.org

Not exporting a class
by Joachim Breitner 11 Nov '12

11 Nov '12

Hi, dear package authors. If you define a useful class, please think twice before you do not export the methods, otherwise something like this will happen: Am Samstag, den 10.11.2012, 21:39 -0800 schrieb David Fox: > It looks like there is an export copy of the whole prettyprint system > in template-haskell. I could switch over to that. Oh, and another > one in uulib. And also in ansi-wl-pprint. I think I will switch to > this last one. > > On Sat, Nov 10, 2012 at 6:17 PM, David Fox <dsf(a)seereason.com> wrote: > It may belong in Text.PrettyPrint. > > On Sat, Nov 10, 2012 at 6:16 PM, David Fox <dsf(a)seereason.com> > wrote: > About as small as a package can get! Yeah, the Pretty > class is in haskell-src, but they don't export the > pretty method. I will drop them a line, but I sort of > assumed they had a reason for doing it that way > > > On Thu, Nov 8, 2012 at 2:02 PM, Joachim Breitner > <mail(a)joachim-breitner.de> wrote: > Hi David, > > your pretty-class package is pretty small. In > the interest of preventing > package proliferation: Have you talked to the > maintainers of pretty or > haskell-src-exts whether they’d include the > class in their code? Thanks, Joachim -- Joachim "nomeata" Breitner Debian Developer nomeata(a)debian.org | ICQ# 74513189 | GPG-Keyid: 4743206C JID: nomeata(a)joachim-breitner.de | http://people.debian.org/~nomeata

2 1

Re: [Haskell-cafe] curl package broken in Windows
by Andrew Pennebaker 10 Nov '12

10 Nov '12

Yes, that's a very good point. Did you know that Strawberry Perl includes a cygwin gcc? http://strawberryperl.com Maybe Haskell Platform could do the same. On Nov 10, 2012 5:28 PM, "Stephen Tetley" <stephen.tetley(a)gmail.com> wrote: > The only short cut would be to have it in the Platform - otherwise it > is DIY I'm afraid. > > This is as much Microsoft's "fault" as Haskell's for not providing a C > compiler, build tools etc. in the baseline environment. >

1 0

Profiling a project with cabal and cabal-dev
by Mike Craig 10 Nov '12

10 Nov '12

Hi all, So I've been working on a project and I'd like to run it with profiling to diagnose the performance hotspots. It's a cabal project and I've been using cabal-dev for sandboxing. Normally I would just run `cabal-dev install` to get everything built and the executables "installed" in ./cabal-dev/bin. Since I want to compile with profiling, I tried this: $ cabal-dev install enable-executable-profiling The build fails with an error log that includes something like "Perhaps you haven't installed the profiling libraries for package `http-types-0.7.3.0.1'?" If I then try ... $ cabal-dev install -p --reinstall --force-reinstalls http-types-0.7.3.0.1 ... I find myself walking down the dependency tree. (http-types -> case-insensitive -> hashable -> etc.) Presumably this would eventually walk me out of the sandbox and I would be reinstalling the base libraries in my sandbox with profiling enabled. This doesn't seem right. (And if it is right, it doesn't seem like I should have to do it manually, package by package.) What am I doing wrong? Cheers, Mike

3 3

Re: [Haskell-cafe] Motion to unify all the string data types
by Tobias Brandt 10 Nov '12

10 Nov '12

On 10 November 2012 04:00, Johan Tibell <johan.tibell(a)gmail.com> wrote: > As for type classes, I don't think we use them enough. Perhaps because > Haskell wasn't developed as an engineering language, some good software > engineering principles (code against an interface, not a concrete > implementation) aren't used in out base libraries. One specific example is > the lack of a sequence abstraction/type class, that all the string, list, > and vector types could implement. Right now all these types try to > implement a compatible interface (i.e. the traditional list interface), > without a language mechanism to express that this is what they do. > Data.Collections<http://hackage.haskell.org/packages/archive/collections-api/1.0.0.0/doc/htm…> has (among others) a Sequence type class and provides instances for the base data types in a separate package. However, it appears that not many people are using it.

1 0

Computed promoted natural
by Arie Peterson 10 Nov '12

10 Nov '12

Hi, I'm trying to use data kinds, and in particular promoted naturals, to simplify an existing program. The background is as follows: I have a big computation, that uses a certain natural number 'd' throughout, which is computed from the input. Previously, this number was present as a field in many of my data types, for instance > data OldA = OldA Integer … . There would be many values of this type (and others) floating around, with all the same value of 'd'. I would like to move this parameter to the type level, like this: > data NewA (d :: Nat) = NewA … The advantage would be, that the compiler can verify that the same value of 'd' is used throughout the computation. Also, it would then be possible to make 'NewA' a full instance of 'Num', because 'fromInteger :: Integer -> NewA d' has a natural meaning (where the value of 'd' is provided by the type, i.e. the context in which the expression is used), while 'fromInteger :: Integer -> OldA' does not, because it is not possible to create the right value of 'd' out of thin air. Is this a sane idea? I seem to get stuck when trying to /use/ the computation, because it is not possible to create 'd :: Nat', at the type level, from the computed integer. Can one somehow instantiate the type variable 'd :: Nat' at an integer that is not statically known? Formulated this way, it sounds like this should not be possible, because all types are erased at compile time. However, it feels as though it might not be unreasonable in this situation, because the computation is polymorphic in the type 'd :: Nat'. I just want to substitute a specific value for 'd'. Maybe there is another way to approach this? Thanks for any advice, Arie

3 7

Re: [Haskell-cafe] GHC for mobile devices?
by Andrew Pennebaker 10 Nov '12

10 Nov '12

Awesome! Jeffrey Scofield has ported OCaml to iOS<http://psellos.com/ocaml/compile-to-iossim.html>, so there's also experience there. On Fri, Nov 9, 2012 at 11:46 PM, Kristopher Micinski <krismicinski(a)gmail.com > wrote: > If you have interest in doing this, I have quite a bit of experience > in Android hacking at the system level and above and would be glad to > talk about what might need to happen. (Though I don't know the GHC > internals / toolchain so well.) > > One potential choice is Scala, though from my limited experience > that's a very rough imitation of the uses for Haskell. (Though, > obviously it works mostly out of the box because of the JVM compiler > target..) > > http://www.haskell.org/haskellwiki/Android > > I've been writing up some thoughts on the Android activity lifecycle > already interpreted with respect to FP, apps are quite functional > already for a variety of reasons. > > kris > > On Fri, Nov 9, 2012 at 8:51 PM, Andrew Pennebaker > <andrew.pennebaker(a)gmail.com> wrote: > > I'd love to use Haskell directly for making mobiles apps. How can we make > > this happen, porting GHC to Android, iOS, and Windows Phone? > > > > -- > > Cheers, > > > > Andrew Pennebaker > > www.yellosoft.us > > > > _______________________________________________ > > Haskell-Cafe mailing list > > Haskell-Cafe(a)haskell.org > > http://www.haskell.org/mailman/listinfo/haskell-cafe > > > -- Cheers, Andrew Pennebaker www.yellosoft.us

3 2

curl package broken in Windows
by Andrew Pennebaker 10 Nov '12

10 Nov '12

When I do "cabal install curl" in Windows, a missing cygwin dll message pops up, and packages that depend on curl, like shpider, fail to install. Specs: * cabal-install 0.14.0 * Cabal 1.14.0 * GHC 7.4.2 * Haskell Platform 2012 * Windows 7 Proffesional x64 -- Cheers, Andrew Pennebaker www.yellosoft.us

2 2

haskell.org committee: Call for nominations
by Jason Dagit 09 Nov '12

09 Nov '12

Dear Haskellers, The second year of the haskell.org committee is drawing to a close and it is therefore time to seek replacements for those members whose term is expiring. This year one member has a term ending, Ganesh Sittampalam. Ganesh has been an excellent contributor and the rest of the committee would like to thank him for his excellent service. To nominate yourself, please send an e-mail to committee at haskell.org by 30 November 2012. As usual, retiring members are eligible to re-nominate themselves. Please feel free to include any information about yourself that you think will help us to make a decision. Being a member of the committee does not require a significant amount of time, but committee members should aim to be responsive during discussions when the committee is called upon to make a decision. More details about the committee's roles and responsibilities are on http://www.haskell.org/haskellwiki/Haskell.org_committee If you have any questions about the process, please feel free to e-mail us at committee at haskell.org or to contact one of us individually. Regards, Jason On behalf of the haskell.org committee

1 1

Waiting for garbage collection can kill parallelism?
by Janek S. 09 Nov '12

09 Nov '12

Today I was reading "Parallel Performance Tuning for Haskell" by Jones, Marlow and Singh and wanted to replicate the results for their first case study. The code goes like this: module Main where import Control.Parallel main :: IO () main = print . parSumFibEuler 38 $ 5300 parSumFibEuler :: Int -> Int -> Int parSumFibEuler a b = f `par` (e `pseq` (e + f)) where f = fib a e = sumEuler b fib :: Int -> Int fib 0 = 0 fib 1 = 1 fib n = fib (n - 1) + fib (n - 2) mkList :: Int -> [Int] mkList n = [1..n-1] relprime :: Int -> Int -> Bool relprime x y = gcd x y == 1 euler :: Int -> Int euler n = length (filter (relprime n) (mkList n)) sumEuler :: Int -> Int sumEuler = sum . (map euler) . mkList sumFibEuler :: Int -> Int -> Int sumFibEuler a b = fib a + sumEuler b This is the version shown on page 3 of the paper, after adding the pseq combinator to enforce correct evaluation order. I compile and run it with: ghc -O2 -rtsopts -threaded -eventlog parallel.hs ./parallel +RTS -s -ls -N2 In the paper authors show that this version does in fact perform computation in parallel and that good speedup is achieved. However, when I run the code what happens is that HEC 1 blocks very quickly requesting GC. HEC 0 (if I am correct the one calculating sumEuler) does not interrupt but instead continues the computations until they are finished. Then the GC is performed and the HEC 1 resumes computation. In this way there is no parallelism, because first HEC 0 does all its computations and after first GC HEC 1 does its computation. My question is why this might be happening? I don't expect the results of the paper to be fully reproducible, because the paper is 3 years old and GHC has developed a lot since then. This however looks like a regression of some sort. I would appreciate if anyone could explain why this. Janek

1 0

Why doesn't GHC optimize recursive functions by converting them into `let`?
by Petr P 09 Nov '12

09 Nov '12

Hi, recently I found out that some recursive functions can be greatly optimized just by rewriting them using `let` so that they're not recursive themselves any more (only the inner let is). To give an example: > fix :: (a -> a) -> a > fix f = f (fix f) isn't optimized, because it's a recursive function and so it isn't inlined or anything. However, defining it as > fix f = let x = f x in x makes it much more efficient, since `fix` is not recursive any more, so it gets inlined and then optimized for a given argument `f`. (See http://stackoverflow.com/questions/12999385/why-does-ghc-make-fix-so-confou… ) Another example is the well-known generic catamorphism function: > newtype Fix f = Fix { unfix :: f (Fix f) } > > catam :: (Functor f) => (f a -> a) -> (Fix f -> a) > catam f = f . fmap (catam f) . unfix is not optimized. When `catam` is used on a data structure such as [] or a tree, at each level `fmap` creates new constructors that are immediately consumed by `f`. But when written as > catam f = let u = f . fmap u . unfix in u this gets inlined and then optimized, and constructor creation/consumption is fused into very effective code. (See http://stackoverflow.com/q/13099203/1333025) As one comment asked, this seems like something GHC could do automatically. Would it be difficult? Is there a reason against it? I suppose in cases where it doesn't help, the additional `let` would get optimized away, so no harm would be done. And in cases like `fix` or `catam` the gain would be substantial. Best regards, Petr

2 2