July 2013 - Haskell-Cafe

Re: [Haskell-cafe] memoization
by wren ng thornton 24 Jul '13

24 Jul '13

On 7/22/13 7:41 PM, David Thomas wrote: > On Mon, Jul 22, 2013 at 4:04 PM, wren ng thornton <wren(a)freegeek.org> wrote: >> In principle, we could translate between these two forms (the f2 ==> f1 >> direction requires detecting that y does not depend on x). However, in >> practice, the compiler has no way to decide which one is better since it >> involves a space/time tradeoff which: (a) requires the language to keep >> track of space and time costs, (b) would require whole-program analysis to >> determine the total space/time costs, and (c) requires the user's >> objective function to know how to weight the tradeoff ratio. > > I, for one, would love to have a compiler do (a) based on (b), my > specification of (c), and the ability to pin particular things... Oh, so would I! But, having worked on this problem in a different language, I'm well aware of how difficult it is to actually implement (a) and (b). -- Live well, ~wren

1 0

CmdArgs non-flag arguments with newtypes
by Michael Orlitzky 24 Jul '13

24 Jul '13

I ran into an issue wrapping a [String] in a newtype with CmdArgs. You're supposed to be able to declare that a field contains a list of non-flag arguments... this works fine: data Cfg = Cfg { <whatever flags>, usernames :: [String] } arg_spec = Cfg { <whatever flags>, usernames = def &= args } ... If I now call my program with, ./foo --flag1 --flag2 arg1 arg2 then usernames = [ "arg1", "arg2" ] as desired. However, I need to wrap usernames in a newtype for an unrelated reason: newtype Usernames = Usernames [String] Now, CmdArgs unexpectedly drops the first non-flag argument. So, ./foo --flag1 --flag2 arg1 arg2 gives me usernames = [ "arg2" ]. Obviously not what I want! Has anyone else run into this? Figured out a workaround?

1 0

Re: [Haskell-cafe] memoization
by wren ng thornton 23 Jul '13

23 Jul '13

On 7/22/13 9:06 AM, Tom Ellis wrote: > On Mon, Jul 22, 2013 at 07:52:06PM +1200, Chris Wong wrote: >> A binding is memoized if, ignoring everything after the equals sign, >> it looks like a constant. >> >> In other words, these are memoized: > [...] >> f = \x -> x + 1 > [...] >> and these are not: >> >> f x = x + 1 > > In what sense is the former memoised? I'm not aware of any difference > between these two definitions. Consider rather, f1 = let y = blah blah in \x -> x + y f2 x = let y = blah blah in x + y The former will memoize y and share it across all invocations of f1; whereas f2 will recompute y for each invocation. In principle, we could translate between these two forms (the f2 ==> f1 direction requires detecting that y does not depend on x). However, in practice, the compiler has no way to decide which one is better since it involves a space/time tradeoff which: (a) requires the language to keep track of space and time costs, (b) would require whole-program analysis to determine the total space/time costs, and (c) requires the user's objective function to know how to weight the tradeoff ratio. -- Live well, ~wren

4 3

catching IO errors in a monad transformer stack
by Eric Rasmussen 22 Jul '13

22 Jul '13

Hello, I am writing a small application that uses a monad transformer stack, and I'm looking for advice on the best way to handle IO errors. Ideally I'd like to be able to perform an action (such as readFile "file_that_does_not_exist"), catch the IOError, and then convert it to a string error in MonadError. Here's an example of what I'm doing now: {-# LANGUAGE FlexibleContexts #-} import Control.Monad.Error import Control.Monad.State import System.IO.Error (tryIOError) catcher :: (MonadIO m, MonadError String m) => IO a -> m a catcher action = do result <- liftIO $ tryIOError action case result of Left e -> throwError (show e) Right r -> return r This does work as expected, but I get the nagging feeling that I'm missing an underlying pattern here. I have tried catch, catchError, and several others, but (unless I misused them) they don't actually help here. The tryIOError function from System.IO.Error is the most helpful, but I still have to manually inspect the result to throwError or return to my underlying monad. Since this has come up for me a few times now, I welcome any advice or suggestions on alternative approaches or whether this functionality already exists somewhere. Thanks! Eric

5 8

News about the GSoC Project : Communicating with mobile devices
by Marcos Pividori 22 Jul '13

22 Jul '13

Hi, I am working this summer on the project "Communicating with mobile devices" for the GSoC. After working with GCM, I developed an API for communicating through APNS with iOS mobile devices [1]<https://github.com/MarcosPividori/GSoC-Communicating-with-mobile-devices/tr…> . For this, I designed a way of building the messages, sending them and handling the result. To use this service, Apple requires you to be an enrolled iOS Developer or something similar, so I decided to test the API with a local server. It simulates the APNS servers and different common scenes. Now, that I have a tested version, I am looking for some iOS developers to test in with real apps. So, I wait for feedback! Also, I have a blog to show the progress. I have posted about a simple usage of the API for GCM, I developed a test example of a Yesod server, where devices can register to receive GCM messages and users can send messages through the web service. For more information: [3]<http://gsoc2013cwithmobiledevices.blogspot.com.ar/> Thanks, Marcos The code is available on GitHub: - The APNS Api : [1]<https://github.com/MarcosPividori/GSoC-Communicating-with-mobile-devices/tr…> - Test example for APNS: [2]<https://github.com/MarcosPividori/GSoC-Communicating-with-mobile-devices/tr…> [1] https://github.com/MarcosPividori/GSoC-Communicating-with-mobile-devices/tr… [2] https://github.com/MarcosPividori/GSoC-Communicating-with-mobile-devices/tr… [3] http://gsoc2013cwithmobiledevices.blogspot.com.ar/

1 0

Wrapping all fields of a data type in e.g. Maybe
by Michael Orlitzky 22 Jul '13

22 Jul '13

I have a common pattern in my command-line programs; I start out with a configuration data type, which over-simplified looks like: data Cfg = Cfg { verbose :: Bool } Now, there's usually a default configuration, default :: Cfg default = Cfg False The user can override the defaults one of two ways, either via a config file, or from the command-line. If both are specified, the command-line takes precedence. The way I do this is with, data OptionalCfg = OptionalCfg { verbose :: Maybe Bool } And then I define I Monoid instance for OptionalCfg which lets me merge two ofthem. Once the two OptionalCfgs are merged, I merge *that* with the default Cfg. This all works great, except that when there's 20 or so options, I duplicate a ton of code in the definition of OptionalCfg. Is there some pre-existing solution that will let me take a Cfg and create a new type with Cfg's fields wrapped in Maybe?

6 9

Re: [Haskell-cafe] [Haskell] [ANN] Initial release of the threepenny-gui library, version 0.1.0.0
by Sergey Mironov 21 Jul '13

21 Jul '13

Hi, I have a Path problem when installing threepenny-gui from Hackage. Probably somtething trivial. Sergey src/BarTab.hs:9:8: Could not find module `Paths' Use -v to see a list of the files searched for. Failed to install threepenny-gui-0.1.0.0 cabal: Error: some packages failed to install: threepenny-gui-0.1.0.0 failed during the building phase. The exception was: ExitFailure 1 2013/7/21 Heinrich Apfelmus <apfelmus(a)quantentunnel.de>: > Dear Haskellers, > > And the shark, he has teeth, > And he wears them in (his) face. > And Macheath, he has a knife, > (But) yes the knife, no-one sees. > > After an obligatory cryptical quotation from a famous writer, I am pleased > to announce the first public release of *threepenny-gui*, a cheap and simple > library to satisfy your immediate GUI needs in Haskell. > > Want to write a small GUI thing but forgot to sacrifice to the giant rubber > duck in the sky before trying to install wxHaskell or Gtk2Hs? Then this > library is for you! > > Threepenny-gui is easy to install (!!) because it uses the web browser as a > display. Internally, we implement a small web server that communicates with > the browser to display GUI elements. Consequently, you can use HTML and CSS > to design the user interface. You can freely manipulate the HTML DOM and > handle browser events by writing Haskell code. > > hackage - http://hackage.haskell.org/package/threepenny-gui > examples - https://github.com/HeinrichApfelmus/threepenny-gui#examples > source - https://github.com/HeinrichApfelmus/threepenny-gui > > Many thanks to Daniel Austin for collaborating with me on this project and > to Chris Done for implementing the Ji library which is the basis for this > effort. > > > On that note, the threepenny API for creating and manipulating GUI elements > departs from earlier traditions. Do you like the new look and feel of the > API? What do you think could be improved? Try it out and send us your > feedback! > > > Best regards, > Heinrich Apfelmus > > -- > http://apfelmus.nfshost.com > > > _______________________________________________ > Haskell mailing list > Haskell(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell

3 2

Choosing an xml parser
by L Corbijn 21 Jul '13

21 Jul '13

Hello Cafe, I am trying to write a library to parse (and process) the OpenGL xml spec into haskell values. The problem is that I don't know what xml library to choose. So far I can think of the following requirements: - Some error reporting, possibly warning for unparsed elements (as that signals that the parser needs updating). - Not too bulky for the not so simple registry schema [1]. - Preferably some way to preprocess some nodes (this could probably be done by any xml library). - Preferably not too memory hungry (the current spec is about 2MB) Does somebody have a recommendation for a xml library to use for this? Regards, Lars [1]: https://cvs.khronos.org/svn/repos/ogl/trunk/doc/registry/public/api/registr…

2 1

Dynamic and equality
by Jose A. Lopes 21 Jul '13

21 Jul '13

Hello, How to define equality for Data.Dynamic ? Cheers, Jose

4 4

ANN: data-fin
by wren ng thornton 21 Jul '13

21 Jul '13

-------------------------------------------- -- data-fin 0.1.0 -------------------------------------------- The data-fin package offers the family of totally ordered finite sets, implemented as newtypes of Integer, etc. Thus, you get all the joys of: data Nat = Zero | Succ !Nat data Fin :: Nat -> * where FZero :: (n::Nat) -> Fin (Succ n) FSucc :: (n::Nat) -> Fin n -> Fun (Succ n) But with the efficiency of native types instead of unary encodings. -------------------------------------------- -- Notes -------------------------------------------- I wrote this package for a linear algebra system I've been working on, but it should also be useful for folks working on Agda, Idris, etc, who want something more efficient to compile down to in Haskell. The package is still highly experimental, and I welcome any and all feedback. Note that we implement type-level numbers using [1] and [2], which works fairly well, but not as nicely as true dependent types since we can't express certain typeclass entailments. Once the constraint solver for type-level natural numbers becomes available, we'll switch over to using that. [1] Oleg Kiselyov and Chung-chieh Shan. (2007) Lightweight static resources: Sexy types for embedded and systems programming. Proc. Trends in Functional Programming. New York, 2--4 April 2007. http://okmij.org/ftp/Haskell/types.html#binary-arithm [2] Oleg Kiselyov and Chung-chieh Shan. (2004) Implicit configurations: or, type classes reflect the values of types. Proc. ACM SIGPLAN 2004 workshop on Haskell. Snowbird, Utah, USA, 22 September 2004. pp.33--44. http://okmij.org/ftp/Haskell/types.html#Prepose -------------------------------------------- -- Links -------------------------------------------- Homepage: http://code.haskell.org/~wren/ Hackage: http://hackage.haskell.org/package/data-fin Darcs: http://community.haskell.org/~wren/data-fin Haddock (Darcs version): http://community.haskell.org/~wren/data-fin/dist/doc/html/data-fin -- Live well, ~wren

1 0