January 2007 - Haskell-Cafe

[Haskell-cafe] Can't figure out how to "show" for type "Int -> (Int, Int)"
by Tom Titchener 11 Jan '07

11 Jan '07

Typing up and running (via Hugs) the examples in Wadler's excellent "Monads for functional programming" (here: http://homepages.inf.ed.ac.uk/wadler/papers/marktoberdorf/baastad.pdf) I hit the inevitable "show" function error: ERROR - Cannot find "show" function for: *** Expression : eval answer *** Of type : Int -> (Int,Int) I can find lots of nice text about extending data for "show" (including the miraculous "Deriving") but... I'm showing my Haskell newbie roots by utterly failing to understand how to do this of a "higher-order" type (my best guess for the right term for the type Int -> (Int, Int)). It'd be a big help if somebody could tell me either: a) It's obvious, you moron, just <insert Haskell code here> or b) It's impossible, you moron, you're trying to violate the <insert Haskell rule here> Here' my code: data Term = Con Int | Div Term Term type M a = State -> (a, State) -- higher-order type, e.g. function type type State = Int -- type synonym eval :: Term -> M Int eval (Con a) x = (a, x) eval (Div t u) x = let (a, y) = eval t x in let (b, z) = eval u y in (a `div` b, z + 1) answer, error :: Term answer = (Div(Div(Con 1972)(Con 2))(Con 23)) error = (Div(Con 1)(Con 0)) I get the "ERROR" message when I type "eval answer" at the Hugs prompt. Thanks! Tom Titchener

3 2

HaskellForge?
by Imam Tashdid ul Alam 11 Jan '07

11 Jan '07

is it a good idea to have HaskellForge? Ruby, Lua and some other languages have already adopted GForge, and I must say, those sites look *impressive*!!! any Ruby programmer on the list? can anyone provide an estimate of the amount of work involved? cheers, and long live the lambda revolution. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

13 14

restricted existential datatypes
by Misha Aizatulin 10 Jan '07

10 Jan '07

hi, I am wondering whether it would be possible to use the existing haskell type system to simulate a certain feature. Namely, I am trying to apply ideas from [1] to existential datatypes. First I will describe the problem and later say something about the context in which it arose. I am using existential boxes like > data Box cxt = forall a . Sat (cxt a) => Box a here Sat is taken from [1]: > class Sat a where dict :: a The result is a box type which can have variable context imposed on its contents. What I noticed is that sometimes I want to write functions that operate on the Box and make use of some part of the context without knowing the context completely. Such a function would look something like this: > f :: (Contains cxt ShowCxt) => Box cxt -> String > f (Box a) = show a The type is meant to say that the context of the box must contain Show as one of the classes. I would imagine the Contains class to be something like > class Contains cxt subCxt where > subDict :: cxt a -> subCxt a Now inside f I would like to have the real (Show a) context available. So I'd have an instance > (Sat cxt a, Contains cxt ShowCxt) => Show a where > show = -- <the show function extracted from (subDict dict)> This instance of course does not work in ghc, because cxt is hanging in the air - I get an error whenever it tries to infer (Show a). The question I'd like to ask is whether there is some trickery to circumvent this problem. In the f function I'd like to be able to hint to the compiler that I want Show to be derived from cxt which is attached to the Box, but I see no way of doing that. (This by the way seems connected to http://www.mail-archive.com/haskell@haskell.org/msg19564.html) Now a couple of words about where I stumbled upon this problem. I am writing a program that is supposed to use plugins. The plugins operate among other things on Boxes as above. Now some plugins would like to demand some additional capabilities from the Box that cannot be foreseen at the spot where Box is defined. Parametrizing over context seems a very natural solution to me - leading to the problem above. I would be very thankful for any suggestions! Cheers, Misha Aizatulin 1. John Hughes, "Restricted Data Types in Haskell", September 4, 1999

3 3

Re: [Haskell] Re: state of HaXml?
by Taral 10 Jan '07

10 Jan '07

On 1/10/07, Malcolm Wallace <Malcolm.Wallace(a)cs.york.ac.uk> wrote: > Actually, I'm pretty sure that most Haskell RTS implementations have a > finalizer attached to all file handles. Once the file handle is no > longer reachable from the program graph (even if its data has not been > fully consumed), the GC will close the file (via the finalizer) before > reaping the memory associated with the handle. That's not the point. The GC will only close the file when the heap is under pressure. It does not aggressively close the file, so the file may stay open for longer than the user likes. For a read-only operation, this shouldn't matter, however on some platforms an open file handle can prevent deletion of the file. -- Taral <taralx(a)gmail.com> "You can't prove anything." -- Gödel's Incompetence Theorem

2 1

What happened to Haskell original goals?
by Maurício 10 Jan '07

10 Jan '07

I remember reading "avoid success at all costs" somewhere. I miss the old days when I could read all messages in this list. Today, I can hardly keep up with a few of the most interesting threads. Good programming for all you guys in 2007. Thanks for your support and comments, and for helping me learning this really cool language. Maurício

1 0

State monad strictness - how?
by Dean Herington 09 Jan '07

09 Jan '07

I can't seem to figure out how to achieve strictness in the context of the State monad. Consider: > import Control.Monad.State > try count = print final > where (_,final) = runState prog 0 > prog = sequence_ (replicate count tick) > tick :: State Int Int > tick = do n <- get > put (n+1) > return n (try 1000000) overflows the stack. It doesn't help to use: > put $! (n+1) The only way I've been able to get the necessary strictness is to add use of ($!) in the definition of (>>=): > data SState s a = SState (s -> (a,s)) > instance Monad (SState s) where > return x = SState (\s -> (x,s)) > m >>= f = SState (\s -> let SState m' = m > (x,s1) = m' s > SState f' = f x > (y,s2) = f' $! s1 > in (y,s2)) > runSState (SState m) s = m s > sget = SState (\s -> (s,s)) > sput s' = SState (\s -> ((),s')) > stry count = print final > where (_,final) = runSState prog 0 > prog = sequence_ (replicate count stick) > stick :: SState Int Int > stick = do n <- sget > sput (n+1) > return n Is there no way to get strictness using the standard State monad? Dean

1 0

hacking ghci
by Walter Moreira 09 Jan '07

09 Jan '07

Hi Haskellers. A short while ago I asked about the possibility to have the output of ghci in color. Thanks to Donald Stewart who set up a wiki page to continue the attempt. Other people suggested HsColour and GUI's (thanks too). I'm not so interested in highlighting the snippets of code which ghci displays, but to use different colors for the different types of output: different colors for the prompt, input, warnings, exceptions, normal output, etc. Are any of these ideas viable?: 1. Hack the code of Interactive.hs in the GHC distribution and substitute the calls to 'putStr[Ln]' by some thin wrapper which can write with colors, according to a command line option (--color or --colour, for example), maybe using HsColour. This looks easy enough and I'd be willing to do it, but it's intrusive, and one has to mantain a different version of ghci in parallel with the main ghci. 2. Use GHC as library (from 6.6 on) and write a similar application as ghci, with more emphasis in the 'interactive' part of the name (in the same way as the relation between Python and the IPython extended shell). This is more work and more duplicate efforts, but it may be worthly, perhaps. 3. Write a wrapper which communicates with ghci via pipes. I read some attempts about this somewhere, and people suggested an approach as in #2. I haven't mentioned the GUI applications, since I would like to just have something in the spirit of ghci (command line oriented), but with colors or with configurable output. Which approach do you people think it is best? Sorry for the long message. Cheers, Walter

3 2

Re: question about database interfaces
by Thomas David Baker 09 Jan '07

09 Jan '07

Eric wrote: > So I'm wondering -- in practice, what do people actually use? Is > there a more-or-less generally accepted standard database library for > Haskell? In practice I am using HDBC for a PostgreSQL database and it is very easy to use and works great. http://gopher.quux.org:70/devel/hdbc T

1 0

question about database interfaces
by Eric Sessoms 09 Jan '07

09 Jan '07

Hi All, I've found the list of database libraries at http://www.haskell.org/haskellwiki/Libraries_and_tools/Database_interfaces but I haven't found much guidance in choosing among them. So I'm wondering -- in practice, what do people actually use? Is there a more-or-less generally accepted standard database library for Haskell? Thanks, Eric -- not about to see your light

2 1

question about "How to Write a Haskell Program" tutorial
by Michael Vanier 09 Jan '07

09 Jan '07

First off, I apologize if this has come up before. As far as I can tell, the mailing list archives don't have a search function. I'm running ghc-6.6 and haddock-0.8, both compiled from source. I'm working my way through the "How to Write a Haskell Program" tutorial (which is a great idea, thanks guys!), and everything works fine until I get to the part about creating the haddock documentation. When I run: $ runhaskell Setup.hs haddock I get this: Preprocessing executables for haq-0.0... Running Haddock for haq-0.0... Warning: cannot use package haq-0.0: ghc-pkg failed With verbosity: $ runhaskell Setup.hs haddock -v I get this: Preprocessing executables for haq-0.0... Running Haddock for haq-0.0... /usr/local/packages/haskell-misc/bin/haddock --html --odir=dist/doc/html/haq --title=haq --use-package=haq-0.0 --use-package=base-2.0 dist/build/tmp/Haq.hs Warning: cannot use package haq-0.0: ghc-pkg failed Note that I'm installing into a non-standard location. I have no idea why haddock (called from cabal) is trying to use a package that is in the process of being created. Nevertheless, the documentation does get generated. Is this just something that should be ignored? Is it a Cabal bug? Mike

3 2