February 2001 - Hugs-Users

FunDeps & Monad transformers
by Jose Emilio Labra Gayo 26 Feb '01

26 Feb '01

I was using monad transformers with definitions like the following: > -- class Monad m => StateMonad s m > -- where update :: (s -> s) -> m s However, if I try to define > test :: ( StateMonad s m, Num s ) => m () > test = do { update (+1) ; return () } I obtain ERROR ... - Ambiguous type signature in type declaration *** ambiguous type : (StateMonad a b, Num a) => b () *** assigned to : test I thought that functional dependencies could be useful to solve those typings. So I wrote: > class Monad m => StateMonad s m | m -> s > where update :: (s -> s) -> m s And now, the 'test' function loads with Hugs. However, when I try to declare the following instances (which previously worked ok): > class (Monad m, Monad (t m)) => MonadT t m where > lift :: m a -> t m a > newtype StateT s m v = S (s -> m (v,s)) > instance ( StateMonad s m > , MonadT t m) => StateMonad s (t m) > where update = lift . update > instance Monad m => StateMonad s (StateT s m) > where update f = S (\s -> return (s, f s)) I obtain the following error ERROR ... - Instances are not consistent with dependencies *** This instance : StateMonad a (StateT a b) *** Conflicts with : StateMonad a (b c) *** For class : StateMonad a b *** Under dependency : b -> a Is it possible to define those declarations? Best regards, Jose Labra http://lsi.uniovi.es/~labra

1 0

Lo mejor de Internet aqu�
by promocionesdosmiluno＠yahoo.es 26 Feb '01

26 Feb '01

Visita esta web http://MundoEspanya.redireccion.com

1 0

Existential quantification
by George Russell 23 Feb '01

23 Feb '01

Excuse me, but is it a bug or a feature that Hugs (98-Feb2001), with the -98 option, rejects: data G b = forall a . H a b => G a b class H a b where with "Variable "b" in constraint is not locally bound"? PS - please CC me with replies as I don't normally subscribe to this list.

3 3

Hugs' behavior
by Koen Claessen 22 Feb '01

22 Feb '01

Hi, I noticed an annoying incompatibility between Hugs (Feb 2000) and Hugs (Feb 2001), namely the fact that when evaluating expressions of type `IO T' in Hugs (Feb 2001), the following happens: 1. when T = (), the IO action is just evaluated. 2. when T /= (), the IO action is evaluated, *and* the result is printed, but only if T is an instance of Show. 3. when T is no instance of Show, just a type error is generated. Hugs (Feb 2000) has behavior 1. in all cases for T. It is discussable which behavior is better. But I think there should at least be a flag specifying what behavior you want. And I also think that it should not be the default! I don't like the new behavior because: * The Haskell 98 report does not say anything about printing the result of the main :: IO T action. * There is a rather arbitrary distinction between T = () and T /= (). * The error in case 3. is totally unacceptable. (Especially because there are interactive systems (such as Lava) which rely on Hugs' (Feb 2000) behavior.) Take a look at the following examples, which illustrate some of the issues. Prelude> return () :: IO () Prelude> return id :: IO (Int -> Int) ERROR - Cannot find "show" function for IO result: *** Expression : return id *** Of type : IO (Int -> Int) This is really strange too: Prelude> return undefined :: IO () Prelude> return undefined :: IO Int Program error: {undefined} And here: Prelude> print "apa" >> return () :: IO () apa Prelude> print "apa" >> return id :: IO (Int -> Int) ERROR - Cannot find "show" function for IO result: *** Expression : print "apa" >> return id *** Of type : IO (Int -> Int) What do you think? /Koen. -- Koen Claessen http://www.cs.chalmers.se/~koen phone:+46-31-772 5424 mailto:koen@cs.chalmers.se ----------------------------------------------------- Chalmers University of Technology, Gothenburg, Sweden

1 0

Feb 2001 MSI available
by Sigbjorn Finne 20 Feb '01

20 Feb '01

Hi, I've packaged up a Windows Installer (.msi) for people that want to easily set up and use Hugs98 February 2001 on a Win32 platform. The MSI can be downloaded from http://www.dcs.gla.ac.uk/~sof/hugs98.msi (1.4M) It contains executables (hugs,runhugs and winhugs), libs (std + win32), docs (html, hlp and pdf) and the demos. The only thing that's (intentionally) missing is the Hugs98 source code. In the unlikely event that your machine doesn't already have the Windows Installer engine/runtime installed, runtime redistributables can be downloaded from http://www.microsoft.com/msdownload/platformsdk/instmsi.htm --sigbjorn

2 1

ANNOUNCE: Hugs98 Feb 2001
by Johan Nordlander 15 Feb '01

15 Feb '01

====================================================================== __ __ __ __ ____ ___ || || || || || || ||__ ||___|| ||__|| ||__|| __|| ||---|| ___|| || || || || Version: Feb 2001 ====================================================================== Dear all, I'm pleased to announce the availability of a new release of Hugs98. This is a major release that incorporates bug fixes as well as several new features and enhancements that have been developed for Hugs over the last year. It is announced with the intentions that it will remain a stable and lightweight implementation of Haskell 98 + extensions for some considerable time. A list of the most important new features looks as follows: - A Foreign Function Interface closely modelled after the one provided by GHC. - Built-in, Hood-like debugging support. - Parallel list comprehensions, a.k.a. zip-comprehensions. - A new syntax for recursive monad bindings. - A new GUI under Windows that doesn't consume all CPU time. - Support for the MacOS platform integrated into the main distriution. - Corrections of all bugs reported for the January 2001 beta release. See the release notes in the "docs" directory of the distribution for more information on the new language features. Send email to hugs-users-request(a)haskell.org to join the hugs-users mailing list. Bug reports should be sent to hugs-bugs(a)haskell.org. Send email to hugs-bugs-request(a)haskell.org to subscribe to the hugs-bugs list. The home page and download site for Hugs is at http://www.haskell.org/hugs/ Enjoy, Johan Nordlander [Hugs maintainer]

1 0

Assertions in Hugs
by Andreas Gruenbacher 13 Feb '01

13 Feb '01

Hi all, I'm looking for a way to assert certain facts in Haskell modules. Is there a way to do this in Hugs? This would be a very nice feature; preferably with a command line switch to control whether assertions are actually run or not. ------------------------------------------------------ module Test where class Test t where f, g :: t -> t instance Test Int where f x = x test1, test2 :: Int test1 = f 3 -- 3 test2 = g 3 -- Program error: Undefined member: g -- this doesn't actually work in Hugs: --assert (test1 == 3) --assert (test2 == 3) ------------------------------------------------------ Thanks, Andreas. ------------------------------------------------------------------------ Andreas Gruenbacher gruenbacher(a)geoinfo.tuwien.ac.at Research Assistant Phone +43(1)58801-12723 Institute for Geoinformation Fax +43(1)58801-12799 Technical University of Vienna Cell phone +43(664)4064789

2 1

Observe and limits
by Patrik Jansson 07 Feb '01

07 Feb '01

With observe in the new hugs beta we have an interesting new toy to play with. I tried observing the state (current character, line and column) of an (inefficient) monadic expression parser to see the amount of backtracking, but I immediately ran into the limit on number of observation tags (400). I guess I used observe in a way it was not inteded for ... observe (show (c,line,col)) c ... producing lots of tag, each with a few (identical) observations, but nevertheless it would be nice to have a simpler way of changing the maximum number of tags than to recompile hugs. (Or at least have a larger default for big hugs.) (For the record, 4000 was not enough either for this example.) When debugging, or performance evaluating using HugsHood it would also be nice to have a possibility of storing something else than just a list of observed values. For the parser example (and in many other cases), just incrementing a tag-specific counter at each evaluation of observe (or a similar extended function) would be very useful. /Patrik Jansson

1 0

ANNOUNCE: Hugs 98 January 2001 beta
by Johan Nordlander 03 Feb '01

03 Feb '01

================================================================== __ __ __ __ ____ ___ || || || || || || ||__ ||___|| ||__|| ||__|| __|| ||---|| ___|| || || || || Version: January 2001 beta ================================================================== Dear all, I'm pleased to announce the availability of a new beta release of Hugs98. This is a beta release that aims at giving as many people as possible a chance to react to the changes and additions that have recently been incorporated, before the next major version is released on February 14 2001. A list of the most important new features looks as follows: - A Foreign Function Interface closely modelled after the one provided by GHC. - Built-in, Hood-like debugging support. - A new syntax for recursive monad bindings. - A new GUI under Windows that doesn't consume all CPU time. See the release notes in the "docs" directory of the distribution for more information on some of these features. Send email to hugs-users-request(a)haskell.org to join the hugs-users mailing list. Bug reports should be sent to hugs-bugs(a)haskell.org. Send email to hugs-bugs-request(a)haskell.org to subscribe to the hugs-bugs list. The home page and download site for Hugs is at http://www.haskell.org/hugs/ Enjoy, Johan Nordlander PS Please don't confuse this release with latest (January 31) release of a new version of O'Hugs, the O'Haskell interpreter!

1 0