October 2008 - Haskell-Cafe

Interesting new user perspective
by Don Stewart 14 Oct '08

14 Oct '08

http://sandersn.com/blog/index.php?title=two_weeks_of_haskell Has some good thoughts on what tripped him up in his first two weeks of Haskell programming. Anything people want to work on there? :-) Worth a read if you're thinking about how to make Haskell yet more accessible. -- Don

20 51

Associated Types and several Classes
by Martin Hofmann 14 Oct '08

14 Oct '08

> {-# OPTIONS_GHC -fglasgow-exts #-} > module Test where >import qualified Data.Set as S Hi. I try to model the following: Hypotheses are build up from Rules, which itself are made of the type Rule. Because I may change the implementation later, I want to use type classes, which define the signature of my functions I will use in other modules. >class CRule r >class (CRule (CRulesRule r) ) => CRules r where > type CRulesRule r >class (CRule (CHypoRule h), CRules (CHypoRules h) ) => CHypo h where > type CHypoRules h > type CHypoRule h > hypo :: > CHypoRules h -> > CHypoRule h -> > h -- | Rule >data Rule = Rule Int deriving(Eq,Ord) >instance CRule Rule -- | Rules >type Rules = S.Set Rule >instance CRules (S.Set Rule) where > type CRulesRule (S.Set Rule) = Rule -- | Hypothese >data Hypo = Hypo { open :: Rules > , closed :: Rules > } >instance CHypo Hypo where > type CHypoRules Hypo = Rules > type CHypoRule Hypo = Rule > hypo ro rc = Hypo { open=ro, closed=(S.singleton rc)} So far so good. But why does now the last of the following lines not type check? It says: Couldn't match expected type `CHypoRules h' against inferred type `S.Set Rule' Couldn't match expected type `CHypoRule h' against inferred type `Rule' >rule1 = Rule 1 >rule2 = Rule 2 >rule3 = Rule 3 >rules = S.fromList [rule1,rule2,rule3] >ahypo = hypo rules rule1 Shouldn't be (CHypoRules Hypo) be associated with Rules and similar (CHypoRule) with Rule? Thanks a lot, Martin -- --------------------------------------------------------------- Dipl.-Wirtsch.Inf. (E.M.B.Sc.) Martin Hofmann Cognitive Systems Group Faculty Information Systems and Applied Computer Science University of Bamberg http://www.cogsys.wiai.uni-bamberg.de/members/hofmann http://www.inductive-programming.org

2 2

couple of questions on monads
by Daryoush Mehrtash 13 Oct '08

13 Oct '08

Question 1: Why are there lazy and strict modules of some monads? (e.g. Control.monad.State) Question 2: If I define a new monad (say xyz), does it have to be as control.monad.xyz module? daryoush

4 4

Query regarding Type classes
by Arun Suresh 13 Oct '08

13 Oct '08

Hello folks, Im kinda new to haskell. Ive only been fiddling around with it for bout 3 - 4 weeks now. And for the life of me... i cant seem to figure out why this doesnt work : class Foo a where fooFunc :: a -> Int data FooData = FData instance Foo FooData where fooFunc _ = 10 class Bar a where barFunc :: (Foo b) => a -> b -> Int data BarData = BData instance Bar BarData where barFunc _ FData = 20 When I compile I get this : Couldn't match expected type `b' against inferred type `FooData' `b' is a rigid type variable bound by the type signature for `barFunc' at Sample.hs:16:20 In the pattern: FData In the definition of `barFunc': barFunc _ FData = 20 In the definition for method `barFunc' Think Im missing something really big... Could somebody kindly help me out here... Regards Arun Could so

5 7

Re: [Haskell-cafe] Still stacking monad transformers
by Tillmann Rendel 13 Oct '08

13 Oct '08

Dino Morelli wrote: > I was wishing I could do this: > > let foo = str `or-if-empty` default > > > If it was a Maybe, this works with mplus: > > (Just "foo") `mplus` (Just "bar") == Just "foo" > Nothing `mplus` (Just "bar") == Just "bar" > > > But not so much for list, mplus just ain't defined that way, instead > doing concatination: > > "foo" `mplus` "bar" == "foobar" > "" `mplus` "bar" == "bar" The difference between Maybe-style MonadPlus and List-style MonadPlus is discussed on http://www.haskell.org/haskellwiki/MonadPlus_reform_proposal. With that proposal, you could use morelse with both Maybe and lists. Personally, I would like to call these beasts <|> (for Maybe-style mplus) and <+> (for List-style mplus). One could even have a <| b = a <|> pure b a <+ b = a <+> pure b a +> b = pure a <+> b in the style of <* and *>. Note that I assume Applicative as a superclass of Monad, while I'm talking about a better world anyway. That said, I would consider using something like the <+> sketched here for Strings as a hack, because I do not see Strings as Chars under the List functor, even if they technically may be exactly that. Since "abc" does not represent multiple results, I don't think "" should represent failure. Instead, I would wrap optional strings in Maybe, and then use fromMaybe to handle the default values: fromMaybe "bar" (Just "foo") = "foo" fromMaybe "bar" Nothing = "bar" Now, with the operators sketched above, fromMaybe = flip (<|), so one could use Just "foo" <| "bar" = "foo" Nothing <| "bar" = "bar" which seems to closely correspond to what you want. Tillmann

1 0

Could not build win32 dll..
by Yakov ZAYTSEV 13 Oct '08

13 Oct '08

Hello, Sadly, I've found --mk-dll option unrecognized by latest stable GHC 6.8.3.. I'm new to win32 development How one can make win32 dll at the moment? I've successfully compiled Adder sources from example from section 11.5.4 of user guide I need to build dll to be called from foreign application The only way I see is to use MinGW's gcc which is shipped with GHC distrib to manually link everything GHC runtime needs Did I missed something? TIA -- Best wishes, Y

3 3

Still stacking monad transformers
by Andrew Coppin 13 Oct '08

13 Oct '08

I am becoming extremely frustrated now. The task I want to perform is simple, yet I simply cannot make Haskell do what I want. I've given up hope of ever getting my program to handle infinite result sets. That means I can make do with just ListT. So I have the following monad: type MyMonad x = StateT MyState (ListT Identity) x Now I'm trying to run two computations, starting from _the same state_, and combine the two resulting lists. The trouble is, I am literally losing the will to live trying to comprehend the whinings of the type checker. The operation I'm trying to perform is perfectly simple; I don't understand why this has to be so damned *difficult*! >_< Any suggestions? I found that by using the brief and easily memorable construction "runIdentity $ runListT $ runStateT foo state" I can get at the result set for each action, and combine them. But nothing in hell seems to transform this from [((), MyState)] back into MyMonad ().

9 12

Overlapping/Incoherent instances
by J. Garrett Morris 13 Oct '08

13 Oct '08

Hello everyone, As part of a project to formalize the theory of overlapping instances, I'm looking for examples of overlapping and incoherent instances and their usage. One such example would be the old version of the Monad Transformer Library, which used overlapping instances together with MonadTrans. Any other examples or suggestions would be greatly appreciated! Thanks, /g

5 6

ANNOUNCE: Graphalyze-0.4 and SourceGraph-0.2
by Ivan Lazar Miljenovic 13 Oct '08

13 Oct '08

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I'd like to announce version 0.4 of my Graphalyze library [1] and 0.2 of my SourceGraph programme [2]. [1] http://hackage.haskell.org/cgi-bin/hackage-scripts/package/Graphalyze [2] http://hackage.haskell.org/cgi-bin/hackage-scripts/package/SourceGraph This should fix the bugs reported by Gwern Branwen, Magnus Therning (thanks to Niklas Broberg for stating the correct version for Haskell-Src-Exts) and Christopher Hinson. No really new features are included in this release. I was planning on fixing this and releasing it sooner but - to utilise what seems to be a current meme [3] - "I accidentally my Gentoo" on Tuesday night and only fixed it yesterday. [3] http://encyclopediadramatica.com/I_accidentally_X Since I'm more awake now than I was when I made the initial release, here's a run-down of what SourceGraph is: SourceGraph is a programme designed to help you analyse the static complexity of your Haskell code when represented as a graph. At the moment, it does the following (M == for each module, I = for imports, C = the whole codebase): * Visualise it {M,I,C} * See a "collapsed" visualisation {M} * Proposed module/directory layout using two different algorithms {I,C} * See the "core" visualisation (recursively strip off roots/leaves) {M} * Calculate the Cyclomatic complexity [4] {M,I,C} * Root analysis (compare what's exported to what actually is a root) {M,I,C} * Determine how many components you have, to see if you should split {M,I,C} * Clique Analysis (find co-recursive functions) {M,C} * Cycle analysis (non-cliques) {M,I,C} * Chain detection (e.g. "straight-line" functions/imports) {M,I,C} [4] http://en.wikipedia.org/wiki/Cyclomatic_complexity Current limitations: * An automatic refactoring tool: SourceGraph gives *you* information on how you might want to possibly refactor your code. It's not smart: it can't tell that you've clumped functions foo and bar in the same module because they do similar things or because it's a utility module, even though they're not related. For automatic refactoring, see something like HaRe [5]. * SourceGraph ignore's "data-based" functions, i.e. record-functions and class/instance declarations, as it's too confusing (IMHO) which actual function you mean (if you see "show" being called, is it for Int, Double, or something else?). * Despite using Haskell-Src-Exts, some extensions (e.g. TH) are ignored, mainly because I have no idea how they work and nothing to test it on. GHC extensions should be supported (read the parser won't choke on them) though. * Reporting output is currently rather limited: - The report will be generated in a subdirectory called "SourceGraph" of the codebase directory; this is currently hardwired in. - It will produce an all-in-one html file report, with no fancy CSS magic to make it look pretty. Ideally, it would produce a split-file, and allow you to choose output format. - Large graphs are shrunk down to being a maximum of 15"x10". Ideally, I'd like to extend this later so that large graphs will have a shrunk version in the graph, which link to a larger version (note though that these graphs get very large very fast). - The output of individual function names, etc. could be improved. [5] http://www.cs.kent.ac.uk/projects/refactor-fp/hare.html SourceGraph can be installed with cabal-install. Once you've done so, you can analyse a cabalized library/application Foo as follows: $ SourceGraph /path/to/codebase/Foo.cabal Report generated at: /path/to/codebase/SourceGraph/Foo.html This has been written as part of my mathematics Honours Thesis, "Graph-Theoretic Analysis of the Relationships in Discrete Data". Since I've actually got to write the thesis up now, I won't be making any more releases for a while. After uni is over for the semester though, I hope to tidy it up and extend it. If you want to check the code out yourselves, there's also darcs repositories for Graphalyze [6] and SourceGraph [7]. [6] http://code.haskell.org/Graphalyze [7] http://code.haskell.org/SourceGraph/ Note that whilst Graphalyze is fully documented, etc., the internals of SourceGraph are a bit fugly (mainly due to time constraints). Enjoy! - -- Ivan Lazar Miljenovic Ivan.Miljenovic(a)gmail.com IvanMiljenovic.wordpress.com -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkjyAacACgkQfEfFJ9JhvyiHEACfVVuRk2FR3ZQiJ6H18FFK/de/ sukAn0tuCLwqxmzlQvWicQKQ3qEJKC1K =HRUK -----END PGP SIGNATURE-----

4 6

Re: [Haskell-cafe] Haskell in Artificial Intelligence
by Martin Hofmann 13 Oct '08

13 Oct '08

Hi Christos, We and a colleague from Japan use Haskell for Inductive Functional Programming, i.e. learn programs from examples. However, we just have started to port our program to Haskell: http://www.cogsys.wiai.uni-bamberg.de/effalip/ Susumu Katayama has already a Haskell library: http://nautilus.cs.miyazaki-u.ac.jp/~skata/MagicHaskeller.html More information about IP can be found here: http://www.inductive-programming.org/ What kind of survey are you doing? Do you have a pointer, too? Greetings, Martin

2 1