- Haskell-Cafe - Haskell.org

symbol $
by Maverick 07 May '07

07 May '07

Hi Do you mind tell what is mean by the symbol $ in this piece of code? Thanks a lot. > mcompile :: Prog -> State Int Code > mcompile (Assign n e) = return $ compileExpr e ++ [POP n] > mcompile (If e p1 p2) = do cur <- fresh > pp1 <- mcompile p1 > pp2 <- mcompile p2 > return $ (compileExpr e ++ (JUMPZ cur):pp1 ++ > JUMP (cur+1):(LABEL cur):pp2 ++ [LABEL (cur+1)]) > mcompile (While e p) = do cur <- fresh > mp <- mcompile p > return $ [LABEL cur] ++ compileExpr e ++ [JUMPZ (cur+1)] ++ mp ++ [JUMP cur,LABEL (cur+1)] -- View this message in context: http://www.nabble.com/symbol-%24-tf3701618.html#a10351407 Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

3 3

Haskell Weekly News: May 07, 2007
by dons＠cse.unsw.edu.au 07 May '07

07 May '07

--------------------------------------------------------------------------- Haskell Weekly News http://sequence.complete.org/hwn/20070507 Issue 62 - May 07, 2007 --------------------------------------------------------------------------- Welcome to issue 62 of HWN, a weekly newsletter covering developments in the [1]Haskell community. This week sees the release of Atom, a hardware description language embedded in Haskell, along with the usual suite of new libraries and tools. In addition, The Monad.Reader Issue 7 was released, and the hackage upload festival continues unabated. 1. http://haskell.org/ Announcements Atom: Hardware Description in Haskell. Tom Hawkins [2]announced the release of [3]Atom, a high-level hardware description language embedded in Haskell, compiles conditional term rewriting systems into conventional HDL. 2. http://article.gmane.org/gmane.comp.lang.haskell.general/15209 3. http://www.funhdl.org/ The Monad.Reader: Issue 7. Wouter Swierstra [4]announced the latest issue of [5]The Monad.Reader. The Monad.Reader is a quarterly magazine about functional programming. It is less-formal than journal, but somehow more enduring than a wiki page or blog post. 4. http://article.gmane.org/gmane.comp.lang.haskell.cafe/22038 5. http://www.haskell.org/haskellwiki/The_Monad.Reader HDBC: Haskell Database Connectivity. John Goerzon [6]announced that [7]HDBC 1.1.2 is now released. HDBC provides an abstraction layer between Haskell programs and SQL relational databases. This lets you write database code once, in Haskell, and have it work with any number of backend SQL databases. 6. http://article.gmane.org/gmane.comp.lang.haskell.general/15227 7. http://software.complete.org/hdbc FileManip: Expressive Filesystem Manipulation. Bryan O'Sullivan [8]announced the [9]FileManip package provides expressive functions and combinators for searching, matching, and manipulating files. 8. http://article.gmane.org/gmane.comp.lang.haskell.cafe/22090 9. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/FileManip-0.1 photoname: manipulate photos using EXIF data. Dino Morelli [10]announced the release of [11]photoname, a command-line utility for renaming and moving photo image files. The new folder location and naming are determined by two things: the photo shoot date information contained within the file's EXIF tags and the usually-camera-assigned serial number, often appearing in the filename. 10. http://article.gmane.org/gmane.comp.lang.haskell.general/15187 11. http://ui3.info/d/proj/photoname.html RSA-Haskell: Command-line Cryptography. David Sankel [12]announced the release of [13]RSA-Haskell, a collection of command-line cryptography tools and a cryptography library written in Haskell. It is intended to be useful to anyone who wants to secure files or communications or who wants to incorporate cryptography in their Haskell application. 12. http://article.gmane.org/gmane.comp.lang.haskell.general/15207 13. http://www.netsuperbrain.com/rsa-haskell.html Haskell modes for Vim. Claus Reinke [14]summarised the various Haskell/Vim support currently available 14. http://article.gmane.org/gmane.comp.lang.haskell.general/15180 French Translation of Gentle Introduction to H98. The haskell-fr team [15]announced a completed a [16]translation into French of the 'Gentle Introduction to Haskell'. 15. http://article.gmane.org/gmane.comp.lang.haskell.general/15193 16. http://gorgonite.developpez.com/livres/traductions/haskell/gentle-haskell/ Haskell' This section covers the [17]Haskell' standardisation process. * [18]Polymorphic strict fields 17. http://hackage.haskell.org/trac/haskell-prime 18. http://thread.gmane.org/gmane.comp.lang.haskell.prime/2192 Hackage This week's new libraries in [19]the Hackage library database. 19. http://hackage.haskell.org/ * BitSyntax-0.2. Adam Langley. [20]A simple function for the construction of binary data. 20. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/BitSyntax-0.2 * filepath-1.0. Neil Mitchell. [21]Library for manipulating FilePath's in a cross platform way. 21. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/filepath-1.0 * Chart-2007.3.5. Tim Docker [22]A library for generating 2D Charts and Plots. 22. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/Chart-2007.3.5 * FileManip-0.1. Bryan O'Sullivan [23]A Haskell library for working with files and directories. 23. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/FileManip-0.1 * hsns-0.5.2. Austin Seipp [24]A network sniffer written in a purely fun language. 24. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hsns-0.5.2 * template-0.1. Johan Tibell [25]Simple string substitution library that supports dollar-based substitution. 25. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/template-0.1 * ASN1-0.0.1. Dominic Steinitz [26]ASN.1 suppport for X.509 identity and attribute certificates, PKCS8, PKCS1v15. 26. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/ASN1-0.0.1 Discussion The Proper Definition of (evaluate :: a -> IO a). Isaac Dupree [27]described a variant of evaluate with modified semantics to the current implementation. 27. http://article.gmane.org/gmane.comp.lang.haskell.libraries/6858 Why is Data.Set not a monad?. Dan Doel [28]documented the reasons why Data.Set is not currently an instance of Monad. 28. http://article.gmane.org/gmane.comp.lang.haskell.libraries/6861 Chaos. Andrew Coppin [29]announced chaos, a fun image generating mystery program. 29. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/22177 The Functional Pearls. Don Stewart [30]collected the functional pearls known to be available online, on to a single page on the Haskell wiki. 30. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/22171 Blog noise [31]Haskell news from the blogosphere. * [32]HUG: Nested Data Parallelism in Haskell * [33]HUG: London Haskell User Group * [34]New York Functional Programmers meeting: roundup * [35]Python-style string split in Haskell * [36]Thinking in types * [37]Using the Haskell package system * [38]Fixed precision, an update * [39]Use of Text.XHtml.Strict for Outputting XHTML * [40]Idiom: Plan for Currying * [41]Bowling in Haskell * [42]The Trivial Monad * [43]Homeland Security Threat Level Monad * [44]Monads as universe helpers * [45]Understanding comonads * [46]Coding in Haskell: conciseness * [47]Functional programming in Wall Street * [48]Phantom Types for Real Problems * [49]Haskell rocks * [50]Advertising the ICFP Programming Contest * [51]Haskell wikibook * [52]Roll Your Own Window Manager: Part 1: Defining and Testing a Model * [53]Haskell and the Type Calculus: or, the Good -> Bad -> Ugliness of Types * [54]FileManip, an expressive Haskell library for manipulating files * [55]Programming in Haskell * [56]Parsing JSON in Haskell * [57]Namespace confusion * [58]A Scheme parser in Haskell * [59]Simple performance analysis * [60]Playing with Haskell unsafely * [61]repeat and sequence * [62]Haskell and C: functions returning more than one value 31. http://planet.haskell.org/ 32. http://www.londonhug.net/2007/05/01/hug-update-nested-data-parallelism-in-h… 33. http://www.phildawes.net/blog/2007/05/02/london-haskell-user-group/ 34. http://privosoft.blogspot.com/2007/05/new-york-functional-programmers-meeti… 35. http://gimbo.org.uk/blog/2007/04/20/splitting-a-string-in-haskell/ 36. http://totherme.livejournal.com/3845.html 37. http://gimbo.org.uk/blog/2007/04/27/haskell-packages-gotcha-global-vs-per-u… 38. http://augustss.blogspot.com/2007/04/fixed-precision-update-so-i-was-bit.ht… 39. http://www.movethemarkets.com/richard/2007/04/14/use-of-textxhtmlstrict-for… 40. http://www.movethemarkets.com/richard/2007/04/28/functional-programming-idi… 41. http://www.randomhacks.net/articles/2007/04/28/bowling-in-haskell 42. http://sigfpe.blogspot.com/2007/04/trivial-monad.html 43. http://sigfpe.blogspot.com/2007/04/homeland-security-threat-level-monad.html 44. http://gelisam.blogspot.com/2006/10/monads-as-universe-helpers.html 45. http://gelisam.blogspot.com/2007/04/i-understand-comonads.html 46. http://defowl.blogspot.com/2007/04/coding-in-haskell.html 47. http://codemiscellany.blogspot.com/2007/04/functional-programming-in-wall-s… 48. http://neilmitchell.blogspot.com/2007/04/phantom-types-for-real-problems.ht… 49. http://errantember.livejournal.com/50627.html 50. http://johanjeuring.blogspot.com/2007/05/advertising-icfp-programming-conte… 51. http://koweycode.blogspot.com/2007/05/haskell-wikibook-now-featured.html 52. http://cgi.cse.unsw.edu.au/~dons/blog/2007/05/01#xmonad_part1_model 53. http://disparatemathematician.blogspot.com/2007/05/haskell-and-type-calculu… 54. http://www.serpentine.com/blog/2007/05/01/filemanip-an-expressive-haskell-l… 55. http://www.regdeveloper.co.uk/2007/05/03/programming_haskell/ 56. http://notes-on-haskell.blogspot.com/2007/05/parsing-json.html 57. http://notes-on-haskell.blogspot.com/2007/05/namespaces-confusion.html 58. http://sami.samhuri.net/articles/2007/05/03/a-scheme-parser-in-haskell-part… 59. http://syntaxfree.wordpress.com/2007/05/04/lra/ 60. http://therning.org/magnus/archives/249 61. http://therning.org/magnus/archives/269 62. http://therning.org/magnus/archives/280 Quotes of the Week * Oleg K: So, `bind' is `let' and monadic programming is equivalent to programming in the A-normal form. That is indeed all there is to monads * kc5tja: Premature evil is the root of all optimization * Tommah: Remember, kids: if you program in a language with side effects, the terrorists win. * ndm: Comments are for people who can't sense what their code does from the indentation * jcreigh: GHC has lots of interesting features above Haskell98, I've noticed. 'You can take the red pill or the blue pill...' 'Hmm. What's the green pill?' 'What? Oh. That's GHC.' * schluehk: It's about a variant of the other big Haskell credo: once it compiles it works. Once you have written a prototype you have also a spec. If this is not agile I don't know what? It is a quite remarkable inversion. Formerly people wanted tools that are so versatile that they let them express almost everything with great ease and where they didn't care a lot about speed optimizations and corner cases in the early iterations. Now people want tools that restricts intentionally their expressivity to let them do big upfront design as source code. They want to be guided to initial perfection. Let's face it: Haskell has quite some momentum in the dialectic move. Code Watch Notable new features and bug fixes to the Haskell compilers. Thu May 3 06:19:55 PDT 2007. Simon Marlow. [63]Add history/trace functionality to the GHCi debugger. The debugger can now log each step of the evaluation without actually stopping, keeping a history of the recent steps (currently 50). When a (real) breakpoint is hit, you can examine previous steps in the history (and their free variables) using the :history, :back and :forward commands. 63. http://article.gmane.org/gmane.comp.lang.haskell.cvs.ghc/20551 Wed May 2 09:34:57 PDT 2007. Simon Peyton-Jones. [64]Make records work properly with type families. This fixes Trac #1204. There's quite a delicate interaction of GADTs, type families, records, and in particular record updates. 64. http://article.gmane.org/gmane.comp.lang.haskell.cvs.ghc/20522 About the Haskell Weekly News Each week, new editions are posted to [65]the Haskell mailing list as well as to [66]the Haskell Sequence and [67]Planet Haskell. [68]RSS is also available, and headlines appear on [69]haskell.org. Headlines are available as [70]PDF. To help create new editions of this newsletter, please see the [71]contributing information. Send stories to dons at cse.unsw.edu.au. The darcs repository is available at darcs get [72]http://www.cse.unsw.edu.au/~dons/code/hwn 65. http://www.haskell.org/mailman/listinfo/haskell 66. http://sequence.complete.org/ 67. http://planet.haskell.org/ 68. http://sequence.complete.org/node/feed 69. http://haskell.org/ 70. http://www.cse.unsw.edu.au/~dons/code/hwn/archives/20070507.pdf 71. http://haskell.org/haskellwiki/HWN 72. http://www.cse.unsw.edu.au/~dons/code/hwn

2 1

Re: [Haskell-cafe] Re: (Chaos) [An interesting toy]
by Andrew Coppin 07 May '07

07 May '07

>> So... if I'm not actually using lazyness for anything, would it do >> any harm to change Double to Double# in the Haskell source code? Or >> is there some reason why this would be undesirable? (The source is >> carefully constructed > > Well, you'd need -fglasgow-exts for it to even compile. Other than > that, you'd have to use all primitive operations and you might well be > preventing the compiler from doing useful optimizations. But this is > really a question for the Simons or other folks who actualy understand > the GHC internals. Hmm... that doesn't sound like such a hot idea...

1 0

Monad and Monad transformer ???
by Maverick 07 May '07

07 May '07

hi I am a newbee to haskell so please forgive me if the question is not appropreate. I am trying to get the idea of using Monad and have to understand the following code. Can you please tell me is how different Monad and Monad transformer is? And what is mean by > newtype State s a = State { runState :: s -> (a, s) } --------------------------------- Define (State s) as a State monad > instance Monad (State s) where > return a = State $ \s -> (a,s) > m >>= k = State $ \s -> let (a,s') = runState m s in runState (k a) s' Thanks lot for your help I really do appreciate that Best regard -- View this message in context: http://www.nabble.com/Monad-and-Monad-transformer-----tf3699943.html#a10346… Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

2 1

Intermediate Haskell Books?
by Adrian Neumann 07 May '07

07 May '07

-----BEGIN PGP SIGNED MESSAGE----- Hash: RIPEMD160 Are there any good books about intermediate to advanced Haskell? The descriptions here http://haskell.org/haskellwiki/Books_and_tutorials aren't very helpful. Adrian -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFGPZCo11V8mqIQMRsRA8nVAJ9fs4c013qKOuEjX//uLW3hLTYtxACfTS/t 5D6I+/OGifs3ltYhx6rwFZA= =sNJP -----END PGP SIGNATURE-----

5 4

c2hs errors when compiling hsGnuTls
by David House 07 May '07

07 May '07

Hey there, I'm getting the following errors when I try to compile hsGnuTls [1]: ~/hs/sandbox/hsgnutls $ c2hs --version C->Haskell Compiler, version 0.14.5 Travelling Lightly, 12 Dec 2005 build platform is "i486-pc-linux-gnu" <1, True, True, 1> ~/hs/sandbox/hsgnutls $ runhaskell Setup.lhs build Setup.lhs: Warning: The field "hs-source-dir" is deprecated, please use hs-source-dirs. Preprocessing library hsgnutls-0.2.3... c2hs: Error in C header file. /usr/include/bits/pthreadtypes.h:69: (column 6) [FATAL] >>> Syntax error! The symbol `;' does not fit here. Setup.lhs: got error code while preprocessing: Network.GnuTLS.GnuTLS c2hs version: I've attached the file it references in case that's relevant. Any tips on how I might address this? Thanks in advance, -David House, dmhouse(a)gmail.com

2 2

The Monad.Reader - Call for copy
by Wouter Swierstra 07 May '07

07 May '07

Call for Copy The Monad.Reader - Issue 8 I would like to welcome articles for the next issue of The Monad.Reader. * The Monad.Reader * The Monad.Reader is a electronic magazine about all things Haskell. It is less-formal than journal, but somehow more enduring than a wiki- page. There have been a wide variety of articles, including: exciting code fragments, intriguing puzzles, book reviews, tutorials, and even half-baked research ideas. * Submission Details * Get in touch with me if you intend to submit something -- the sooner you let me know what you're up to, the better. Please submit articles for the next issue by e-mail (wss at cs.nott.ac.uk) to me before ** July 27, 2007 ** Articles should be written according to the guidelines available from http://www.haskell.org/haskellwiki/The_Monad.Reader Please submit your article in PDF, together with any source files you used. The sources will be released together with the magazine under a BSD license. If you would like to submit an article, but have trouble with LaTeX please let me know and we'll sort something out. Looking forward to your submission, Wouter This message has been checked for viruses but the contents of an attachment may still contain software viruses, which could damage your computer system: you are advised to perform your own checks. Email communications with the University of Nottingham may be monitored as permitted by UK legislation.

1 0

Kleene star in types
by Matthew Sackman 07 May '07

07 May '07

Apologies for sending this twice. Mutt obviously outsmarted mailman and put a useful mime type header in. Not so this one! This email is a .lhs. > {-# OPTIONS -fglasgow-exts #-} > {-# OPTIONS -fallow-undecidable-instances #-} > {-# OPTIONS -fallow-overlapping-instances #-} Let's implement the Kleene * in the type system. First some phantom types: > data At n > data Bt n > data Ct n > data Endt Now a chain to hold it together: > data Chain :: * -> * where > Ac :: (Chain n) -> Chain (At n) > Bc :: (Chain n) -> Chain (Bt n) > Cc :: (Chain n) -> Chain (Ct n) > Endc :: Chain Endt > > instance Show (Chain n) where > show (Ac n) = "Ac ~> " ++ (show n) > show (Bc n) = "Bc ~> " ++ (show n) > show (Cc n) = "Cc ~> " ++ (show n) > show (Endc) = "Endc" Now, first things first, implement identity. I don't really care about the actual functions right now, but let's put it in anyway. > class IdOp a b | a -> b, b -> a where > idOp :: a -> b > instance IdOp (Chain Endt) (Chain Endt) where > idOp x = x > instance (IdOp (Chain l) (Chain r)) => > IdOp (Chain (At l)) (Chain (At r)) where > idOp (Ac n) = Ac (idOp n) > instance (IdOp (Chain l) (Chain r)) => > IdOp (Chain (Bt l)) (Chain (Bt r)) where > idOp (Bc n) = Bc (idOp n) > instance (IdOp (Chain l) (Chain r)) => > IdOp (Chain (Ct l)) (Chain (Ct r)) where > idOp (Cc n) = Cc (idOp n) Now, I'm sure many of you are wondering why I didn't just write: instance IdOp a a where idOp = id I'll get back to that. Now, if I go zero steps, how hard is it to go zero or one steps? > class ZeroOneStep a b where > step :: a -> b > instance ZeroOneStep (Chain (At l)) (Chain l) where > step (Ac n) = n > instance ZeroOneStep (Chain (Bt l)) (Chain l) where > step (Bc n) = n > instance ZeroOneStep (Chain (Ct l)) (Chain l) where > step (Cc n) = n > instance (IdOp (Chain l) (Chain r)) => > ZeroOneStep (Chain l) (Chain r) where > step = idOp That's quite nifty. I now need the allow-overlapping-instances, but it works fine: > fooZeroOne :: (ZeroOneStep a b) => a -> b -> Bool > fooZeroOne _ _ = True $ Main> fooZeroOne (Ac $ Bc Endc) (Ac $ Bc Endc) $ True $ Main> fooZeroOne (Ac $ Bc Endc) (Bc Endc) $ True $ Main> fooZeroOne (Ac $ Bc Endc) (Endc) $ ...big exposion... $ Main> fooZeroOne (Endc) (Bc Endc) $ ...big exposion... $ Main> fooZeroOne (Endc) (Endc) $ True So now I want zero or more steps, i.e. the Kleene *: > class ZeroMoreSteps a b where > stepN :: a -> b > instance (ZeroMoreSteps (Chain l) (Chain r)) => > ZeroMoreSteps (Chain (At l)) (Chain r) where > stepN (Ac n) = stepN n > instance (ZeroMoreSteps (Chain l) (Chain r)) => > ZeroMoreSteps (Chain (Bt l)) (Chain r) where > stepN (Bc n) = stepN n > instance (ZeroMoreSteps (Chain l) (Chain r)) => > ZeroMoreSteps (Chain (Ct l)) (Chain r) where > stepN (Cc n) = stepN n I.e. if we can already get from l to r then it's not very hard to go one step further. Now I need to bring in single step case: > instance (ZeroOneStep (Chain l) (Chain r)) => > ZeroMoreSteps (Chain l) (Chain r) where > stepN = step So now it should be done right? Because ZeroOneStep allows for zero steps (via IdOp) so that should be it right? > fooZeroMore :: (ZeroMoreSteps a b) => a -> b -> Bool > fooZeroMore _ _ = True $ Main > fooZeroMore Endc Endc $ True $ Main > fooZeroMore (Ac Endc) Endc $ True $ Main > fooZeroMore (Ac Endc) (Ac Endc) $ ...big exposion... Now what I think is happening is this: the rules for overlapping instances say that the instance which is chosen is the one that specifies the most information matching the particular context. In this case, I think it's choosing and trying to use the "wrong" instances because of this rule. Even if I try: > instance ZeroMoreSteps (Chain l) (Chain l) where > stepN = id It still blows up. So I have to do it really manually: > instance ZeroMoreSteps (Chain Endt) (Chain Endt) where > stepN = id > instance ZeroMoreSteps (Chain (At l)) (Chain (At l)) where > stepN = id > instance ZeroMoreSteps (Chain (Bt l)) (Chain (Bt l)) where > stepN = id > instance ZeroMoreSteps (Chain (Ct l)) (Chain (Ct l)) where > stepN = id $ Main > fooZeroMore (Ac Endc) (Ac Endc) $ True Now it works. So am I right - is the reason for this the selection of the instance to use given overlapping instances? If so, can anything be done to make it "better"? I can think of two ideas, which I suspect should both be laughed out of town asap: i) backtrack - if you can't make the instance work then go back to where you last had a choice and try the next instance; ii) whilst selecting the instance which is most specific seems like the "sensible" choice, there are cases, eg guards, where the vertical order in the program determines the ordering of things to try. Maybe allow the programmer to specify an order to try things in? Note however how cool this now is: > data Cell :: * -> * where > Cell :: x -> Cell a -- muh ha Ha HAA! > data Section :: * -> * -> * where > SectionTo :: (ZeroMoreSteps a b) => > (Cell a) -> b -> Section a b > SectionFrom :: (ZeroMoreSteps a b) => > a -> (Cell b) -> Section a b > fooCoolness :: (ZeroMoreSteps a b) => > a -> Section a b -> Section a b -> b > fooCoolness chain _ _ = stepN chain $ Main > fooCoolness (Bc . Ac $ Endc) (SectionTo (Cell True) (Ac $ Endc)) (SectionFrom (Bc . Ac $ Endc) (Cell 1)) $ Ac ~> Endc And stepN just goes the "right" number of steps. The point is that I don't need to use pattern matching to decompose the various constructorsof Section in order to find (and perhaps fail) the unused section of the chain. Cheers, Matthew

1 0

Kleene star in types
by Matthew Sackman 07 May '07

07 May '07

This email is a .lhs. > {-# OPTIONS -fglasgow-exts #-} > {-# OPTIONS -fallow-undecidable-instances #-} > {-# OPTIONS -fallow-overlapping-instances #-} Let's implement the Kleene * in the type system. First some phantom types: > data At n > data Bt n > data Ct n > data Endt Now a chain to hold it together: > data Chain :: * -> * where > Ac :: (Chain n) -> Chain (At n) > Bc :: (Chain n) -> Chain (Bt n) > Cc :: (Chain n) -> Chain (Ct n) > Endc :: Chain Endt > > instance Show (Chain n) where > show (Ac n) = "Ac ~> " ++ (show n) > show (Bc n) = "Bc ~> " ++ (show n) > show (Cc n) = "Cc ~> " ++ (show n) > show (Endc) = "Endc" Now, first things first, implement identity. I don't really care about the actual functions right now, but let's put it in anyway. > class IdOp a b | a -> b, b -> a where > idOp :: a -> b > instance IdOp (Chain Endt) (Chain Endt) where > idOp x = x > instance (IdOp (Chain l) (Chain r)) => > IdOp (Chain (At l)) (Chain (At r)) where > idOp (Ac n) = Ac (idOp n) > instance (IdOp (Chain l) (Chain r)) => > IdOp (Chain (Bt l)) (Chain (Bt r)) where > idOp (Bc n) = Bc (idOp n) > instance (IdOp (Chain l) (Chain r)) => > IdOp (Chain (Ct l)) (Chain (Ct r)) where > idOp (Cc n) = Cc (idOp n) Now, I'm sure many of you are wondering why I didn't just write: instance IdOp a a where idOp = id I'll get back to that. Now, if I go zero steps, how hard is it to go zero or one steps? > class ZeroOneStep a b where > step :: a -> b > instance ZeroOneStep (Chain (At l)) (Chain l) where > step (Ac n) = n > instance ZeroOneStep (Chain (Bt l)) (Chain l) where > step (Bc n) = n > instance ZeroOneStep (Chain (Ct l)) (Chain l) where > step (Cc n) = n > instance (IdOp (Chain l) (Chain r)) => > ZeroOneStep (Chain l) (Chain r) where > step = idOp That's quite nifty. I now need the allow-overlapping-instances, but it works fine: > fooZeroOne :: (ZeroOneStep a b) => a -> b -> Bool > fooZeroOne _ _ = True $ Main> fooZeroOne (Ac $ Bc Endc) (Ac $ Bc Endc) $ True $ Main> fooZeroOne (Ac $ Bc Endc) (Bc Endc) $ True $ Main> fooZeroOne (Ac $ Bc Endc) (Endc) $ ...big exposion... $ Main> fooZeroOne (Endc) (Bc Endc) $ ...big exposion... $ Main> fooZeroOne (Endc) (Endc) $ True So now I want zero or more steps, i.e. the Kleene *: > class ZeroMoreSteps a b where > stepN :: a -> b > instance (ZeroMoreSteps (Chain l) (Chain r)) => > ZeroMoreSteps (Chain (At l)) (Chain r) where > stepN (Ac n) = stepN n > instance (ZeroMoreSteps (Chain l) (Chain r)) => > ZeroMoreSteps (Chain (Bt l)) (Chain r) where > stepN (Bc n) = stepN n > instance (ZeroMoreSteps (Chain l) (Chain r)) => > ZeroMoreSteps (Chain (Ct l)) (Chain r) where > stepN (Cc n) = stepN n I.e. if we can already get from l to r then it's not very hard to go one step further. Now I need to bring in single step case: > instance (ZeroOneStep (Chain l) (Chain r)) => > ZeroMoreSteps (Chain l) (Chain r) where > stepN = step So now it should be done right? Because ZeroOneStep allows for zero steps (via IdOp) so that should be it right? > fooZeroMore :: (ZeroMoreSteps a b) => a -> b -> Bool > fooZeroMore _ _ = True $ Main > fooZeroMore Endc Endc $ True $ Main > fooZeroMore (Ac Endc) Endc $ True $ Main > fooZeroMore (Ac Endc) (Ac Endc) $ ...big exposion... Now what I think is happening is this: the rules for overlapping instances say that the instance which is chosen is the one that specifies the most information matching the particular context. In this case, I think it's choosing and trying to use the "wrong" instances because of this rule. Even if I try: > instance ZeroMoreSteps (Chain l) (Chain l) where > stepN = id It still blows up. So I have to do it really manually: > instance ZeroMoreSteps (Chain Endt) (Chain Endt) where > stepN = id > instance ZeroMoreSteps (Chain (At l)) (Chain (At l)) where > stepN = id > instance ZeroMoreSteps (Chain (Bt l)) (Chain (Bt l)) where > stepN = id > instance ZeroMoreSteps (Chain (Ct l)) (Chain (Ct l)) where > stepN = id $ Main > fooZeroMore (Ac Endc) (Ac Endc) $ True Now it works. So am I right - is the reason for this the selection of the instance to use given overlapping instances? If so, can anything be done to make it "better"? I can think of two ideas, which I suspect should both be laughed out of town asap: i) backtrack - if you can't make the instance work then go back to where you last had a choice and try the next instance; ii) whilst selecting the instance which is most specific seems like the "sensible" choice, there are cases, eg guards, where the vertical order in the program determines the ordering of things to try. Maybe allow the programmer to specify an order to try things in? Note however how cool this now is: > data Cell :: * -> * where > Cell :: x -> Cell a -- muh ha Ha HAA! > data Section :: * -> * -> * where > SectionTo :: (ZeroMoreSteps a b) => > (Cell a) -> b -> Section a b > SectionFrom :: (ZeroMoreSteps a b) => > a -> (Cell b) -> Section a b > fooCoolness :: (ZeroMoreSteps a b) => > a -> Section a b -> Section a b -> b > fooCoolness chain _ _ = stepN chain $ Main > fooCoolness (Bc . Ac $ Endc) (SectionTo (Cell True) (Ac $ Endc)) (SectionFrom (Bc . Ac $ Endc) (Cell 1)) $ Ac ~> Endc And stepN just goes the "right" number of steps. The point is that I don't need to use pattern matching to decompose the various constructorsof Section in order to find (and perhaps fail) the unused section of the chain. Cheers, Matthew

1 0

RE: [Haskell] Monad and Monad transformer???
by Simon Peyton-Jones 07 May '07

07 May '07

As someone mentioned, don't worry about being new to Haskell, but DO send your messages to haskell-cafe(a)haskell.org, not the main Haskell list. I'm redirecting this so that the Helpful Denizens of the Haskell Cafe can help you. Simon | -----Original Message----- | From: haskell-bounces(a)haskell.org [mailto:haskell-bounces@haskell.org] On Behalf Of Duong Thai Ha | Sent: 06 May 2007 17:03 | To: haskell(a)haskell.org | Subject: [Haskell] Monad and Monad transformer??? | | hi | I am a newbee to haskell so please forgive me if the question is not | appropreate. | | I am trying to get the idea of using Monad and have to understand the | following code. | | Can you please tell me is how different Monad and Monad transformer is? | And what is mean by | | > newtype State s a = State { runState :: s -> (a, s) } | | --------------------------------- | | Define (State s) as a State monad | | > instance Monad (State s) where | > return a = State $ \s -> (a,s) | > m >>= k = State $ \s -> let (a,s') = runState m s in runState (k | a) s' | | Thanks lot for your help | I really do appreciate that | | Best regard | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/haskell

1 0