- Haskell-Cafe - Haskell.org

Problem with lazy IO
by Maciej Piechotka 19 Jul '09

19 Jul '09

Hello. I've tried to combine lazy IO and parsec. The hole process is done by network. Currently I have implemented 'short parsers' so I enter them on need. To update state I have following code: parser2nntp :: Monad m => NntpParser m a -> NntpT m a parser2nntp p = do s <- NntpT (gets $ input . connection) e <- runParserT (do v <- p i <- getInput return (v, i)) () "" s case e of Left er -> error $ show er Right (v, i') -> (NntpT (modify (pNI i')) >> return v) where pNI :: Monad m => ByteString ->NntpState m -> NntpState m pNI i s = s {connection = (connection s) {input = i}} However the 4 line (i <- getInput) blocks the execution as trace indicated. String returned have no input available so it should block on evaluation - and here I pass only a reference to it (or rather I think so). What's wrong? PS. Full code is in darsc nntp repository http://code.haskell.org/nntp/ - please note that it seems to require network compiled against parsec 3 - not 2.

1 1

Why is there no Zippable class? Would this work?
by Job Vranish 19 Jul '09

19 Jul '09

I was needing a way to zip generic data structures together today and was very annoyed to find that there is no Zippable class, or variant there of. So I made my own: class (Foldable f, Functor f) => Zippable f where fmaps :: (Foldable g) => g (a -> b) -> f a -> f b fmaps' :: [a -> b] -> f a -> f b -- to save a step on instance implementation zipWith :: (a -> b -> c) -> f a -> f b -> f c zip :: f a -> f b -> f (a, b) unzip :: f (a, b) -> (f a, f b) fmaps fs a = fmaps' (toList fs) a fmaps' fs a = fmaps fs a zipWith f a b = fmaps (fmap f a) b zip = zipWith (,) unzip a = (fmap fst a, fmap snd a) instance Zippable [] where fmaps' (fx:fs) (x:xs) = fx x : fmaps' fs xs fmaps' _ _ = [] --The fmaps function is also quite handy as a replacment for zipWith3, zipWith4, etc... --For example: x = [1, 3, 5, 7, 3] y = [6, 9, 3, 1, 4] z = [2, 4, 0, 8, 2] test = fmap (,,) x `fmaps` y `fmaps` z -- > [(1,6,2),(3,9,4),(5,3,0),(7,1,8),(3,4,2)] --you can also throw in a functor instance to remove the dependency on the Functor class, but it -- might not be worth it: instance (Zippable f) => Functor f where fmap f a = fmaps (repeat f) a Is there any good reason that there isn't something like this in the standard libraries? Or, as far as I can tell, on hackage? If not, then maybe I'll stick it on hackage. - Job Vranish

10 12

A voyage of undiscovery
by Andrew Coppin 19 Jul '09

19 Jul '09

I've been working hard this week, and I'm stumbled upon something which is probably of absolutely no surprise to anybody but me. Consider the following expression: (foo True, foo 'x') Is this expression well-typed? Astonishingly, the answer depends on where "foo" is defined. If "foo" is a local variable, then the above expression is guaranteed to be ill-typed. However, if we have (for example) foo :: x -> x as a top-level function, then the above expression becomes well-typed. I had never ever noticed this fact before. I'm still trying to bend my mind around exactly why it happens. As best as I can tell, top-level functions (and value constructors, for that matter) seem to get a "new" set of type variables each time they're used, but local variables each get a single type variable, so every mention of a local variable must be of the exact same type. Could this be what GHC's weird "forall" syntax is about? Does "forall x." mean that "x" gets replaced with a unique type variable each time? It's an interesting hypothesis... Anyway, not that anybody is likely to care, but I've just spent an entire week writing a program which can type-check simple Haskell expressions. As in, you type in an expression and give types to any free variables it involves (including value constructor functions), and it tells you the type of the expression and all its subexpressions. (Or tells you that it's ill-typed.) It turns out that this is radically less trivial than you'd imagine. (The ramblings above being just one of the issues I blundered into. Others include the subtleties of writing an expression parser, building a pretty-printer with bracketing that works correctly, and the fact that expression processing malfunctions horribly if you don't make all the variable names unique first...) Other issues were mostly related to the difficulty of constantly refactoring code because you're not quite sure what you're trying to do or how you're trying to do it. (And obscure type checker warnings about GADTs...) Well there we are. I don't suppose anybody will be overly impressed, but I'm glad to have finally got it to work. Now, if it could parse more than 10% of Haskell's syntax sugar, it might even be useful for something...

9 15

Debugging methods for haskell
by Fernan Bolando 18 Jul '09

18 Jul '09

Hi all I recently used 2 hours of work looking for a bug that was causing Program error: Prelude.!!: index too large This is not very informative. It did not give me a hint which function was causing this. In C adding a few printf would have helped me, but in haskell I was not sure how to do that. Can anybody point me to some debuggin method everyone uses. After 2 hours I did find the bug eventually. The code can be viewed here. Maybe some reformatting of the code would make finding bugs easier? http://plan9.bell-labs.com/sources/contrib/fernan/escomma/ fernan -- http://www.fernski.com

10 14

Haskell Zippers on Wikibooks: teasing! :)
by Peter Verswyvelen 18 Jul '09

18 Jul '09

After my colleague explained me about zippers and how one could derive the datatype using differential rules, I had to read about it. So I started reading http://en.wikibooks.org/wiki/Haskell/Zippers#Mechanical_Differentiation This page contains the sentence: *"For a systematic construction, we need to calculate with types. The basics of structural calculations with types are outlined in a separate chapter **Generic Programming*<http://en.wikibooks.org/w/index.php?title=Haskell/Generic_Programming&actio…> * and we will heavily rely on this material"* * * However, the generic programming link does not exist yet :-) So although I now have a rough idea about it, I don't understand the details yet, e.g. notation like [image: \mathit{Node}\,A = \mu X.\,\mathit{NodeF}_A\,X] doesn't make much sense to me yet, although I suspect I can read the mu as a lambda on types? Thanks, Peter * *

6 5

uncommon IMO problem - toilet management
by Henning Thielemann 18 Jul '09

18 Jul '09

In the last two days I was invigilator at the International Mathematic Olympics 2009 in Bremen, Germany. There we got a problem different from the official math problems. :-) Eventually I solved it using Haskell. Read the detailed description at http://hackage.haskell.org/package/toilet-0.0.1 I think the problem is simple enough to be used in education of programming. (I also think it would have been better to avoid monads and use lazy list processing.)

2 1

Haskell Weekly News: Issue 126 - July 18, 2009
by Brent Yorgey 18 Jul '09

18 Jul '09

--------------------------------------------------------------------------- Haskell Weekly News http://sequence.complete.org/hwn/20090718 Issue 126 - July 18, 2009 --------------------------------------------------------------------------- Welcome to issue 126 of HWN, a newsletter covering developments in the [1]Haskell community. [2]Hac phi is next weekend! With almost 30 people already registered, it looks like we're going to have a fantastic time hacking in Philadelphia. It's still not too late to [3]register! Announcements GHC 6.10.4. Ian Lynagh [4]announced a new patchlevel release of GHC, 6.10.4. This version has very few changes over 6.10.3, but fixes some bugs that could be critical for a few users. See the [5]release notes for details. shelltestrunner 0.6 released. Simon Michael [6]announced the first release of [7]shelltestrunner, a small tool for testing any command-line program by running it through "shell" tests defined with a simple file format. generator 0.5.1. Yair Chuchem [8]announced the release of the [9]generator package, which implements an alternative list monad transformer, a list class, and related functions. GLURaw 1.0.0.0. Sven Panne [10]announced a new [11]GLURaw package, containing full support for all GLU functionality and similar in spirit to the OpenGLRaw package: it is a 1:1 mapping of the C interface, no libraries or headers are needed at build time, and the GLU API entries are resolved dynamically at runtime. OpenGLRaw 1.0.1.0. Sven Panne [12]announced a new version of the [13]OpenGLRaw package, which adds support for a number of OpenGL extensions. ObjectName 1.0.0.0. Sven Panne [14]announced a (tiny) new package, [15]ObjectName, which contains a class corresponding to the general notion of explicitly handled identifiers for API objects, e.g. a texture object name in OpenGL or a buffer object name in OpenAL. StateVar 1.0.0.0. Sven Panne [16]announced the [17]StateVar package, which further modularizes the OpenGL/OpenAL packages. It implements state variables, which are references in the IO monad, like IORefs or parts of the OpenGL state. data-ordlist-0.0.1 and NumberSieves-0.0. Leon Smith [18]announced the release of two new packages: [19]Data.OrdList offers a convenient way for efficiently dealing with lists that you happen to know are ordered, and includes operations such as union, merge, exclusive union, intersection, and difference. [20]NumberSieves includes the Sieve of O'Neill, from "The Geniune Sieve of Eratosthenes" by Melissa O'Neill, which offers an incremental primality sieve based on priority queues. Also included are two array-based generalizations of the Sieve of Eratosthenes: one for factoring a large quantity of small numbers, and another for calculating the phi function for a large quantity of small numbers. graphviz-2999.0.0.0. Ivan Lazar Miljenovic [21]announced a new release of the [22]graphviz package for Haskell, which provides bindings to the [23]GraphViz suite of tools. The biggest and most important change in this release is that all 152 attributes utilised/supported by GraphViz are now specified and supported. uncommon IMO problem - toilet management. Henning Thielemann [24]announced a [25]Haskell package for managing toilet use at the International Mathematical Olympiad. darcs 2.3 beta 4. Petr Rockai [26]announced another darcs 2.3 beta release, which features better Windows support. If you're on Windows, you should be able to install it with 'cabal install darcs-beta' -- give it a try! Google Summer of Code Progress updates from participants in the 2008 [27]Google Summer of Code. space profiling. Gergely Patai has been working on a [28]heap profile manager. fast darcs. Petr Rockai put out another [29]another darcs 2.3 beta release, and made a [30]bunch of other progress including getting darcs up and running on win32, working on hashed-storage, and optimizing 'darcs show contents'. Discussion is closing a class this easy? Conor McBride [31]asked for feedback on some code intended to effectively create a closed type class. laziness blowup exercise. Thomas Hartman [32]challenged readers to squash a memory leak. Blog noise [33]Haskell news from the [34]blogosphere. Blog posts from people new to the Haskell community are marked with >>>, be sure to welcome them! * Gergely Patai: [35]Introducing the heap profile manager. * FP Lunch: [36]Folding Statistics. * Petr Rockai: [37]soc progress 8. * Greg Bacon: [38]Monadic takeWhile. * Petr Rockai: [39]darcs 2.3 beta 4. * David Amos: [40]Counting symmetries using transversals. * Magnus Therning: [41]XML prettifier in Haskell. * Petr Rockai: [42]soc progress 7. Quotes of the Week * Berengal: For me, understanding the basics/reasoning behind haskell's type system was just a minute meditating on the phrase "what's the square root of hello?" * bitwize: The oleg is to functional studliness as the farad is to capacitance: a hopelessly large base unit. * maartenm: euclidate: to promote a conjecture to an axiom just for the sake of simplicity * RobertGreaye: Some suggest the original English remained in Britain when the North American colonies were founded; others claim it was brought to the Americas by the British settlers, leaving a pale imitation back in Britain. The truth is much stranger: the original English was actually smuggled out of Britain to the West Indies in a wardrobe belonging to General Sir Ralph Abercromby, where it ended up on the island of Trinidad after Sir Ralph took possession of that territory in the name of the British Crown. About the Haskell Weekly News New editions are posted to [43]the Haskell mailing list as well as to [44]the Haskell Sequence and [45]Planet Haskell. [46]RSS is also available, and headlines appear on [47]haskell.org. To help create new editions of this newsletter, please see the information on [48]how to contribute. Send stories to byorgey at cis dot upenn dot edu. The darcs repository is available at darcs get [49]http://code.haskell.org/~byorgey/code/hwn/ . References 1. http://haskell.org/ 2. http://haskell.org/haskellwiki/Hac_%CF%86 3. http://haskell.org/haskellwiki/Hac_%CF%86/Register 4. http://www.haskell.org//pipermail/haskell/2009-July/021508.html 5. http://haskell.org/ghc/docs/6.10.4/html/users_guide/release-6-10-4.html 6. http://article.gmane.org/gmane.comp.lang.haskell.general/17367 7. http://hackage.haskell.org/package/shelltestrunner 8. http://article.gmane.org/gmane.comp.lang.haskell.general/17366 9. http://hackage.haskell.org/package/generator 10. http://article.gmane.org/gmane.comp.lang.haskell.general/17364 11. http://hackage.haskell.org/package/GLURaw 12. http://article.gmane.org/gmane.comp.lang.haskell.general/17363 13. http://hackage.haskell.org/package/OpenGLRaw 14. http://article.gmane.org/gmane.comp.lang.haskell.general/17362 15. http://hackage.haskell.org/package/ObjectName 16. http://article.gmane.org/gmane.comp.lang.haskell.general/17361 17. http://hackage.haskell.org/package/StateVar 18. http://article.gmane.org/gmane.comp.lang.haskell.general/17355 19. http://hackage.haskell.org/package/data-ordlist 20. http://hackage.haskell.org/package/NumberSieves 21. http://article.gmane.org/gmane.comp.lang.haskell.cafe/61494 22. http://hackage.haskell.org/package/graphviz 23. http://www.graphviz.org/ 24. http://article.gmane.org/gmane.comp.lang.haskell.cafe/61448 25. http://hackage.haskell.org/package/toilet-0.0.1 26. http://article.gmane.org/gmane.comp.lang.haskell.cafe/61302 27. http://hackage.haskell.org/trac/summer-of-code/wiki/SoC2008 28. http://just-bottom.blogspot.com/2009/07/introducing-heap-profile-manager.ht… 29. http://web.mornfall.net/blog/darcs_2.3_beta_4.html 30. http://web.mornfall.net/blog/soc_progress_8.html 31. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/61456 32. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/61283 33. http://planet.haskell.org/ 34. http://haskell.org/haskellwiki/Blog_articles 35. http://just-bottom.blogspot.com/2009/07/introducing-heap-profile-manager.ht… 36. http://sneezy.cs.nott.ac.uk/fplunch/weblog/?p=232 37. http://web.mornfall.net/blog/soc_progress_8.html 38. http://feedproxy.google.com/~r/gbacon/~3/Q-Y-KapB8iI/monadic-takewhile.html 39. http://web.mornfall.net/blog/darcs_2.3_beta_4.html 40. http://haskellformaths.blogspot.com/2009/07/counting-symmetries-using-trans… 41. http://therning.org/magnus/archives/695 42. http://web.mornfall.net/blog/soc_progress_7.html 43. http://www.haskell.org/mailman/listinfo/haskell 44. http://sequence.complete.org/ 45. http://planet.haskell.org/ 46. http://sequence.complete.org/node/feed 47. http://haskell.org/ 48. http://haskell.org/haskellwiki/HWN 49. http://code.haskell.org/~byorgey/code/hwn/

1 0

cabal version issue
by Vasili I. Galchin 18 Jul '09

18 Jul '09

Hello, vigalchin@ubuntu:~/FTP/Haskell/06052009.Swish-0.2.1$ cabal --version cabal-install version 0.5.1 using version 1.4.0.1 of the Cabal library vigalchin@ubuntu:~/FTP/Haskell/Swish-0.2.1$ cabal configure --user --prefix=$HOME Warning: swish.cabal: A package using section syntax should require "Cabal-Version: >= 1.2" or equivalent. Warning: swish.cabal: A package using section syntax should require "Cabal-Version: >= 1.2" or equivalent. What is the "Cabal-Version" in the latter? Why I am getting this warning? Thanks, Vasili

2 1

Importance of identity element in finger trees?
by Daniel Peebles 18 Jul '09

18 Jul '09

Hi all, I was looking at finger trees and the tricks for getting priority queues out of them seemed a little hackish, with a distinguished infinity element or maxBound. But it seems (although I have not yet tried it) like in many cases the monoid's identity element wouldn't be necessary (a bit like the difference between fold* and fold*1). Could a finger tree be applied to an arbitrary semigroup? Or is the identity more fundamental than it looks? Thanks, Dan

2 1

catchSTM and asynchronous exceptions
by Peter Robinson 18 Jul '09

18 Jul '09

I couldn't find any information on whether catchSTM catches asynchronous exceptions so I tried to run the following: import Control.Concurrent.STM import Control.Concurrent import Control.Exception import Prelude hiding (catch) test = do tid <- myThreadId forkIO (threadDelay 5000000 >> throwTo tid (AssertionFailed "Exception in forked thread!")) (atomically $ retry `catchSTM` stmHandler) `catch` (ioHandler tid) where stmHandler (e::SomeException) = throw $ AssertionFailed ("Caught Exc. in STM; Rethrowing exception: "++ show e) ioHandler tid (e::SomeException) = print (tid,"Caught Exception in IO: ",e) Yielding the following output: # (ThreadId 6942,"Caught Exception in IO: ",Exception in forked thread!) Apparently the exception is caught by "catch" and not by "catchSTM". So the point is that catchSTM is only meant to be used for non-async exceptions, right? Regards, Peter

1 0