- Haskell-Cafe - Haskell.org

ANNOUNCE managing cabal dependencies using nix and hack-nix
by Marc Weber 25 Nov '09

25 Nov '09

First of all: This is a release (very) early and often release announcment. So expect that there are some minor glitches [5]. Linux is only supported at this moment. [1] summary: ======== main goal: provide an environment (= list dependencies) to build a haskell package with minimal efforts. The two repositories git://github.com/MarcWeber/haskell-nix-overlay.git git://github.com/MarcWeber/hack-nix.git enable you to build dependencies required to build a specific cabal project automatically. In some way its similar to what cabal-install provides. However haskell packages are installed using the nix package manager which allows you installing arbitrary versions of the same packages at the same time on the same system. Follow their READMEs to get latest updates. what is nix? ============ See http://nixos.org. Brief: its a package manager utilizing a lazy, interpreted language called nix to describe package dependencies. Packages can be installed into user environments. Atomic updates and rollbacks are some of its key features as well as allowing the user / or system to install additional versions of a package without breaking previously installed software. [2] I won't go into details here. repo haskell-nix-overlay: ===================== This contains a sketchy implementation of a dependency solver for cabal packages using a brute force algorithm [6]. It also contains a set of patches to make hackage packages compile. hack-nix creates a copy of the hackage index file 00-index.tar called hackage/hackage-nix-db.nix which can be read by the nix language applying some optimizations. You run the solver by telling it about your target packages and their cabal flags. [3] globalFlags = { base4 = true; } packageFlags = { Cabal-1.4.4.splitBase = true; }; targetPackages = [ "happy" "HAppS-Server" "haskell-src" ]; I only pasted the most important parameters here. So what is this all for? Many existing package managers assume that there is only one true way to install packages - using the versions which got packaged. You want to switch from base-3 to base-4 ? You want to use both? Some package build with parsec-2 others require parsec-3? Out of luck(?) Maybe you end up using different users, installing all packages manually into local (user) databases. Also determining which packages must be recompiled is a tedious job. So you updated a package and you have recompiled depending packages. Then you notice that you want to fix a small bug in an app you didn't update for ages.. however the old versions of your libraries have been replaced with newer ones.. usage example ============== Let me introduce you to a small hack-nix workflow showing you how I think you should use it: # get a cabal package. vcs clone xxxx # tell hack-nix about package profiles. A profile is a name associated # with cabal flags: echo "default:-split-base" > .hack-nix-cabal-config # run hack-nix to build all dependencies required by xxxx [3] # and create a file exporting PATH and GHC_PACKAGE_PATH # you should source that in your bash or zsh session hack-nix --build-env default source hack-nix-envs/default/source-me/haskell-env ghc --make Setup.hs ./Setup configure # optionally repeat the cabal flags here [4] ./Setup build # the package bulids? fine. Now you can register it at your nix # configuration file ~/.nixpkgs/config.nix after creating a dist file # and an unmodified representation of the cabal file: hack-nix --to-nix cat >> ~/.nixpkgs/config.nix << EOF {pkgs, ... }: { hackNix.additionalPackages = [ (import $PATH_TO_CABAL_REPO/dist/$CABAL_PROJECT_NAME.nix) # repeat the line above for each package ]; } EOF >From now on nix will know about your (dev version?) of the pkg xxxx and it can be used when running hack-nix --build-env in other cabal directories. Let's compare it to cabal install --user here: base -> lib -> otherlib -> target you build a dev version of lib. When running hack-nix --build-env default in the target cabal directory otherlib will be rebuild automatically. nix can also create tag files for you. This is implemented but it's not supported by hack-nix --build-env yet. (TODO) Now you want to clean those many builds get rid of them and tidy up. Just run nix-collect-garbage to remove all packages managed by nix which are no longer referenced by any environment or their previous generations. Generations? This means that if updating an environment causes more trouble to than it solves you can temporarily switch back to the environment you were using before. How to get started? =================== See [5], read those READMEs. Tell me about your problems. Right now HAppS-Server and hack-nix can be built easily. More work has to be done to make all packages compile. This usally means replacing "base < 4" by "base < 5" or such. Another cause of failure is that C-lib dependencies such as gtk, zlib or happ & alex have to be added manually once only. issues: nix will install its own glibc which dosen't cope with all kernels. You an use kind of native builds. I don't have experiences using them. I'm using nixos so I don't care anyway. speed: for speed reasons I propase only adding latest hackage packages. However you can have different databases or a filter can be implemented to disable certain packages to speed up calculations. Last resort: reimplement it in C putting code aside nix interpreter code. random notes: you can build many packages at the same time utilizing multiple cores using the nix-env -j flag of course many processes are started to build each package. So it definitely is a lot slower than using cabal-install which doesn't spawn processes for each cabal built. Maybe cabal-install and haskell-nix-overlay can be made sharing some patches. [..] If you're interested reply or contact me on irc to make me post about updates and major changes. Thanks for listening and for any feedback in advance. Marc Weber NOTES: [1]: Of course nix runs on cygwin as well. However nothing has been tested so quite some effort has to be spent making this work on Windows as well. [2]: If you have a dependency chain glibc -> lib -> executable ` -> executable2 and lib got some security reinstalling executable will yield: glibc -> lib -> executable `-> lib' -> executable2 so you have to update executable to make it use executable2 as well. This all is done by putting each packgae into its own /nix/store/hash-name/ directory where hash is based on all build inputs. You can think about those hashes similar you do when thinking about git hashes. If a patch / package changes all hashes of depending packages / patches will change. If you want to have more details read about nix on the official homepage.G [3] hack-nix --build-env does a lot of stuff: - it reads the cabal file and writes a nix file containing the cabal contents - it creates more nix files in hack-nix-envs/nix Example: http://github.com/MarcWeber/hack-nix/tree/master/hack-nix-envs/nix/ I committed those so that you can bootstrap hack-nix. default9.nix is the nix representation of the cabal package. However the version has been changed to 9999. This is done to identify your development version of this package. Thus default.nix wants to build this version: targetPackages = [{ n = "hack-nix"; v = "99999"; }]; hack-nix itself isn't built. But it's used to gather its dependencies. - then it realises that env derivation. Right now the solver will output many debugging lines to help you find out why it failed finding a solution. You can this off by debugS = false; [4] this can be done more convenient using scion. [5] help about nix: try #nixos on irc.freenode.net or its mailinglist. help about hack-nix: contact me on irc or by mail. I will try to keep the READMEs up to date. [6] At the first glance this sounds bad because: Each package can have different flag settings defining a different sets of dependencies. Each dependency declaration be satisfied by different versions of a package. So the amount of solutions to be inspected explodes. Nix isn't the best fit to write such a solver so it's quite slow. However limiting the ways a package can be built by - only adding latest package versions from hackage - setting some global flags (such as use base-4.*) seems to make the system fast enough for everydays use. I'm sorry that I didn't manage to find a better algorithm in reasonable time. If I had to say thank you to some people this list would include dccouts (for his support on Cabal), Saizan (who helped me by talking to me), the nix(os) dev team, nominolo (for his work on scion), and many others I don't even know about.

1 0

Where is `newTVarIO` defined ?
by zaxis 25 Nov '09

25 Nov '09

I cannot hoogle it. It appears in Pugs: run' ("-d":rest) = do info <- fmap Just (io $ newTVarIO Map.empty) let ?debugInfo = info run' rest Sincerely! ----- fac n = foldr (*) 1 [1..n] -- View this message in context: http://old.nabble.com/Where-is-%60newTVarIO%60-defined---tp26504967p2650496… Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

3 3

Haskell Weekly News: Issue 140 - November 22, 2009
by jfredett＠gmail.com 24 Nov '09

24 Nov '09

--------------------------------------------------------------------------- Haskell Weekly News http://sequence.complete.org/hwn/20091122 Issue 140 - November 22, 2009 --------------------------------------------------------------------------- Welcome to issue 140 of HWN, a newsletter covering developments in the [1]Haskell community. Apologies for the somewhat late edition, I got back late from the NES/MAA conference yesterday in Springfield, MA, and was generally too exhausted from all the math competing and giving of talks that HWN could not compete with the appeal of sleeping... This week there was a new edition of the HCAR, plenty of good discussion about Iteratee's and the Type Directed Name Resolution proposal, altogether a busy week. So here it is, your Haskell Weekly News! Announcements [BostonHaskell] Next meeting: November 24th at MIT (32-G882). Ravi Nanavati [2]announced the next meeting of BAHUG. Haskell Communities and Activities Report (17th ed., November 2009). Janis Voigtlaender [3]announced the new edition of the Haskell Communities and Activities Report. Call for Participation - PEPM'10 (co-located with POPL'10). Janis Voigtlaender [4]announced a call for participation for PEPM 2010. LambdaCube engine and Bullet physics binding. Csaba Hruska [5]announced a binding to the LambdaCube and Bullet engines. ICFP '10: Second call for workshop proposals. Wouter Swierstra [6]announced a second call for workshop proposals for ICFP 2010. deepseq-1.0.0.0. Simon Marlow [7]announced version 1.0.0.0 of `deepseq` wcwidth-0.0.1. Jason Dusek [8]announced a small package which provides binding to wchar.h, which assigns a column width to unicode characters. gnome-keyring 0.1 (bindings to libgnome-keyring). John Millikin [9]announced a set of bindings to the GNOME keyring libraries. attempt. Michael Snoyman [10]announced a new release of the `attempt` package. control-monad-failure and safe-failure. Michael Snoyman [11]also announced a new version of `control-monad-failure` and `safe-failure`. Announcing the GHC Bug Sweep. Simon Marlow [12]announced the GHC bug sweep, to help weed out the GHC Trac of old bugs, and also to get warm fuzzy feelings from helping everyone's favorite compiler devs. New Industrial Haskell Group membership options. Duncan Coutts [13]announced some new membership options for the the Industrial Haskell Group (IHG) bindings-SDL 1.0.2, the domain specific language for FFI description. Mauricio Antunes [14]announced a new version of the bindings-SDL package. wxHaskell 0.12.1.2. Jeremy O'Donoghue [15]announced a release of the wxHaskell package, including new improved support for installation via cabal on any system, with only a minor caveat on Windows. TFP 2010 - Call for Papers. TFP 2010 [16]announced a call for papers for TFP 2010, the 11th symposium on Trends in Functional Programming. Reminder: Fun in the afternoon, MSR Cambridge, 26 Nov. Simon Marlow [17]announced a final reminder for the `Fun in the Afternoon` meeting, which will be at MSR Cambridge on the 26th of November (ED: Thanksgiving for us Americans, if only there were some way to combine turkey-oriented gluttony with Functional programming...). Job at the University of Technology in Cottbus. Wolfgang Jeltsch [18]announced a job opening at the University of Technology in Cottbus. Scottish Category Theory Seminar. Conor McBride [19]announced the first meeting of Scottish Category Theory Seminar, a forum for discussion of all aspects of Category Theory, be they pure or applied. (ED: I am fighting very hard to not make some sort of Braveheart Joke...) Discussion Iteratee question. Valery V. Vorotyntsev [20]asked about using iteratee's in his binary data parser code. Haskell as an alternative to Java. Philippos Apolinarius [21]wondered whether Haskell would make for a good Java alternative. Status of TypeDirectedNameResolution proposal? Levi Greenspan [22]asked about the status of the TDNR proposal. Typef*ck: Brainf*ck in the type system. Johnny Morrice [23]showed us his implementation of everyone's favorite profane programming language... in the type system. Could someone teach me why we use Data.Monoid? Magicloud Magiclouds [24]requested some insight to why we use monoids so much in Haskell, leading to a fantastic discussion of all the myriad places Monoids pop up in both Haskell and in Math in general. Blog noise [25]Haskell news from the [26]blogosphere. Blog posts from people new to the Haskell community are marked with >>>, be sure to welcome them! * Neil Brown: [27]The Operators and Monoids of CHP. * Philip Wadler: [28]A list is an odd creature, take 2. * Darcs: [29]darcs hacking sprint 3 report. * Mikael Vejdemo Johansson (Syzygy-): [30][MATH198] Lecture 9 posted and lectured. * Gergely Patai: [31]LambdaCube and Bullet on Hackage at last. * David Amos: [32]Three new modules in HaskellForMaths. * Darcs: [33]darcs weekly news #46. * Christophe Poucet (vincenz): [34]Setting up iptables to throttle incoming ssh. * Ivan Lazar Miljenovic: [35]Waddaya know, testing WORKS!. * Neil Mitchell: [36]Reviewing View Patterns. * Neil Brown: [37]An Introduction to Communicating Sequential Processes. * Joachim Breitner: [38]Darcs Hacking Sprint: Mission Complete. * Ivan Lazar Miljenovic: [39]Past, Present and PEPM. * Erik de Castro Lopo: [40]Hacking DDC.. * Dan Piponi (sigfpe): [41]Haskell Monoids and their Uses. * Joachim Breitner: [42]Arrived at the Darcs hacking sprint. Quotes of the Week * Apocalisp: You can't have your baby and eat it too * tensorpudding: so you boil lisp for an hour to sift out the parentheses and impurities, make a whitespace sauce with liberal syntactic sugar, and you have haskell a la mode * ddarius: I'm not aware of anything (including C++) that can seamlessly talk to C++ code. * ksf: is Data.Data.Data some kind of reference to swedish chefs? * IceDane: [on escaping an imperative mindset]: <kmc> i recommend heavy drinking <IceDane> I've tried that. I just have fun and wake up and feel like shit the day after. but still think in loops. * jpet: Ok, after studying the generated core a bit, I can conclude that generated core is somewhat hard to follow. * Adamant: [on the update complexity of Data.Map] I read that as 'Oleg(n)' * skorpan: I did not have impure relations with that language 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 jfredett . at . gmail . dot . com. The darcs repository is available at darcs get [49]http://patch-tag.com/r/jfredett/HWN2/pullrepo HWN2 . References 1. http://haskell.org/ 2. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66609 3. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66594 4. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66530 5. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66523 6. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66510 7. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66445 8. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66441 9. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66440 10. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66413 11. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66408 12. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66396 13. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66395 14. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66383 15. http://article.gmane.org/gmane.comp.lang.haskell.cafe/66279 16. http://thread.gmane.org/gmane.comp.lang.haskell.general/17631 17. http://article.gmane.org/gmane.comp.lang.haskell.general/17627 18. http://article.gmane.org/gmane.comp.lang.haskell.general/17625 19. http://article.gmane.org/gmane.comp.lang.haskell.general/17623 20. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/66601 21. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/66511 22. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/66431 23. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/66398 24. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/66223 25. http://planet.haskell.org/ 26. http://haskell.org/haskellwiki/Blog_articles 27. http://chplib.wordpress.com/2009/11/20/the-operators-and-monoids-of-chp/ 28. http://wadler.blogspot.com/2009/11/list-is-odd-creature-take-2.html 29. http://blog.darcs.net/2009/11/darcs-hacking-sprint-3-report.html 30. http://blog.mikael.johanssons.org/archive/2009/11/math198-lecture-9-posted-… 31. http://just-bottom.blogspot.com/2009/11/lambdacube-and-bullet-on-hackage-at… 32. http://haskellformaths.blogspot.com/2009/11/three-new-modules-in-haskellfor… 33. http://blog.darcs.net/2009/11/darcs-weekly-news-46.html 34. http://blog.poucet.org/2009/11/setting-up-iptables-to-throttle-incoming-ssh/ 35. http://ivanmiljenovic.wordpress.com/2009/11/17/waddaya-know-testing-works/ 36. http://neilmitchell.blogspot.com/2009/11/reviewing-view-patterns.html 37. http://chplib.wordpress.com/2009/11/16/an-introduction-to-communicating-seq… 38. https://www.joachim-breitner.de/blog/archives/350-Darcs-Hacking-Sprint-Miss… 39. http://ivanmiljenovic.wordpress.com/2009/11/15/past-present-and-pepm/ 40. http://www.mega-nerd.com/erikd/Blog/CodeHacking/DDC/hacking_ddc.html 41. http://blog.sigfpe.com/2009/01/haskell-monoids-and-their-uses.html 42. https://www.joachim-breitner.de/blog/archives/349-Arrived-at-the-Darcs-hack… 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://patch-tag.com/r/jfredett/HWN2/pullrepo%20HWN2

7 9

Status of TypeDirectedNameResolution proposal?
by Levi Greenspan 24 Nov '09

24 Nov '09

What's the status of the TDNR proposal [1]? Personally I think it is a very good idea and I'd like to see it in Haskell'/GHC rather sooner than later. Working around the limitations of the current record system is one of my biggest pain points in Haskell and TDNR would be a major improvement. Thus I wonder if someone is actively working on this proposal? Thanks, Levi. ---------------- [1] http://hackage.haskell.org/trac/haskell-prime/wiki/TypeDirectedNameResoluti…

21 37

Re: [Haskell] Announcing Haskell 2010
by Simon Marlow 24 Nov '09

24 Nov '09

On 24/11/2009 15:19, David Leimbach wrote: > First off congratulations everyone! > > Second, Oh shit! Graham Hutton's excellent Haskell introduction book is > now not valid Haskell 2010 due to N+K patterns? Right, but it's still valid Haskell 98, and we have no immediate plans at least in GHC to drop support for Haskell 98. It will probably not be the default in 6.14.1, though. Cheers, Simon > I loaned that book to my boss and he's really enjoyed it. I guess I'll > have to buy a revised copy. Can we get an update to it? :-) > > I realize N+K was considered dangerous > > On Tue, Nov 24, 2009 at 2:50 AM, Simon Marlow <marlowsd(a)gmail.com > <mailto:marlowsd@gmail.com>> wrote: > > I'm very proud to announce a new revision of the Haskell language, > Haskell 2010. Over the last couple of months the committee has been > making final decisions about which extensions should be a part of > this revision. The final list is: > > DoAndIfThenElse > HierarchicalModules > EmptyDataDeclarations > FixityResolution > ForeignFunctionInterface > LineCommentSyntax > PatternGuards > RelaxedDependencyAnalysis > LanguagePragma > NoNPlusKPatterns > > You can read more about each one, including rationale for and > against, on its relevant wiki page, which are linked from the tickets: > > http://hackage.haskell.org/trac/haskell-prime/query?status=new&status=assig… > <http://hackage.haskell.org/trac/haskell-prime/query?status=new&status=assig…> > > Haskell 2010 is a small but significant step on the road that was > started by the Haskell' committee 4 years ago, The process has not > been a smooth one, and there have been several changes of direction, > but the current process is actually producing concrete results that > let us move the language forward in positive steps, so I feel we're > on the right track. > > We all owe the current committee a big thank-you for sticking with > the process this long: most of them didn't realise the magnitude of > what they were signing up for at the beginning. The short list of > changes above tells only a small part of the story, there is a > wealth of wiki content and mailing-list discussion that future > language revisions can draw on. > > So what now? > > * We will produce a revised version of the Haskell language report > incorporating these changes. That will happen over the next few > months. > > * Compilers can start implementing the changes, and flags to > select the Haskell 2010 revision. In GHC we expect to have > support in the next major release, i.e. 6.14.1. > > * Right now, we will start forming a Haskell 2011 committee to > mange the process of deciding on changes for next year's revision. > The current committee is still discussing how to go about > finding a new committee (the plan is to at least have open > nominations) but I expect to be able to announce more details > very soon. > > * Everyone can participate in the Haskell 2011 process, by discussing > and refining proposals. Information about how to do that is on > the Haskell Prime wiki: > http://hackage.haskell.org/trac/haskell-prime/wiki > > Remember: this is a community effort. The changes that get > adopted in each revision are drawn from the pool of fully-specified > proposals, and those proposals can be written by anyone. > > Thanks, > > Simon, on behalf of the Haskell 2010 committee > > _______________________________________________ > Haskell mailing list > Haskell(a)haskell.org <mailto:Haskell@haskell.org> > http://www.haskell.org/mailman/listinfo/haskell > >

1 0

Re: Haskell-Cafe Digest, Vol 75, Issue 51
by Johnny Morrice 24 Nov '09

24 Nov '09

> 1) the uncensored version > 2) Monadam* > 3) Monada** > 4) Monad*** > 5) Mona**** Putting stars in place of letters in no way makes a word less offensive. Consider the word '****ing'. It's about context. I think it's wise that such a word have a star put in it in the weekly news or a journal article as it shows the author is less bombastic and doesn't want to use loud language - which counts for a lot in other's opinions. Johnny

1 0

Idea for a very simple GUI llibrary
by Maurício CA 24 Nov '09

24 Nov '09

Hi, Here is a sketch for a library with these properties: -> Easy to test. All Haskell code can be tested in a text terminal. Also, testing code that uses the library can also be done without using a GUI. -> Extremely easy to document and use. -> Not even close to Gtk2hs power, but enough for small applications. -> Could be the first GUI to build on hackage :) What we need is: -> MyState. A user suplied type for application state. -> WidId. A user suplied type for widget identifiers. -> Gui wi. A type capable of describing an interface with all of its state. It's an instance of Eq. -> Event wi. A type for events. -> Prop. A type for properties than can related to a WidId. Running an application would be like this: main = runGUI initState -- An initial MyState. event -- :: MyState -> DiffTime -> Event WidId -> MyState props -- :: WidId -> [Prop] action -- :: MyState -> DiffTime -> IO (Maybe (MyState,Gui WidId)) timeout -- :: DiffTime DiffTime parameters for callbacks are always the time elapsed since application started. From initState and event, the implementation of runGUI can save a state that optionally changes with time. From props, it can get details on what to present in widgets associated with a WidId (selected state, picture to draw etc.). action presents a chance for using IO, and optionally change state and GUI description. timeout is the maximum time runGUI implementation is allowed to wait between calls to action. Examples for those types: newtype MyState = { lastUpdate :: DiffTime, builtGui :: Bool, earthCoordinates :: (Double,Double), map :: SVG, ... } data WidId = XCoord | YCoord | MapWindow | ReloadButton ... data Gui widid = TitleWindow (Gui widid) | Tabs [(String,Gui widid)] | PressButton String widid | Selection [String] widid | ... deriving Eq {- Eq is needed by runGUI to detect if GUI has changed after the last call to action. -} data Event widid = ButtonPressed widid | FileSelected String widid | OptionSelected String widid | ... data Prop widid = Active Bool | Text String | Draw SVG | ... I believe this can represent most kinds of simple applications, and be efficient enough for practical use. It's interesting that all of this can be designed, implemented and tested independent of runGUI implementation. Actually, if you want a pet project and want to write and design the Haskell part, I may probably be able to write runGUI for you :) Best, Maurício

9 12

Some useful TH templates
by Yair Chuchem 23 Nov '09

23 Nov '09

Hi, I wrote some Template Haskell templates that I think may be of use to others. The first generates "in" and "with" functions for newtypes. For example, using it one can replace this code (from TypeCompose): > inFlip :: ((a~>b) -> (a' ~~> b')) -> (Flip (~>) b a -> Flip (~~>) b' a') > inFlip = (Flip .).(. unFlip) > > inFlip2 :: ((a~>b) -> (a' ~~> b') -> (a'' ~~~> b'')) > -> (Flip (~>) b a -> Flip (~~>) b' a' -> Flip (~~~>) b'' a'') > inFlip2 f (Flip ar) = inFlip (f ar) > > inFlip3 :: ((a~>b) -> (a' ~~> b') -> (a'' ~~~> b'') -> (a''' ~~~~> b''')) > -> (Flip (~>) b a -> Flip (~~>) b' a' -> Flip (~~~>) b'' a'' -> Flip (~~~~>) b''' a''') > inFlip3 f (Flip ar) = inFlip2 (f ar) with this code: > {-# LANGUAGE TemplateHaskell #-} > import Data.Newtype > $(mkInNewtypeFuncs [1..3] ''Flip) The second template is for accessing ADTs in the Maybe monad. For example: > {-# LANGUAGE TemplateHaskell #-} > import Data.ADT.Getters > data Blah a = NoBlah | YesBlah a | ManyBlah a Int > $(mkADTGetters ''Blah) Generates: > gNoBlah :: Blah a -> Maybe () > gYesBlah :: Blah a -> Maybe a > gManyBlah :: Blah a -> Maybe (a, Int) I'm more than willing to upload these templates to hackage (or rather, split them out of an unrelated package), but all I need is a suggestion for a package name :) For now you can find these here: http://hackage.haskell.org/package/peakachu Hoping someone else will find it useful, cheers, Yair

2 2

Wiki software?
by Günther Schmidt 23 Nov '09

23 Nov '09

Hi, I'm finally about to organize myself, somewhat. And am going to use a wiki for it. Does there a good one exist that's written in Haskell? Günther

10 12

Re: ANNOUNCE: deepseq-1.0.0.0
by Simon Marlow 23 Nov '09

23 Nov '09

On 19/11/2009 11:52, Sittampalam, Ganesh wrote: > Yitzchak Gale wrote: >> Simon Marlow wrote: >>> So then what shall we call the a -> () version? >>> One possibility is to go back to calling it rnf. >> >> In light of apfelmus' comment, I vote for rnf. >> >> And in that case, how about the analogous alternative for seq itself: >> >> hnf :: a -> () > > I think it would be whnf since it doesn't evaluate under lambdas. > > I also vote for rnf, because we should have a good reason for changing > names of things. Various people would prefer rnf. Ok, unless there are any further objections, I'll change the names back to class NFData a where rnf :: a -> () and also add deepseq :: a -> b -> b but I'll leave the module name as Control.DeepSeq. Cheers, Simon

1 1