October 2009 - Haskell-Cafe

Designing a DSL?
by Günther Schmidt 05 Oct '09

05 Oct '09

Hi, there are numerous examples on how to implement a DSL, but I haven't been able to figure out how to "design" one. I mean I have a pretty good idea of the problem domain, I've coded it over and over again until I got it right. Now I'd like to express that part as a DSL instead of "hard coding" it as before so I can be more prepared if I need to change or amend it. Ideally I'd only have to change the interpreter, right? And that I find to be the really tricky part, how do I *design* a DSL? Günther

4 4

System.Random.Mersenne - no randomR variants
by Paul Moore 05 Oct '09

05 Oct '09

I notice that System.Random.Mersenne has no randomR variants. Is there a reason for this, or is it just an oversight? It makes porting code from System.Random a little more complicated than it needs to be. I know it's possible to write a function mapping a Double (or other value, as appropriate) to a range, effectively replacing randomR, but it's fiddly to get the edge cases right. A standard implementation would be helpful. Paul

2 2

suggestions for doing date/time values with Data.Generics ?
by Thomas Hartman 05 Oct '09

05 Oct '09

I want to patch the time package, which I darcs got head on. Goal is to be able to do newtype deriving for UTCTime which I reckon is the base time value, for use in happstack applications state. basically, need instances for UTCTime, for Data and Typeable. ********** on #haskell: <patch-tag> I just darcs got time package and attempted to install. cabal install didn't work because missing configure fle. there's a .ac file there, so I tried first autoconf, then cabal install. configure file got created, but failed with config.status: error: cannot find input file: include/HsTimeConfig.h.in. now what? <patch-tag> my goal is to patch time package so can do data generics deriving. <Berengal> patch-tag, use System.Time instead? * Berengal shrugs and sighs [12:20] <patch-tag> could try that... *** o-regalia (n=o-regali(a)host-0-130.mimpvbg.clients.pavlovmedia.com) has quit: "Lost terminal" <Berengal> It already has a Serializeable instance defined in one of the happstack modules, so there's probably a Typeable and Data instance somewhere as well <Berengal> patch-tag, unfortunately, UTCTime has a component of a hidden datatype which isn't an instance of Typeable or Data, and since it's not exported it's impossible to create the instance as well <Berengal> If it were exported, it'd be pretty easy <patch-tag> hw can I use UTCTime, deriving Typeable and Data? (want to use UTCTime in happstack). <patch-tag> http://moonpatio.com/fastcgi/hpaste.fcgi/view?id=4278#a4278 <patch-tag> I guess I should have said I'd like to use newtype deriving rather than define instance myself if that's possible. <Berengal> patch-tag, that's possible, but could break other things. The correct solution is to patch the existing package, but that takes time <Berengal> patch-tag, Day, or something. Can't remember off the top of my head *** paolino (n=paolino(a)87.7.161.162) has quit: "Leaving." <ski> Philonous : if you can't implement both operations (satisfying laws), though, it is probably better to make a new class [12:09] <patch-tag> Berengal: a package is hidden if when you do :info in ghci, it comes out as fully qualified, right? <patch-tag> pity also that newtype deriving doesn't give you a nicer error message like "deriving choked at type blah" <Berengal> patch-tag, it comes fully qualified if it's not imported, so you'll have to try to import it to see if it's really hidden [12:10] ********* Seems like the right thing to do is patch the time package anyway. File Edit Options Buffers Tools Haskell Help {-# LANGUAGE DeriveDataTypeable #-} module TypeableTime where import Data.Time import Data.Data import Data.Generics import Data.Typeable -- doesn't work --data MyUTCTime = MyUTCTime UTCTime -- deriving (Typeable,Data) -- should I work up to UTCTime by doing this...? data MyDay = MyDay {toMyDayModifiedJulianDay :: Integer} deriving (Typeable,Data) newtype MyString = MyString String deriving (Read,Show,Ord,Eq,Typeable,Data) -- should I just modify time package? is this easier?

1 1

Re: [Haskell-cafe] Num instances for 2-dimensional types
by Brad Larsen 05 Oct '09

05 Oct '09

On Mon, Oct 5, 2009 at 10:36 AM, Miguel Mitrofanov <miguelimo38(a)yandex.ru> wrote: [...] > Of course, it's OK to call anything "numbers" provided that you stated > explicitly what exactly you would mean by that. But then you have to drop > all kind of stuff mathematicians developed for the usual notion of numbers. > In the same way, you shouldn't use the "Num" class for your "numbers". > > On the other hand, people can (ab)use the "Num" class as they wish, and it's > their business until they ask a question about it somewhere outside - which > makes the business not only theirs. [...] The Num class has `negate' as part of its definition. Natural numbers are numbers, but I don't believe there is any sensible definition of `negate' for them. Haskell 98's numeric hierarchy combines many operations which should be separate. As further evidence, every bit of Haskell I have seen that does symbolic manipulation of numeric expressions either leaves out instances that would make the syntax more convenient, or else defines partial instances because certain class functions have no sensible definition for symbolic expressions. Sincerely, Brad

3 2

apply function arguments in a list
by Michael Mossey 05 Oct '09

05 Oct '09

If I have a list containing the arguments I want to give to a function, is there a general way to supply those arguments in a compact syntax? In other words, I could have args = [1,2,3] f x y z = ... I would write t = f (args!!0) (args!!1) (args!!2) but there may be a neater, more general syntax. I tried writing a fold with $, but I don't think that works because the type of each step is different. I.e. f :: a -> a -> a -> a f x :: a -> a -> a f x y :: a -> a f x y z :: a This would seem to preclude a general way of writing a function that supplies arguments from a list.

3 2

ICFP videos now available
by Wouter Swierstra 05 Oct '09

05 Oct '09

I am happy to announce that videos of all talks at ICFP and some of the associated workshops this year have made available online: http://www.vimeo.com/user2191865/albums I'm sure you'll join me in thanking Malcolm Wallace for the time and effort he put into making this possible. Thank you Malcolm! Wouter

1 0

Is c2hs able to generate FunPtr rather than just the function?
by Rafal Kolanski 05 Oct '09

05 Oct '09

Hi Everyone, I'm trying to figure out an elegant way of letting the Haskell garbage collector deal with structures allocated from C. So I have the C pair: fluid_event_t* new_fluid_event(void); void delete_fluid_event(fluid_event_t* evt); I've handled the new_fluid_event thing nicely, ForeignPtr and all. Now if I want to add a finalizer to the ForeignPtr via newForeignPtr, and for that I need a FunPtr. So what I need is: foreign import ccall safe "fluidsynth.h &delete_fluid_event" deleteFluidEvent'' :: FunPtr (((Ptr (Event)) -> (IO ()))) (if I use that as a finalizer, everything works nicely) But the closest I can get to generating something similar with c2hs is: {# fun delete_fluid_event as ^ {fromEvent `Event'} -> `()' #} which results in: foreign import ccall safe "Event.chs.h delete_fluid_event" deleteFluidEvent'_ :: ((Ptr (Event)) -> (IO ())) Is there a neat way to have c2hs generate the FunPtr version for me? Sincerely, Rafal Kolanski.

2 1

ANN: TxtSushi 0.4.0
by Keith Sheppard 04 Oct '09

04 Oct '09

Hello Cafe, I've just uploaded TxtSushi 0.4.0 to hackage. TxtSushi is a collection of command line utilities for processing comma-separated and tab-delimited files. I posted details on my blog (along with an advertisement to see if others are interested in hacking TxtSushi): http://blog.keithsheppard.name/2009/10/txtsushi-040.html Best Regards Keith -- keithsheppard.name

1 0

Fwd: [Haskell-cafe] killer app sought
by Alberto G. Corona 04 Oct '09

04 Oct '09

Yes, Maybe The piece of the web that desperately need a boost in performance, declarativeness, safety, static typing threading, modularity etc etc etc is the Web Browser. 2009/10/4 John A. De Goes <john(a)n-brain.net> > With few exceptions, no such thing as a killer server-side app. > > The Web 3.0 paradigm is simple: all work except sharing and persistence of > data is done on the client. > > Regards, > > John A. De Goes > N-Brain, Inc. > The Evolution of Collaboration > > http://www.n-brain.net | 877-376-2724 x 101 > > > On Oct 3, 2009, at 9:08 PM, Mark Wotton wrote: > > Hi, >> >> I've been writing a little binding from Ruby to Haskell called Hubris ( >> http://github.com/mwotton/Hubris) which I think has some potential both >> for making Haskell web apps easier to write, and also for bringing the more >> adventurous Ruby programmers into the Haskell community. Code-wise it's >> coming along nicely, and once 6.12 is out it'll run without modifications at >> least on Linux (remains to be seen how long it'll take to get the Mac >> patches into shape). My real problem is marketing: I need a killer app that >> shows it's easy either to >> >> 1. wrap a kickarse Haskell library in a convenient Ruby web app shell >> 2. speed up a poorly performing Ruby web app >> >> I've been badgering the Ruby guys in Sydney that I know on the second >> point, but either none of them have performance problems, or none of them >> want to admit it. The first is entirely possible - if you only attack the >> subset of problems where your runtime is dominated by the database and >> network latency, language performance is moot. Conversely, if that's your >> worldview, the other problems that could be attacked won't ever come to mind >> (to monstrously abuse the Sapir-Whorf hypothesis). >> >> So, I'm asking you guys. What are some really nice Haskell libraries or >> apps that could benefit from being shown off in one of the plethora of >> slick, mature web frameworks that exist in Ruby? Manuel Chakravarty >> suggested something with vector operations in order to take advantage of his >> 'accelerate' library (once it gets a GPU backend, of course), and more >> generally, something taking advantage of Haskell's support for multicore >> would be cool. (The standard edition of Ruby is still unicore, I believe.) >> >> Parenthetically yours, >> Mark >> _______________________________________________ >> Haskell-Cafe mailing list >> Haskell-Cafe(a)haskell.org >> http://www.haskell.org/mailman/listinfo/haskell-cafe >> > > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe >

2 1

killer app sought
by Mark Wotton 04 Oct '09

04 Oct '09

Hi, I've been writing a little binding from Ruby to Haskell called Hubris (http://github.com/mwotton/Hubris ) which I think has some potential both for making Haskell web apps easier to write, and also for bringing the more adventurous Ruby programmers into the Haskell community. Code-wise it's coming along nicely, and once 6.12 is out it'll run without modifications at least on Linux (remains to be seen how long it'll take to get the Mac patches into shape). My real problem is marketing: I need a killer app that shows it's easy either to 1. wrap a kickarse Haskell library in a convenient Ruby web app shell 2. speed up a poorly performing Ruby web app I've been badgering the Ruby guys in Sydney that I know on the second point, but either none of them have performance problems, or none of them want to admit it. The first is entirely possible - if you only attack the subset of problems where your runtime is dominated by the database and network latency, language performance is moot. Conversely, if that's your worldview, the other problems that could be attacked won't ever come to mind (to monstrously abuse the Sapir-Whorf hypothesis). So, I'm asking you guys. What are some really nice Haskell libraries or apps that could benefit from being shown off in one of the plethora of slick, mature web frameworks that exist in Ruby? Manuel Chakravarty suggested something with vector operations in order to take advantage of his 'accelerate' library (once it gets a GPU backend, of course), and more generally, something taking advantage of Haskell's support for multicore would be cool. (The standard edition of Ruby is still unicore, I believe.) Parenthetically yours, Mark

4 4