December 2011 - Haskell-Cafe

Parallel Matrix Multiplication
by mukesh tiwari 16 Dec '11

16 Dec '11

Hello all I am trying to learn parallel Haskell and I have gone through couple of resources ( Real world Haskell and http://research.microsoft.com/en-us/um/people/simonpj/papers/parallel/AFP08… ). I understood par and pseq and I wrote matrix multiplication using these two function but it does not look. Although I did not get all these details but none of the sparks converted which is really important for parallel programming. Could some one please tell me how to improve this. Also kindly recommend me some literature for parallel programming in Haskell. Regards Mukesh Tiwari import Data.List import Control.Parallel parHelp :: ( Num a ) => [ a ] -> [ a ] -> a parHelp [] [] = 0 parHelp ( x : xs ) ( y : ys ) = ret where ret = par a ( pseq a ( a + parHelp xs ys ) ) where a = x * y helpMult :: ( Num a ) => [ a ] -> [ [ a ] ] -> [ a ] helpMult _ [] = [] helpMult x ( y : ys ) = ret where ret = par a ( pseq a ( a : helpMult x ys ) ) where a = parHelp x y mult :: ( Num a ) => [ [ a ] ] -> [ [ a ] ] -> [ [ a ] ] mult [] _ = [] mult ( x : xs ) ys = ret where ret = par a ( pseq a ( a : mult xs ys ) ) where a = helpMult x ys main = print $ mult [[1 .. 4 ] , [ 1 .. 4 ] , [ 1 .. 4 ] , [ 1 .. 4] ] ( transpose [[1 .. 4 ] , [ 1 .. 4 ] , [ 1 .. 4 ] , [ 1 .. 4] ]) [user@haskell Programming]$ ghc -O2 -threaded -rtsopts Matpar.hs [1 of 1] Compiling Main ( Matpar.hs, Matpar.o ) Linking Matpar ... [user@haskell Programming]$ ./Matpar +RTS -N2 -s ./Matpar +RTS -N2 -s [[10,20,30,40],[10,20,30,40],[10,20,30,40],[10,20,30,40]] 85,480 bytes allocated in the heap 5,216 bytes copied during GC 47,328 bytes maximum residency (1 sample(s)) 22,304 bytes maximum slop 2 MB total memory in use (0 MB lost due to fragmentation) Generation 0: 0 collections, 0 parallel, 0.00s, 0.00s elapsed Generation 1: 1 collections, 0 parallel, 0.00s, 0.00s elapsed Parallel GC work balance: -nan (0 / 0, ideal 2) MUT time (elapsed) GC time (elapsed) Task 0 (worker) : 0.00s ( 0.00s) 0.00s ( 0.00s) Task 1 (worker) : 0.00s ( 0.00s) 0.00s ( 0.00s) Task 2 (bound) : 0.00s ( 0.00s) 0.00s ( 0.00s) Task 3 (worker) : 0.00s ( 0.00s) 0.00s ( 0.00s) * SPARKS: 84 (0 converted, 0 pruned)* INIT time 0.00s ( 0.00s elapsed) MUT time 0.00s ( 0.00s elapsed) GC time 0.00s ( 0.00s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 0.00s ( 0.00s elapsed) %GC time 33.3% (13.5% elapsed) Alloc rate 42,761,380 bytes per MUT second Productivity 33.3% of total user, 45.0% of total elapsed gc_alloc_block_sync: 0 whitehole_spin: 0 gen[0].sync_large_objects: 0 gen[1].sync_large_objects: 0 [user@haskell Programming]$

2 2

smt solver bindings
by Dimitrios Vytiniotis 16 Dec '11

16 Dec '11

I've a quick question: Are there Haskell wrappers for the Z3 C API around? Thanks! d-

4 3

[Alternative] some/many narrative
by Gregory Crosswhite 15 Dec '11

15 Dec '11

Okay, so how about the following as a user narrative for some and many? many v = repeatedly execute the action v and save each obtained result until v fails; at that point, *succeed* and return a list with all of the results that had been collected some v = like many v, but if v does not succeed even once then *fail* instead of return the empty list Note: If v *never* fails, then at best many/some will return infinite lists, and at worst they will repeat your action forever. Cheers, Greg P.S.: And yes, I know that many of you will complain that these definitions are not mathematically precise or what not, but it is important to be able to build up narratives at a sufficiently high level that they explain the essentials of what is going on for the benefit of users who are unfamiliar with the precise mathematical language used to precisely define their semantics.

5 7

How to get a file path to the program invoked?
by dokondr 15 Dec '11

15 Dec '11

Hi, When my program starts it needs to know a complete path to the directory from which it was invoked. In terms of standard shell (sh) I need the Haskell function that will do equivalent to: #!/bin/sh path=$(dirname $0) How to get this path in Haskell? getProgName :: IO String defined System.Environment only returns a file name of the program without its full path. Thanks!

12 21

indentation blues
by Martin DeMello 15 Dec '11

15 Dec '11

The vim autoindent for haskell is really bad :( Is there a better indent.hs file floating around somewhere? Alternatively, is the emacs haskell mode better enough that it's worth my time learning my way around emacs and evil? martin

8 11

ANNOUNCE: keera-network-sms-diamondcard
by Ivan Perez 14 Dec '11

14 Dec '11

My fellow haskellers, I wrote a very simple library to send SMSs using the DiamondCard VoIP provider. At the present time, it uses the HTTP interface to contact the provider (there's a SOAP interface, but I haven't been able to make it play along with shoap). The current API is quite straightforward: sendSMS :: String -> String -> String -> String -> [ String ] -> IO SMSDeliveryResult sendSMS accountID pinCode message fromNumber toNumbers I don't like submitting alpha-stage software to hackage, so it's only in github for now: https://github.com/ivanperez-keera/keera-network-sms-diamondcard In a few days I plan to release a small GTK frontend that I've been using for a few months. If you give either of them a try, I'd be glad to know where it got you. Cheers, Ivan Perez

1 0

A Mascot
by heathmatlock 14 Dec '11

14 Dec '11

I liked Go's mascot, and I figure it couldn't hurt to have our own. I spent the past hour making this: http://i.imgur.com/Mib6Q.png What do you think? -- Heath Matlock +1 256 274 4225

46 82

Off The Beaten Track: Call for Participation
by David Walker 13 Dec '11

13 Dec '11

Off The Beaten Track: Underrepresented Problems for Programming Language Researchers http://www.cs.princeton.edu/~dpw/obt/ A Workshop Co-located with POPL 2012 Philadelphia, USA January 28, 2012 Come and join us for OBT -- we have a great program filled with a diverse set of problems and ideas for programming language researchers. And at the end of the day, we will have an open mic session to discuss directions for the PL community. Our program is now up here: http://www.cs.princeton.edu/~dpw/obt/obt_program.html And you can register here: https://regmaster3.com/2012conf/POPL12/register.php A broader explanation of the workshop goals: Programming language researchers have the principles, tools, algorithms and abstractions to solve all kinds of problems, in all areas of computer science. However, identifying and evaluating new problems, particularly those that lie outside the typical core PL problems we all know and love, can be a significant challenge. Hence, the goal of this workshop is to identify and discuss problems that do not often show up in our top conferences, but where programming language researchers can make a substantial impact. The hope is that by holding such a forum and associating it directly with a top conference like POPL, we can slowly start to increase the diversity of problems that are studied by PL researchers and that by doing so we will increase the impact that our community has on the world. While many workshops associated with POPL have become more like mini-conferences themselves, this is not the goal for Off the Beaten Track. The workshop will be informal and structured to encourage discussion. It will also be centered around problems and problem areas as opposed to fully-formed solutions.

1 0

acid state
by Anatoly Yakovenko 13 Dec '11

13 Dec '11

So I am trying to understand how acid state works. The HelloWorld example has a type Message = String data Database = Database [Message] $(deriveSafeCopy 0 'base ''Database) -- Transactions are defined to run in either the 'Update' monad -- or the 'Query' monad. addMessage :: Message -> Update Database () addMessage msg = do Database messages <- get put $ Database (msg:messages) It seems to me that since the Dababase is a list of messages every update would require acid-state to rewrite the list into the file, so each update would get slower as the list gets bigger, but what I am seeing is that updates are constant time regardless of the size of the list. So how does it work?

2 2

AFRP is not (necessarily) imperative
by Ertugrul Söylemez 13 Dec '11

13 Dec '11

Hello fellows, after a few discussions on IRC and via private mail I feel obligated to point out that arrows and in particular AFRP do not force you to use an imperative style in any way. You can use a style very similar to SHE's idiom brackets. I will demonstrate this using the Netwire library. The following code has a very imperative feel to it and also looks quite ugly: myWire = proc _ -> do fps <- avgFpsInt 1000 100 -< () t <- time -< () let x = 3 + t y <- integral 0 -< t returnA -< printf "%8.2f %8.2f %8.2f" fps x y Let's improve this code. The magic lies in identifying behaviors from classic FRP. The arrow variables from the above code can be seen as the behaviors, but that's not very useful for getting rid of the imperative style. A better way to look at it is that every wire that ignores its input (i.e. has a fully polymorphic input type) is a behavior, so let's find them. First of all it is impossible to write a proper Num instance for wires. The underlying problem is the same as for writing a Num instance for functions. However, the Wire type forms a vector space, and the next release of Netwire will include the corresponding instances (see the vector-space package by Conal Elliot). With them we can write: x = constant 3 ^+^ time The x wire is our first behavior. Passing x to a generic wire is simply regular arrow composition, giving you behaviors as functions of other behaviors, hence: y = integral 0 <<< x Also fps is just a simple behavior: fps = avgFpsInt 1000 100 To get to the final output there are multiple ways. Perhaps the nicest way is to exploit the Applicative instance, giving you: myWire = liftA3 (printf "%8.2f %8.2f %8.2f") fps x y where fps = avgFpsInt 1000 100 x = constant 3 ^+^ time y = integral 0 <<< x Looks much more declarative, no? For more complicated compositions or generic wires use banana brackets or the combinator syntax in arrow notation. Examples: myWire = proc x -> (someWire -< x) ^+^ (otherWire -< 15) myWire = proc x -> (| f (someWire -< x) (otherWire -< 15) |) This closely resembles idiom bracket notation, but allows you to be explicit about inputs, takes care of creating side channels and gives you the full power of arrows, including ArrowChoice and ArrowLoop. Greets, Ertugrul -- nightmare = unsafePerformIO (getWrongWife >>= sex) http://ertes.de/

3 2