June 2008 - Haskell-Cafe

Issues of the "mailto" link on the list archive page
by Xiao-Yong Jin 08 Jun '08

08 Jun '08

Hi, I had private conversation with Andrzej Jaworski about the fact that his reply to Alberto Ruiz's post is off thread. What he did was clicking on the "mailto" link beside the author's name on the list archive web page [1]. [1] http://www.haskell.org/pipermail/haskell-cafe/2008-June/044023.html I am not quite sure about what this link is supposed to do. But, if it is meant for people to click and reply to the thread, it is not doing this correctly, as the "Subject" and "In-Reply-To" fields in the link are obtained from these fields in the head of current email. It would be useful if the link shows "Subject" with an added "Re: " if there is not already one, and fill in the "in-Reply-To" field with the value from the "Message-Id" field of current email head. An optional "Reference" field is informative for most of the email client, too. Sorry if this topic has the least connection to the glorious Haskell. Best, Xiao-Yong -- c/* __o/* <\ * (__ */\ <

2 1

Data.Map traversal.
by Serguey Zefirov 08 Jun '08

08 Jun '08

I found that I often need predecessor and successor of some key in Data.Map.Map. Just like that: predKey, succKey :: Ord k => Data.Map.Map k a -> k -> Maybe k predKeyElem, succKeyElem :: Ord k => Data.Map.Map k a -> k -> Maybe (k,a) Data.Map has operations like that on key indexes, but it is slightly unnatural to work with three entities (index, key and element) where two suffices. So I propose to include those operations into next version of Data.Map. If anyone could point me in the right direction I could do any necessary modifications myself (just because I need it).

2 1

Mersenne Build Problem
by Dominic Steinitz 08 Jun '08

08 Jun '08

I'm getting errors (see below) trying to build the tests in http://hackage.haskell.org/cgi-bin/hackage-scripts/package/mersenne-random-… I built the package itself using ./Setup configure -f use_sse2 I thought I had an intel core duo (also see below). I think I may be missing a library but I'm not sure which one. Thanks, Dominic. > dom@lagrange:~/mersenne-random-0.1.1> cat /proc/cpuinfo > processor : 0 > vendor_id : GenuineIntel > cpu family : 6 > model : 15 > model name : Intel(R) Core(TM)2 Duo CPU E4500 @ 2.20GHz > stepping : 13 > cpu MHz : 1200.000 > cache size : 2048 KB > physical id : 0 > siblings : 2 > core id : 0 > cpu cores : 2 > fdiv_bug : no > hlt_bug : no > f00f_bug : no > coma_bug : no > fpu : yes > fpu_exception : yes > cpuid level : 10 > wp : yes > flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe lm constant_tsc pni monitor ds_cpl est tm2 ssse3 cx16 xtpr lahf_lm > bogomips : 4392.14 > clflush size : 64 > > processor : 1 > vendor_id : GenuineIntel > cpu family : 6 > model : 15 > model name : Intel(R) Core(TM)2 Duo CPU E4500 @ 2.20GHz > stepping : 13 > cpu MHz : 1200.000 > cache size : 2048 KB > physical id : 0 > siblings : 2 > core id : 1 > cpu cores : 2 > fdiv_bug : no > hlt_bug : no > f00f_bug : no > coma_bug : no > fpu : yes > fpu_exception : yes > cpuid level : 10 > wp : yes > flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe lm constant_tsc pni monitor ds_cpl est tm2 ssse3 cx16 xtpr lahf_lm > bogomips : 4388.99 > clflush size : 64 > dom@lagrange:~/mersenne-random-0.1.1/tests> make > ghc -O2 -ddump-simpl-stats -no-recomp Unit.hs --make > [1 of 1] Compiling Main ( Unit.hs, Unit.o ) > > ==================== FloatOut stats: ==================== > 154 Lets floated to top level; 35 Lets floated elsewhere; from 40 Lambda groups > > > > ==================== FloatOut stats: ==================== > 133 Lets floated to top level; 14 Lets floated elsewhere; from 34 Lambda groups > > > > ==================== Grand total simplifier statistics ==================== > Total ticks: 8378 > > 2022 PreInlineUnconditionally > 1760 PostInlineUnconditionally > 991 UnfoldingDone > 132 RuleFired > 1 *# > 15 +# > 25 ++ > 1 <# > 3 ==#->case > 2 SC:a0 > 1 SC:a_s2sW0 > 1 SPEC GHC.Num.- > 2 SPEC GHC.Real.$p1Integral > 2 SPEC GHC.Real.$p1Real > 2 SPEC GHC.Real.$p2Real > 3 SPEC Main.speed > 3 SPEC System.Random.random > 1 eftInt > 9 fold/build > 14 foldr/app > 1 fromIntegral/Word->Int > 1 int2Double# > 1 int2Word# > 3 map > 2 mapList > 1 minimumInt > 2 remInt# > 2 take > 2 takeList > 17 unpack > 6 unpack-append > 9 unpack-list > 212 LetFloatFromLet > 1 EtaReduction > 2874 BetaReduction > 27 CaseOfCase > 341 KnownBranch > 3 CaseElim > 2 CaseIdentity > 13 FillInCaseDefault > 22 SimplifierDone > > > Linking Unit ... > Unit.o: In function `s4Da_info': > (.text+0x1b21): undefined reference to `genrand_real2' > Unit.o: In function `s4RA_info': > (.text+0x3e75): undefined reference to `genrand_real2' > Unit.o: In function `s4S4_info': > (.text+0x3f61): undefined reference to `genrand_real2' > Unit.o: In function `s5su_info': > (.text+0x40bc): undefined reference to `genrand_real2' > /usr/local/lib/mersenne-random-0.1.1/ghc-6.9.20080517/libHSmersenne-random-0.1.1.a(Mersenne.o): In function `mersennezmrandomzm0zi1zi1_SystemziRandomziMersenne_zdwa2_info': > ghc13223_0.hc:(.text+0x1a3): undefined reference to `gen_rand64_mix' > /usr/local/lib/mersenne-random-0.1.1/ghc-6.9.20080517/libHSmersenne-random-0.1.1.a(Mersenne.o): In function `s2J1_info': > ghc13223_0.hc:(.text+0x91d): undefined reference to `gen_rand64_mix' > /usr/local/lib/mersenne-random-0.1.1/ghc-6.9.20080517/libHSmersenne-random-0.1.1.a(Mersenne.o): In function `s2JZ_info': > ghc13223_0.hc:(.text+0xb3d): undefined reference to `genrand_real2' > /usr/local/lib/mersenne-random-0.1.1/ghc-6.9.20080517/libHSmersenne-random-0.1.1.a(Mersenne.o): In function `s2LJ_info': > ghc13223_0.hc:(.text+0xf8d): undefined reference to `gen_rand64_mix' > /usr/local/lib/mersenne-random-0.1.1/ghc-6.9.20080517/libHSmersenne-random-0.1.1.a(Mersenne.o): In function `s35i_info': > ghc13223_0.hc:(.text+0x1397): undefined reference to `genrand_real2' > /usr/local/lib/mersenne-random-0.1.1/ghc-6.9.20080517/libHSmersenne-random-0.1.1.a(Mersenne.o): In function `s36A_info': > ghc13223_0.hc:(.text+0x1517): undefined reference to `gen_rand64_mix' > collect2: ld returned 1 exit status > make: *** [all] Error 1

3 3

Newbie Q: Overloading and type classes
by Dmitri O.Kondratiev 08 Jun '08

08 Jun '08

{-- I try to define class Store that will group types implmenting different storage mechanisms . All types should support two functions: 1) put (key, value) pair into storage 2) get value from storage corresponding to the given key As an example I start with the following storage types: --} -- simple memory cell data OneCell k v = Cell(k, v) -- list of cells data CellList k v = CLst [(k, v)] {-- First, without grouping these types into any class, just to illustrate what these types should do, I define the following functions: --} {-- put::(k, v) -> OneCell k v -> OneCell k v put (k1, v1) (Cell(k, v)) = Cell(k1, v1) get :: Eq k => k -> OneCell k v -> Maybe v get k (Cell (_, v)) = Just v c1 ::OneCell String Int c1 = Cell("one", 1) putLst :: (k,v) -> CellList k v -> CellList k v putLst (k,v) (CLst xs) = CLst ((k,v) : xs) getLst :: Eq k => k -> CellList k v -> Maybe v getLst k (CLst xs) = lookup k xs cl1 = CLst [("one",1),("two",2),("three",3)] --} {-- These work as expected. Now I try to define Store class that should allow me to overload put & get functions: --} class Store s where put :: Eq k => (k, v) -> s -> s get :: k s -> Maybe v {-- instance Store OneCell where put (k1, v1) (Cell(k, v)) = Cell(k1, v1) get k (Cell (_, v)) = Just v --} {-- I get the following error: `OneCell' is not applied to enough type arguments Expected kind `*', but `OneCell' has kind `* -> * -> *' In the instance declaration for `Store OneCell' --} instance Store CellList where put (k,v) (CLst xs) = CLst ((k,v) : xs) get k (CLst xs) = lookup k xs {-- I get the following error: `CellList' is not applied to enough type arguments Expected kind `*', but `CellList' has kind `* -> * -> *' In the instance declaration for `Store CellList' --} What should I do to create Store class? Thanks! -- Dmitri O. Kondratiev dokondr(a)gmail.com http://www.geocities.com/dkondr

2 4

[ANN] Bindings to Xen Control (xenctrl.h)
by Thomas M. DuBuisson 08 Jun '08

08 Jun '08

All, I'm just getting started with hsXenCtrl [1] as both a fun way to play with Xen and become proficient with Haskell FFI. Once I get my community.haskell.org account squared away I'll likely setup a public darcs repo (and a homepage somewhere). As for modules: I intend to expand on the trival FFI bindings in System.Xen.CBindings and there'll be a higher level interface in System.Xen (functions will standardize on returning Either or throwing exceptions, hiding the xc_handle open/close, etc). Typical Disclaimers: The _only_ thing I've actually done with this, just tonight, is pause / unpause domains. API is subject to change! Only my exact build of Xen is sure to work, and no promise even then. TomMD [1] http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hsXenCtrl

2 3

Re: Fwd: installing happy 1.17
by Duncan Coutts 07 Jun '08

07 Jun '08

On Sat, 2008-06-07 at 09:36 -0700, Thomas Hartman wrote: > cabal issue? Yes and no. What is happening is that there are several versions of the Cabal library installed. In particular Cabal-1.3.12 and Cabal-1.5.2. The happy package uses build-type Custom so cabal-install compiles the Setup.lhs script. The question is what version of the Cabal library should it use? The happy.cabal file specifies cabal-version: >= 1.2 so we'd better respect that. But even then we have several choices. Currently it just picks the highest one of the remaining choices. In this case that meant Cabal-1.5.2 -- the latest development version. This version turns out to have API changes that means that we get a type error when compiling Setup.lhs. As it happens, if cabal-install had picked one of the other choices then it would have compiled ok because the 1.2.x and upcoming 1.4.x series do not have those api changes. The immediate workarounds are: * unregister Cabal-1.5.2 * cabal install happy --cabal-lib-version 1.3.12 Perhaps a better solution is for cabal-install to choose the version of the Cabal library differently. Of course it has to be within the hard constraints specified in the .cabal file and on the command line. But after that perhaps it should prefer the version of the Cabal library that it itself was built with and at second preference the latest version in the same major series. The last preference would be as now, the latest version available. Sound sensible? Duncan > ---------- Forwarded message ---------- > From: Thomas Hartman <tphyahoo(a)gmail.com> > Date: 2008/6/7 > Subject: installing happy 1.17 > To: haskell <haskell-cafe(a)haskell.org> > > > I had a problem installing happy 1.17 (same result with happy head). > This appears to be required for installing happs-hsp-template via > cabal install. > > thanks for any advice! > > Thomas. > > ***** > > thartman@thartman-laptop:~/haskellInstalls/smallInstalls>darcs get > --partial http://darcs.haskell.org/happy/ > ********************** > Copying patch 87 of 87... done. > Applying patch 86 of 86... done. > Finished getting. > > thartman@thartman-laptop:~/haskellInstalls/smallInstalls>cd happy > > thartman@thartman-laptop:~/haskellInstalls/smallInstalls/happy>runghc > Setup.lhs configure > > Setup.lhs:30:43: Not in scope: `buildVerbose' > thartman@thartman-laptop:~/haskellInstalls/smallInstalls/happy>cabal --version > cabal-install version 0.4.9 > using version 1.3.12 of the Cabal library > thartman@thartman-laptop:~/haskellInstalls/smallInstalls/happy>ghc-pkg > list | grep -i cabal > Cabal-1.2.3.0, Cabal-1.3.11, Cabal-1.3.12, Cabal-1.5.2,

3 5

Re: [Haskell-cafe] example of FFI FunPtr
by Galchin, Vasili 07 Jun '08

07 Jun '08

Hi Claude, I want to do an incremental "experiment". I want just want to pass a C function to "callback" to a Haskell function. ??? Thanks, Vasili On Thu, Jun 5, 2008 at 4:21 AM, Claude Heiland-Allen < claudiusmaximus(a)goto10.org> wrote: > Galchin, Vasili wrote: > >> Hello, >> >> I want to model a Haskell function that is a callback from C. I have >> only found one example in the unix package's Semaphore.hsc, which >> apparently >> is not used. I want to be able to marshall a Haskell function that is a >> first class citizen residing in a Haskell data type and pass to a C >> function >> via FFI. Are there examples of this? >> > > Attached is a simple example. > > The main thing to note is 'foreign import ccall "wrapper"' which gives you > a factory for turning Haskell functions into foreign function pointers. > > More information: > > http://www.cse.unsw.edu.au/~chak/haskell/ffi/<http://www.cse.unsw.edu.au/%7Echak/haskell/ffi/> > > > Claude > -- > http://claudiusmaximus.goto10.org > > > CallBacker: CallBacker.hs callerback.c callerback.h > ghc -O2 -Wall -fffi -o CallBacker CallBacker.hs callerback.c > >

2 1

Patrick Perry's BLAS package
by Andrzej Jaworski 07 Jun '08

07 Jun '08

Hi, Indeed, this has been long awaited. Long live Patrick!!! And continue the good work:-) However, such essential work shouldn't be dependent on heroic effort of an individual. If Haskell is to remain non-commercial a disciplined community effort should be taken akin to Pythonian. Perhaps also thouse of you who teach Haskell could better use cheap labour of students, after all Haskell's module system is not all that weak;-) Cheers, -A.J.

2 2

Dynamic curiosity
by Gökhan San 07 Jun '08

07 Jun '08

Hello All, I've recently modified my existing project to support dynamic data (using 'Data.Dynamic'), and despite my earlier benchmarks on 'Dynamic', the program slowed down by a factor of 10. Profiler says that 'fromDynamic' is consuming 60% of the CPU time (with -O2). Using the below function instead of 'fromDynamic' seems to make coercion 20 times faster (thereby solving my problem): > {-# OPTIONS -fglasgow-exts #-} > import GHC.Prim > data Dynamic' = Dynamic' TypeRep Any > fromDynamic' :: forall a. Typeable a => Dynamic -> Maybe a > fromDynamic' d = > let Dynamic' t v = unsafeCoerce# d > in if t == typeOf (undefined :: a) > then Just (unsafeCoerce# v) > else Nothing I was able to reproduce the same result from scratch. The only difference with the standard implementation (I put it below for reference), other than the first ugly unsafeCoerce#, seems to be existential quantification (i.e. When I replace "if t == typeOf (undefined :: a) ..." with the "case ..." below, performance goes down again.). > fromDynamic :: Typeable a => Dynamic -> Maybe a > fromDynamic (Dynamic t v) = > case unsafeCoerce v of > r | t == typeOf r -> Just r > | otherwise -> Nothing So what is causing the improvement, or am I missing something? Thanks. -- Gokhan

1 0

Teaching Monads
by Ronald Guida 07 Jun '08

07 Jun '08

Monads in Haskell are a topic that I, like most beginners, find difficult and "mind-twisting". Now that I think I understand monads, they seem to be very simple; I've read that this is a common experience. So I wonder, what would it take to help beginners catch on with a minimum of fuss or frustration? The plethora of monad tutorials out there is clear evidence that plenty of others have wondered the same thing. What made monads "click" for me is when I understood the following things: 1. When monads are being used, closures are almost always involved. 2. These closures naturally arise when desugaring do-syntax. do x1 <- m1 m1 >>= (\x1 -> x2 <- m2 m2 >>= (\x2 -> [Eq1] x3 <- m3 m3 >>= (\x3 -> return (f x1 x2 x3) return (f x1 x2 x3)))) 3. These closures are extremely similar to the closures that arise when desugaring let-syntax. let x1 = f1 in f1 -$ (\x1 -> Where: let x2 = f2 in f2 -$ (\x2 -> (-$) :: a -> (a -> b) -> b let x3 = f3 in f3 -$ (\x3 -> x -$ f = f x f x1 x2 x3 f x1 x2 x3))) 4. While I can think of a monad as a fancy container that holds an element of type "t", it might be more accurate to think of a monad as a container that merely displays the type "t" as a label for its contents. The container might hold one object of type "t", or it might hold several. It might not be holding any at all. Perhaps the container /never/ holds an object of type "t", but instead it holds a set of instructions to produce such an object. (e.g. Reader, State). Naturally, it's hard to illustrate nested closures, higher-order functions, and objects that aren't really there. It's easy to illustrate a sequential scheme where a single "thing" passes through a series of operations, while some related data travels in parallel. m1 >>= f1 >>= f2 >>= f3 [Eq2] In any case, the extreme similarity between desugared "do" and desugared "let" leads me to think of the concepts of a manual plumbing system and a monadic plumbing system. Basically, a manual plumbing system is what I have if I'm threading information down through a series of nested function calls in a certain stereotypical way. A monadic plumbing system is what I get when I introduce the appropriate monad to encapsulate my threading. In fact, if I look at Wadler [*], there are three examples of an evaluator that use what I'm calling "manual plumbing". In section 2.5, the operations required of a monad (return, bind) pretty much just drop right out. Wadler even points out the similarity between "bind" and "let". Now that I finally "get it", I feel that the Wadler paper, section 2.5 in particular, is probably a better introduction than many of the monad tutorials out there. Moreover, I feel that for /some/ of the tutorials out there, they spend too much time and too many illustrations explaining things like [Eq2], and then they quickly present do-notation and gloss over [Eq1]. For me, I found that that the concepts of "manual plumbing" and "monadic plumbing" were key to actually grasping the Wadler paper and understanding what monads are. In particular, I feel that these two concepts might be a way to help other beginners catch on as well. OK, so before I attempt to write a monad tutorial based on "manual plumbing" and "monadic plumbing", I would like to know, does anyone else think this is a good idea? [*] "Monads for Functional Programming". (http://homepages.inf.ed.ac.uk/wadler/papers/marktoberdorf/baastad.pdf)

4 3