- Haskell-Cafe - Haskell.org

unsafeInterleaveST (and IO) is really unsafe [was: meaning of "referential transparency"]
by oleg＠okmij.org 12 Apr '13

12 Apr '13

One may read this message as proving True === False without resorting to IO. In other words, referential transparency, or the substitution of equals for equals, may fail even in expressions of type Bool. This message is intended as an indirect stab at lazy IO. Unfortunately, Lazy IO and even the raw unsafeInterleaveIO, on which it is based, are sometimes recommended, without warnings that usually accompany unsafePerformIO. http://www.haskell.org/pipermail/haskell-cafe/2013-March/107027.html UnsafePerformIO is known to be unsafe, breaking equational reasoning; unsafeInterleaveIO gets a free pass because any computation with it has to be embedded in the IO context in order to be evaluated -- and we can expect anything from IO. But unsafeInterleaveIO has essentially the same code as unsafeInterleaveST: compare unsafeInterleaveST from GHC/ST.lhs with unsafeDupableInterleaveIO from GHC/IO.hs keeping in mind that IO and ST have the same representation, as described in GHC/IO.hs. And unsafeInterleaveST is really unsafe -- not just mildly or somewhat or vaguely unsafe. In breaks equational reasoning, in pure contexts. Here is the evidence. I hope most people believe that the equality on Booleans is symmetric. I mean the function (==) :: Bool -> Bool -> Bool True == True = True False == False = True _ == _ = False or its Prelude implementation. The equality x == y to y == x holds even if either x or y (or both) are undefined. And yet there exists a context that distinguishes x == y from y ==x. That is, there exists bad_ctx :: ((Bool,Bool) -> Bool) -> Bool such that *R> bad_ctx $ \(x,y) -> x == y True *R> bad_ctx $ \(x,y) -> y == x False The function unsafeInterleaveST ought to bear the same stigma as does unsafePerformIO. After all, both masquerade side-effecting computations as pure. Hopefully even lazy IO will get recommended less, and with more caveats. (Lazy IO may be superficially convenient -- so are the absence of typing discipline and the presence of unrestricted mutation, as many people in Python/Ruby/Scheme etc worlds would like to argue.) The complete code: module R where import Control.Monad.ST.Lazy (runST) import Control.Monad.ST.Lazy.Unsafe (unsafeInterleaveST) import Data.STRef.Lazy bad_ctx :: ((Bool,Bool) -> Bool) -> Bool bad_ctx body = body $ runST (do r <- newSTRef False x <- unsafeInterleaveST (writeSTRef r True >> return True) y <- readSTRef r return (x,y)) t1 = bad_ctx $ \(x,y) -> x == y t2 = bad_ctx $ \(x,y) -> y == x

6 6

Web servers: Running them multiple times in a ghci session
by Niklas Hambüchen 12 Apr '13

12 Apr '13

I'm writing a web server app, which I run in ghci: :main localhost 8000 Unfortunately, after Ctrl-C and :reload, running it again: ** Exception: bind: resource busy (Address already in use) This is pretty annoying, because quitting-and-restarting takes a lot of time since I have many dependencies. How do you deal with this? Can you propose some working code that can be wrapped around my main function to make it work? My first idea is running main in a separate process. If you have a working idea, please also post it as an answer on http://stackoverflow.com/questions/15890912/how-do-i-terminate-a-socket-ser… Thanks

3 2

cabal-install 1.16.0.2 on Mac
by Richard A. O'Keefe 12 Apr '13

12 Apr '13

Machine: an Intel Core 2 Duo desktop Mac. OS: Mac OS X 10.7.4 Xcode: 4.6.1 (including command line tools) Haskell: "Haskell Platform 2012.4.0.0 64bit.pkg" downloaded today (GHC 7.4.2) cabal update advised me to install a new cabal-install. m% cabal install cabal-install Resolving dependencies... Downloading Cabal-1.16.0.3... [ 1 of 65] Compiling Distribution.Compat.Exception ( /var/folders/_n/5bc02hb51d361crtq41g_gc4000g8p/T/Cabal-1.16.0.3-85340/Cabal-1.16.0.3/Distribution/Compat/Exception.hs, /var/folders/_n/5bc02hb51d361crtq41g_gc4000g8p/T/Cabal-1.16.0.3-85340/Cabal-1.16.0.3/dist/setup/Distribution/Compat/Exception.o ) <<< snip >>> [52 of 67] Compiling Distribution.Simple.Build.PathsModule ( Distribution/Simple/Build/PathsModule.hs, dist/build/Distribution/Simple/Build/PathsModule.o ) Distribution/Simple/Build/PathsModule.hs:210:19: Warning: Pattern match(es) are non-exhaustive In a case alternative: Patterns not matched: PPC PPC64 Sparc Arm ... [53 of 67] Compiling Distribution.Simple.GHC ( Distribution/Simple/GHC.hs, dist/build/Distribution/Simple/GHC.o ) <<< snip >>> [65 of 65] Compiling Main ( /var/folders/_n/5bc02hb51d361crtq41g_gc4000g8p/T/Cabal-1.16.0.3-85340/Cabal-1.16.0.3/Setup.hs, /var/folders/_n/5bc02hb51d361crtq41g_gc4000g8p/T/Cabal-1.16.0.3-85340/Cabal-1.16.0.3/dist/setup/Main.o ) Linking /var/folders/_n/5bc02hb51d361crtq41g_gc4000g8p/T/Cabal-1.16.0.3-85340/Cabal-1.16.0.3/dist/setup/setup ... Configuring Cabal-1.16.0.3... Building Cabal-1.16.0.3... Preprocessing library Cabal-1.16.0.3... [ 1 of 67] Compiling Paths_Cabal ( dist/build/autogen/Paths_Cabal.hs, dist/build/Paths_Cabal.o ) <<< snip >>> [56 of 65] Compiling Distribution.Client.SetupWrapper ( Distribution/Client/SetupWrapper.hs, dist/build/cabal/cabal-tmp/Distribution/Client/SetupWrapper.o ) Distribution/Client/SetupWrapper.hs:51:12: Warning: In the use of `ghcVerbosityOptions' (imported from Distribution.Simple.GHC): Deprecated: "Use the GhcOptions record instead" [57 of 65] Compiling Distribution.Client.Upload ( Distribution/Client/Upload.hs, dist/build/cabal/cabal-tmp/Distribution/Client/Upload.o ) <<< snip >>> [65 of 65] Compiling Main ( Main.hs, dist/build/cabal/cabal-tmp/Main.o ) Linking dist/build/cabal/cabal ... Installing executable(s) in /home/cshome/o/ok/.cabal/bin /Developer/usr/bin/strip: object: /home/cshome/o/ok/.cabal/bin/cabal malformed object (unknown load command 15) cabal: Error: some packages failed to install: cabal-install-1.16.0.2 failed during the final install step. The exception was: ExitFailure 1 m% file ~/.cabal/bin/cabal /home/cshome/o/ok/.cabal/bin/cabal: Mach-O 64-bit executable x86_64 The strip(1) page ends with this, which may be relevant: LIMITATIONS Not every layout of a Mach-O file can be stripped by this program. But all layouts produced by the Apple compiler system can be stripped. m% otool -l ~/.cabal/bin/cabal <<< snip >>> Load command 14 cmd LC_FUNCTION_STARTS cmdsize 16 dataoff 12743064 datasize 204136 Load command 15 cmd ?(0x00000029) Unknown load command cmdsize 16 00c58f00 00000000 So something is definitely putting something in there that the Xcode 4.6.1 tools do not like.

3 6

exp implementation
by Christopher Howard 11 Apr '13

11 Apr '13

Hi. For my own learning, I wanted to see how the exp function is implemented in GHC. I have GHC 7.4.1 source code open, but I'm having trouble figuring out which file the actual function definition is in. I see expFloat (F# x) = F# (expFloat# x) in libraries/base/GHC/Float.lhs. -- frigidcode.com

2 3

Which generic zipper?
by Luke Evans 11 Apr '13

11 Apr '13

It's sometimes difficult to figure out a good choice when you're searching for a new feature/library (when you go beyond safety of the HP). I'm looking for a generic zipper to (hopefully) navigate over an AST… so a hetrogeneous structure. It's probably sufficient to make this traversable in a fixed way, but a zipper seemed like a good idea. Google says there are lots of options (various papers and their related implementations etc). I see a generic zipper in an 'syz' package that is based on generics, which looks like a nice choice as it's now easy to add the extension for deriving generics on ADTs. However, that package looks pretty old and unmaintained. Does anyone have any suggestions/advice? Cheers -- Luke

3 5

Numerics and Warnings
by Barak A. Pearlmutter 11 Apr '13

11 Apr '13

In fiddling around with some numeric code in Haskell, I noticed some issues. Basically, you get warnings if you write energy mass = mass * c^2 but not if you write energy mass = mass * c * c which seems a bit perverse. Some more examples are below. I understand the inference issues that cause this, but common innocuous cases could---and I would argue, should---be addressed in ad-hoc ways. -- Barak A. Pearlmutter Hamilton Institute & Dept Comp Sci, NUI Maynooth, Co. Kildare, Ireland http://www.bcl.hamilton.ie/~barak/ ---------------- transcript ---------------- $ cat two.hs main :: IO () main = do print 1 print ((2::Float)^3) print 44444444444444444444444444444444 print (555555555555555555555555::Int) -- Each of the four numeric expressions printed above gives a -- compile-time warning. Except the last, which is the only one that is -- actually incorrect. Maybe ghc could eliminate some of these warnings -- by adding extra reasonableness guards. E.g., if the 2nd argument of -- (^) is a constant that fits in an Int, just do that. Basically, if -- used as an argument to a function which has no danger of overflow, -- default to Int without warning. And if a constant doesn't fit in an -- Int, jamming it into one is an error. Or at least a warning! $ ghc --version The Glorious Glasgow Haskell Compilation System, version 7.6.2 $ ghc -Wall two.hs [1 of 1] Compiling Main ( two.hs, two.o ) two.hs:3:9: Warning: Defaulting the following constraint(s) to type `Integer' (Num a0) arising from the literal `1' at two.hs:3:9 (Show a0) arising from a use of `print' at two.hs:3:3-7 In the first argument of `print', namely `1' In a stmt of a 'do' block: print 1 In the expression: do { print 1; print ((2 :: Float) ^ 3); print 44444444444444444444444444444444; print (555555555555555555555555 :: Int) } two.hs:4:20: Warning: Defaulting the following constraint(s) to type `Integer' (Integral b0) arising from a use of `^' at two.hs:4:20 (Num b0) arising from the literal `3' at two.hs:4:21 In the first argument of `print', namely `((2 :: Float) ^ 3)' In a stmt of a 'do' block: print ((2 :: Float) ^ 3) In the expression: do { print 1; print ((2 :: Float) ^ 3); print 44444444444444444444444444444444; print (555555555555555555555555 :: Int) } two.hs:5:10: Warning: Defaulting the following constraint(s) to type `Integer' (Num a0) arising from the literal `44444444444444444444444444444444' at two.hs:5:10-41 (Show a0) arising from a use of `print' at two.hs:5:3-7 In the first argument of `print', namely `44444444444444444444444444444444' In a stmt of a 'do' block: print 44444444444444444444444444444444 In the expression: do { print 1; print ((2 :: Float) ^ 3); print 44444444444444444444444444444444; print (555555555555555555555555 :: Int) } Linking two ... $ ./two 1 8.0 44444444444444444444444444444444 -5035712355010463517

5 10

Bug in Network package
by Florian Hofmann 11 Apr '13

11 Apr '13

I might be mistaken, but is there a bug in the Show instance of PortNum? λ PortNum 1 256 λ PortNum 2 512 λ PortNum 3 768 λ let (PortNum x) = PortNum 10 λ x 10 Tested with network-2.4.1.2

5 5

Haskell Weekly News: Issue 265
by Daniel Santa Cruz 11 Apr '13

11 Apr '13

Welcome to issue 265 of the HWN, an issue covering crowd-sourced bits of information about Haskell from around the web. This issue covers the week of March 31 to April 6, 2013. Quotes of the Week * Jafet: Haskell is an abstract research language used only in academia, education, banking, stock trading, circuit design, embedded systems, cryptography, operating systems research, bioinformatics, phone apps, and web services. * monochrom: primop is where the turtle tower ends and the russian dolls begin! * sclv: haha. i love this stuff but if i bury myself in it my head will start spinning around like exorcist and i will start spitting blood Top Reddit Stories * Haskell beats C using generalised stream fusion - Geoff Mainland, Simon Peyton Jones, Simon Marlow, Roman Leshchinskiy. Submitted to ICFP 2013 [PDF] Domain: research.microsoft.com, Score: 106, Comments: 47 On Reddit: [1] http://goo.gl/Ui7bS Original: [2] http://goo.gl/es51B * Functor is now a superclass of Monad in GHC HEAD Domain: darcs.haskell.org, Score: 87, Comments: 33 On Reddit: [3] http://goo.gl/B1LYf Original: [4] http://goo.gl/JiunP * Yet Another Lambda Blog » Haskell as fast as C: A case study Domain: lambda.jstolarek.com, Score: 61, Comments: 25 On Reddit: [5] http://goo.gl/cemyl Original: [6] http://goo.gl/uCbTd * Beautiful Concurrency - a series of articles by Simon Peyton-Jones - on FP Complete School of Haskell Domain: fpcomplete.com, Score: 60, Comments: 20 On Reddit: [7] http://goo.gl/pWAA4 Original: [8] http://goo.gl/ynFG8 * Web Woes for Haskell Domain: self.haskell, Score: 43, Comments: 42 On Reddit: [9] http://goo.gl/vl13n Original: [10] http://goo.gl/vl13n * Cap'n Proto: faster protocol buffers, written in Haskell & C++ Domain: kentonv.github.com, Score: 42, Comments: 11 On Reddit: [11] http://goo.gl/6Qwza Original: [12] http://goo.gl/94iip * Resource limits for Haskell Domain: blog.ezyang.com, Score: 38, Comments: 1 On Reddit: [13] http://goo.gl/01WPS Original: [14] http://goo.gl/KjwNx * Dependent Types - A New Paradigm? by Paul Callaghan Domain: pragprog.com, Score: 33, Comments: 1 On Reddit: [15] http://goo.gl/Gjs3K Original: [16] http://goo.gl/4wbYa * So I accidentally a profunctors tutorial Domain: fpcomplete.com, Score: 31, Comments: 43 On Reddit: [17] http://goo.gl/QnzoR Original: [18] http://goo.gl/VDlYh * UPenn CIS 194: Intro to Haskell, Instructor: Brent Yorgey Domain: seas.upenn.edu, Score: 31, Comments: 4 On Reddit: [19] http://goo.gl/zXgrM Original: [20] http://goo.gl/OC3x1 * Sorting with Agda Domain: mazzo.li, Score: 30, Comments: 33 On Reddit: [21] http://goo.gl/rkq58 Original: [22] http://goo.gl/8GRCF * Announcing The Pragmatic Haskeller Domain: cakesolutions.net, Score: 29, Comments: 7 On Reddit: [23] http://goo.gl/3ahrb Original: [24] http://goo.gl/AsOKQ * Flavours of free applicative functors Domain: ro-che.info, Score: 27, Comments: 6 On Reddit: [25] http://goo.gl/oE2sk Original: [26] http://goo.gl/sE5VD * Free Applicative Functors (ICFP submission) Domain: paolocapriotti.com, Score: 25, Comments: 11 On Reddit: [27] http://goo.gl/2dB7B Original: [28] http://goo.gl/VWnHG * Handling out of memory? Domain: self.haskell, Score: 25, Comments: 17 On Reddit: [29] http://goo.gl/mGwUt Original: [30] http://goo.gl/mGwUt * hydra-print - visualizing multiple texts streams Domain: parfunk.blogspot.se, Score: 24, Comments: 7 On Reddit: [31] http://goo.gl/uToAZ Original: [32] http://goo.gl/lc70w * Lazy IO breaks equational reasoning Domain: okmij.org, Score: 24, Comments: 124 On Reddit: [33] http://goo.gl/48f0h Original: [34] http://goo.gl/m8Vbr * Haskell on Windows Domain: gundersen.net, Score: 23, Comments: 14 On Reddit: [35] http://goo.gl/oRV4B Original: [36] http://goo.gl/0hOL4 * Constrained Monads (GADTs indexed by Constraint) Domain: ittc.ku.edu, Score: 21, Comments: 7 On Reddit: [37] http://goo.gl/lVRpl Original: [38] http://goo.gl/jtbU3 Top StackOverflow Questions * Simple examples to illustrate Category, Monoid and Monad? votes: 28, answers: 1 Read on SO: [39] http://goo.gl/HPE8k * Which techniques have contributed the most to Haskell's improving performance? [closed] votes: 18, answers: 1 Read on SO: [40] http://goo.gl/tz904 * Why does Parsec not use Control.Applicative operators votes: 13, answers: 1 Read on SO: [41] http://goo.gl/0mR3f * ghc 7.4.2, Dynamically calling modules votes: 10, answers: 1 Read on SO: [42] http://goo.gl/MZD8q * What's wrong with this implementation of quicksort using Arrows? votes: 10, answers: 2 Read on SO: [43] http://goo.gl/zwhfy * Initializing algebraic data type from list votes: 9, answers: 4 Read on SO: [44] http://goo.gl/nbyUP * Scoped type variables require explicit foralls. Why? votes: 9, answers: 1 Read on SO: [45] http://goo.gl/SUlCf * Parsing UnicodeSyntax with haskell-src-exts votes: 8, answers: 1 Read on SO: [46] http://goo.gl/Yi8Vd Until next time, +Daniel Santa Cruz References 1. http://research.microsoft.com/en-us/um/people/simonpj/papers/ndp/haskell-be… 2. http://www.reddit.com/r/haskell/comments/1br0ls/haskell_beats_c_using_gener… 3. http://darcs.haskell.org/libraries/base/GHC/Base.lhs 4. http://www.reddit.com/r/haskell/comments/1bfojn/functor_is_now_a_superclass… 5. http://lambda.jstolarek.com/2013/04/haskell-as-fast-as-c-a-case-study/ 6. http://www.reddit.com/r/haskell/comments/1bikvs/yet_another_lambda_blog_has… 7. https://www.fpcomplete.com/user/simonpj/beautiful-concurrency 8. http://www.reddit.com/r/haskell/comments/1bm3ep/beautiful_concurrency_a_ser… 9. http://www.reddit.com/r/haskell/comments/1bqsvi/web_woes_for_haskell/ 10. http://www.reddit.com/r/haskell/comments/1bqsvi/web_woes_for_haskell/ 11. http://kentonv.github.com/capnproto/install.html 12. http://www.reddit.com/r/haskell/comments/1bjt4f/capn_proto_faster_protocol_… 13. http://blog.ezyang.com/2013/04/resource-limits-for-haskell/ 14. http://www.reddit.com/r/haskell/comments/1bkjj8/resource_limits_for_haskell/ 15. http://pragprog.com/magazines/2013-04/dependent-types 16. http://www.reddit.com/r/haskell/comments/1bmqba/dependent_types_a_new_parad… 17. https://www.fpcomplete.com/user/liyang/profunctors 18. http://www.reddit.com/r/haskell/comments/1bg21c/so_i_accidentally_a_profunc… 19. http://www.seas.upenn.edu/~cis194/lectures.html 20. http://www.reddit.com/r/haskell/comments/1bqax8/upenn_cis_194_intro_to_hask… 21. http://mazzo.li/posts/AgdaSort.html 22. http://www.reddit.com/r/haskell/comments/1biasi/sorting_with_agda/ 23. http://www.cakesolutions.net/teamblogs/2013/04/05/announcing-the-pragmatic-… 24. http://www.reddit.com/r/haskell/comments/1brbac/announcing_the_pragmatic_ha… 25. http://ro-che.info/articles/2013-03-31-flavours-of-free-applicative-functor… 26. http://www.reddit.com/r/haskell/comments/1be2iv/flavours_of_free_applicativ… 27. http://paolocapriotti.com/blog/2013/04/03/free-applicative-functors/ 28. http://www.reddit.com/r/haskell/comments/1bnql3/free_applicative_functors_i… 29. http://www.reddit.com/r/haskell/comments/1bsucw/handling_out_of_memory/ 30. http://www.reddit.com/r/haskell/comments/1bsucw/handling_out_of_memory/ 31. http://parfunk.blogspot.se/2013/04/hydra-print.html 32. http://www.reddit.com/r/haskell/comments/1bsc9u/hydraprint_visualizing_mult… 33. http://okmij.org/ftp/Haskell/index.html#lazyIO-not-True 34. http://www.reddit.com/r/haskell/comments/1bsitm/lazy_io_breaks_equational_r… 35. http://www.gundersen.net/haskell-on-windows/ 36. http://www.reddit.com/r/haskell/comments/1bsm6v/haskell_on_windows/ 37. http://www.ittc.ku.edu/csdl/fpg/papers/Sculthorpe-13-ConstrainedMonad.html 38. http://www.reddit.com/r/haskell/comments/1bktx3/constrained_monads_gadts_in… 39. http://stackoverflow.com/questions/15726733/simple-examples-to-illustrate-c… 40. http://stackoverflow.com/questions/15750998/which-techniques-have-contribut… 41. http://stackoverflow.com/questions/15767978/why-does-parsec-not-use-control… 42. http://stackoverflow.com/questions/15824756/ghc-7-4-2-dynamically-calling-m… 43. http://stackoverflow.com/questions/15849550/whats-wrong-with-this-implement… 44. http://stackoverflow.com/questions/15734573/initializing-algebraic-data-typ… 45. http://stackoverflow.com/questions/15800878/scoped-type-variables-require-e… 46. http://stackoverflow.com/questions/15784076/parsing-unicodesyntax-with-hask… 47. https://plus.google.com/105107667630152149014/about

1 0

Automated Differentiation of Matrices (hmatrix)
by Dominic Steinitz 10 Apr '13

10 Apr '13

Hi Cafe, Suppose I want to find the grad of a function then it's easy I just use http://hackage.haskell.org/package/ad-3.4: import Numeric.AD import Data.Foldable (Foldable) import Data.Traversable (Traversable) data MyMatrix a = MyMatrix (a, a) deriving (Show, Functor, Foldable, Traversable) f :: Floating a => MyMatrix a -> a f (MyMatrix (x, y)) = exp $ negate $ (x^2 + y^2) / 2.0 main :: IO () main = do putStrLn $ show $ f $ MyMatrix (0.0, 0.0) putStrLn $ show $ grad f $ MyMatrix (0.0, 0.0) But now suppose I am doing some matrix calculations http://hackage.haskell.org/package/hmatrix-0.14.1.0 and I want to find the grad of a function of a matrix: import Numeric.AD import Numeric.LinearAlgebra import Data.Foldable (Foldable) import Data.Traversable (Traversable) g :: (Element a, Floating a) => Matrix a -> a g m = exp $ negate $ (x^2 + y^2) / 2.0 where r = (toLists m)!!0 x = r!!0 y = r!!1 main :: IO () main = do putStrLn $ show $ g $ (1 >< 2) ([0.0, 0.0] :: [Double]) putStrLn $ show $ grad g $ (1 >< 2) ([0.0, 0.0] :: [Double]) Then I am in trouble: /Users/dom/Dropbox/Private/Whales/MyAD.hs:24:21: No instance for (Traversable Matrix) arising from a use of `grad' Possible fix: add an instance declaration for (Traversable Matrix) In the expression: grad g In the second argument of `($)', namely `grad g $ (1 >< 2) ([0.0, 0.0] :: [Double])' In the second argument of `($)', namely `show $ grad g $ (1 >< 2) ([0.0, 0.0] :: [Double])' /Users/dom/Dropbox/Private/Whales/MyAD.hs:24:26: Could not deduce (Element (ad-3.4:Numeric.AD.Internal.Types.AD s Double)) arising from a use of `g' from the context (Numeric.AD.Internal.Classes.Mode s) bound by a type expected by the context: Numeric.AD.Internal.Classes.Mode s => Matrix (ad-3.4:Numeric.AD.Internal.Types.AD s Double) -> ad-3.4:Numeric.AD.Internal.Types.AD s Double at /Users/dom/Dropbox/Private/Whales/MyAD.hs:24:21-26 Possible fix: add an instance declaration for (Element (ad-3.4:Numeric.AD.Internal.Types.AD s Double)) In the first argument of `grad', namely `g' In the expression: grad g In the second argument of `($)', namely `grad g $ (1 >< 2) ([0.0, 0.0] :: [Double])' What are my options here? Clearly I can convert my matrix into a list (which is traversable), find the grad and convert it back into a matrix but given I am doing numerical calculations and speed is an important factor, this seems undesirable. I think I would have the same problem with: http://hackage.haskell.org/package/repa http://hackage.haskell.org/package/yarr-1.3.1 although I haven'¯t checked. Thanks, Dominic.

3 5

abs on Float/Doubles
by Levent Erkok 10 Apr '13

10 Apr '13

Hello all, I definitely do not want to create a yet another thread on handling of floating-point values in Haskell. My intention is only to see what can be done to make it more "compilant" with IEEE754. (Also, my interest is not a theoretical one, but rather entirely practical: I'm interested in using SMT solvers for reasoning about Haskell programs, in particular using the new SMT logics that cover IEEE-754 semantics. So, any discrepancy between Haskell and IEEE-754 is a cause for concern as we strive to build powerful tools around the Haskell eco-system.) In IEEE-754, there are two zero values: +0, and -0; inhabiting all floating point types. IEEE-754 requires these two compare equal, but be distinguished when used as arguments to functions as they can produce different results. (Unfortunately the standard is not freely available, but there's a nice discussion in wikipedia: http://en.wikipedia.org/wiki/Negative_zero.) As far as I know, Haskell's floating-point handling does *not* explicitly claim to be IEEE-754 compliant, but I think it's safe to assume that any serious implementation will take advantage of the underlying CPU's floating-point facilities; and thus will use IEEE-754 semantics. In particular, Haskell supports both +0 and -0; and you can distinguish them using the isNegativeZero function of the RealFloat class. Furthermore, the show instance for Float/Double will turn these two values to strings differently, preserving the minus sign if it receives a negative-zero. Comparisons follow the IEEE rules; so all these behaviors seem to be in accordance with IEEE754. Unfortunately, the same isn't true for the abs function. In Haskell, the call: "isNegativeZero (abs (-0::Double))" evaluates to "True". IEEE754 requires abs to always return "+0" in this case. The same also holds for the semantics of the newly proposed SMT-Lib logic for IEEE754: See section 3.3.1 (page 8) of: http://www.philipp.ruemmer.org/publications/smt-fpa.pdf The fix is easy. A one line change in the Double/Float instances of the Num class in the Prelude would guarantee the IEEE754 semantics. The reason I'm not simply posting this as a "library" issue is that it appears the topic came up in stack-overflow a while ago; without any clear resolution: http://stackoverflow.com/questions/10395761/absolute-value-of-negative-zero… It appears that the consensus is that this is a historical definition dating back to the times when IEEE754 itself wasn't quite clear on the topic itself, and "so nobody thought that hard about negative zeroes." (The quote is from a comment from Lennart.) I think it's a safe bet to assume IEEE754 semantics (for better or worse) is here to stay, and it would be nice to reduce any discrepancy we have in the Haskell libraries with respect to it. So, if there's consensus, I'd like to propose a library change to implement the proper IEEE754 semantics for the abs function for Double and Float types. Note that the argument is not really about semantics here, but rather with compliance to the accepted industrial standard. So, it would be best if we kept the discussion purely to the IEEE-compliance here, as opposed to relative merits of the semantics as defined. -Levent.

2 2