November 2011 - Haskell-Cafe

1p function for NonEmpty
by Yitzchak Gale 15 Nov '11

15 Nov '11

Hi Edward, I started @telling you on #haskell, but it has become too long for that. :) The following functions would be nice additions to Data.List.NonEmpty in the semigroups package: replicate1p :: Whole n => n -> a -> NonEmpty a take1p :: Whole n => n -> NonEmpty a -> NonEmpty a splitAt1p :: Whole n => n -> NonEmpty a -> (NonEmpty a, [a]) Also, your warnings in the haddocks about blowing up on negative arguments don't seem to be needed for take, drop, and splitAt. We could avoid the problem with negative arguments for (!!) too by providing an alternative for Whole numbers. But in that case the function would still be partial, so I'm not sure if it's worth it. I wish there were some way to get Whole to default to Natural. Thanks, Yitz

1 1

[Announcement] New release of SBV (0.9.22)
by Levent Erkok 14 Nov '11

14 Nov '11

New release of SBV (0.9.22) is out: http://hackage.haskell.org/package/sbv Major changes in this release are: - Support for explicit quantification (including alternating existentials and universals) - Ability to use Microsoft's Z3 SMT solver (in addition to Yices). Full release notes: http://github.com/LeventErkok/sbv/blob/b75fb082c013ae9612ef09b90d5feeaec652… Thanks, -Levent.

1 0

Can't establish subprocess communication
by Poprádi Árpád 14 Nov '11

14 Nov '11

Hi all! I have found a simple program on the web: --code begin: copierer.hs module Main (main) where main = interact id --code end I compiled it with ghc -threaded --make copierer.hs If i start it from a terminal,it behaves like the "cat" program without arguments: simply copies the stdin to stdout line by line. I wanted it to use from another Haskell program as subprocess: --code begin: twowaysubprocesscomm.hs module Main where import System.IO import System.Process import Control.Concurrent main :: IO () main = do (hin,hout,p) <- start_subprocess send_and_receive (hin,hout) "boo" send_and_receive (hin,hout) "foo" terminateProcess p start_subprocess :: IO (Handle,Handle,ProcessHandle) start_subprocess = do -- (hin, hout, _, p) <- runInteractiveProcess "cat" [] Nothing Nothing -- This line works as expected (hin, hout, _, p) <- runInteractiveProcess "copierer" [] Nothing Nothing -- This line doesn't work hSetBuffering hin LineBuffering hSetBuffering hout LineBuffering return (hin, hout, p) send_and_receive :: (Handle,Handle) -> String -> IO () send_and_receive (hin,hout) indata = do forkIO $ hPutStrLn hin indata outdata <- hGetLine hout putStrLn $ "outdata: " ++ outdata --code end I compiled it with: ghc -threaded --make twowaysubprocesscomm.hs then ran: ./twowaysubprocesscomm twowaysubprocesscomm: fd:7: hGetLine: end of file twowaysubprocesscomm: fd:6: hPutChar: resource vanished (Broken pipe) Copierer doesn't work as subprocess! If i compile twowaysubprocesscomm.hs using the original "cat", it works as expected: ./twowaysubprocesscomm outdata: boo outdata: foo What is wrong here? The "copierer.hs", its usage in "twowaysubprocesscomm.hs" or both? I use GHC 6.12.3 on a 64 bit linux. Thanks, Árpád

3 3

ANN: Netwire 2.0.0
by Ertugrul Soeylemez 14 Nov '11

14 Nov '11

Hello everybody, it's finally done. I have just released Netwire 2.0.0, which is a complete rewrite from scratch. It has been generalized to arbitrary automata with inhibition, so FRP is now just a special use case. First of all the top module is now: import Control.Wire Wire is now an arrow transformer instead of an arrow-to-monad transformer and the inhibition type is polymorphic instead of the hard-coded SomeException type: Wire e (>~) a b Wire represents the following underlying type: a >~ (Either e b, Wire e (>~) a b) Inhibition uses any monoid 'e'. To get the old behaviour, you can use Last SomeException as 'e'. One significant change is that all features which were builtin in Netwire 1 are now addons, including even time and random number generation. These addons are available in the form of type classes (Control.Wire.Classes) for the underlying arrow, including for example: * ArrowIO, * ArrowRandom, * ArrowTime. Also there are no explicit sessions anymore. The current state of a wire is simply the wire itself. Everything else is captured by the underlying arrow (>~), which you can construct the way you like. There is no hard-coded WireState anymore. To get going really fast, just use the following type: type SimpleWire = Wire () (Kleisli IO) with the testWire function. You can find a small demo here: <http://hpaste.org/54025> There are instances for all addon classes for Kleisli IO, so you can write your first application very quickly. For more serious applications, you would usually write or compose your own underlying arrow or even leave it polymorphic. Please don't hesitate to give me feedback. Greets, Ertugrul -- nightmare = unsafePerformIO (getWrongWife >>= sex) http://ertes.de/

2 3

Do type classes have a partial order?
by Patrick Browne 14 Nov '11

14 Nov '11

2 1

Re: [Haskell-cafe] German names for kinds and sorts
by John Lato 14 Nov '11

14 Nov '11

> From: Jerzy Karczmarczuk <jerzy.karczmarczuk(a)unicaen.fr> > > So, even more seriously, I propose to use more often Latin and Greek. Jerzy's arguments make a lot of sense to me, but why not use an index-based notation? T0 = Type T1 = Kind T2 = Sort etc. this seems easier to me than an arbitrary hierarchy, particularly as you go higher up the structure. John Lato

1 0

Identifying thunks from GHC heap profiles
by Tristan Ravitch 14 Nov '11

14 Nov '11

I am trying to debug a stack overflow that seems to be caused by a large chain of thunks being evaluated all at once. I have some profiles generated using +RTS -hd -RTS that show some promising candidates, but they have generated names like Module.sat_s9et, Module.sat_s9jk, etc Is there any convenient way to map these generated names back to something recognizable in the source? I tried using +RTS -hc -hdsat_s9et -RTS to filter SCCs as suggested in http://hackage.haskell.org/trac/ghc/ticket/1639, but the resulting profile is always empty. I also tried to use -fext-core and -ddump-simpl to dump some intermediate representations but none of the thunk names seem to appear in those outputs.

2 1

German names for kinds and sorts
by Robert Clausecker 14 Nov '11

14 Nov '11

Hi all! I want to write my Facharbeit (kind of an essay you have to write on a specific topic you can choose yourself for highschool graduation) about the type-system of Haskell. It is required in our school to write this document in German language. Most time, it is not really difficult to find an appropriate term for concepts of Haskell, like types (Typen) or type classes (Typklassen). But I really don't know how to call kinds and sorts in German. Any ideas? Yours, Robert Clausecker

10 10

ANNOUNCE: iteratee-compress 0.3.0.0
by Maciej Marcin Piechotka 13 Nov '11

13 Nov '11

Iteratee-compress provides compressing and decompressing enumerators including flushing (using John Lato's implementation). Currently only gzip and bzip is provided but LZMA is planned. Changes from previous version: - Move API from enumerator to enumeratee Next goals: - LZMA support - Generic interface for flushing To think about: - Should inner iteratee be able to request flushing Regards

1 0

Getting started on Mac OSX
by hstenstrom 12 Nov '11

12 Nov '11

I have a book on Haskell, and I've downloaded and installed Haskell Platform for Mac OS X. What do I do now? XCode is a requirement, and I have it, but I don't know how to run it. To begin with, I want to test small examples from the book, using ghc or ghci. I have read file:///Library/Haskell/doc/start.html. - Where is the software I've just installed? - How is ghci started? Or the "platform", whatever that is? Many thanks if you can help me! Helge -- View this message in context: http://haskell.1045720.n5.nabble.com/Getting-started-on-Mac-OSX-tp4983597p4… Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

3 4