August 2010 - Haskell-Cafe

Re: [Haskell] ANNOUNCE: jhc 0.7.4
by Antoine Latter 09 Aug '10

09 Aug '10

On Sat, Jul 10, 2010 at 1:47 AM, John Meacham <john(a)repetae.net> wrote: > Announcing jhc 0.7.4! There have been a few major changes, the main one being > that there is now support for a garbage collector. This drastically increases > the number of programs which are feasable to compile with jhc. > > http://repetae.net/computer/jhc/ > Hi John, This looks fantastic and fun to play with. A few issues: * running DrIFT on src/E/TypeCheck.hs fails with an illegal bytesequence in hGetContents. I'm guessing that this is only an issue when building DrIFT with GHC 6.12+, and that the file contains bytes illegal in UTF8. I deleted everything funny looking in the file and then it went smooth * The way you use sed doesn't work with the BSD sed that ships with my Mac Book. Installing GNU sed and using it works. Similarly, BSD find doesn't know about '-name', so make hl-clean results in sadness. * jhci works great, but jhc crashes when I try to compile something: >>>>> jhc test1.hs jhc 0.7.4 (tokfekyuvi-27) Finding Dependencies... Using Ho Cache: '/Users/alatter/.jhc/cache' Main [test1.hs] Typechecking... [1 of 1] Main (.............................................) test1.hs:9 - Warning: defaulting: t93 => Jhc.Basics.Integer Compiling... [1 of 1] Main <..................................................> Collected Compilation... -- typeAnalyzeMethods -- BoxifyProgram -- Boxy WorkWrap -- LambdaLift E jhc: <stdout>: hPutChar: invalid argument (Illegal byte sequence) <<<<< Again, this seems like the handle is in UTF8 mode and we're trying to output something non-UTF8. * cabal install has a --jhc flag, but it doesn't seem to work: >>>>> cabal install byteorder --jhc Resolving dependencies... cabal: internal error: impossible <<<<< I have jhc installed in a non-standard location (under ${HOME}/usr) so I may need to have some environment variables set up. This is wil Cabal 1.8.0.4 and cabal-install 0.8.2 Keep up the good work! I have an implementation for STRefs I threw together this afternoon for jhc if you're interested. I can't test it properly, though, with the compiler crash above. Antoine

7 22

Odd parallel haskell observations
by Alexander Kotelnikov 09 Aug '10

09 Aug '10

Hello. I am exploring haskell features for parallel and cocurrent programming and see something difficult to explain. In brief - asking RTS to use more threads results in awfull drop of performance. And according to 'top' test programm consumes up to N CPUs power. Am I doing something wrong? I attached the code, but I am just issuing thousands of HTTP GET requests in 1-4 forkIO threads. And since it looks like local apache is faster than haskell program (which is a pity) I expected that using more OS threads should improve performance. Just in case: ghc --version The Glorious Glasgow Haskell Compilation System, version 6.12.1 import Data.List import System.IO import qualified System.IO.UTF8 import System.Environment (getArgs) import Network.HTTP import Network.URI import System.Time import System.IO.Unsafe import Control.Monad import Control.Exception import Control.Concurrent import Control.Concurrent.MVar secDiff :: ClockTime -> ClockTime -> Float secDiff (TOD secs1 psecs1) (TOD secs2 psecs2) = fromInteger (psecs2 - psecs1) / 1e12 + fromInteger (secs2 - secs1) -- single get get :: Int -> IO(String) get id = do res <- simpleHTTP $ getRequest "http://127.0.0.1" case res of Left err -> return(show err) Right rsp -> return(show $ rspCode rsp) -- perform GET per each list element using c threads doList :: [Int] -> Int -> IO() doList ids 0 = return() doList [] c = return() doList ids c = do forkChild $ forM_ todo get doList later (c-1) where (todo, later) = splitAt (length ids `div` c) ids {- Copied from http://haskell.org/ghc/docs/6.12.2/html/libraries/base-4.2.0.1/Control-Conc… Terminating the program -} children :: MVar [MVar ()] children = unsafePerformIO (newMVar []) waitForChildren :: IO () waitForChildren = do cs <- takeMVar children case cs of [] -> return () m:ms -> do putMVar children ms takeMVar m waitForChildren forkChild :: IO () -> IO ThreadId forkChild io = do mvar <- newEmptyMVar childs <- takeMVar children putMVar children (mvar:childs) forkIO (io `finally` putMVar mvar ()) -- end of copied code main = do [c', n'] <- getArgs let c = read c' :: Int n = read n' :: Int start <- getClockTime doList [1..n] c waitForChildren end <- getClockTime putStrLn $ show(c) ++ " " ++ show(secDiff start end) ++ "s" 20:31 sacha@loft4633:/tmp 21> ghc --make -threaded get.hs [1 of 1] Compiling Main ( get.hs, get.o ) Linking get ... 20:31 sacha@loft4633:/tmp 22> ./get 1 10000 1 3.242352s 20:31 sacha@loft4633:/tmp 23> ./get 2 10000 2 3.08306s 20:31 sacha@loft4633:/tmp 24> ./get 2 10000 +RTS -N2 2 6.898871s 20:32 sacha@loft4633:/tmp 25> ./get 3 10000 3 2.950677s 20:32 sacha@loft4633:/tmp 26> ./get 3 10000 +RTS -N2 3 7.381678s 20:32 sacha@loft4633:/tmp 27> ./get 3 10000 +RTS -N3 3 14.683548s 20:32 sacha@loft4633:/tmp 28> ./get 4 10000 4 3.332165s 20:33 sacha@loft4633:/tmp 29> ./get 4 10000 +RTS -N4 -s ./get 4 10000 +RTS -N4 -s 4 57.17923s 2,147,969,912 bytes allocated in the heap 49,059,288 bytes copied during GC 736,656 bytes maximum residency (98 sample(s)) 486,744 bytes maximum slop 5 MB total memory in use (0 MB lost due to fragmentation) Generation 0: 949 collections, 948 parallel, 76.73s, 25.67s elapsed Generation 1: 98 collections, 98 parallel, 7.70s, 2.56s elapsed Parallel GC work balance: 2.17 (6115428 / 2822692, ideal 4) MUT time (elapsed) GC time (elapsed) Task 0 (worker) : 1.43s ( 27.76s) 6.31s ( 2.12s) Task 1 (worker) : 0.00s ( 28.13s) 10.62s ( 3.56s) Task 2 (worker) : 0.37s ( 28.63s) 11.06s ( 3.69s) Task 3 (worker) : 0.00s ( 28.95s) 6.29s ( 2.10s) Task 4 (worker) : 20.73s ( 28.95s) 9.68s ( 3.24s) Task 5 (worker) : 0.00s ( 28.95s) 0.60s ( 0.20s) Task 6 (worker) : 21.81s ( 28.95s) 11.91s ( 3.97s) Task 7 (worker) : 18.59s ( 28.95s) 13.04s ( 4.36s) Task 8 (worker) : 17.24s ( 28.96s) 14.92s ( 4.99s) SPARKS: 0 (0 converted, 0 pruned) INIT time 0.00s ( 0.00s elapsed) MUT time 79.23s ( 28.95s elapsed) GC time 84.43s ( 28.23s elapsed) EXIT time 0.00s ( 0.01s elapsed) Total time 162.49s ( 57.19s elapsed) %GC time 52.0% (49.4% elapsed) Alloc rate 27,513,782 bytes per MUT second Productivity 48.0% of total user, 136.5% of total elapsed gc_alloc_block_sync: 15006 whitehole_spin: 0 gen[0].steps[0].sync_large_objects: 7617 gen[0].steps[1].sync_large_objects: 35 gen[1].steps[0].sync_large_objects: 1400

4 5

Suggestions For An Intro To Monads Talk.
by aditya siram 09 Aug '10

09 Aug '10

Hi all, I am doing an "Intro To Monads" talk in September [1]. The audience consists of experienced non-Haskell developers but they will be familiar with basic functional concepts (closures, first-class functions etc.). I am looking for suggestions on how to introduce the concept and its implications. I'd also like to include a section on why monads exist and why we don't really see them outside of Haskell. Has anyone here done a talk like this? And if so what parts of your presentation were successful and what would you stay away from. Thanks for the feedback. -deech [1] It's in St.Louis, Missouri at the St.Louis Perl Mongers meeting<http://St.Louis%20Perl%20Mongers%20meeting>so come on by if you're around!

17 20

preparing for Haskell 2011
by Malcolm Wallace 09 Aug '10

09 Aug '10

Dear all, Although the Haskell 2010 Language Report has only been published recently, it will soon be time for the Committee to make decisions on the next version, Haskell 2011. I am aiming for the committee to make decisions around the end of Sept or beginning of October 2010. Can I therefore encourage any people who have made proposals, either informally on mailing lists, or formally in the Haskell-prime ticket system, to consider what they need to do to bring those proposals to a state where the committee can vote on them. Perhaps you have not made such a proposal yourself, but are very keen that someone else's proposal be adopted. Work with the proposer to polish it! Here is what you need to know about the proposal process: http://hackage.haskell.org/trac/haskell-prime/wiki/Process Please note especially that the key requirement that will bring your proposal to the attention of the committee for a decision is the "Report delta", which describes the changes with exact precision. Regards, Malcolm

1 0

Socket not released
by Jonathan Geddes 09 Aug '10

09 Aug '10

Cafe, I'm writing a network application that uses static configuration a la xmonad and yi. When the app receives a certain command it recompiles its source, closes the socket it is using and runs its newly compiled "predecessor" as a new process. The problem I'm having is that the port that the parent process was using is not available to the child process. Even though the parent process has terminated, the port is unusable until the child process also terminates. Can anyone give me a clue about what's going on here? I'm using Network and System.Process modules and ruining on Ubuntu linux. I'm using the sClose function to close the socket. Thanks for any tips. --Jonathan

3 6

Re: [Haskell] ANNOUNCE: jhc 0.7.4
by Simon Marlow 09 Aug '10

09 Aug '10

On 27/07/2010 01:54, John Meacham wrote: > For each type I can statically generate an "optimal" layout based on its > structure. For instance, maybe benefits from two of these optimizations, > first of all, nullary constructors (Nothing) need never appear in the > heap, so they are given values that pack directly into a word, this > happens for all nullay constructors in general. Then, since there is > only one constructor left (Just) we can discard the tag field, because > if something of type Maybe, is in WHNF, and is not Nothing then in must be > a Just. so the Just is a single heap allocated word (which are packed > end to end in a page with no overhead) > > I am refining the optimization algorithm, on 64 bit machines I have > another few bits to play with for instance that I can take advantage of. > > A particularly nice case is that characters can be fully integreated > into the word, so the entire space usage of > > "Hello" is 10 words as lists benefit from the same packing benefits as > Maybe. I should point out that in GHC "Hello" requires 15 words: 3 for each list cell, and the Chars are free because they're statically allocated in the RTS. A Just requires 2 words (one more than JHC, I guess). Tantalisingly, it's almost possible to represent a Just with zero words - the pointer to the Just points directly to the element - but there's nowhere to put the tag bits for the element pointer (the tag bits are for the Just). A Nothing in GHC takes zero words, as do all nullary constructors. I wrote up a summary of the GHC heap representation on Stack Overflow recently: http://stackoverflow.com/questions/3254758/memory-footprint-of-haskell-data… Cheers, Simon > compare this to 30 or so words used in a traditional "everything is on > the heap and tagged" model. > > > The manual has a section describing the RTS, it doesn't describe the GC > though, but talks about how I pack things in the pointers. > > http://repetae.net/computer/jhc/manual.html#the-run-time-system > > John >

1 0

Typo or on purpose? http://haskell.org/ghc/docs/6.12.2/html/users_guide/lang-parallel.html
by Alexander Kotelnikov 09 Aug '10

09 Aug '10

This + 1 in (n1 + n2 + 1) what is it doing there? import Control.Parallel nfib :: Int -> Int nfib n | n <= 1 = 1 | otherwise = par n1 (pseq n2 (n1 + n2 + 1)) where n1 = nfib (n-1) n2 = nfib (n-2) BTW, same code can be found in "Seq no more: Better Strategies for Parallel Haskell" article.

2 1

a random numbers generator with a good 'split'
by deb＠pudlak.name 09 Aug '10

09 Aug '10

Hi cafe. recently I was looking at various possibilities of generating random numbers in Haskell. Apparently, the main problem is the 'split' function in the System.Random.RandomGen class. Since it is rarely needed in imperative programs, not much is known how to implement it safely. It's documentation says: > The split operation allows one to obtain two distinct random number > generators. This is very useful in functional programs (for example, > when passing a random number generator down to recursive calls), but > very little work has been done on statistically robust implementations > of split ([System.Random, System.Random] are the only examples we know > of). I consulted it with a friend, who has done much work in the area of random number generators. He suggested that this might be implemented using a random number generator with a very large period, such as Mersene Twister (MT19937 has a period of 2^19937-1). A pair of splitted generators states could be positioned sufficiently far enough from each other so that in practice they never reach each other, and in such a way that successive splits (in practice) never meet. I searched Hackage, where I found a few implementations of Mersene Twister, but AFAIK no one actually implements neither RandomGen nor 'split' alone. I wonder why -- is RandomGen inconvenient or obsolete in some way? Or just nobody tried? With best regards, Petr Pudlak PS: Sorry if you get multiple copies, I had problems sending the mail.

2 1

Please take the State of Haskell, 2010 survey
by Johan Tibell 09 Aug '10

09 Aug '10

Hi all, I've put together a quick, 9-question State of Haskell, 2010 survey: http://blog.johantibell.com/2010/08/state-of-haskell-2010-survey.html The survey will hopefully give us some insight into how people use Haskell and perhaps also some ideas on how Haskell tools and libraries could be improved. P.S. Please direct replies to this email to haskell-cafe(a)haskell.org. Cheers, Johan

1 1

gkt2hs seg fault issue
by briand＠aracnet.com 09 Aug '10

09 Aug '10

I reported a seg-fault in chart a while back. so I was experimenting with cairo and rendering to to the screen, which if my understanding is correct, uses gtk. Certainly looks like it. So I compiled the cairo example Drawing.hs and that worked fine. I then tried running the example from ghci, i.e. loading it and typing main. I get a seg-fault when it exits: Ok, modules loaded: Main. *Main> main Loading package array-0.3.0.0 ... linking ... done. Loading package bytestring-0.9.1.5 ... linking ... done. Loading package filepath-1.1.0.3 ... linking ... done. Loading package old-locale-1.0.0.2 ... linking ... done. Loading package old-time-1.0.0.3 ... linking ... done. Loading package unix-2.4.0.0 ... linking ... done. Loading package directory-1.0.1.0 ... linking ... done. Loading package process-1.0.1.2 ... linking ... done. Loading package time-1.1.4 ... linking ... done. Loading package random-1.0.0.2 ... linking ... done. Loading package haskell98 ... linking ... done. Loading package mtl-1.1.0.2 ... linking ... done. Loading package cairo-0.11.0 ... linking ... done. Loading package containers-0.3.0.0 ... linking ... done. Loading package glib-0.11.0 ... linking ... done. Loading package gio-0.11.0 ... linking ... done. Loading package pretty-1.0.1.1 ... linking ... done. Loading package pango-0.11.0 ... linking ... done. Loading package gtk-0.11.0 ... linking ... done. *Main> Segmentation fault I included all the messages in case someone knows that one of the libraries is out of date and could be the source of the problem. Is it to be expected that it would seg-fault under ghci ??

3 5