August 2001 - Haskell

RE: lexical description problem in language report?
by Simon Peyton-Jones 17 Aug '01

17 Aug '01

1 0

RE: foldl'
by Simon Peyton-Jones 17 Aug '01

17 Aug '01

No good reason -- but it's too late for H98. Adding it to the Prelude would break some existing programs (ones which define their own foldl'), and it's an unforced change. Alas. Simon | -----Original Message----- | From: Jon Fairbairn [mailto:Jon.Fairbairn@cl.cam.ac.uk] | Sent: 28 July 2001 11:28 | To: haskell(a)haskell.org | Subject: foldl' | | | Unless I'm mistaken, foldl' (the strict version of foldl) | doesn't appear in (the export list of) the standard prelude | or the list library. | | Is there a good reason for this? New users quite quickly | find that they need it. | | Jón | | | -- | Jón Fairbairn | Jon.Fairbairn(a)cl.cam.ac.uk | 31 Chalmers Road | jf(a)cl.cam.ac.uk | Cambridge CB1 3SZ +44 1223 570179 (after 14:00 | only, please!) | | | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell |

1 0

Mathematics of Program Construction, 1st CFP
by mpc02 17 Aug '01

17 Aug '01

[Apologies for multiple copies] MPC 2002 6th International Conference on MATHEMATICS OF PROGRAM CONSTRUCTION ----------------------------------- http://www.cs.ukc.ac.uk/conf/mpc2002/ Organised in conjunction with WCGP '02 8--10 July, 2002 Dagstuhl, Germany CALL FOR PAPERS This conference aims to promote the development of mathematical principles and techniques that are demonstrably useful and usable in the process of constructing computer programs (whether implemented in hardware or software). The focus of the conference is on techniques that combine precision with conciseness, enabling programs to be constructed by formal calculation. Within this theme, the scope of the conference is very diverse. We welcome contributions to programming methodology (for example, formal methods for program specification and transformation), to programming paradigms (for example, generic programming techniques and type systems) and to language design (for example, programming calculi and programming language semantics). Theoretical contributions are welcome provided their relevance to program construction is evident; discussion of applications is welcome provided the mathematical basis is evident. The conference will be organised in conjunction with the IFIP TC2 Working Conference on Generic Programming, WCGP '02. There will also be a number of co-located workshops, including CMPP. IMPORTANT DATES Deadline for submission of papers: 6th January, 2002 Notification of acceptance/rejection: 4th March, 2002 Final papers due: 25th April, 2002 Full papers should be submitted in Postscript or pdf format by e-mail to mpc02(a)ukc.ac.uk by 6th January, 2002. PROGRAMME COMMITTEE Roland Backhouse (UK) Eerke Boiten (UK, co-chair) Michael Butler (UK) Ernie Cohen (USA) Jules Desharnais (Canada) Jeremy Gibbons (UK, GP liaison) David Gries (USA) Lindsay Groves (New Zealand) Ian Hayes (Australia) Eric Hehner (Canada) Zhenjiang Hu (Japan) John Hughes (Sweden) Bart Jacobs (The Netherlands) Johan Jeuring (The Netherlands, GP liaison) Dick Kieburtz (USA) Dexter Kozen (USA) K. Rustan M. Leino (USA) Christian Lengauer (Germany, CMPP liaison) Erik Meijer (The Netherlands/USA) Bernhard Moeller (Germany, co-chair) David Naumann (USA) J.N. Oliveira (Portugal) Alberto Pardo (Uruguay) Peter Pepper (Germany) Kaisa Sere (Finland) Mark Utting (New Zealand) FURTHER INFORMATION Please refer to the web page for further details. http://www.cs.ukc.ac.uk/conf/mpc2002/ -- Mathematics of Program Construction '02 at Dagstuhl Eerke Boiten, Computing Laboratory, University of Kent at Canterbury, UK +44.1227.827615 (fax 762811)

1 0

ANN: linkchk-0.00 - a gtk based network link heartbeat monitor
by Jens-Ulrik Petersen 17 Aug '01

17 Aug '01

Hello, I am happy to announce the first release of linkchk, a small program whose sole job is simply to monitor a ping heartbeat between its host and a gateway (router) on a local network link. Sounds simple? Well it is really, except for some first hurdles of pipes, lazy fileRead'ing, etc. Source code and a Linux binary are available from <http://www.01.246.ne.jp/~juhp/haskell/linkchk/>. ghc and gtk+hs are required to build linkchk. This is the first Haskell program I am releasing on the net, so I would appreciate any feedback I can get whether it be in the form of comments on the code, questions, or patches. Happy pinging! Jens

1 0

series
by Luis Pablo Michelena 16 Aug '01

16 Aug '01

hello, i just want to ask a simple question: does somebody have or know where to find a haskell program that calculates the number e, that is the list of infinite digits? Because i think it may be possible to do it, but i haven´t find the way to do it. what i am looking for is something like the ertostenes sifts, that prints every prime number until it run out of memory... thank you, Luis

4 3

How do I export Haskell-functions to C under Linux or Unix?
by Robbert de Haan 15 Aug '01

15 Aug '01

Hi, I'm currently working on a project where it would save me a lot of time if I could use the FFI's 'foreign export'-statement, enabling me to call Haskell-functions from C. Unfortunately, I can't get it to work... After reading about various tools and compilers and experimenting with some, I came to the conclusion that the most (and perhaps only) viable option would be to use HDirect in combination with the Glasgow Haskell Compiler. But even the most simple example, exporting Int -> Int-functions, (which doesn't even appear to use the HDirect library) just won't work. When using ghc-4.08.2 everything compiles and links just fine, but when running the result I get an 'Illegal instruction' error (RedHat Linux 7.1) or a 'Segmentation Fault' under Solaris 8. When using ghc-5.00.2 the object files are created, but I get the following error when linking (using ghc-5.00.2 -fglasgow-exts -static -no-hs-main -i../../lib -L../../lse -lhdirect -o tst tst.o libHStst.a, see the HDirect server-example): Undefined first referenced symbol in file Main_main_closure /home/rdehaan/ghc/lib/ghc-5.00.2/libHSstd.a(PrelMain__2.o) __init_Main /home/rdehaan/ghc/lib/ghc-5.00.2/libHSstd.a(PrelMain__1.o) ld: fatal: Symbol referencing errors. No output written to tst Has anyone else succeeded in exporting Haskell-functions and using them in C-modules under Linux or Unix? Any help would be appreciated! Robbert.

1 0

Re: series
by John Hughes 15 Aug '01

15 Aug '01

hello, i just want to ask a simple question: does somebody have or know = where to find a haskell program that calculates the number e, that is = the list of infinite digits? Because i think it may be possible to do = it, but i haven=B4t find the way to do it. what i am looking for is something like the ertostenes sifts, that = prints every prime number until it run out of memory... thank you, Luis It's a nice problem, which I encountered many years ago as one of the first examples I saw of lazy evaluation! Ah dear, I managed to make an off-by-one error in the code! "2.2718..." indeed! A corrected version is below. John ------------------------------------------------------------------------ Here's a way to compute all the digits of e. We use the series e = 2 + 1 + 1 + 1 + ... -- -- -- 2! 3! 4! which we can think of as representing e as 2.11111... in a strange number system with a varying base. In this number system, the fraction 0.abcd... represents a + b + c + d + ... -- -- -- -- 2! 3! 4! 5! To convert such a fraction to decimal, we multiply by 10, take the integer part for the next digit, and continue with the fractional part. Multiplying by 10 is easy: we just multiply each "digit" by 10, and then propagate carries. The hard part is knowing how far carries might propagate: since we carry leftwards in an infinite expansion, we must be careful to avoid needing to inspect the entire fraction in order to decide on the first carry. But each fraction we work with is less than one, so after multiplying by 10, it is less than 10. The "carry out" from each digit can be at most 9, therefore. So if a carry of 9 from the next digit would not affect the carry out from the current one, then that carry out can be emitted immediately. Since the base soon becomes much larger than 10, then this is likely to happen quickly. No doubt there are much better ways than this of solving the problem, but this one works. > carryPropagate base (d:ds) > | carryguess == (d+9) `div` base > = carryguess : (remainder+nextcarry) : fraction > | otherwise > = (dCorrected `div` base) : (dCorrected `mod` base) : fraction > where carryguess = d `div` base > remainder = d `mod` base > nextcarry:fraction = carryPropagate (base+1) ds > dCorrected = d + nextcarry > e = ("2."++) $ > tail . concat $ > map (show.head) $ > iterate (carryPropagate 2 . map (10*) . tail) $ > 2:[1,1..]

2 1

Re: series
by John Hughes 15 Aug '01

15 Aug '01

hello, i just want to ask a simple question: does somebody have or know = where to find a haskell program that calculates the number e, that is = the list of infinite digits? Because i think it may be possible to do = it, but i haven=B4t find the way to do it. what i am looking for is something like the ertostenes sifts, that = prints every prime number until it run out of memory... thank you, Luis It's a nice problem, which I encountered many years ago as one of the first examples I saw of lazy evaluation! The point is, of course, to produce the infinite stream of digits of e itself, not some approximation to it, as a couple of solutions posted earlier do. I'm attaching my solution (a literate Haskell script) which produces about 250 digits before hugs runs out of memory. John Hughes ----------------------------------------------------------------------------- Here's a way to compute all the digits of e. We use the series e = 2 + 1 + 1 + 1 + ... -- -- -- 2! 3! 4! which we can think of as representing e as 2.11111... in a strange number system with a varying base. In this number system, the fraction 0.abcd... represents a + b + c + d + ... -- -- -- -- 2! 3! 4! 5! To convert such a fraction to decimal, we multiply by 10, take the integer part for the next digit, and continue with the fractional part. Multiplying by 10 is easy: we just multiply each "digit" by 10, and then propagate carries. The hard part is knowing how far carries might propagate: since we carry leftwards in an infinite expansion, we must be careful to avoid needing to inspect the entire fraction in order to decide on the first carry. But each fraction we work with is less than one, so after multiplying by 10, it is less than 10. The "carry out" from each digit can be at most 9, therefore. So if a carry of 9 from the next digit would not affect the carry out from the current one, then that carry out can be emitted immediately. Since the base soon becomes much larger than 10, then this is likely to happen quickly. No doubt there are much better ways than this of solving the problem, but this one works. > carryPropagate base (d:ds) > | carryguess == (d+9) `div` base > = carryguess : (remainder+nextcarry) : fraction > | otherwise > = (dCorrected `div` base) : (dCorrected `mod` base) : fraction > where carryguess = d `div` base > remainder = d `mod` base > nextcarry:fraction = carryPropagate (base+1) ds > dCorrected = d + nextcarry > e = ("2."++) $ > concat $ > map (show.head) $ > iterate (carryPropagate 2 . map (10*) . tail) $ > 2:[1,1..]

1 0

series. Reply
by S.D.Mechveliani 15 Aug '01

15 Aug '01

Luis Pablo Michelena <lmichele(a)multi.com.uy> writes on the subject `series' > where to find a haskell program that calculates the number e, that is > the list of infinite digits? > [..] > what i am looking for is something like the ertostenes sifts, that > prints every prime number until it run out of memory ... In what way the Heratosphenes sieve for prime numbers may relate to finding approximations of number e ? As to finding the infinite list of digits for e = lim (1 + 1/n)^n, n -> infinity here the program is suggested for finding (eAppr n) :: Rational such that |e - (eAppr n)| < 1/2^(n-3) : import Ratio eAppr :: Integer -> Rational eAppr n = appr 0 (1%1) (0%1) where appr k member res = --- member = 1/(k!), if --- res = sum [1/i! | i <- [0..k]] k==n then res else appr (k+1) (member/(fromInteger (k+1))) (res+member) Several decimal digits can be obtained, then, like this: fromRational (eAppr 6) :: Double --> 2.716666666666667 fromRational (eAppr 20) :: Double --> 2.718281828459045 Explanation. According to Calculus, we have e = lim (eApp n), where eAppr n = sum [1/k! | k <- [0..n]] n -> infinity and |e - (eAppr n)| < 3/n! <= 1/2^(n-3). Therefore, eAppr(n+3) differs from e in less than 1/2^n. I believe, this fact will help us to find first true k digits of e for any given k. If people would not give other good solution and if you ask me to complete this task, then I'll try to do this. Regards, ----------------- Serge Mechveliani mechvel(a)botik.ru

1 0

Haskell workshop: 10-minute slots
by Ralf Hinze 14 Aug '01

14 Aug '01

Folks, The Haskell Workshop in Firenze on 2nd September is now only a couple of weeks away. If you'd like to attend but haven't yet registered, please do so as soon as possible. For further details, see: http://music.dsi.unifi.it/pli01/registration/index.html The workshop programme is available here (including the proceedings): http://www.cs.uu.nl/~ralf/hw2001.html In addition to longer presentations on accepted papers, a number of slots for participants to make 10-minute informal presentations are available. If you'd like to request such a slot, please drop me an email. Note: the talk need not be polished, the wackier the better. Cheers, Ralf

1 0