- Haskell - Haskell.org

Why are strings linked lists?
by Ben Escoto 30 Nov '03

30 Nov '03

Hi, can someone tell me why Haskell strings are linked lists? I have had some problems with Haskell strings: 1. Today I spend a few hours trying to track down a memory leak. It turns out I just didn't realize how much space a string takes up. On my machine "replicate 5000000 'a'" will use 90MB of space! 2. They are extremely slow for most operations like writing to disk, adding something to the end, concatenation, etc. 3. They make learning Haskell harder. Lazy IO functions like hGetContents are kind of slick in a shallow way, but I was confused about the Haskell IO model because I thought of this as the normal way IO was done, not something you could only set up through unsafeInterleaveIO or similar. Python's strings are also immutable, but they work great and are really convenient. They are apparently implemented as variable length arrays. -- Ben Escoto

20 33

Incomplete output when hPutStr is lifted
by andrew cooke 29 Nov '03

29 Nov '03

Hi, If I compile and run the code below, the file "foo" contains 10 lines of output (as I would expect), but the file "bar" contains just 9 - the final line is missing. If I add a "join", as in the comment, then all 10 lines appear. I don't understand why, completely - my best guess is that it's related laziness and the final result of the fold not being used. However, in that case, I'm worried that the "join fix" isn't guaranteed to work. Please could someone explain what is happening and how I should guarantee that the output is "complete". Thanks, Andrew $ cat test.hs main :: IO() main = do h <- openFile "foo" WriteMode foldM (printer h) 1 [1..10] hClose h h <- openFile "bar" WriteMode foldM (liftM2 (printer h)) (return 1) $ map return [1..10] -- join $ foldM (liftM2 (printer h)) (return 1) $ map return [1..10] hClose h printer :: Handle -> Int -> Int -> IO Int printer h n i = do hPutStr h (show n) hPutStr h ":" hPutStr h (show i) hPutStr h "\n" return (n+1) $ ghc test.hs $ ./a.out $ diff foo bar 10d9 < 10:10 -- personal web site: http://www.acooke.org/andrew personal mail list: http://www.acooke.org/andrew/compute.html

2 2

System.Posix (symbolic links)
by Johannes Goetz 29 Nov '03

29 Nov '03

Hi! Sorry for posting this message twice. Last message had wrong subject. Calling isSymbolicLink always returns False... (ghc-6.0.1linux binary tarball) It doesn't make a difference whether the symbolic link points to a regular file or a directory. Test code: #ln -s test link #ghc Test.hs -o test #./test False # Test.hs: module Main(main) where import System.Posix main = do status <- getFileStatus "link" print (isSymbolicLink status) Johannes

2 1

RE: How to best add logging/debugging code?
by Simon Marlow 28 Nov '03

28 Nov '03

> Manuel M T Chakravarty wrote: > > > There is `Debug.Trace.trace' for this: > > > > http://haskell.org/ghc/docs/latest/html/base/Debug.Trace.html > > > > However, if you want to log as opposed to debug, you may > > want to have the output go to somewhere else but stdout. > > Hence, it might be useful to have a variant of the `trace' > > that takes an explicit file handle in the library. > > Is there a list of problems anywhere with using trace? For > example does it affect evaluation order? You can think of the meaning of trace like this: trace e1 e2 <=> foldr seq () e1 `seq` e2 in other words, it deepSeqs the first argument and then behaves as the second argument. If you insert a trace with a literal string into your program, eg. trace "calling f" (f x) then it has no effect on the meaning of the program. Trace will only have an effect on the semantics of the program when you print out a value in the program: trace ("calling f with " ++ show x) (f x) in which case it will cause the value (x in this case) to be evaluated earlier than it would otherwise. If x turned out to be _|_, then the program might fail when it otherwise might not have. Cheers, Simon

3 2

Functional dependencies interfere with generalization
by Ken Shan 27 Nov '03

27 Nov '03

Hello, Consider the following code, which uses type classes with functional dependencies: {-# OPTIONS -fglasgow-exts #-} module Foo where class R a b | a -> b where r :: a -> b -- 1 rr :: (R a b1, R a b2) => a -> (b1, b2) rr a = (r a, r a) -- 2 data RAB a = RAB (forall b. (R a b) => (a, b)) mkRAB :: (R a b) => a -> b -> RAB a mkRAB a b = RAB (a, b) Neither 1 nor 2 passes the type-checker (GHC 6.0.1). The error messages are similar: Inferred type is less polymorphic than expected Quantified type variable `b2' is unified with another quantified type variable `b1' When trying to generalise the type inferred for `rr' Signature type: forall a b1 b2. (R a b1, R a b2) => a -> (b1, b2) Type to generalise: a -> (b1, b1) When checking the type signature for `rr' When generalising the type(s) for `rr' Inferred type is less polymorphic than expected Quantified type variable `b' escapes It is mentioned in the environment: b :: b (bound at Foo.hs:17) In the first argument of `RAB', namely `(a, b)' In the definition of `mkRAB': mkRAB a b = RAB (a, b) In both cases, the compiler is failing to make use of functional dependencies information that it has at its disposal. Specifically, it seems to me that, if two type variables b1 and b2 have been unified due to functional dependencies, making two constraints in the context identical, then the inner constraint ("inner" with respect to the scope of quantified type variables) should be ignored. Is there a technical reason why the type checker should reject the code above? Would it be possible to at least automatically define a function like equal :: forall f a b1 b1. (R a b1, R a b2) => f b1 -> f b2 for every functional dependency, with which I would be able to persuade the type checker to generalize (Baars and Swierstra, ICFP 2002)? I suppose I can use unsafeCoerce to manually define such a function... is that a bad idea for some reason I don't see? Thank you, Ken -- Edit this signature at http://www.digitas.harvard.edu/cgi-bin/ken/sig Tax the rich! new journal Physical Biology: http://physbio.iop.org/ What if All Chemists Went on Strike? (science fiction): http://www.iupac.org/publications/ci/2003/2506/iw3_letters.html

2 2

Test
by onemind555 27 Nov '03

27 Nov '03

....

1 0

a type question
by rui yang 27 Nov '03

27 Nov '03

Suppose I have a function: funcmap :: a->b->c can I use type synonyms to describe a new type like this: Type Funcmap = a->b>c ? it will give a syntax error message: syntax error in input (unexpected '->'), does anyone know how to define this Funcmap type? rui ---------------------------------------- This mail sent through www.mywaterloo.ca

3 3

How to best add logging/debugging code?
by Ben Escoto 26 Nov '03

26 Nov '03

Hi all, does anyone have any tips on how to insert debugging or logging statements through a program? Here are two possibilities: 1. Use some kind of logging monad. But then all your types become more complicated and it seems you have to use lots of ->'s and >>='s instead of ordinary function application. It seems this can make the code much less elegant. Also it seems that every time you want to add logging to a function, you have to change its type, and then change the type of all the functions that call it, etc. Should it really be this complicated? 2. Use unsafePerformIO or similar. But then your code is regularly defeating the type system, which presumably isn't good. Also the ghc manual has special instructions for compiling code that uses unsafePerformIO, intimating that it isn't used very often. It seems this must come up a lot when writing Haskell programs. What do most pople do? -- Ben Escoto

7 8

[JFPLC'04]First Call for Papers
by jfplc2004 25 Nov '03

25 Nov '03

--------------------------------------------------------------- Call for Papers -- JFPLC'2004 Thirteenth French Conference on Logic and Constraint Programming http://www.info.univ-angers.fr/jfplc2004/ Angers, France, June 21-23, 2004 --------------------------------------------------------------- The thirteenth French Conference on Logic and Constraint Programming of the AFPLC (http://www.afplc.org) JFPLC'2004 is co-localized with JNPC'04 (http://www.info.univ-angers.fr/jnpc2004/appelJNPC.html) JFPLC'2004 follows the series of previous meetings which took place in Amiens (2003), Nice (2002), Paris (2001), Marseille (2000), Lyon (1999), Nantes (1998), Orleans (1997), Clermont-Ferrand (1996), Dijon (1995), Bordeaux (1994), Nimes (1993) and Lille (1992). The aim of the meeting is to review recent research works on logic and constraint programming including theoretical aspects, applications, and teaching issues. --------------------------------------------------------------- Important dates (hard deadlines): Abstract February 22, 2004 Submission February 29, 2004 Acceptation April 5, 2004 Final Version April 18, 2004 The proceedings will be published by Hermes. Official conference languages are French and English. Papers may be submitted in English, but should be published in French if one of the authors speaks French fluently. Simultaneous submissions to JFPLC and JNPC is not allowed. Scope : Papers are welcome on all theoretical, experimental and pedagogical aspects of logic programming and/or constraint programming. Submissions : Authors are invited to send papers in PDF or Postscript format not exceeding 15 pages, formatted according to the Hermes style (see http://www.editions-hermes.fr/fr/cons_actes.html) Papers should contain original, previously unpublished results (neither in a conference nor in a journal). Accepted papers should be presented at the conference. Electronic submissions should be made before: - February 22, 2004 as a standard text email including the title, the authors, an abstract, its category (theory, application, pedagogy) - February 29, 2004 as a standard postscript or PDF file, together with the possibly updated data described above and send to Frederic.Mesnard(a)univ-reunion.fr. Please ensure that the subject of your email is [jfplc04]. ------------------------------------------------------------------- Program Chair: Fred Mesnard, IREMIA, universite de La Reunion Program Committee: Abder Aggoun, COSYTEC Roberto Bagnara, Universita degli Studi di Parma Pascal Brisset, ENAC Veronica Dahl, université Simon Fraser Pierre Deransart, INRIA Rocquencourt Yves Deville, université catholique de Louvain François Fages, INRIA Rocquencourt Carmen Gervet, IC Parc Isabelle Gnaedig, LORIA Arnaud Gotlieb, IRISA Jin-Kao Hao, université d'Angers Narendra Jussien, École des mines de Nantes François Laburthe, Bouygues Arnaud Lallouet, université d'Orléans Juliette Mattioli, Thales Eric Monfroy, université de Nantes Catuscia Palamidessi, INRIA Futurs Laurent Perron, ILOG Andreas Podelski, Max-Planck-Institut für Informatik Alexander Serebrenik, Katholieke Universiteit Leuven Christine Solnon, université de Lyon ---------------------------------------------------------------

1 0

ghc, FFI and autotools
by Lloyd Allison 25 Nov '03

25 Nov '03

We are trying to build and call some Haskell code in and from from the open-src spreadsheet "gn'meric". It seems that it could be a good basis for experiments in FP + SSs. We have worked through the simple FFI examples and can create small C+Haskell stand-alone programs. However, gnu'eric is quite a large program and autoconf|automake|...|libtools are used to build it. Consequently one does not have an entirely free hand to just compile and link the pieces in any old way. It has to fit in with the general build scheme. There is a general pattern to 'eric's "plugins": add a directory, e.g. ...../plugins/haskell/Makefile.am --->Mafilefile.in--->Makefile /functions.c -- the plugin's code /.... and we can make C-only plugins. The next step, or "impasse" to give it the technical name, is to add /functionsH.hs -- Haskell called from functions.c ? Please has anybody used ghc in this type of environment before, and ? have advice on the exercise in general or on ? creating a *shared library* from C + Haskell code in particular ? We[*] have tried a few things, so far with less than complete success, and often featuring undefined labels,-( e.g. E Unable to open module file "/home/lloyd/gnumeric/install//lib/gnumeric/1.2.1-bonobo/plugins/haskell/plugin.la". E /home/lloyd/gnumeric/install//lib/gnumeric/1.2.1-bonobo/plugins/haskell/plugin.so: undefined symbol: GHCziInt_I8zh_con_info Lloyd -- Lloyd ALLISON, CSSE, Monash University, Victoria, Australia 3800. web: http://www.csse.monash.edu.au/~lloyd/ tel: +61 3 9905 5205 use: http://www.linux.org/ OpenOffice: http://www.openoffice.org/ PS. [*] Not a royal we.

3 3