- Haskell-Cafe - Haskell.org

Re: [Haskell-cafe] Who generates Haskell code and uses type information at runtime? (fwd)
by Chris Brown 18 Mar '09

18 Mar '09

I'm not sure if you got my previous message, as I was having some problems posting to the list... > Putting in a nutshell, I generalize an extensional defined function > definition into a recursive one. This is done in a number of steps by > modifying expressions and exploiting type information of > sub-expressions. For example: > > rev [] = [] > rev [a] = [a] > rev [a,b] = [b,a] > rev [a,b,c] = [c,b,a] > > ~~> > > rev x = y > > ~~> > > rev [] = [] > rev (x:xs) = (y:ys) > > ~~> > > rev [] = [] > rev (x:xs) = (last (x:xs)) : (reverse (x:xs)) > > > The initial set of rules is given by the user (in a file, via IO, ...). > The problem later is to infer the type of an intermediate variable, e.g. > 'y'. I'm still not entirely sure what you want to do here. But it sounds like HaRe could already do most of this for you via a sequence of folds, unfolds, introduce pattern matching and generative folding. HaRe already has built-in support for some symbolic evaluation, which is already used in the generative fold refactoring, and has type checking support too. http://www.cs.kent.ac.uk/projects/refactor-fp/hare.html If it doesn't do exactly what you want out of the tin, it does have a large API for designing transformations over Haskell code. http://www.cs.kent.ac.uk/projects/refactor-fp/hare/haddock/RefacUtils.html Chris. > > Thanks, > > Martin > > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe > -- Chris Brown Visualization Software Engineer, Peiriannydd Meddalwedd Delweddu. Cast Ltd., Technium CAST, Ffordd Penlan, Parc Menai, Bangor, Gwynedd UK. LL57 4HJ. Tel: +44 (0)1248 675038 Fax: +44 (0)1248 675012 Mobile: +44 (0)7917 763712 -- Centre for Advanced Software Technology Limited is a limited company registered in England and Wales. Registered Number: 04473521, Registered Office: Finance Office, Bangor University, College Road, Bangor, Gwynedd. LL57 2DG.

2 1

encoding for least fixpoint
by ben 18 Mar '09

18 Mar '09

Hello, I am trying to understand the definition of (co)inductive types in Haskell. After reading the beginning of Vene's thesis [1] I was happy, because I understood the definition of the least fixpoint: newtype Mu f = In (f (Mu f)). But this definition coincides with his definition of the greatest fixpoint newtype Nu f = Wrap (f (Nu f)), which in reality is not true (e. g. for f x = a * x -- the 'stream functor' the type Mu f should be empty, isn't it?). Then I stumbled over a blog entry of Shin-Cheng Mu [2] and from there over an article of Wadler [3], where the least fixpoint is encoded as Lfix X. F X = All X. (F X -> X) -> X. and the greatest fixpoint as Gfix X. F X = Exists X. (X -> F X) * X. I would like to understand these definitions, or get an intuition about their meaning. Could somebody give some explanation or a pointer to a place where I can find one? Thanks a lot in advance, ben [1] http://www.cs.ut.ee/~varmo/papers/thesis.pdf [2] http://www.iis.sinica.edu.tw/~scm/2007/encoding-inductive-and-coinductive-t… [3]http://homepages.inf.ed.ac.uk/wadler/papers/free-rectypes/free-rectypes.txt

7 11

Need an overview of FP-related research topics
by Yitzchak Gale 18 Mar '09

18 Mar '09

I spoke to a faculty member in a decent Computer Science department in which no one has ever done anything related to FP. (You may say that is an inherent contradiction, but what can I do, the department does have a good reputation. I am withholding names to protect the innocent.) This faculty member happens to be the one who was forced to teach the single programming languages course offered by the department, due to his having the least seniority in the department. As such, he is now actually starting to become a bit interested in FP and Haskell. I gave him a few links to Haskell resources, which he appreciated. But coming from more of a pragmatic non-research point of view, I didn't really know what to say. The best I could do was to point him to the Haskell Wikibook, and to the "Why FP Matters" paper. I would like some links that would give such a person a nice overview of the various active areas of FP-related research these days, leaning towards Haskell. I want to give him a fairly broad view of what is interesting and exciting, why various topics are important, where to find ideas for collaboration and applications to other areas, etc. I actually know about a few departments like that. This could be a good strategy for drumming up more research interest in Haskell. In some cases, the person in question has already been influenced somewhat by the lisp fanatics, so I would like some help in how to deal with that also. Thanks, Yitz

4 4

libgmp for GHC 6.10.1 on Mac OS X 10.5
by Dean Herington 18 Mar '09

18 Mar '09

I'm trying to install GHC 6.10.1 on Mac OS X 10.5 (PowerPC). I installed Xcode 3.1.2. I built libgmp 4.2.4 and installed it in /usr/local/lib. When I do "./configure" in GHC's dist directory, however, I get: bash-3.2$ ./configure checking build system type... powerpc-apple-darwin9.6.0 checking host system type... powerpc-apple-darwin9.6.0 checking target system type... powerpc-apple-darwin9.6.0 Which we'll further canonicalise into: powerpc-apple-darwin checking for path to top of build tree... dyld: Library not loaded: /usr/local/lib/libgmp.3.dylib Referenced from: /Users/family/Desktop/Downloads/ghc-6.10.1/dist/utils/pwd/pwd Reason: no suitable image found. Did find: /usr/local/lib/libgmp.3.dylib: mach-o, but wrong architecture /usr/local/lib/libgmp.3.dylib: mach-o, but wrong architecture configure: error: cannot determine current directory Any ideas what I'm doing wrong? Thanks.

3 4

UNPACK multiple constructors
by Louis Wasserman 18 Mar '09

18 Mar '09

{-# UNPACK #-} is specified to work on single-constructor types, but would there be advantages to, for instance, compiling data BinTree e = Node e {-# UNPACK #-} !(Maybe (BinTree e)) {-# UNPACK #-} !(Maybe (BinTree e)) into four constructors, one for each combination of Maybe constructors? Thoughts? Louis Wasserman wasserman.louis(a)gmail.com

1 0

symbolic evaluator for Haskell?
by Tim Newsham 18 Mar '09

18 Mar '09

Is there a symbolic evaluator for Haskell that will perform all applications except on specified functions? Ie. I would love something that would take foldr (+) (6 `div` 5) [1,2,3*4] and "(+) (*)" and return 1 + (2 + (3*4 + 1)) by performing all the applications except for (+) and (*). (Something that supports ghc extensions is preferred :) Tim Newsham http://www.thenewsh.com/~newsham/

4 4

Re: [Haskell-cafe] OpenGL and Cabal installation
by Henk-Jan van Tuyl 18 Mar '09

18 Mar '09

On Wed, 18 Mar 2009 04:34:29 +0100, Mark Spezzano <mark.spezzano(a)chariot.net.au> wrote: > > configure: error: no GLUT header found, so this package cannot be built > > See `config.log' for more details. > > > Where is it looking for these glut.h files? I’ve tried putting them > everywhere in my PATH but msys won’t find them. Add an environment variable C_INCLUDE_PATH and assign it the value of the directory where you keep the header files; for instance, if the glut.h file is in: C:\usr\local\include\GLUT then do Set C_INCLUDE_PATH=C:\usr\local\include before compiling, or set this variable in Windows with System properties. Similarly, set LIBRARY_PATH for the libraries to link. -- Regards, Henk-Jan van Tuyl -- http://functor.bamikanarie.com http://Van.Tuyl.eu/ --

1 0

Re: [Haskell-cafe] Re: Query on list comprehension
by Kalman Noel 18 Mar '09

18 Mar '09

Jon Fairbairn schrieb: > Melanie_Green <jac_legend_sas(a)hotmail.com> writes: >> What are the limitations of list comprehension. [...] >> a >> aa >> aaa > > I'm not clear what you mean by the question. Why do you want > to use list comprehensions? What if they aren't the best way > of getting the result you want? > > You can write > > [a | b <- [replicate n 'a' | n <- [1..]], a <- b ++ "\n"] > > but does that "replicate" fail to meet your specification? Since we're »holfing« again: fix $ \xs -> [ 'a':xs | xs <- []:xs ]

1 0

ANN: regex-tdfa-1.1.0
by ChrisK 18 Mar '09

18 Mar '09

I have just uploaded the new regex-tdfa-1.1.0 to hackage. This version is a small performance update to the old regex-tdfa-1.0.0 version. Previously all text (e.g. ByteString) being search was converted to String and sent through a single engine. The new version uses a type class and SPECIALIZE pragmas to avoid converting to String. This should make adding support for searching other Char containers easy to do. The new version includes six specialized engine loops to take advantage of obvious optimizations of the traversal. The previous version had only a couple of such engines. The new code paths have been tested for correctness and no performance degradations have shown up. -- Chris

1 0

Re: [Haskell-cafe] Design Patterns by Gamma or equivalent
by wren ng thornton 18 Mar '09

18 Mar '09

Mark Spezzano wrote: > Because Haskell is not OO, it is functional, I was wondering if there is > some kind of analogous “design pattern”/”template” type concept that > describe commonly used functions that can be “factored out” in a general > sense to provide the same kind of usefulness that Design Patterns do for > OOP. Basically I’m asking if there are any kinds of “common denominator” > function compositions that are used again and again to solve problems. If > so, what are they called? Most of the (particular) problems OO design patterns solve are non-issues in Haskell because the language is more expressive. The issues giving rise to patterns like Visitor and Factory are non-issues, so their "solutions" are trivial. A number of other patterns can actually be written down once and for all (in higher-order functions like foldr, map,...) instead of needing repetition. But that's not to say we don't have our own expressiveness deficiencies ;) The analogous term for the genre is "functional pearl", though the individual pearls don't tend to be as codified as in OOP. One example is using coproducts for open unions[1] which solves the dual problem to Visitor. Another is using open recursion[2]. A third example along the same track is using two-level types[3]. But again a lot of the patterns, once discovered, get turned into libraries that can be used off the shelf: e.g. the two-continuation solution for efficient and fair backtracking search[4], and list-fusion techniques[5]. There also a number of "idioms" which are similar in scope to the idioms that arise in other languages: using tail recursion, accumulators, continuation-passing transformations, closures over recursion[6], Schwartzian transforms, etc. And then there are some things like monoids which fall somewhere between idiom and pearl. Some specific OO patterns have direct analogues in "defunctionalization"[7]. For the most part defunctionalization is something best left to the compiler, but on occasion we want to be explicit about it and this too is an idiom/pearl border case IMO. And, of course, there's the entire cottage industry of using category theory for fun and profit. [1] http://www.cs.nott.ac.uk/~wss/Publications/DataTypesALaCarte.pdf [2] http://www.cs.utexas.edu/~wcook/Drafts/2006/MemoMixins.pdf [3a] http://web.cecs.pdx.edu/~sheard/papers/generic.ps [3b] http://web.cecs.pdx.edu/~sheard/papers/JfpPearl.ps [4a] http://okmij.org/ftp/papers/LogicT.pdf [4b] http://hackage.haskell.org/cgi-bin/hackage-scripts/package/logict [5a] http://www.cse.unsw.edu.au/~dons/papers/CSL06.html [5b] http://hackage.haskell.org/cgi-bin/hackage-scripts/package/bytestring [5c] http://www.cse.unsw.edu.au/~dons/papers/CLS07.html [5d] http://hackage.haskell.org/cgi-bin/hackage-scripts/package/stream-fusion [5e] http://homepages.inf.ed.ac.uk/wadler/papers/vanish/vanish.pdf [6] For lack of a better name. I mean doing this: > -- Here the a,b,c,d variables are captured in the closure for go > recursive a b c d = go > where > go 0 = ...a...b... > go n = ...c...d...go (n-1) instead of this: > -- ...instead of needing to be passed through the recursion each time > recursive a b c d 0 = ...a...b... > recursive a b c d n = ...c...d...recursive a b c d (n-1) [7a] http://blog.plover.com/prog/defunctionalization.html -- Live well, ~wren

8 13