June 2010 - Haskell-Cafe

Haddock Problem
by Dominic Steinitz 24 Jun '10

24 Jun '10

..\ThirdParty\Haskell_Platform\2010.1.0.0\bin\haddock.exe BackendC\Core.hs haddock.exe: can't find a package database at E:\ghc\ghc-6.12.1lib\package.conf.d But if I do haddock --help there is no option to set the package database and I don't even have an E: drive. I'm on windows in case that wasn't obvious and we have a custom package database which we point e.g. ghc and cabal at. Can someone tell me what I am doing wrong? Many thanks, Dominic.

4 14

Re: Installing Haskell on OSX
by Giuseppe Luigi Punzi 24 Jun '10

24 Jun '10

Sorry for my ignorance but..why install with +universal? i didn't use this in any port. Sudish Joseph <sudish(a)gmail.com> wrote: >Hi, > >I ran into one problem worth mentioning while recompiling gtk and its >dependencies as universal libraries using macports: the pango port >refused to build with +universal (it fails when trying to merge the >i386 & x86_64 libs). The pango-devel port does work with one small >tweak as described at https://trac.macports.org/ticket/22801. > >Essentially, I had to comment out the "PortGroup muniversal 1.0" line >in the Portfile to get a working universal library for pango. > >-Sudish > >On Mon, Jun 21, 2010 at 5:09 AM, Giuseppe Luigi Punzi ><glpunzi(a)lordzealon.com> wrote: >> I died in more dependencies. One, ige-mac-integration. To avoid possible problems I uninstall all ports last night to reinstall full gtk2 with quartz. >> >> For now, i'm solving all of this and I hope to get it working this night. >> >> Sudish Joseph <sudish(a)gmail.com> wrote: >> >>>Antoine Latter <aslatter(a)gmail.com> writes: >>>> On Sun, Jun 20, 2010 at 12:53 PM, Giuseppe Luigi Punzi Ruiz >>>> <glpunzi(a)lordzealon.com> wrote: >>>>> Hi again, >>>>> >>>>> Yes, you are right, but now, "cabal install leksah" I get: >>>>> >>> >>>[...] >>> >>>>> Undefined symbols: >>>>> "_iconv_close", referenced from: >>>>> _hs_iconv_close in libHSbase-4.2.0.0.a(iconv.o) >>>>> "_iconv_open", referenced from: >>>>> _hs_iconv_open in libHSbase-4.2.0.0.a(iconv.o) >>>>> "_iconv", referenced from: >>>>> _hs_iconv in libHSbase-4.2.0.0.a(iconv.o) >>>>> ld: symbol(s) not found >>> >>>> This one is a bummer, and I see it all the time when I try to build a >>>> package linked against macports. >>> >>>This is caused by the two libiconv's in the system being >>>ABI-incompatible, sadly. The ghc pkg available for download from >>>haskell.org is linked against the system /usr/lib/libiconv.dynlib, which >>>has iconv_open() as a function. Macports has GNU libiconv which >>>#defines iconv_open to libiconv_open() in /opt/local/include/iconv.h. >>> >>>This then blows up as above when linking against other libraries in >>>macports - the linker pulls in GNU libiconv which lacks the symbols >>>needed as you see above. >>> >>>One workaround is to link GHC itself against the macports version of >>>libiconv (and libgmp) and have cabal-install link all subsequent >>>libraries against macports. I did just that this weekend and now have a >>>working threadscope using the Quartz backend for gtk2, which is very >>>nice (no need to run the X server). >>> >>>Recipe for reproducing this build is included below (and I can also >>>provide the resulting ghc-6.12.3 pkg file if needed). >>> >>>Another possible option is to use Homebrew to install gtk2 and other >>>dependencies. Homebrew prefers to use the system-provided libraries and >>>will not pull in GNU libiconv, so this should work with the existing GHC >>>pkg in theory. I didn't pursue this since the homebrew 'gtk+' package >>>didn't seem to have the option to use Quartz instead of X11 as its >>>backend. >>> >>>Steps for linking the ghc runtime against macports: >>> >>>- Specify EXTRA_CABAL_CONFIGURE_FLAGS in mk/build.mk as mentioned at >>> the bottom of http://www.haskell.org/ghc/download_ghc_6_12_3.html >>> >>> EXTRA_CABAL_CONFIGURE_FLAGS = --extra-include-dirs=/opt/local/include \ >>> --extra-lib-dirs=/opt/local/lib >>> >>>- Use the --with-iconv-* and --with-gmp-* flags when configuring ghc. >>> >>> ./configure --with-iconv-includes=/opt/local/include \ >>> --with-iconv-libraries=/opt/local/lib \ >>> --with-gmp-includes=/opt/local/include \ >>> --with-gmp-libraries=/opt/local/lib >>> >>>- This produces a GHC runtime and libraries linked against macports: >>> >>> % otool -L ghc >>> ghc: >>> /opt/local/lib/libncurses.5.dylib (compatibility version 5.0.0, current version 5.0.0) >>> /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 125.2.0) >>> /opt/local/lib/libiconv.2.dylib (compatibility version 8.0.0, current version 8.0.0) >>> /opt/local/lib/libgmp.10.dylib (compatibility version 11.0.0, current version 11.1.0) >>> >>> % nm HSbase-4.2.0.2.o | fgrep iconv >>> 001ff7f0 T _hs_iconv >>> 001ff7e0 T _hs_iconv_close >>> 001ff800 T _hs_iconv_open >>> U _libiconv >>> U _libiconv_close >>> U _libiconv_open >>> >>> ghc and included libraries use the macports libiconv, so linking >>> against other libraries in macports (gtk2!) will work. >>> >>>- Have cabal-install use macports as well for packages it installs by >>> editing ~/.cabal/config and setting: >>> >>> extra-include-dirs: /opt/local/include >>> extra-lib-dirs: /opt/local/lib >>> >>>- This gives, for e.g., threadscope linked against macports: >>> >>> % otool -L ~/.cabal/bin/threadscope | egrep '(gtk|iconv)' >>> /opt/local/lib/libgtk-quartz-2.0.0.dylib (compatibility version 2001.0.0, current version 2001.1.0) >>> /opt/local/lib/libiconv.2.dylib (compatibility version 8.0.0, current version 8.0.0) >>> >>>- Note that since the ghc runtime is still in 32-bit i386 mode, we >>> need universal versions of most libraries in macports. Recent >>> versions of macports (1.9 for sure, maybe 1.8) make it simple to >>> switch from an x86_64 library to a universal one: >>> >>> % port install gtk2 +universal >>> >>> This will recompile gtk2 and *all* dependent libraries as universal >>> libraries which is exactly what you need. You can then eliminate any >>> inactive 64-bit libraries with: >>> >>> % port -f uninstall inactive >>> >>>> Here's the last thread about with, with more links and discussion: >>>> >>>> http://thread.gmane.org/gmane.comp.lang.haskell.general/18064/ >>>> >>>> The response by Jean-Marie Gaillourdet has worked for me in the past. >>>> >>>> Antoine >>> >>>Thanks for that link, I didn't think of overriding libiconv on a >>>per-package basis. >>> >>>-Sudish >>

3 7

Relating generated asm to Core
by Rami Mukhtar 23 Jun '10

23 Jun '10

Hi, Can anyone tell me a way to identify the generated assembly (as found in the intermediate files produced by GHC) corresponding to a particular fragment of Core code. Thanks, Rami ________________________________ The information in this e-mail may be confidential and subject to legal professional privilege and/or copyright. National ICT Australia Limited accepts no liability for any damage caused by this email or its attachments.

3 2

Accounting Engine in Haskell
by Amiruddin Nagri 23 Jun '10

23 Jun '10

My current project is about making an accounting engine that handles all the journal entries, transactions, portfolios etc. The communication with the engine is based on simple protocol, the things to be taken care of in the order are consistency, handling large data(performance) and availability. I came across a video lecture by Simon Peyton Jones where he gives an example from Financial domain (derivatives etc) to explain how haskell is being used and the advantages provided. I am interested in knowing if Haskell will be the right fit for my project. My requirements are transactional nature, which I believe is one of the strengths of functional programming, also handling large data set and being available. there is no such requirement for partitioning of data and the application is going to be centrally hosted on a single server. AFAIK OCaml and other functional languages are heavily used in financial domain, some of the reason are same as features I am looking for. I wanted some insight as to how Haskell is going to help me with my project. Also there has been some concerns because of lazy evaluation in Haskell and memory leaks associated with it. http://jlouisramblings.blogspot.com/2010/04/haskell-vs-erlang-for-bittorent… Also, if you have any suggestions of the choice of programming language, we have been looking into other functional languages like Scala and Clojure. But we have not dig deep on the performance aspects of these languages, if someone can shed a light on the pros- cons of these languages, it will help us very much to come to a decision. -Regards, Amir

6 6

offtopic - installing archlinux as a domU
by Michael Litchard 23 Jun '10

23 Jun '10

I know some haskell people out there have done what I am trying to do, and I hope you can help me out. I've got access to a debian server which is running a Xen Hypervisor 3.0.3 I believe. I would like to install archlinux as a paravirtual machine (or HVM if I must) in order to get a more supportive haskell environment than my current OS. Could someone point me to some updated documentation on this process? Much thanks. Michael Litchard

1 0

Haskell Weekly News?
by aditya siram 23 Jun '10

23 Jun '10

Hi all, I haven't seen a "Haskell Weekly News" in a while. Is there anything I can do to help? -deech

9 16

In other words is there a good place to post stuff like the following?
by caseyh＠istar.ca 23 Jun '10

23 Jun '10

-- Algorithms From: "Selected Papers on Design of Algorithms", Donald Knuth, 2010 -- Chapter 10 Addition Machines -- Haskell version by Casey Hawthorne -- Note this is only a skeleton of the chapter, -- so as to wet your interest to buy the book. -- Addition Machine -- The only operations allowed are the following: -- read x -- x <- y -- x <- x + y -- x <- x - y -- if x >= y -- write x -- Can all functional versions be tail recursive? ----------------------------------------------------------------------- ----------------------------------------------------------------------- -- Modulus -- Assuming without loss of generality x>=0 and y>0, -- since one can use various identities for other cases. -- Performs floor(x/y) subtractions. modulusNaive x y | x >= y = modulusNaive (x-y) y | otherwise = x -- Can we do better? -- Uses a doubling procedure to subtract larger multiplies of y. -- Bounded by O(log(x/y))^2 modulusByDoubling x y | x >= y = helper y y | otherwise = x where helper w z | x < z = helper z z+z | otherwise = modulusByDoubling (x-w) y -- Can we do better? -- Want bounded by O(log(x/y)). -- Addition Machine, so cannot divide by 2. -- Implicitly use the Fibonacci representation of floor(x/y) -- instead of its the binary representation. -- F0 = 0; F1 = 1; Fn = F(n-1) + F(n-2), n >=2 -- Every nonnegative integer n can be uniquely represented in the form -- n = F(m1) + F(m2) + ... + F(mt), m1 >> m2 >> ... >> mt >> 0 -- where t >= 0 and m >> m' means that m - m' >= 2 -- If n > 0 this representation can be found by choosing m1 such that -- F(m1) <= n < F(m1+1) ----------------------------------------------------------------------- -- Furthermore Fibonacci numbers grow exponentially, -- about 69% as fast as powers of 2. -- They have been used as power-of-2 analogs in a variety of algorithms. ----------------------------------------------------------------------- modulusFib x y | x >= y = helper x y y | otherwise = x where helper x y z | x < z = helper x z y+z | otherwise = helper2 x y z helper2 x y z | x >= y = helper2 (x-y) y z | y > z = helper2 x (z-y) y | otherwise = x

2 1

In other words is there a good place to post stuff like the following?
by caseyh＠istar.ca 23 Jun '10

23 Jun '10

-- Algorithms From: "Selected Papers on Design of Algorithms", Donald Knuth, 2010 -- Chapter 10 Addition Machines -- Haskell version by Casey Hawthorne -- Note this is only a skeleton of the chapter, -- so as to wet your interest to buy the book. -- Addition Machine -- The only operations allowed are the following: -- read x -- x <- y -- x <- x + y -- x <- x - y -- if x >= y -- write x -- Can all functional versions be tail recursive? ----------------------------------------------------------------------- ----------------------------------------------------------------------- -- Modulus -- Assuming without loss of generality x>=0 and y>0, -- since one can use various identities for other cases. -- Performs floor(x/y) subtractions. modulusNaive x y | x >= y = modulusNaive (x-y) y | otherwise = x -- Can we do better? -- Uses a doubling procedure to subtract larger multiplies of y. -- Bounded by O(log(x/y))^2 modulusByDoubling x y | x >= y = helper y y | otherwise = x where helper w z | x < z = helper z z+z | otherwise = modulusByDoubling (x-w) y -- Can we do better? -- Want bounded by O(log(x/y)). -- Addition Machine, so cannot divide by 2. -- Implicitly use the Fibonacci representation of floor(x/y) -- instead of its the binary representation. -- F0 = 0; F1 = 1; Fn = F(n-1) + F(n-2), n >=2 -- Every nonnegative integer n can be uniquely represented in the form -- n = F(m1) + F(m2) + ... + F(mt), m1 >> m2 >> ... >> mt >> 0 -- where t >= 0 and m >> m' means that m - m' >= 2 -- If n > 0 this representation can be found by choosing m1 such that -- F(m1) <= n < F(m1+1) ----------------------------------------------------------------------- -- Furthermore Fibonacci numbers grow exponentially, -- about 69% as fast as powers of 2. -- They have been used as power-of-2 analogs in a variety of algorithms. ----------------------------------------------------------------------- modulusFib x y | x >= y = helper x y y | otherwise = x where helper x y z | x < z = helper x z y+z | otherwise = helper2 x y z helper2 x y z | x >= y = helper2 (x-y) y z | y > z = helper2 x (z-y) y | otherwise = x

1 0

FastCGI, new maintainer
by Christopher Done 23 Jun '10

23 Jun '10

I'm the new maintainer of the FastCGI package[1]. Please upgrade to the latest version, 3001.0.2.3. It includes a fix for a special but real case in which the library throws an exception when the httpd server unexpectedly closes the connection[2], and the base version and exceptions have been upgraded to the latest versions.[3] Any problems, get in touch, or post an issue on this github project: http://github.com/chrisdone/fastcgi If you have patches preferably fork the fastcgi project on github and do a "Pull request". Alternatively just send me a diff. Cheers [1]: http://hackage.haskell.org/package/fastcgi [2]: http://github.com/chrisdone/fastcgi/commit/de8d303b30dec097a1d24a8991da85d3…

1 0

Control.Alternative --- some and many?
by Gregory Crosswhite 23 Jun '10

23 Jun '10

Hey everyone, Could someone explain to me (or point me to a reference explaining) the purpose of the "some" and "many" methods of the Alternative class? Also, there is a link posted in the documentation for Control.Applicative to a paper which describes the motivation behind the Applicative class; is there similarly a paper explaining the motivation behind the Alternative class? (The problem with Googling for "Alternative" is that this word is used a whole lot of the time, and very rarely does it refer to the Alernative class. :-) ) Cheers, Greg

3 2