November 2005 - Libraries

Data.Map API
by Iavor Diatchki 13 Nov '05

13 Nov '05

Hello, I have been using 'Data.Map' from the library distributed with GHC/Hugs, and I quite like it. Often though, I find myself needing a function that behaves like this: > import Data.Map (Map); import qualified Data.Map as Map hiding (Map) > > insert1 :: Ord k => k -> (Maybe v -> v) -> Map k v -> (v, Map k v) > insert1 k f m = let v = f (Map.lookup k m) > in (v, Map.insert k v m) In words, we use a function to determine what to place in the map, depending on what is already in the map. In the result, we return the new map, and the value that was inserted in the map. Here is an example of how we may use this function: > example m = insert1 0 v m > where v Nothing = 0 > v (Just n) = n + 1 Clearly, I can implement this function myself as above. However, I was wondering if I can achieve the same, without searching in the map twice. I came up with the following: > insert2 :: Ord k => k -> (Maybe v -> v) -> Map k v -> (v, Map k v) > insert2 k f m = (v,m') > where > v = f r > (r, m') = Map.insertLookupWithKey g k v m > g _ x _ = x Is there a more direct way to implement this function? The use of lazyness in this case seems like an overkill. If not, I would like to suggest that we add it to the API. Also, it seems that some of the functions in the library are quite easy to implement in terms of this function, so perhaps they don't need to be in the API? > insertWith' :: Ord k => (v -> v -> v) -> k -> v -> Map k v -> Map k v > insertWith' f k v m = snd (insert2 k (maybe v (f v)) m) > > insertWithKey' :: Ord k => (k -> v -> v -> v) -> k -> v -> Map k v -> Map k v > insertWithKey' f k v m = snd (insert2 k (maybe v (f k v)) m) -Iavor PS: How do people import 'Data.Map'? The above seems clunky...

7 9

Re[6]: Data.* collections maintenance
by Bulat Ziganshin 10 Nov '05

10 Nov '05

Hello Jean-Philippe, Friday, October 28, 2005, 6:56:25 PM, you wrote: >> i propose the following class hierarchy: >> >> Collection >> Sequential (lists, arrays) >> UpdatableSequential (updatable arrays) >> NonSequentional >> KeyVal (AVLTree, Map) >> UpdatableKeyVal (HashTable) >> KeyOnly (Set) >> UpdatableKeyOnly JPB> If anything, arrays can be seen as a KeyVal map too... We don't JPB> necessarily want a tree-like hierarchy. In any case, please post your JPB> suggestions on the list, preferably with the corresponding source JPB> code. The problem is not straightforward, and we certainly want to JPB> hear a few more opinions before deciding anything. ;) well, i am as user of libraries just wants consistency of using different data structures for the same tasks. for example, i want to have one indexing operator instead of !, !! and find. so, it's my hope-list: 1) all collections are divided into 3-4 classes: arrays/lists, maps and sets. arrays/lists have sequential indexes in some range while maps have sparse indexes. also each class have an Updatable subclass 2) collections in one class can be collected to each other with one (better universal) operator, such as: let list = cvt array let tree = cvt map 3) collections can be converted to/from Updatable subclass with help of usual freeze/thaw operators 4) all operations which are currently implemented in Data.List, Array.*, Map, Set, HashTable modules must be bound to one of these classes, with duplicates (such as !/!!) removed. now i see the following class hierarchy: Collection (map, size, values) SetLike (union, diff) Set Map Indexed (indexes, !) Map Sequential (reverse, head) Array List (tail) i give in parentheses examples of operations which are supported on each level of hierarchy this will give possibility to write datastructure-independent algorithms and easily convert data to the data type which are best for each part of the program, smthg like: a :: Map Int String <- read str algorithm1 a let b :: Hash Int String <- cvt a algorithm2 b c :: HashTable Int String <- thaw b algorithm3 c -- Best regards, Bulat mailto:bulatz@HotPOP.com

2 2

RE: System.Posix.* not haddocked
by Simon Marlow 10 Nov '05

10 Nov '05

On 08 November 2005 08:17, Isaac Jones wrote: > Björn Bringert <bringert(a)cs.chalmers.se> writes: > >> Is there some special reason why the System.Posix.* modules are not >> haddocked? There seems to be documentation available in the hslibs >> docs [1]. Is it just that noone has bothered to do the copy-and-paste >> yet? If this is the case, I might try to find time to do some of it >> myself. > > I did the haddock-ization, but then couldn't get the whole fptools > tree to build for a while and gave up; maybe I should have just > checked it in and let someone with a working tree fix it. Here's the > link to the archives. Might be a good starting point: > > http://www.haskell.org/pipermail/libraries/2005-July/004077.html Do you have more patches or just that one? I'll be happy to test/fix/checkin anything you have. You can 'make html' without having to build GHC. You do need a full GHC build tree, configured, and probably 'make boot' in libraries. Then you can do this: $ make html HC=ghc in any libraries/<dir>. Cheers, Simon

2 1

RE: System.Posix.* not haddocked
by Simon Marlow 08 Nov '05

08 Nov '05

On 07 November 2005 21:07, Björn Bringert wrote: > Is there some special reason why the System.Posix.* modules are not > haddocked? There seems to be documentation available in the hslibs > docs [1]. Is it just that noone has bothered to do the copy-and-paste > yet? If this is the case, I might try to find time to do some of it > myself. Please do - I just never got around to it. Bear in mind there are various differences between the old Posix and System.Posix. Cheers, Simon

1 0

Cabal Problem
by Russell, Daniel J 08 Nov '05

08 Nov '05

I am trying to install the Haskell Crypto package using Cabal. I have a problem when using Cabal. I have installed NewBinary-0.1 but when trying to configure the Crypto library using Setup.exe configure --prefix=h:\NewHaXml I receive the message: Configuring Crypto-2.0.3... configure: Dependency base-any: using base-1.0 configure: Dependency mtl-any: using mtl-1.0 configure: Dependency QuickCheck-any: using QuickCheck-1.0 configure: Dependency HUnit-any: using HUnit-1.1 Setup.exe: cannot satisfy dependency NewBinary-any This suggests that a NewBinary package does not exist but ghc-pkg list returns: h:\Cfgs\ie\AppData\ghc\i386-mingw32-6.4\package.conf: Cabal-1.1.4, NewBinary-0.1 Any suggestions? Many thanks. Dan Russell This email has been scanned for all viruses by the MessageLabs Email Security System.

2 1

System.Posix.* not haddocked
by Björn Bringert 08 Nov '05

08 Nov '05

Is there some special reason why the System.Posix.* modules are not haddocked? There seems to be documentation available in the hslibs docs [1]. Is it just that noone has bothered to do the copy-and-paste yet? If this is the case, I might try to find time to do some of it myself. /Björn [1] http://haskell.org/ghc/docs/latest/html/hslibs/sec-posix.html

2 1

ANN: network-alt 0.3 and hsgnutls 0.2.1
by Einar Karttunen 07 Nov '05

07 Nov '05

Hello network-alt 0.3 featuring: * New improved build system * Recvfrom and sendto fixes * Windows support Network-Alt is an alternative networking library for Haskell supporting a nicer API, better performance and IPv6 support. http://www.cs.helsinki.fi/u/ekarttun/network-alt hsgnutls 0.2.1 featuring: * New memory model making automatic memory management work better hsgnutls is a TLS/SSL layer built on top of the gnutls library, supporting both client and server applications. http://www.cs.helsinki.fi/u/ekarttun/hsgnutls/ - Einar Karttunen

4 7

ANNOUNCE: Gtk2Hs version 0.9.10 released
by Duncan Coutts 07 Nov '05

07 Nov '05

Gtk2Hs - A GUI Library for Haskell based on Gtk+ Version 0.9.10 is now available from: http://haskell.org/gtk2hs/download/ An installer for Windows is available. A package is available for Gentoo. Packages for other platforms should become available soon (hopefully including Fedora Core, Debian, Darwin, FreeBSD and OpenBSD). Changes since 0.9.9: * New bindings for the cairo vector graphics library. This provides a easy-to-use vector graphics API (using a PDF-style drawing model) with high quality output for multiple backends (screen, print and image files). * Completion of Pango, the text layout and rendering engine of Gtk +. We now provide all functions that end users should ever need for rendering text, ranging from type-setting whole paragraphs down to breaking up attributed text into several runs. All offsets into Haskell strings are transparently translated into UTF-8 offsets used in Pango, thereby alleviating much of the grief of dealing with Unicode. * new Gtk+ 2.8 APIs are bound * many bugs fixed * various minor documentation improvements * an installer for Windows is available again * the mozembed HTML rendering component can now be built against the FireFox libraries if the appropriate developer files are present. Other news since the last release announcement: * Paolo Martini won a $4500 grant under the Google Summer of Code programme for a project to add bindings for the cairo vector graphics library to Gtk2Hs. * Gtk2Hs is being used as part of the practical element of Functional Programming courses at the University of Oxford in England and the University of Jyväskylä in Finland. Gtk2Hs feature highlights: * automatic memory management * Unicode support * nearly full coverage of Gtk+ 2.8 * bindings for the cairo vector graphics library * support for the Glade visual GUI builder * bindings for several Gnome modules: GConf, Gnome's system for storing app preferences SourceView, code editor widget with syntax highlighting the Mozilla browser rendering engine in a widget * extensive API reference documentation * multi-platform support with native look * LGPL licence Platforms and requirements: * builds from source on Linux, Windows, MacOS X, FreeBSD, OpenBSD and Solaris * builds with GHC 6.4.x, 6.2.x, 6.0.x and 5.04.x * works with Gtk+ version 2.0 through to 2.2, 2.4, 2.6 and 2.8 * optional cairo support requires cairo 1.0 or later, however using cairo in conjunction with Gtk+ requires Gtk+ 2.8.x or later * optional Gnome support requires Gnome 2.0 or better * optional Mozilla support requires Mozilla 1.4 or FireFox 1.0 or later with the appropriate development files. Goals for the 1.0 release: * a stable API for the 1.0.x series * greater coverage of the Gtk+ API (we're about 90% there now) * support for the OpenGL 3D rendering API * introductory and other tutorials * packages for all popular platforms Please report all problems to gtk2hs-devel(a)sourceforge.net Contributions and feedback are also most welcome. To lower the barrier to contributions, the development version of Gtk2Hs is now available via darcs: darcs get http://haskell.org/gtk2hs/darcs/gtk2hs Duncan (on behalf of the Gtk2Hs team)

2 1

RE: [Bug#337909: ghc6: regression in Cabal] [vela@debian.org]
by Simon Marlow 07 Nov '05

07 Nov '05

On 07 November 2005 11:59, Ian Lynagh wrote: > Does the attached patch look right? Looks reasonable, though I haven't tested it. The Cabal test case in the testsuite doesn't use hs-source-dir, which explains why it slipped through. hs-source-dir is deprecated, and the code in the HEAD is quite different. Cheers, Simon

1 0

Fwd: [Bug#337909: ghc6: regression in Cabal] [vela@debian.org]
by Ian Lynagh 07 Nov '05

07 Nov '05

Hi cabal people, Does the attached patch look right? Thanks Ian

2 1