This project contains the database used to generate an ABS tree of ArchLinux PKGBUILDs. These PKGBUILDs can then be used to compile a binary repository of Haskell packages. If you want to use the official build of this repository then add the following two lines to your /etc/pacman.conf file:

[haskell-core]
Server =  http://xsounds.org/~haskell/core/$arch

If you'd like to build the repository yourself, then you need to install the following packages:

• devtools
• cblrepo (available in [haskell-core], and from hackageDB)
• git

The packages and database are signed with the following key:

pub   2048D/4209170B 2012-12-26
      Key fingerprint = F310 4992 EBF2 4EB8 72B9  7B9C 32B0 B453 4209 170B
uid                  ArchHaskell (Magnus Therning) <[email protected]>
sub   2048D/A418C0FE 2012-12-26

Please note that this will build all the packages in the repo, which probably is a waste of time for most people. It's much easier to add the ArchHaskell repo of pre-built packages. There is also no need to build all packages in order to build a single one, please see the documenation of cblrepo for more details on how to add and update packages.

First clone the database in this repository:

% git clone git://github.com/archhaskell/habs.git
% cd habs

Then sync the cabal index:

% cblrepo sync

Then create all the PKGBUILDs for the packages:

% cblrepo pkgbuild $(cblrepo build base|tail -n +2)

that should fill up the directory with directories containing PKGBUILDs, one for each package. To build packages use the script makeahpkg. There's a builtin help:

% ./makeahpkg -h
Usage: makeahpkg [options] -- [packages]

Run this script in your HABS dir to build the named packages inside a
chroot.

The chroot dir consists of two dirs, a pristine chroot (root), and a build
chroot (build).  This script handles both of them.

Options:
-h          This help
-c          Clean the build chroot before building
-x          Update the pristine chroot first (implies -c)
-l <dir>    Location of chroot (default .)
-a <arch>   Architecture to build for i686 or x86_64 (default is local arch)
-b <dir>    Location of your HABS dir (default .)

To build all the packages:

% ./makeahpkg -c -- $(cblrepo build base | tail -n +2)

The script will create a clean ArchLinx installation in a chroot environment at ./${arch}-chroot, where ${arch} is either i686 or x86_64, depending on your hardware. Use the option -l <dir> to put the chroot in a different location. It is also possible to cross-compile the i686 repository on 64-bit machines by using the -a option.

All PKGBUILDs in the project must be generated with the latest release of cblrepo.

The main goal is to maintain a binary repository of up-to-date packages taken from hackageDB. To achieve this we also maintain several repositories of tools, libraries, and scripts:

• cblrepo is the main tool used to maintain a consistent database of package versions and generate PKGBUILDs.
• cabal2arch is a tool which converts a CABAL file into a PKGBUILD, and optionally a .install. (Currently only used for packages not included in the database, e.g. for generating source packages to upload to AUR.)
• archlinux is the main dependency of cabal2arch.
• archlinux-web contains a library and a set of scripts that help with various web-related tasks such as updating the information on hackageDB on what versions of packages that are available for ArchLinux.

All discussion around our HABS tree and other packages take place on the ArchHaskell mailinglist.

Currently the repository contains some 200 packages, which is only a small fraction of the packages on hackageDB. We would of course like to cover 100% of hackageDB, but our resources, tools and processes don't allow it yet.

The current set of packages was chosen in a very unstructured way, and we are still discussing how to decide what packages we take on. Please get involved by joining the mailinglist if you have opinions on how we should pick the packages we provide.

Suggestions for adding individual packages to the set are always welcome. Use the github issue tracker to make sure your request isn't lost. Be aware though that each added package adds to the burden of the team, so there is a real risk that your request is denied. That risk is reduced if you show an interest in contributing to the project, e.g. by providing already built packages together with your request..

1. Fork it.
2. (Maybe) Create a new branch.
3. Make the required changes. (Adding packages, patches etc.)
4. Build packages for both i686 and x86_64.
5. Upload the built packages to someplace where the maintainers can get them. (e.g. Dropbox)
6. Push the changes to your fork.
7. Create a pull request mentioning the url from which the package can be downloaded.