[arch-haskell] GHC 7.4.1 or HP 2011.4.0.0??

[Matthew Cox]

A a long-time lurker on this list, here are my thoughts.

On Fri, Feb 3, 2012 at 10:28 AM, Magnus Therning <magnus at therning.org> wrote:
> It's been out for a day now... this would be a good time to decide
> whether ArchLinux should be bold and move to a haskell-platform-free
> state, or trudge on with HP and the ache it causes.

Let me put in a vote to cut the HP cord. The approach of the HP is
fundamentally at odds with what Arch Linux provides to its users.

Quoting https://wiki.archlinux.org/index.php/Arch_Linux#Modernity:
> Arch Linux strives to maintain the latest stable version of its software, and is based on a rolling-release system, which allows a one-time installation and continuous seamless upgrades,
> without ever having to reinstall or perform elaborate system upgrades from one version to the next. By issuing one command, an Arch system is kept up-to-date and on the bleeding edge.

The precise definition of stable matters. There is the Debian
definition, and then there's the Arch version which has consistently
been interpreted by the rest of the Arch developers as the latest
release from the upstream developers. Is the 3.2 kernel stable? Will
GCC 4.7 be stable the week of its release? Arch users expect that
there will be occasional breakage but that the fast moving nature of
our distribution answers these by emphasizing the rapid delivery of
upstream bug fixes.

The Haskell Platform takes an inherently more conservative approach.
It follows the Debian definition of stable. Some advanced users prefer
to concentrate on their own code and appreciate the long release
cycles of the Platform that doesn't introduce exciting new bugs on a
regular basis. New users appreciate that the Platform insulates them
from breakage or churn of new features and ideas that are introduced
as libraries evolve. I agree that the Haskell Platform serves a
purpose, but I don't think that it's one that is of much appeal to the
typical person who chooses to run Arch in the first place.

I believe this community is less effective than it could be because
many people who use Arch and Haskell don't participate or contribute.
I freely admit that I don't. I have my own PKGBUILD for GHC 7.2 (I'll
get around to 7.4 this weekend) and build my own libraries with
cabal-install in HOME. On a fresh install, I use the [extra] ghc
package to build my own GHC package and also use [community]'s xmonad
until I get around to building it myself. After that I don't come into
contact with the rest of the activities of arch-haskell and so my
experience as a user of Arch and Haskell doesn't provide anything I
can usefully contribute back to the community.

> If we're moving to 7.4.1 there's a lot of work with getting
> ArchHaskell in shape for it, so the longer notice the better :)

I hope that if we drop the Platform we'll attract a few more hands
since more of the lurkers on this list will take a more active
interest. I'm not too worried about breakage since the people who
encounter it are usually Haskell developers and are ready to go to
work on a problem that blocks them from working.

Let's start breaking things.

Matt.