Hello beloved release fanatics,
you may have already heard of aif. (Arch Linux Installation Framework). A project of mine that originally begun to allow me to do automated installations, but soon was extended to contain all things needed to do full "interactive" installations (a la /arch/setup) . It's actually really a framework where all backend code (ui, partitioning, ...) is available as libraries, making it possible to cleanly implement different kinds of installers (automated, interactive etc) (lots of the code in those libraries is actually a refactoring of the code in /arch/setup) It also supports dm_crypt and lvm.
It got much interest from the Arch devs, and later they told me they want to use it as the next official installer.
That's great :).
Now, it would be very good if we could shift to it asap. I'm not saying this to promote my own stuff, rather I think we all agree on the technical advantages (old installer is hard to maintain, it's "spaghetticode", we don't want to maintain 2 installers at the same time, aif is a better longterm solution etc)
But, we need to start testing and patching. In it's current state, aif is actually quite usable and most of the time produces installations just as well as /arch/setup does. But there are probably bugs that I'm not aware of yet. I want to get most, if not all of the bugs ironed out before 2.6.29 is out. So that when we start releasing images for the new release, aif does not become the bottleneck.
How can you test? Simple: aif will be included on 2009.02. That's a good starting point. However: there are also git packages on aur. I recommend you install aif-experimental-git which contains the latest features and fixes. You can easily install a git package on the installation cd : see http://github.com/Dieterbe/aif/tree/experimental/HOWTO
I also have a TODO file, which contains a list of issues (most of them minor) and some thoughts. http://github.com/Dieterbe/aif/tree/experimental/TODO