I've been running Fedora ARM for a short while, and am really impressed
by the work that's been done. It seems that this will become even more
important as new ARM-based devices -- tablets, home automation,
netbooks, XO 1.75 -- hit the market.
I'd like to help with this initiative. I'm at Seneca College, and we're
willing to host ARM builders and a Koji instance, with a goal of
tracking kernels, packages, and updates as close to the primary archs as
possible. I've got a small group working on this infrastructure, and we
should have it fully up and running in March (next week is our study
break).
I have a couple of questions:
(1) What's a good representative sample of systems to test on, in
addition to a SheevaPlug, OpenRD, BeagleBoard, and HawkBoard?
(2) Since most ARM systems are initially booted with a prebuilt rootFS,
Anaconda has little value. However, it seems that building the rootFS
for ARM is not really any different from building a Live CD for
i386/x86_64 except for the final output format (obviously, this would
have to be done on an ARM system or in emulation so package scriptlets
could run). I'm wondering if we can adapt the livecdcreator to have
selectable output for various ARM devices, building images for SD cards
(and so forth, as appropriate to each target device) that have the root
FS plus the kernel with the right naming/file sequencing in the image.
This would make it easier to build a rootFS with a particular package
set. Any feedback on this idea?
-Chris (ctyler on freenode)