On Sun, 6 Jan 2013 23:38:24 -0800, Tracy Reed <[email protected]>
wrote:
On Fri, Jan 04, 2013 at 01:58:48AM PST, Gordan Bobic spake thusly:
At some point in the not too distant future I hope to get around to
getting
an automated build system in place, but since this project isn't
sponsored by
anyone, I have this thing called a "job" that really eats into my
free time
available to work on RS. ;)
I suspect the only sort of sponsorship that would help is cash,
right?
If you ever need ARM hardware, bandwidth, hosting, whatever, I may be
able to
help.
Thank you for your kind offer. Hardware and hosting/bandwidth aren't an
issue at the moment. The mirror is up on mirrorservice.org which
covers the bandwidth/hosting issue to any extent that might be required
for the foreseeable future, and I have a stack of ARM hardware that I
want to get RS running on that I haven't had a chance to even touch in
months.
Anaconda doesn't work on ARM yet. I'm not even sure they have it
fully
working on the latest Fedora yet. Because there is very little in
terms of
common standards on ARM, most devices require a proprietary kernel
and a
proprietary installation method.
My ODROID boards are from hardkernel.com which has a customized
Ubuntu. I may
be able to use their bootloader, kernel and installation method but
put the
RedSleeves userland on it to get what I want. If that works I'll post
here. I
can also probably also use their bootloader setup, kernel source and
.config to
roll my own for use with RedSleeves if necessary.
Indeed, that is pretty much what is expected to be required at the
moment - either use a known good 3rd party kernel or roll your own
based
on it.
The upstream PNAELV kernel predates support for most currently popular
devices
being upstream. I've been working on getting better and more complete
Marvell
Kirkwood support into it (to avoid having to lie to the kernel about
the
machine ID), but it is quite time consuming because there are over a
hundred
patches that need merging, testing and sometimes fixing/dropping.
Combined
with it taking several hours to build a kernel from scratch, it's a
problem.
If anybody wishes to lend a hand with this, I could probably get my
work-in-progress on the kernel put somewhere publicly accessible in a
few
days. This is one of the things I really wanted to get out of the way
before putting all my Marvell Kirkwood devices together into a koji
farm.
>Is the Alpha release from Feb 2012 still the latest?
Yes, but some of the packages have been updated since then.
It might be good for marketing purposes to move it from Alpha to beta
or even
release if people are having success with it as it is. Is there
really anything to worry about?
I wanted to at least get some kind of automated building in place as
well
as package signing before I bump it up from Alpha. Considering the
frequency of bug reports coming in, there's probably not much in there
to worry about. Having said that, I would rather people approach this
approach with caution and expectation that they will have to put some
work in to get it working for them and get pleasantly surprised.
Under-promising and over-delivering is much better marketing, IMO,
than over-hyping and under-delivering.
Gordan
_______________________________________________
users mailing list
[email protected]
http://lists.redsleeve.org/mailman/listinfo/users
