Live from Afghanistan!
Viridior
new assignment and ready to get back to work. The good thing is that
I have some internet connectivity, the bad being that its very very
slow and I don't have direct access to any ARMv7a boards, not that it
would really matter since I couldn't upload binaries anyway. However
this does allow me to focus on building a lot of the framework and
learn some programming which I had been pushing off for a while.
My general TODO list
-------------------------------
* review metro
* finish the install system
* more ebuilds
* DSP
* graphics
* kernels
* other embedded projects
* automated build system with web-frontend
* documentation
James Dominy
Hey there. Firstly, good luck in Afghanistan. Secondly, I'm a programmer
and I do have have some spare time for this. I just need some help with
getting the tool chain and a virtual machine setup. I've tried before,
but the tool chain produced broken binaries, and I didn't need a virtual
machine before because I had a working touchbook. The touchbook is since
kaput, hence the need for a VM.
So if there's something that needs doing, programming wise, I'm happy to
help out.
Cheers,
sirlark
Viridior
Programming project:
I had a prototype bash script install program that I had messed around
with months ago. I think I'm going to go back and revise it, make it
modular, and see about supporting custom profile/ trees. Thinking
about switching it over to Python while I'm at it.
metro & ebuild:
This always needs work, the problem with this will be that I don't
have a good internet connection and testing this would be time-
consuming and painful.
web-based automated system:
This would be another good project to kick off, I might work on this
on the side when I get stuck with the "installer" project. There
might even be some synergy between the two.
documentation:
javajake and I went through a few times to make the pages consistant.
Probably should be reviewed again. I still have a catalog of sites
that I would routinely visit and gather information, need to start
that again.
Very Respectfully,
Jake
>
> - Show quoted text -
Dennis.Yxun
It's really really loooooong time.. and good to see you back again
I thought about getting a really usable gentoo system while not care about how to get this
either use cross-compile or native-compile, I still doubt about the cross-compile ways
web-based automated build system would be a good idea,
make me think about the SUSE abs system, every one can setup a
build system use prebuilt package then just build the only packages they want.
Dennis
bear.sh
On 25 Mai, 15:50, "Dennis.Yxun" <dennis.y...@gmail.com> wrote:
> I thought about getting a really usable gentoo system while not care about
> how to get this
> either use cross-compile or native-compile, I still doubt about the
> cross-compile ways
will become the easiest way to build a system
but still, cross-compiling as long as it works will be faster until
arm catches up with intel...
and there are also other cases where there's no much choice about the
target cpu (too expensive, already available, etc)
for example at work I have to work with a 190MHz arm9. nowbody will
compile natively on that device... so I've set up a dev chroot where I
do the cross-compilation from there I can chroot into the compiled
system using qemu-user for packages which don't cross-compile. as soon
as that package is built, I can switch back to cross-compilation...
so for now a mixed system would make most sense, I'm pretty sure...
would be great to have a build frontend which delegates the job to
either a cross-compile instance or to a native (qemu or target board)
compiler...
/bearsh
Jacob Galbreath
Lama (with one L)
http://www.gumstix.com/press/Gumstix-strongbox-TI-Tech-Days-2011.pdf
Gentoo needs!
On May 25, 10:52 am, Jacob Galbreath <jacobgalbre...@gmail.com> wrote:
> I'm really glad to hear from you all, great to be back.
>
> I'm seriously considering buying a OMAP4 system or two to help with native
> compiling. I think that with 4+ cpus (with icecream/distcc) that the
> compile time would be short enough to make it feasible. Started coding a
> automated build system, may be a little while before I get anything posted
> but definately could throw some ideas around.
>
Alec Moskvin
Flameeyes' Tinderbox[1] scripts, or whatever the official tinderbox[2]
uses?
Alec
[1] http://blog.flameeyes.eu/2009/12/06/opening-up-the-tinderbox
[2] http://tinderbox.dev.gentoo.org/
Viridior
On May 26, 12:32 am, "Lama (with one L)" <lucas.yamani...@gmail.com>
wrote:
> Check this out:http://www.gumstix.com/images/cluster-1.jpghttp://www.gumstix.com/press/Gumstix-strongbox-TI-Tech-Days-2011.pdf
Viridior
On May 26, 12:53 am, Alec Moskvin <al...@gmx.com> wrote:
> Are you planning to do it from scratch? Have you considered adapting
> Flameeyes' Tinderbox[1] scripts, or whatever the official tinderbox[2]
> uses?
>
> Alec
>
> [1]http://blog.flameeyes.eu/2009/12/06/opening-up-the-tinderbox
> [2]http://tinderbox.dev.gentoo.org/
>
> On Wednesday 25 May 2011 19:22:12, Jacob Galbreath wrote:
>
>
>
> > I'm really glad to hear from you all, great to be back.
>
> > I'm seriously considering buying a OMAP4 system or two to help with native
> > compiling. I think that with 4+ cpus (with icecream/distcc) that the
> > compile time would be short enough to make it feasible. Started coding a
> > automated build system, may be a little while before I get anything posted
> > but definately could throw some ideas around.
>
>
> > > Hi all
>
> > > On 25 Mai, 15:50, "Dennis.Yxun" <dennis.y...@gmail.com> wrote:
> > > > I thought about getting a really usable gentoo system while not care
> > > about
> > > > how to get this
> > > > either use cross-compile or native-compile, I still doubt about the
> > > > cross-compile ways
>
> > > looking at the power of todays and tomorrows cpus, native compilations
> > > will become the easiest way to build a system
> > > but still, cross-compiling as long as it works will be faster until
> > > arm catches up with intel...
> > > and there are also other cases where there's no much choice about the
> > > target cpu (too expensive, already available, etc)
> > > for example at work I have to work with a 190MHz arm9. nowbody will
> > > compile natively on that device... so I've set up a dev chroot where I
> > > do the cross-compilation from there I can chroot into the compiled
> > > system using qemu-user for packages which don't cross-compile. as soon
> > > as that package is built, I can switch back to cross-compilation...
> > > so for now a mixed system would make most sense, I'm pretty sure...
> > > would be great to have a build frontend which delegates the job to
> > > either a cross-compile instance or to a native (qemu or target board)
> > > compiler...
>
Lama (with one L)
the configuration would be. In my experience, usbnet is too finicky
and slow to function as a solid IPC bus, not to mention heat and
energy issues.
If Gumsitx were to make something similar using something like TI's
AM3892 (http://focus.ti.com/docs/prod/folders/print/am3892.html), it
might prove more cost-effective-- certainly faster. Until then, I'm
looking at Trimslice, myself.
BTW, stay safe in the 'stan.
On May 25, 11:25 pm, Viridior <jacobgalbre...@gmail.com> wrote:
> I've looked at this, its on the list of things to choose from.
>
> On May 26, 12:32 am, "Lama (with one L)" <lucas.yamani...@gmail.com>
> wrote:
>
Viridior
variety of ways and still prefer ethernet to it.
Before I left I rebuilt my local network of PCs (13 total with a mix
of Intel and AMDs) to a diskless node farm. Mostly to save power but
also as a precurser to working with nfs root fs based images with the
eventually use with Neuvoo images. I would like to make it so I can
just build an image, reboot a board and test it, also this would be
much easier to manage remotely.
Thoughts?
Jake
On May 26, 9:31 am, "Lama (with one L)" <lucas.yamani...@gmail.com>
Jacob Godserv
> Are you planning to do it from scratch? Have you considered adapting
> Flameeyes' Tinderbox[1] scripts, or whatever the official tinderbox[2]
> uses?
I've actually looked into this, and talked to flameeyes about it. He
basically did not recommend tinderbox for our project, and told us we
should build our own tools for the kind of project we're working on.
(I don't remember the details, but I have his e-mail somewhere.)
The result is this project I worked on a year ago called Farmerge:
http://neuvoo.org/wiki/index.php?title=Farmerge
It is the beginning of a distributed binary repository maintenance
system, that will hopefully allow for the following:
* Consistent up-to-date binary repositories for popular profiles and
architectures
* User emerge instances can request binaries on-demand
* Parallel distributed building of packages and package dependencies
Presently, the first is satisfied. My next goal was going to be
parallel distribution. After that, I was going to use my portage hook
feature for emerge to allow emerge to request, report progress on,
download, and install up-to-date binaries.
This goes beyond tinderbox's feature set, I believe, and hopefully
will be useful for more people than just us. :)
--
Jacob
"For then there will be great distress, unequaled
from the beginning of the world until now — and never
to be equaled again. If those days had not been cut
short, no one would survive, but for the sake of the
elect those days will be shortened."
Are you ready?
Jacob Godserv
> The good thing is that
> I have some internet connectivity, the bad being that its very very
> slow and I don't have direct access to any ARMv7a boards
This also applies to me. My beagleboard died last Fall (I didn't have
a case; lesson learned) and I've been thinking about replacing it.
I'm going to resume work on the portage hook patches asap. They are so
close to being done that the official portage people have started to
look favorably upon it, and have seriously considered merging them.
This would replace existing half-baked hook-like functionality in
/etc/portage/post-sync.d so I think it's good for portage to finish
that feature off.
After that, I want to finish farmerge.
On the side, I want to get back into helping #gentoo-embedded with
cross-compilation. The day we can boot from a completely
cross-compiled stage4, desktop and all, will be a landmark moment for
this project, I'd say.
As far as time goes, I should have some. I'm finishing up some smaller
stuff over the next month, but I will be here on the mailing list, and
I'll try to idle on IRC as often as possible. The biggest roadblock is
the fact that I lost my Gentoo installation to a hard-drive failure,
and then my optical drive died, which kept me from reinstalling. All
those blocks are gone now, so there should be progress.
I'm really excited about getting back to work on Neuvoo. This is the
first time a "startup" open-source project I've been involved in has
actually obtained a community, and that tells me what we're doing here
is not only useful for us, but for other areas of the Gentoo community
as well. :)
Jacob Godserv
I forgot to reply to this one. Metro's had a major update since we
last used it, so we're going to need to take the time to make the
right changes.
We should probably release a new up-to-date image as well, and get
ourselves back on our feet in terms of supporting releases.