Hello, On Mon, 18 Feb 2008 08:30:19 -0300 Daniel Ribeiro <[EMAIL PROTECTED]> wrote:
> -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Paul Sokolovsky wrote: > > Angstrom was made to rectify that situation. So, once you're > > interested, and ready, it is there. > > And feel free to consider this as the downstream distro coming to > > you, begging for collaboration ;-). > Is Angstrom really calling us to maintain the EZX devices? Yes. In Angstrom terms, that's called "device mentorship", with rough responsibilities described at http://linuxtogo.org/gowiki/MachineMentors Failing complete "mentorship", just regular testing would be good start. In this regard, the news port on http://openezx.org/ is just the right thing except one thing - it doesn't mention that we're expecting feedback on those unstable/autobuild images! > Why dont you guys start with a more gentle approach? I'm > already doing some "distro integrator" work for OpenEZX on > svn.openezx, why dont Angstrom just uses the defconfigs and patches > from svn.openezx? Just to point some benefits: > 1. Immediate support to Angstrom users from the OpenEZX > kernel hackers. 2. Kernel builds which are actively tested on all > OpenEZX supported devices (not only a780). > 3. Avoid duplicate work in both directions. Sorry, but that puts it upside down. Following is my personal account on the stuff, don't take it as official wording from Angstrom, but: Angstrom doesn't need more machines! That would be far too easy! OE has some hundred of machines, we could crunch hundreds of images for them and announce that, bwahaha, Angstrom supports hundred machines. Supports? No. No real support here. So, what we need are not more machines added, but more people involved, bringing actual support to the devices they care about. Because otherwise, we get exactly what's there now - someone not directly involved with OpenEZX adds configs for it, and you wonder what the crap is that, and instead suggest, that you, kernel hackers, know better how to maintain distro and suggest to possibly violate its integrity by importing your config, which from distro's point of view can be just a hack. (Please think about the last clause with some attention - do you imagine someone coming to Debian and say "Hey, here's my cool machine, 'Black box with white keyboard'. Now, it's special. You ship kernel with builtin ext3, modular reiser and no XFS? Do builtin reiser and XFS for my piece, and yes, tha-a-a-t one module builtin too." ?) > > You (and other OErs) can review our builds/defconfigs/patches > and make suggestions here, im sure most of your requests would be > accepted. I cannot guarantee that I will have enough time to play with > OE/Angstrom but if OE built kernels behave at least similar to the > kernels that i build, i will recommend OE and stop asking users to > build kernels outside of OE. > Its that simple! Isnt it great? If we keep talking about kernels, we won't get far with this. The talk is about aiming towards providing a complete user experience, and that's possible only on *distro* level. Preferrably, active and device-neutral distro, so while you hack on kernel, it improves too, and when you get another device, you still can use it. > > - -- > Daniel Ribeiro -- Best regards, Paul mailto:[EMAIL PROTECTED]
