Re: 2.6.13, experimental and 2.6.14-rc ...
On Wed, 2005-10-05 at 17:28 +0200, Jonas Smedegaard wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> On Wed, 5 Oct 2005 09:24:17 +0200
> Sven Luther <email@example.com> wrote:
> > That was initramfs-tools, we should evaluate also yaird.
> Is anyone (besides you, sven) interested in yaird at all?
I've yet to try out either tool, but I do like the fact that initramfs
will be getting heavy testing (and associated fixing) by our Ubuntu
friends. Are any other distros using initramfs or yaird?
I agree with Sven that, at the end of the day, this should be a
user-selectable option. We just need to pick a default that has the
best chance of working for most users. And, we should be able to change
this default if a better option comes along.
So, maybe the best approach is to come up with a matrix of test cases,
and see where the problems in each lie.
Something like this, maybe?
And also, what will be the best way to migrate? Should we make a new
initrd-tools package that provides a /usr/sbin/mkinitrd script that
selects a backend and performs any necessary argument munging?
(I think someone suggested this before...). This would be a clean
upgrade path, and would work with existing kernel debs that are out
there, and wouldn't require asking the user to update their
kernel-img.conf file (but they could do so, if they wanted). It also
wouldn't require a kernel-package mods.