Thanks to everyone for feedback. The situation is somewhat more clear now.
In summary:
* d-i is fine as far as we keep the traditional initrd support for a
while. As Bastian mentioned, it is pretty much guaranteed that it is
going to be supported until 2.6.15, which gives the d-i team some time
to make the transition.
* yaird vs initramfs for initrd generation - both seem to be in a somewhat
similar state. Both appear to be usable on mainstream architectures,
with the status on other architectures largely unknown. Advantage of
yaird is that it's capable of including needed modules only, while
initramfs requires a predefined list. I would say that the only
criterion here is support for larger amount of architectures. And
probably a sensible thing to do at that point is to test both on as much
architectures as possible to figure out what still needs to be done.
* libc issue: klibc is about to enter archive. However Bastian mentions
that the tools which are supposed to use it when running off initrd
are not very happy with it. Maximilian, could you comment on that
please? I believe that stabilizing a new libc version (porting it to all
arches, making sure that executables linked against it work, etc) will
take a lot of time, unless such an effort is already under way. For me
it looks reasonable to use glibc on initrd for 2.6.13 and work on
improving klibc usability in the meantime.
One thing which I am not too comfortable with yet, is how we are going to
switch to an alternative initrd generator in the kernel deb postinsts.
The command to generate initrd is currently configurable in
/etc/kernel-img.conf, which is owned by kernel-package, so the actual
kernel packages cannot easily fiddle with it. A suggestion by Jonas
for yaird to provide mkinitrd script and conflict with initrd-tools is not
too good either, as people might want to have 2.6.12 and 2.6.13 installed
at the same time. Anybody has any thoughts on that?
Please keep your comments coming.
Best regards,
Jurij Smakov [EMAIL PROTECTED]
Key: http://www.wooyd.org/pgpkey/ KeyID: C99E03CC
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
