KernelFreeze, 2.6.14, UBZ agenda, and other bits

Fabio Massimo Di Nitto fabbione at
Mon Oct 10 06:07:21 UTC 2005

Ben Collins wrote:
> On Mon, Oct 03, 2005 at 09:06:13AM +0200, Jean-Eric Cuendet wrote:
>>>Generally, you can install the kernel .deb's from one release to the
>>>prior. I can't guarantee it will install, but if it does install, it will
>>>almost certainly boot.
>>Yes, normally....
>>But for Hoary, Breezy kernels won't install. Because it needed newer 
>>modutils which needed newer glibc...
>>That's just a dependency problem, nothing more. Probably a rebuild on 
>>Breezy will do the trick if Dapper kernel doesn't install.
>>It would be nice if there could be an apt repo just containing the 
>>Dapper kernel, so it's easy to put it in sources.list on Breezy. What do 
>>you think?
>>Or am I alone in willing to test the Dapper kernel on Breezy?
> I'm more than willing to entertain the idea of making the development
> kernel easier to test. My fear is that we may encounter bugs when running
> the kernel on breezy, that are simple because of supporting programs
> needed from dapper (API changes for kernel/userspace communications, like
> IEEE-1394, USB, and such). Some things may need to be recompiled with
> recent kernel headers in order to work properly.
> I don't want to be slowed down by bugs like this when the kernel works as
> expected under dapper.

During my maintainance time, i did always strongly unsuggest people to do backport
exactly because of problems that can't be easily solved like userland/kernel misallignment.
We really don't have the resources to handle backports or to dig into bugs generated by them.

IMHO we need to make this policy more strong and our users more aware of the problems that
they will face using such backports.


I'm going to make him an offer he can't refuse.

More information about the kernel-team mailing list