Mailinglist Archive: opensuse-packaging (138 mails)

< Previous Next >
Re: [opensuse-packaging] problematic new path naming structure
  • From: Robert Schiele <rschiele@xxxxxxxxx>
  • Date: Tue, 20 Mar 2007 17:29:58 +0100
  • Message-id: <20070320162958.GL4664@xxxxxxxxxxxxxxxxxx>
On Tue, Mar 20, 2007 at 04:47:02PM +0100, Jurzitza, Dieter wrote:
> Dear Robert,
> dear listmembers,
> > > Call the directory /lib/modules/persistent from the very
> > begin. Maybe
> > > one would like to use persistent-default / persistent-bigsmp.
> >
> > This does not work because it does not allow installation of
> > multiple kernels with different ABIs.
> well, how would this work right now? The packages provided binary are provided for one single kernel, probably for default / bigsmp, that's about it.

You just install one package for each kernel you need this module for. It's
that simple. Nothing magic about that.

> The question is: why make the majority of people using a single kernel configuration suffer from a somewhat exotic concept of people using different kernels in parallel (rumor on the list ;-))?

Nobody really suffers from the current situation. It's just that you don't
like it the way it works now but until now you didn't provide a reasobable
argument that something else is better.

> In my understanding the SuSE concept is straightforward and monolithic to a certain point: you get a kernel with the distribution, you use it, you may upgrade "in line", but just try to upgrade from i.e. 2.6.11 to 2.6.13, a decent part of things like the automount functionality (among other discrepancies) is gone because things do not match any more.

Well, you can just keep the old version for safety reasons, just for the case
something breaks when booting the new kernel.

> If you want to do what you suggest you will have to rebuild all the kernel module extra packages from scratch for the new kernel anyway, no improvement here. The current concept does not work for this either.

No. It seems you just did not understand how the whole kmp packaging system
works. I'd suggest you read the white paper about it.

> I can only support what Herbert Graeber said. But as initially stated, these are just my 2 cents here :-). We won't solve the issue here.

There is no real problem and thus there is nothing to solve.


Robert Schiele
Dipl.-Wirtsch.informatiker mailto:rschiele@xxxxxxxxx

"Quidquid latine dictum sit, altum sonatur."
< Previous Next >