Kernel module packages (was - Re: Pre-Review: Asterisk)

[Michal Jaegermann]

On Sat, Apr 02, 2005 at 04:56:01PM -0600, Tom 'spot' Callaway wrote:
> 
> Yes. Let me expand on my example.
> 
> I have three kernel packages installed:
> 
> kernel-2.6.10-2
> kernel-smp-2.6.10-2
> kernel-2.6.10-3
> 
> Each of these has a kernel-module-foo package installed alongside it:
> 
> kernel-module-foo-2.6.10_2-1
> kernel-module-foo-2.6.10_2smp-1
> kernel-module-foo-2.6.10_3-1

So far so good.  The only thing is that you want this kernel
version id squeeze into a package version part while as for
now people doing that have to make it a part of a name.

There is, of course, this problem that during development a module
source code may have its own version as well but one can possibly
overload a release part to get around the issue.

> 
> We need to teach rpm to perform the following transaction (ignoring
> normal dependency checks):

That is this tricky step.  It is not only a question of various
rpm releases which users will try to apply to that all over the
world but also various releated more-or-less official tools which
are widely used.  Seems quite a touchy issue to me and I am not
sure how realistic are expectation that it is possibly to have
not too rough transition (if any at all).

I do not mind to be convinced otherwise but I do have doubts.

> 
>   - if %{release} of pending update is greater than existing, then 
>     upgrade ONLY that existing package with same %{name}-%{update}
                                                             ^^^^
                                   Hm, some new entity showed up here.

> Does that make it more clear? :)

You goals are clear but now you have to find a way to distinguish
packages which require such special treatment from those who
are handled as upto now.  Here we start into some twisty passages
as far as I can tell.  A special tag?  Some extra part of a name?

   Michal