1 |
Hi, |
2 |
|
3 |
On Fri, 2004-10-29 at 17:36, Peter Johanson wrote: |
4 |
> The goal of kernel-mod eclass is to enable kernel module ebuilds to have |
5 |
> an easier time doing what they do. If non-kernel module ebuilds are |
6 |
> using kernel-mod, we should investigate what functionality *those* |
7 |
> ebuilds use, and move that into a seperate eclass that kernel-mod.eclass |
8 |
> can then inherit from. Things like the kernel version variable |
9 |
> detection, etc, is what i'm guessing is used by other ebuilds (i don't |
10 |
> have time to dig currently) and we could certainly move that into |
11 |
> kernel-info.eclass or somesuch, and let kernel-mod.eclass inherit that. |
12 |
> Then we are free to do things like brix suggests, without annoying those |
13 |
> few ebuilds that think they are kernel module ebuilds, but are really |
14 |
> imposters. (: |
15 |
|
16 |
I've moved most of the current support functions from kernel-mod.eclass |
17 |
to kernel-info.eclass and added functions for calling depmod and |
18 |
functions for substituting SUBDIR= with M= to kernel-mod.eclass. |
19 |
|
20 |
My proposal for kernel-info.eclass and the new kernel-mod.eclass is |
21 |
attached to this email (a patch isn't well-suited because of the major |
22 |
changes and moving of functions). |
23 |
|
24 |
Please comment. If nobody objects I will commit the kernel-info.eclass |
25 |
and kernel-mod.eclass files along with the necessary changes to the |
26 |
affected ebuilds in 5 days from now. |
27 |
|
28 |
Sincerely, |
29 |
Brix |
30 |
-- |
31 |
Henrik Brix Andersen <brix@g.o> |
32 |
Gentoo Linux |