1 |
On Friday, March 16, 2012 02:54:16 PM Pandu Poluan wrote: |
2 |
> On Fri, Mar 16, 2012 at 14:32, Walter Dnes <waltdnes@××××××××.org> wrote: |
3 |
> > On Fri, Mar 16, 2012 at 08:41:48AM +0700, Pandu Poluan wrote |
4 |
> > |
5 |
> > > Hmmm... are you planning to host an overlay? |
6 |
> > > |
7 |
> > > If so, I'll be willing to donate some of my time to provide some |
8 |
> > > patched ebuilds for packages that can function without udev but |
9 |
> > > lazily specify DEPEND=sys-fs/udev... |
10 |
> > |
11 |
> > I wouldn't call it "lazy". Before the hulabaloo about udev/initramfs, |
12 |
> > I don't think anybody was running mdev on Gentoo. So there was no need |
13 |
> > for mdev in the ebuilds. The only non-embedded distro to use mdev was |
14 |
> > Alpine linux. And they also use uclibc. |
15 |
> |
16 |
> Ah yes, sorry. That was originally tongue-in-cheek, but I now see it |
17 |
> may be too disparaging. My bad. |
18 |
> |
19 |
> > I'm not familiar with the server side of things. I can follow |
20 |
> > instructions if supplied. I don't know if the hosting provider I'm |
21 |
> > thinking of does rsync. I never theought to ask. |
22 |
> > |
23 |
> > I'll check on the dev list about the etiquitte regarding contacting |
24 |
> > upstream. Even if a package works today with mdev, there's no guarantee |
25 |
> > about tomorrow. It'll help if upstream knows that people are using |
26 |
> > their packages with mdev, and they take that into account when updating |
27 |
> > the software. |
28 |
> > |
29 |
> > Note that my request for updating virtual/dev-manager went through OK. |
30 |
> > Once we test a "udev-required" package with mdev, and confirm it works, |
31 |
> > we should post a request on the Gentoo bugzilla to update Gentoo's |
32 |
> > ebuild. |
33 |
> |
34 |
> Good idea. |
35 |
> |
36 |
> > > ... and while at it, let's see if I can make a package containing |
37 |
> > > scripts to ease transitioning from udev to mdev. Maybe call it, |
38 |
> > > sys-utils/mdev-helper? |
39 |
> > |
40 |
> > The kernel reconfig and rebuild, and sticking "init=/sbin/linuxrc" |
41 |
> > into the append line are user-specific. I dual-boot 2 kernels |
42 |
> > ("production" and "experimental"), and I run lilo. Somebody with only |
43 |
> > one kernel, and/or running GRUB will need to do things differently. So |
44 |
> > a script won't help. This is simple enough to copy+paste from docs |
45 |
> > to your terminal. |
46 |
> |
47 |
> Well... as to the kernel requirement... nothing's stopping one from |
48 |
> emerging sys-fs/reiserfsprogs even when the kernel doesn't support |
49 |
> reiserfs ;) |
50 |
> |
51 |
> The "init=/sbin/linuxrc" can be automated using script (and sed), |
52 |
> which we can imbue with the intelligence necessary to edit LILO/GRUB |
53 |
> conf. I do have a collection of my own scripts to make it easier to |
54 |
> install new Gentoo systems; one of them I whup up to automatically add |
55 |
> a new kernel into menu.lst and (optionally) modify the default kernel |
56 |
> [1]. |
57 |
> |
58 |
> What I have in mind for "helper scripts" would be (for example) a |
59 |
> script to ensure that, on boot, ethernet devices will maintain their |
60 |
> relative order. This needs to be stuck into /etc/mdev.conf (already |
61 |
> part of stage3). |
62 |
> |
63 |
> (And if someone's well-versed enough in Linux, maybe he/she will |
64 |
> convert the shellscript into a simple -- and faster -- binary with |
65 |
> exact same functionality). |
66 |
|
67 |
I think you are talking about a script that handles a more "dynamic" database |
68 |
to force renames/softlinks for devices keeping names identical? |
69 |
I haven't played with mdev yet, but isn't that already in mdev? |
70 |
Or does mdev require it to be set manually? |
71 |
|
72 |
Btw, the "keep same devicename" is rather annoying when having to replace the |
73 |
network card and the network then doesn't come back up... |
74 |
|
75 |
-- |
76 |
Joost |