1 |
On Sun, Jun 02, 2013 at 08:48:23PM +0200, Fabio Erculiani wrote: |
2 |
> On Sun, Jun 2, 2013 at 8:20 PM, Steven J. Long |
3 |
> <slong@××××××××××××××××××.uk> wrote: |
4 |
> > |
5 |
> |
6 |
> [...] |
7 |
> |
8 |
> > The whole symlink/boot/fallback thing is simply a waste of technical effort. |
9 |
> > And blanket "your opinion" and "you didn't comment a week ago, so I don't |
10 |
> > have to deal with the substance of your points" don't change that. |
11 |
> |
12 |
> Waste? We have multiple use cases for that as stated in several places |
13 |
> (here, bugzilla, IRC, etc). |
14 |
> If such use cases are of no interest for you, then you shouldn't be bothered. |
15 |
|
16 |
The specific idea of reimplementing the kernel fallback mechanism is a waste of |
17 |
technical effort. Not the whole idea of eselect init, as I stated several times. |
18 |
|
19 |
If you lose that idiotic idea, you have a lot less complexity to worry about and |
20 |
can instead get on with the *necessary* complexity: handling whether it is safe |
21 |
to switch boot, and how to make it so given the 'from' init and the 'to' init, |
22 |
which might require write access to the rootfs, eg to swap inittab, or to |
23 |
mkdir -p a necessary path. |
24 |
|
25 |
It could need anything, there's simply no way of knowing, and it will require |
26 |
maintenance over time as init-systems change. That's why the Unix inventors came |
27 |
up with sh. And the best people to maintain it over time, are the users in |
28 |
collaboration with the init-system devs, since they are the ones who have the |
29 |
testcases in front of them, and the motivation to make the software work. While |
30 |
the devs have a broader view, and an interest in keeping things portable (aka: |
31 |
'generic'.) |
32 |
|
33 |
Please try to read and consider my whole argument, instead of selectively quoting |
34 |
one part and using it to justify YAF ad-hominem. |
35 |
|
36 |
-- |
37 |
#friendly-coders -- We're friendly, but we're not /that/ friendly ;-) |