1 |
On Sunday 13 November 2011 13:04:57 Chí-Thanh Christopher Nguyễn wrote: |
2 |
> Mike Frysinger schrieb: |
3 |
> > until we have replacement for all of its tools, it's always going to be |
4 |
> > there. |
5 |
> |
6 |
> After net-tools is no longer needed for basic setups (which I understand |
7 |
> will be still the case after the proposed changes), why should it remain |
8 |
> in the system set? |
9 |
|
10 |
you snipped the following sentence where i explained why net-tools can't be |
11 |
removed. iproute2 replaces *most* of net-tools, not *all*. |
12 |
|
13 |
> >> openrc can already use busybox udhcpc instead of dhcpcd, so there is a |
14 |
> >> precedent. |
15 |
> > |
16 |
> > that's not the same thing at all. `udhcpc` is not intended to be a |
17 |
> > drop-in replacement for `dhcpcd`. we have a dedicated module to work |
18 |
> > with udhcpc, and we have another dedicated module to work with dhcpcd. |
19 |
> |
20 |
> So if someone creates a "busybox_ip" module (which would essentially be |
21 |
> a copy of the iproute2 module) then it will suddenly become the same thing? |
22 |
|
23 |
no. i don't know why you're trying to force fit an analogy that makes no |
24 |
sense. for busybox/embedded users, they'll have a system where `ip` is a |
25 |
symlink to `busybox`, and so the existing iproute module will be used |
26 |
unchanged. if there are problems, they can report. but that is a system |
27 |
where we know up front is drastically different vs an otherwise "normal" Linux |
28 |
system silently switching between the standard iproute2 and the simplified |
29 |
busybox replacement. |
30 |
|
31 |
the only way we could drop back to busybox's iproute2 is if we have the module |
32 |
issue quite a bit of verbose warnings, but even then i'm not sure. if people |
33 |
are creating network setups that need iproute2, and then then unmerge |
34 |
iproute2, then i'm not terribly sympathetic to their plight. |
35 |
-mike |