Hotplug patches from Debian

Nathan Coulson conathan at conet.dyndns.org
Fri Jun 4 09:28:43 PDT 2004


> On Fri, 4 Jun 2004, Alexander E. Patrakov wrote:
>
>> >
>> >  My understanding, at least for the udev side, was that Greg K-H was
>> > using gentoo on one of his boxes and was mostly happy with the gentoo
>> > implementation of it.  Apologies if I've confused unrelated issues.
>>
>> If someone is happy, that means nothing. Someone is even happy with
>> Windows, despite all shareware crap, artificial limitations, pirated CDs
>> with non-working cracks, mail worms,...
>>
>  Yeah, but the udev maintainer probably counts as a little more than
> somebody.
>
>> My hotplug patches currently in the book are created by fixing the
>> problems I encountered, without any reference to Debian. I am not happy
>> with unpatched hotplug, and I can point to problems. And I _do_
>> understand what Debian patches accepted by me do.
>>
>
>  Good.  You're in advance of my understanding, I'm just trying to point
> out that sometimes there are different approaches to deciding where the
> error is, and how it should be fixed.
>
>> As for udev, it is probably not buggy by itself. At least I found
>> nothing (but that means nothing). It just changes the way of thinking
>> and requires some rewriting of bootscripts (most affected: ALSA).
>>
>> >
>> > [ mostly snipped ]
>> >
>> >>
>> >>050_net.agent_ifupdown: debian specific network management with
>> >>ifupdown. We have to do something similar sooner or later.
>> >>
>> >
>> >
>> >  The debian boot configuration management has to be the least-common
>> way
>> > of doing things, in my opinion.  I'd be very reluctant to follow their
>> > lead in anything to do with bootscripts.  (not generally anti-debian,
>> I
>> > like a lot of what they do, but some of it is over the top).
>>
>> The problem is that for every network device found, hotplug logs a
>> message: "How do I bring interfaces up on this distro?" We must either
>> teach hotplug to bring interfaces up on LFS or declare that we don't
>> support that.
>>
>>
>
>  Ah, right.

wont be fun, we dont want it in rcsysinit.d, or rc1.d, or rc0/2.d [I'm
assuming it goes up/down, based on if it's plugged in or not].

> Ken
> --




More information about the lfs-dev mailing list