Re: Installing packages to a foreign system
- To: Phillip Susi <email@example.com>
- Cc: Goswin von Brederlow <firstname.lastname@example.org>, email@example.com
- Subject: Re: Installing packages to a foreign system
- From: Goswin von Brederlow <firstname.lastname@example.org>
- Date: Mon, 05 Oct 2009 11:08:20 +0200
- Message-id: <email@example.com>
- In-reply-to: <4AC26641.firstname.lastname@example.org> (Phillip Susi's message of "Tue, 29 Sep 2009 15:55:45 -0400")
- References: <4AB3F39A.email@example.com> <firstname.lastname@example.org> <4ABBDEE3.email@example.com> <firstname.lastname@example.org> <4AC26641.email@example.com>
Phillip Susi <firstname.lastname@example.org> writes:
> On 9/28/2009 7:06 AM, Goswin von Brederlow wrote:
>> Phillip Susi<email@example.com> writes:
>>> This looks like it does the trick, but I am curious; is this how the
>>> installer does it? When the installer is first installing the system
>>> it also needs to install the packages to the hard disk, but without
>>> having them interfere with the running state of the installing system.
>>> Is this how it does that?
>> It is what policy dictates for every package. Would be stupid to
>> invent something else.
> So is a package broken if its configure script invokes tools that rely
> on a running daemon, and the configure fails if the tools can't
> contact the daemon? Like say, udevinfo?
It gets even worse. What if it contacts the daemon runing outside the
chroot and gets totaly the wrong information?