Re: Essentiality of Bash
On Sun, 27 Jun 2010 10:31:32 +0200, Josselin Mouette <firstname.lastname@example.org>
>Le samedi 26 juin 2010 à 22:30 +0200, Marc Haber a écrit :
>> On Sat, 26 Jun 2010 15:20:46 +0200, Josselin Mouette <email@example.com>
>> >Furthermore, I’d be interested to know how to fix such a “shortcoming”
>> >in our software. If both A and B depend on each other, A.postinst must
>> >be executed before B.postinst, and vice versa.
>> That's only one kind of circular dependency, and the one making the
>> problems. When you look at the circular dependencies, for example
>> within the exim4 or the aide packages, you see that the dependencies
>> defined in there aren't _that_ strict.
>You could say it’s not problematic all the time - that’s the case of
>fcron/exim4, which is only a problem for users installing fcron. But it
>is a real one: if fcron requires exim to be running to work correctly,
>while exim requires fcron to be running to work correctly, you’re
>In this case, there’s a solution: exim needs (f)cron installed on the
>system, but it doesn’t need it to be configured and running. But there’s
>simply no way for dpkg to know that, since the semantics of dependencies
>don’t express that order.
The problem is that we don't properly distinguish between "foo needs
bar during installation or foo's installation will fail" and "foo
needs bar to be installed or foo will not work". One could express
this via Recommends, but the sheer wording is too lose for this.
>For aide, I just don’t see the point: it’s the simplest, straight
>example of an unjustified 2-packages circular dependency.
explain "unjustified". This implies that the aide maintainers didn't
think when they established the dependencies as they are.
-------------------------------------- !! No courtesy copies, please !! -----
Marc Haber | " Questions are the | Mailadresse im Header
Mannheim, Germany | Beginning of Wisdom " | http://www.zugschlus.de/
Nordisch by Nature | Lt. Worf, TNG "Rightful Heir" | Fon: *49 621 72739834