Le mardi 24 novembre 2009 à 13:18 +0100, David Kalnischkies a écrit : > The problem simply is that the package libesd-alsa0 has a higher > scoring than libesd0 and therefore apt thinks it is a bad idea to remove > libesd-alsa0 (you can see this with the debug option for the resolver > yourself) in your case -- this should fix itself after a few packages > remove the or-group as this will lower the scoring (and depending on > non-exisiting packages is ++ungood as you can see now). OK I get it. So one of the possibilities is to binNMU most reverse dependencies to get rid of the libesd-alsa0 dependency. > btw: The package doesn't have a source anymore, but it is installable > (if dependencies wouldn't forbid it) as it is still available in the archive - > but don't worry, it should disappear automatically which could improve > the situation a bit (Not much as the scoring penalty for this is low). > And as i try to describe above: As APT don't know that it is a transition > it think these are two packages which conflicts - maybe because of a > simple file and therefore doesn't provide the same functionality. > (And no, Replace doesn't have this transition semantic either) > It would therefore need to check if this Conflict stands for a transition by > checking every dependency in which both packages are involved - > and this for every Conflict which exists in the archive... … and the other possibility is to introduce a transitional libesd-alsa0 (which I wanted to avoid). Would a Provides+Replaces+Conflicts, as policy allows, help? Thanks a lot for your deep analysis BTW. Cheers, -- .''`. Josselin Mouette : :' : `. `' “I recommend you to learn English in hope that you in `- future understand things” -- Jörg Schilling
Attachment:
signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?=