[Aptitude-devel] Bug#844300: Bug#844300: nvidia-driver-libs:amd64: upgrade failure due to dependency issue

[Axel Beckert]

Control: tag -1 + moreinfo

Hi Vincent,

Vincent Lefevre wrote:
> On 2016-11-22 09:10:01 +0000, Luca Boccassi wrote:
> > On Tue, 2016-11-22 at 04:27 +0100, Vincent Lefevre wrote:
> > > On 2016-11-22 00:37:14 +0000, Luca Boccassi wrote:
> > > > In the end, you shouldn't have let aptitude remove the packages. It can
> > > > happen from time to time on unstable to have temporary inconsistent
> > > > state in the apt tree (that's why it's called unstable), for example in
> > > > this case it was probably because the new amd64 version was up in the
> > > > repo but the i386 was still being built/published.
> > > 
> > > The problem here is that aptitude said that the packages were
> > > no longer used, i.e. there were no dependencies on them. This
> > > is very misleading.
> > 
> > I have no control over what text aptitude outputs, I suggest contacting
> > the aptitude maintainers if you have a suggestion regarding that.
> > 
> > > Still, there are missing Breaks.
> > 
> > No, there are not.
> 
> So, this is a bug in aptitude.

Not necessarily.

> In any case, if the dependencies are correct, the package system
> should never be put in a broken state.

I'm sorry but that's wrong. Maintainer scripts can still put packages
in a broken state even if dependencies are correct.

>From an earlier mail of you:
> But an upgrade should not fail even when the apt sources are not up
> to date: that's the goal of Depends and Breaks to make sure that the
> package system remains in a consistent state.

Also wrong. Example: Not uptodate package lists can lead to download
failures (aptitude asks then if it should continue anyways) and hence
(if the user decides to continue) to packages where dependencies are
missing as they were not downloadable.

Anyways, in the whole correspondence I saw nowhere that you also tried
the same with apt or apt-get. Can you please check if upgrading these
packages with apt or apt-get leads to the same issue? (If so, it's
clearly no issue in aptitude.)

Additionally it could be helpful to provide a state bundle (man
aptitude-create-state-bundle) of the situation, either before the
issue appears (if you can reproduce it) or directly after it has
happened. TIA!

		Regards, Axel
-- 
 ,''`.  |  Axel Beckert <abe at debian.org>, http://people.debian.org/~abe/
: :' :  |  Debian Developer, ftp.ch.debian.org Admin
`. `'   |  4096R: 2517 B724 C5F6 CA99 5329  6E61 2FF9 CD59 6126 16B5
  `-    |  1024D: F067 EA27 26B9 C3FC 1486  202E C09E 1D89 9593 0EDE