Bug#519469: apt-get proposes to autoremove packages not yet installed
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512
It looks like this happens when:
* One version of a package is installed.
* A newer version of that package is available from the repositories.
* The newer version depends on a package which the currently-installed
version does not.
* The new-dependency package is not currently installed.
* The original package is for some reason marked as "kept back" by the
dist-upgrade logic, and so will not actually be upgraded.
In this circumstance, apt-get incorrectly selects the new dependency for
installation as a new package by dist-upgrade, even though the package
version that depends on it is not actually going to be installed. It
then correctly lists the newly installed package for autoremoval,
because nothing that's actually installed depends on it.
Here's the details of my case, from which I came to this conclusion:
I run an amd64 system, with i386 enabled on multiarch.
jackd1, libjack0, libjack0:i386, and libjack-dev are installed on my
system.
Both libjack-dev and jackd1 depend on libjack0.
The newest available version of jackd1 adds a dependency on
libzita-alsa-pcmi0 and libzita-resampler1, and the newest version of
libjack-dev adds a dependency on uuid-dev. None of those three packages
are currently installed.
Due to bug 740708, the current testing version of libjack0 is not
coinstallable with the current testing version of libjack0:i386.
As a result, libjack0 is "kept back" on dist-upgrade, causing
libjack-dev and jackd1 to also be kept back.
However, the new dependencies of those packages are still selected for
installation by dist-upgrade.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/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=cAO7
-----END PGP SIGNATURE-----
Reply to: