On Sat, Jul 10, 2021 at 12:37 AM Andreas Beckmann <a...@debian.org> wrote: > > On 10/07/2021 09.16, Otto Kekäläinen wrote: > > Seems this issue only happens with apt – not when using apt-get. > > Yuck. Can you ask the apt developers whether this is intended? I always > expected apt-get and apt to be functionally equivalent (i.e. to always > take the same decisions when installing/upgrading/removing packages > since they use the same resolver in the background). > (If they are not, testing apt, too, means doubling the number of > piuparts tests...)
Filed https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=990922 > Have you tried default-mysql-server + (roundcube-core (or roundcube) or > cacti)? This yields the unwanted behavior (removing default-mysql-server > + mariadb-server-10.3 without installing mariadb-server-10.5)) with > apt-get in my piuparts tests both with and without --install-recommends. > I can find more examples that only work with --install-recommends. Tested Roundcube + default-mysql-server upgrades with both apt and apt-get, and with and without --no-install-recommends in https://salsa.debian.org/mariadb-team/mariadb-10.5/-/commit/c3f53fe664517bbce7f8ae67244e018f9086534c All passed fine.
