I'm not sure why, but AFAIR I've never done a dist-upgrade
that was
entirely successful. That is true to my Debian system at home as well
as to various other Ubuntu systems I've laid hands on.
My Debian system was a mix of stable and testing packages, and somehow
aptitude
came to a point where it'd just not be able to resolve some
dependencies, meaning I could not install nor remove any new packages.
So I opted for a dist-upgrade
to Testing. Here's how it went, roughly:
aptitude
tries installing libc
,
where the installation process will go into a deadlock. Error
message and solution is posted
here
(yes, you have to manually patch a Perl module). linux-base
has
the same problem.aptitude
won't do any
further upgrading. apt-get dist-upgrade
continues, though.pcspkr
module is appearently now called
snd_pcsp
and is thus loaded, strangely overriding all other sound
drivers. I blacklist it manually, reboot.libcaca0
wants to overwrite
a file which is also present in libcucul0
. Purging libcucul0
will reinstall libcaca0
, which in turn will fail. Turn to manual
override and uninstall all packages libcaca*
and libcucul*
using
dpkg
.xterm
. Inspect the
situation: rxvt
and rxvt-unicode
are missing. Well, thanks, so
here I go, aptitude install rxvt
. But guess what: apt-get
uninstalled aptitude
! Yeah, great.apt-get install rxvt-unicode
, finally a decent terminal. Fire up
mplayer. It's not installed any more. Like, what?!mplayer
, start it on a file. No sound. After various
tries, a pattern aplay
works just fine.mplayer -ao alsa
works. Put it into the config. Happy now with
music.Now, what I ask myself: Manually patching a Perl module, manually
resolving dependencies and invoking dpkg
, being (momentarily)
deprived of the few programs I use on a daily basis ... how will this
packaging system ever be remotely feasible for someone who's not an
expert of sorts and well-versed in debugging in a Unix wold?
Update: Some crap program said it depended on DECnet
stuff. After a reboot, this caused the MAC addresses of all my
interfaces (yes, both wireless and wired!) to be the same, i.e.:
aa:00:04:00:0a:04. Solution is removing libdnet*
and dnet-common
. – You kidding me?!