On Thu, Oct 02, 2003 at 08:40:40PM +0200, David Fokkema wrote: > On Thu, 2003-10-02 at 16:58, Colin Watson wrote: > > On Thu, Oct 02, 2003 at 09:43:26AM -0500, Ron Johnson wrote: > > > Don't go directly from stable to unstable!! Too radical > > > > It's not significantly more radical than stable to testing. I don't > > think there's much benefit in trying to go from stable to unstable via > > testing, especially considering that there are things broken in testing > > that work in unstable for one reason or another (e.g. apt-listchanges). > > I'm sorry, I have to disagree with you on this one. Several people on > this list have complained about broken dependencies and uninstallable > packages when upgrading from stable to unstable. Upgrading to testing > and then to unstable solved their problems. I was one of them and IIRC, > I've responded to at least two others on this list with the same > problems. When I bought my new laptop only a few weeks ago I had to deal > with the same issues. Going through testing solved them all. >
FWIW, I go from stable->unstable all the time, perhaps in a "better way". I install the base system only, during setup I don't run tasksel or dselect. Then, I install patch and aptitude from Woody, which picks a strange libstdc++ dependency. Then, I change my sources list to unstable, and upgrade the base only. Finally, I only install the rest of the system from unstable. I bet if I had a whole *bunch* of Woody installed and did an upgrade I'd see what you're seeing, and sometimes unstable is just broken anyway. But a couple times a week I blow away my system and do a fresh reinstall of everything (I've automated it to 30 minutes, including personal settings, everything). I believe in disposable systems (ala google or clusters).... -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
