As we noticed from the filesystem thread ( I rule ) that I tested the initscripts out this morning. I ran into a lot of retarded problems for simple reasons, udev 118-2 should depend on the new initscripts and the new initscripts should depend on netcfg. Let's watch what happened! # pacman -Syu # updated to new [testing] reboot % ping google.com # no response! % sudo /etc/rc.d/network restart # restarts, but doesn't ask for the menu? % vim /etc/rc.conf /etc/rc.conf.pacnew # rename the variable correctly % sudo /etc/rc.d/network restart # restarts, but still doesn't ask for the menu? % sudo pacman -U /var/cache/pacman/pkg/initscripts-2007.11-2-i686.pkg.tar.gz % sudo /etc/rc.d/network restart # asks menu, connects, yippie kay-yay! <.. > computer use % reboot Bad superblock on /dev/sda4? bad superblock on /dev/sda2? # ls /dev console null zero Udev didn't restart correctly because I was running the old initscripts with the new udev, and the new udev gets rid of /etc/start_udev (which is fine) except that the new udev REQUIRES the new initscripts to work. Get on the phone with Aaron, he helps me figure this hoopla out. pacman -S initscripts netcfg And now were at the present and me writing this email. Add the netcfg dependency to initscripts, and the initscripts>=2008.02-1 to udev 118, and I'm happy. // jeff -- . : [ + carpe diem totus tuus + ] : .

On Fri, Feb 22, 2008 at 12:22:32PM -0600, Aaron Griffin wrote:

On Fri, Feb 22, 2008 at 11:46 AM, Jeff Mickey wrote:

...

And now were at the present and me writing this email. Add the netcfg dependency to initscripts, and the initscripts>=2008.02-1 to udev 118, and I'm happy.

This is a little off. a) netcfg isn't required, so it'd be an optdepend. Still, an install message would be a good idea for this. b) udev 118 doesn't depend on initscripts. It's the other way around. Sure adding that dep would have fixed your issue, but it's still not the right way to fix this... the real fix is to re-release a udev package that still contains start_udev.

What about JGC's suggestion of using versioned conflict? I think it makes sense. http://www.archlinux.org/pipermail/arch-dev-public/2008-February/004899.html