Of course we don't have to approach that, the answer is simple:
Don't
rely on device-names and/or physical position of the hardware. Here are
two reason: a) users don't understand it; and b) sysadmins don't need to
understand it. That's why e.g. hal/fstab-sync put /media/<fslabel>
in /etc/fstab so you can just label your disks and expect them to be
mounted at the appropriate mount point.
I'd wish people would file bugs against software, especially desktop
software, that forces people to rely on device names. We need to move
away from that.
1. sw-raid arrays - how do you not rely on device names when trying to
fail-out a drive?
2. adding a new disk to a system. The new disk has been a boot disk for
another older system. Suddenly you have two partitions labeled /
and /var. How do you get out of that w/o booting up and making fstab
mount via device name?
-sv