On Wed, Apr 14, 1999 at 05:01:18AM -0400, Adam Di Carlo wrote: > Enrique Zanardi <[EMAIL PROTECTED]> writes: > > > Yes, liloconfig step on dbootstrap is one of the "rough spots". > > If someone has a little time to spare improving it, feel free to join us > > in debian-boot and help us make it better. > > Should we work maybe on just using/adapting/extending the existing > /usr/sbin/liloconfig script? If not, why not?
I'd say no: - As of lilo 21-4 /usr/sbin/liloconfig basically does the same that "liloconfig on dbootstrap": 1.- detect which partition is the root one and write a lilo.conf with a single "Linux" entry using that partition for boot and root 2.- install mbr on the MBR of the same disk where the root partition is. 3.- activate the root partition But: - It doesn't detect "wierd cases" (IDE and SCSI on the same machine, root on the second hard disk, ...). - It doesn't use our "windowing libraries" (newt or Ben's graphics library). - It's not internationalized. So, there's no point in working on that, as it does less than "liloconfig on dbootstrap". IMO, it would be better to extract "liloconfig on dbootstrap" to a stand-alone tool, that replaces /usr/sbin/liloconfig > FYI, there are many lilo bugs in boot-floppies, search for lilo. This > is a pretty good log of the missing functionality. It seems to me > that /usr/sbin/liloconfig just needs to be hacked to ask for an > optional 'append' string, and maybe adapted for our needs (i.e., > alternative mount point, i.e., /target, for installling lilo prior to > booting into the new base system, and we're done. Look again. Every lilo "bug" (most of them are wishlist items) in boot-floppies is also in /usr/sbin/liloconfig, and there are a few ones fixed on boot-floppies that are still present on /usr/sbin/liloconfig -- Enrique Zanardi [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
