After a recent problem with my Debian testing system, I have to question the thought process behind two build decision in the Debian packages.
First off, in the kernel, why is net-pf-1 included as a module? Just about every system needs Unix Sockets to run well. And in my case, I was using the Debian CD as a rescue disk to boot my own system where net-pf-1 was required my another kernel option to boot. When the rescue disk couldn't find net-pf-1 attempted to write to a ksymoops log, but because the root partition was still in it's first read-only pass, I got caught in an infinite loop of (roughly): modprobe: Can't locate module net-pf-1 modprobe: cannot create /var/log/ksymoops/20010323.log Secondly, I'm surprised the default install of LILO is not statically built, considering it role in booting, and therefore often fixing a bad MBR. The current testing LILO (21.7) now won't run from the 2.2r2 rescue disks, which is a real pain. Can anyone explain or support these design decisions? Anm
