--- Toshio Kuratomi <a.badger(a)gmail.com> wrote:
> Here are a couple thoughts I had while skimming
> 1) inst /sbin/busybox.anaconda should probably be inst
> $rootdir/sbin/busybox.anaconda (which would require it to be installed on
> target system)
Bah -- Since kadischi knows the location of the repository, how about
extracting busybox.anaconda from the repository rpm and inst'ing it?
Yup, an rpm -i --nodeps --noscripts --root=/tmppath oughtta do it.
> 2) inst $INSTALLDIR/initrd/find-live-cd
> makes me think this could be your problem. I.e. this would appear to me to
> (in your case) installing a find-live-cd binary which was compiled on a
> system into an initrd which is meant to be run on a 32-bit system.
Well, you and AutOPSY both had the same idea. Unfortunately, it's not
the problem. I replaced both find-live-cd and scanswap with i386
versions and I still get the same error.
Yup, I think the answer here, would possibly be requiring a kadischi rpm to be
available for the target arch in addition to the target arch's repository.
Then an rpm call like the one above could be used to pull it.
I've also burned an actual ISO and tried to boot that on a 32bit
computer and received the same error. I'll try hacking something
together for the busybox issue sometime tomorrow and see if that gets
As you can guess, I chalk the above issues up to the same issue I've been
harping on about the installer having many bugs(as I call them) pertaining to
it's incorporating the host system's state (system files, hardware status) into
the target system.
I.e, I seriously doubt that enough (if any) focus has been put into seeing just
how well behaved anaconda --rootpath will be when it is run on an x86_64
system, but with a 32bit repository intended for a 32bit system. Just
speculation for now though.
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around