Date:        Tue, 24 Oct 2000 16:23:02 +0200 (MET DST)
    From:        Paul Kranenburg <pk@cs.few.eur.nl>
    Message-ID:  <200010241423.QAA21700@kaa.cs.few.eur.nl>

  | I'd say that any error generated by readdisklabel() could just be
  | silently ignored,

It probably should be on any kind of removable media (and arguably,
all the time).

  | So, one possible `unexpected side effect' might be that the sector size
  | in the label does not match the actual sector size used by the device.
  | A cleverly constructed label would allow this to be fixed by applying
  | a simple `geometry translation'.

But do you really want to try and cram all this in before 1.5 ?
A working sparc CD boot method is needed there, and I'd think, without
mass revision of all kinds of other (mostly) unrelated code.

  | You can't make it "huge" if you want to use the "floopy image" approach
  | since you're still constrained by the amount of memory available for
  | bootstrapping, which currently maxes out at approximately 3MB.

Sure - but for present purposes, that should be enough, right?

kre