Re: Boot-floppies hacking
On Thu, 16 Sep 1999, Martin Schulze wrote:
> One advantage of such a meeting+weekend would be that n people
> would meet, discuss and implement techniques face-to-face. The
> current situation is worst. There were some ideas, they would
> have to be tried out. Unfortunately one person cannot do this
I can't find from the July mail archives of this list any resolution to
the threads about 'strategy for boot floppies'. What's the plan?
Currently, I don't see how anyone can be building i386 boot floppies with
the standard Makefile from CVS and an up-to-date potato installation; for
instance, the build process depends on having a GIF-enabled libgd, which
no longer exists in potato (I have a patch for PNG).
With 2.2 kernels it's no longer feasible to follow the kernel + root
filesystem on one disk model. This makes floppy-only installations even
more of a pain. Would it therefore be sensible to admit defeat, and pursue
a less integrated set of install options? [floppy only, floppy + CD,
floppy + NFS for i386]. One thing which would be particularly useful (from
a user's point of view - from a maintainer's point of view it's an
annoyance) would be to use something other than an utterly stripped down
miniroot filesystem for installs and rescues; this is much more palatable
if we're not (as I'm suggesting) going to be constrained to what can be
crammed onto a floppy.
I'm going to hack away until I actually get a clean build on i386 (which
is currently breaking on unmet dependencies in an obviously
slink-flavoured task/profile list).
Mk
Reply to: