> 	Yep. I am still fiddling with trying to compile the NetBSD
> filesystem code under MacOS. I have some crashing related to the
> memory allocation pool stuff, but I cannot track it down.

Gag.  I remember that in the installer, we basically disabled the disk
buffers because we couldn't get them to work--that's one reason the
Installer's painfully slow.

Are you still using CW?

> 	From memory, it still has the old 'zlib' code you hacked in
> a few years ago.

That's what I figured.

> > I'd also like to play around with some
> > ideas for bootstrapping through a boot block so that we can build the
> > bootblock with libsa and be done with ffs version incompatibilties.
> > Then the booter just boots the boot block and we are closer to having
> > the option of a vastly simplified booter.
> 
> (Thought -
>  even better would be writing an A/UX partition containing the
>  bootblock, so that we may be able to have a Mac boot from it
>  _without_ a Booter.
>  It is all documented in Inside Macintosh, vol. V, in the Disk
>  Partitioning section, pp.576-582, but I haven't had any luck yet)

Yeah.  That's the direction I'd like to eventually go.  Going to an
installboot (awesome, Bill!) is step one.  Getting as much information
out of the booter/kernel interface as possible is step two.  Getting
the booter to boot an intermediate bootblock is step three, and eliminating
the booter (optionally) is step four.

Another reason for doing the intermediate bootblocks is to be able to
net-boot test kernels over DHCP/BOOTP/TFTP...  :-)

-allen