"John F. Woods" says:
    > Has anyone ever looked into having BSD (generically, not just NetBSD)
    > properly understand multiply-zoned disks?

But would that really help all that much?  With the large caches on modern
SCSI and IDE drives, I'm not sure that the old FFS optimizations based on
geometry are meaningful.  The simplicity of not caring about the geometry
might be a win these days.  Does the LFS in 4.4 care about geometry?  I confess
total ignorance on that.

On another topic, it looks like I may have to run Linux in addition to NetBSD
in the near future, as I want to run some software that will likely be ported
to Linux but not to *bsd.  It would be great if I could share at least one
partition between the two systems.  As far as I know, there is no filesystem,
other than MS-DOS (eeek!), that is common to both.  This is most sad.  How hard 
would it be to hack the linux filesystem into the BSD kernel?  I might be
tempted to try this.  I've also wondered if it might be possible to add
the hacks to the MS-DOS filesystem that Microsoft has put into NT 3.5 and that
will be in Windows 4.0^H^H^H95 that allow for long, multicase file names.
This would be just barely tollerable.  These changes are documented in the
new book on the NT filesystem or in the Windows 95 book, I forget which.

This is all probably not appropriate for current-users, so if you have ideas on
this, please send me mail.  No criticism intended to any of the free Unix
developers but the interoperability, and nearly all levels, of all the current
free unicies (and BSDI's products) really sucks rocks.  It would be nice to
start fixing that.

Jim