> >From what I understand (from talking with people who were at CMU when
> the mach work was happening), CMU Mach3 wasn't anything you ever
> wanted to run on your alpha, for speed and stability reasons.

As I said, it requires some minor fixes (otherwise it won't even
boot).  As for speed, since there are no fully functional systems
freely available (Lites sort of works a little), it's not really
relevant (besides, any microkernel version of Mach isn't blazingly
fast on any platform that I know of).  In any case, you'd have to be
doing moderately low level development work to want to run a Mach
microkernel.

My point was that it can work, whether it works well is a different
issue.  ;--)

Of course it may be a better idea to take Mach4 from Utah (or perhaps
even some GNU Mach release) and work on Alpha support for that (I
assume it's still mostly 64-bit clean, even without support for 64-bit
architectures, since someone claimed to have done a 64-bit MIPS port
of the Hurd on Mach4).  It depends on what you really want to work on
first (servers running under Mach or Mach itself).

Once again, I don't know about what exactly is available from OSF,
AFAIK nothing free (which may very well be relevant to someone posting
to a NetBSD list).  If restrictive licensing doesn't matter, it
probably is the best source for a supported microkernel.