Re: Microkernel Q

[nisse@lysator.liu.se (Niels Möller)]

"B. Douglas Hilton" <doug.hilton@engineer.com> writes:

> I am thinking that the Mach uKernel is severely hampering
> the Hurd project. As lond as Hurd only relies on mach.h it
> would be very effective to rewrite oskit-mach in Ada.

I don't think rewriting any or all pieces in a different language
would help much. I think there is some agreement that Mach is
suboptimal, in more fundamental ways. Mach is the µ-kernel that works
with the Hurd today. If you want to replace it, I would recommend that
you try helping the hurd-on-l4 project.

To get hurd-on-l4 moving, I think the most urgent tasks (please
correct me if I'm wrong) are, in order:

  x Porting libports to l4, and/or figuring out how to change the
    Hurd's ports abstraction to be less mach-centric and still fit the
    Hurd's needs.

  x Implement a threads library to work around the l4 limitations on
    the number of threads. The thread implementation on Solaris (which
    multiplexes n user pthreads on top of m kernel threads, n > m)
    should provide some ideas.

  x Implement similar multiplection for user tasks on top of kernel
    tasks.

The debian-hurd list is probably not the right place for discussion
about the l4 track, though.

As for language wars, I believe C is the language of choice for
systems programming on the GNU system, and I don't think that will
change anytime soon.

Not that I want to stop you from rewriting HURD-related stuff using
your language of choice, but I believe there are better ways to help
moving the Hurd away from its current µ-kernel, Mach.

Best regards,
/Niels