Re: Hurd/L4 active?

[olafBuddenhagen]

Hi,

On Sat, Apr 14, 2007 at 02:14:10PM +0200, Martin Barth wrote:

> what can anyone do if he or she wants to help at the micro kernel
> thingie?

You could help out with ngHurd, once Marcus and/or Neal take up the ball
again.

You could revive the existing (partial) Hurd/L4 port, moving it to one
of the new L4 variants; or start a new port of the existing codebase to
some modern microkernel.

You could join the x15mach development. This is a new variant of Mach
being implemented from scratch, but cleaner, more modern, and stripped
of any functionality not actually needed by the Hurd. The idea is to get
a kernel that is better maintainable than gnumach, and makes it easier
to make fundamental improvments. It's still far from coplete, however.

Or you could just help with improving the existing gnumach -- which is
still the only kernel actually used in a working Hurd implementation,
and thus the only one where improvements will immediately benefit the
Hurd...

> Spending time for the mach-kernel seems not very clever, if mach is
> going to be replaced.

Not at all. The Mach-based Hurd implementation won't go away any time
soon.

And even when one day the mainline Hurd moves to a different
microkernel, or fundamental changes are made to Mach (effectively
turning it into a different microkernel), any experience with earlier
improvments will be extremely valuable. After all, we can't do better
with a new variant, without first getting a thorough understanding what
to watch out for :-)

-antrik-