Re: Hurd/L4 active?

[olafBuddenhagen]

Hi,

On Sat, Apr 14, 2007 at 02:38:53PM +1200, Shams wrote:

> I read somewhere that instead of L4, Hurd could be using L4.Sec as the
> microkernel? Is this still a possibility?

Yes, L4.Sec is among the kernels that could be used for ngHurd.

It should also be possible to port the existing Hurd implementation to
L4.Sec or some similar variant (either picking up the Hurd/L4 work done
so far, or starting a new attempt), though currently nobody is planning
to do so AFAIK.

> > rather an experimental/research subproject.
> 
> Why another subproject, why not just develop/experiment with the
> existing Hurd?

The ideas proposed for ngHurd are not minor changes, but totally new
concepts. While personally I do believe that most of these things could
be implemented on top of the current Hurd/Mach code, experimenting with
totally new concepts can be easier when starting from scratch. Of
course, if some of the ideas turn out to work well, it's possible and
desirable to implement them in the mainline Hurd also -- or even make
ngHurd the mainline.

> Btw who calls the shots at which microkernel Hurd is going to be using
> and the development path for Hurd.

Well, the original Hurd/L4 port, and the ngHurd concepts that sprang
from it, are the brainchild of Marcus Brinkmann and Neal Walfield; so
they are the ones pointing the direction here.

As for the existing Hurd/Mach implementation, those people who presently
work on it, are focussing on specific improvments rather then
fundamental changes; so the question of switching to a new microkernel
doesn't really arise currently.

But if anyone starts working on a port of the existing codebase to a new
microkernel again, there is nobody to dictate the direction to go. It's
really up to individual developers -- if they consider it a worthwhile
effort, they will join; otherwise, they will stick with Mach or ngHurd.

> Is it people like RMS, Thomas, Roland or who else or is no one
> incharge of this project?

The role of these people is a bit hard to describe. RMS was never
involved in Hurd development, except that he participated in the
original design a bit I believe. Consequently, he has no power to make
any decisions for the Hurd developers. Being the leader of the GNU
project however, he is the one setting the goals for GNU as a whole and
the Hurd as it's kernel. Thus he indirectly has some influence on the
direction the Hurd takes.

Thomas is the main architect of the original design, and did a large
part of the implementation; but he is retired from the project now. He
only gives advice from time to time, and he is not involved in any of
the new developments like Hurd/L4 or ngHurd.

Roland is the official Hurd maintainer, i.e. he ulitimately decides what
goes into CVS HEAD of the mainline implementation. Aside from that
however nowadays he also has more of an advisory role. He isn't involved
in Hurd/L4 or ngHurd either.

In general, nobody has the power to say what Hurd developers will work
on. After all, this is a voluntary project.

-antrik-