Re: The idea of an own L4

[Jonathan S. Shapiro]

On Tue, 2005-10-11 at 13:21 +0200, Ludovic Courtès wrote:
> "Alfred M. Szmidt" <address@hidden> writes:
> 
> > I just hate all this redesign crap, and I'm sure there would be more
> > progress and easier progress if things were done incrementally.
> 
> That's what has been happening over the last years.  There exists an
> implementation on top of Mach out of which lessons were learnt.  Marcus,
> Neal, and other people started to redesign various low-level things on
> top of the L4 of the time (look at the `doc' directory in the
> repository).  Eventually, people started to implement it.  Eventually,
> it turned out that some of the decisions made were arguably flawed.  And
> this is where the Hurd on L4 is: trying to fix these flaws.
> 
> I consider this to be incremental actually.

Ludovic: I understand why you say this, but self-delusion does not serve
us. I am sure that many of the existing hurd mechanisms will move with
only small change, but we are certainly discussing a complete rewrite of
the foundation. It does not serve us to pretend otherwise.

The question is whether it is worth it. I think the answer is yes.
Anyone who has read Ford's work on improving Mach will understand that
there is only so far that incremental change can take you in a given
design. At some point, it is necessary to make a major change.

The question for Hurd is: is now the time? This is something that, of
course, I cannot answer.

shap