[coyotos-dev] Status and roadmap

[Jonathan S. Shapiro]

On Wed, 2007-01-17 at 13:06 +0100, Pierre THIERRY wrote:
> Scribit Jonathan S. Shapiro dies 16/01/2007 hora 09:46:
> > The kernel is something that is mostly specified
> 
> Does that mean that the kernel's specification is almost finished, and
> when it will be, it will have to be implemented?

The kernel specification has been public for quite some time, and the
implementation is already underway. We have nearly completed the
specification review.

However, "finished" has some caveats:

1. At this point we still reserve the right to move bit positions around
for optimization reasons.

2. We are still clarifying which invariants are enforced in which
places. This is a major output of the review.

> > If support for LISP-like languages is the place where you want to
> > help, and you know LISP, one obvious candidate is to work on getting
> > BitC re-implemented in BitC to the point where it is self-bootstrapped
> > and has a stand-alone runtime.
> 
> Well, Lisp languages self-bootstrap is a domain that I'm definitely
> interested in, so I'll take a look at that.
> 
> What part of Coyotos, in short and long term, is planned to be written
> or rewritten in BitC? Everything?

It's very hard to answer this until we see how much BitC can do.
Speaking for myself, I don't think it makes sense to try to rewrite the
great bulk of code out there. There is a lot of C code that is horrible
but relatively non-sensitive. The place where the effort of BitC makes
sense is where we want to be able to analyze the program for some
reason. This would certainly include kernel and primordial domains, but
also reference monitors.

shap