Re: Microkernel architecture?

[Kip Macy <kmacy@xxxxxxxxxxx>]

L4 is really cool until you realize how much work it would be to
anything interesting in its context. The current L4Linux doesn't have
anywhere near the performance relative to native Linux that the one
discussed in the paper did. The amount of optimization and tuning that
they had to do get it to that point just seems hard to justify in the
absence of a "killer app" or a large user base.


				-Kip

On Tue, 28 Oct 2003, David Leimbach wrote:

>
> On Oct 28, 2003, at 4:37 AM, Bill Huey (hui) wrote:
>
> > On Tue, Oct 28, 2003 at 11:14:35AM +0100, Michael Neumann wrote:
> >> I'll forward this post to the L4 mailing list, as I am not enough
> >> competent yet
> >> in answering these questions (only two weeks since I first heard
> >> about L4).
> >
> > Don't take it too seriously. They were just passing thoughts designed
> > to proke interesting conversation on this list and is not a commitment
> > to do anything about it on anybodys part. :)
> >
>
> It might be interesting to have an L4 emulator which provides the APIs
> so one
> can do work on a microkernel without a microkernel :).
>
> I think DragonFlyBSD would probably be one of the only non uKernel
> OSes that
> could provide a fail-safe environment for that.
>
> I don't know that I am interested in implementing such stuff though.  I
> have barely
> ever used L4 aside from the demo boot disks
>
>
> > bill
> >
>
>