On Thu, May 04, 2006 at 06:48:41PM +0200, Arnaud Lacombe wrote:
> On Tue, May 02, 2006 at 08:28:13PM -0500, David Young wrote:
>
> > I strongly prefer an in-kernel implementation, because I believe it
> > will perform better on low-end machines (486-class, like the Soekris
> > net45xx) than a userland implementation.  I have something in mind like
> > the pseudo-device, stf(4).
> ok, in fact I was looking for somethings portable. stf(4) device could
> also be used :). Directly using device will also ease the work.

Let me try to make things more clear: Miredo is *already* a userland
implementation that *already* works on NetBSD.  Thus, it is hard to
see how we could approve (or Google could approve) a Summer of Code
project to do what Miredo already does.

An in-kernel implementation would be very worthwhile, and, as David
points out, could share code with some of the several other protocl
translators and tunnel encapsulations in the kernel.

Thor