On Sat, Apr 07, 2001 at 02:57:25AM +0900, itojun@iijlab.net wrote:
> 
> >I think that would be incredibly dumb, confusing, and unnecessary.
> >All of those magic IPv6 routes are added in the SAME script that handles
> >$flushroutes.
> 
> 	there are "magic IPv6 routes" that will be added prior to
> 	/etc/rc.d/network, by the kernel.  as jinmei pointed out, with
> 	the netbsd-current code at this moment it does not raise problems.
> 	in the near future, with latest updates from kame tree, there will
> 	be an problem.

So, this is new behaviour in KAME?  If these routes are required, why
does the network stack allow them to be deleted at all?  And why is the
network stack adding routes like this in the first place, when it can
be done just as well by user-level code?