On Thu, Aug 08, 2002 at 01:54:26PM +0100, Rasputin wrote:
> * Greg A. Woods <woods@weird.com> [020807 22:21]:
> > 
> > I think there's enough code in user-land that makes certain assumptions
> > about what privileges are required in order to bind a socket (either for
> > listening or for the source port) to a port <= 1024 that such an option
> > really should always only be a compile-time option, and one documented
> > with grave warnings attached at that.  SSH for example makes many such
> > assumptions.  I.e. you really must know exactly what you're doing before
> > giving up this protection on a given machine.
> 
> Does this change win anything over using something like ipnat to
> redirect inbound port 80 traffic -> 8888 , for example?

IIRC ipnat doesn't allow redirecting traffic between ports on the same
interface.

	Cheers,

	-- Johnny Lam <jlam@netbsd.org>