") Fcc: +sent-mail On Thu, Apr 07, 2005 at 12:59:19PM +1000, Anand Kumria wrote: > On Wed, Apr 06, 2005 at 10:35:36AM +0100, Mark Brown wrote:
> > That's not buggy: > Yes it is, if this only existed in /etc/network/interfaces > iface eth0 inet6 static Oh, you're thinking of the Debian configuration done by the package - I was thinking about the program itself. > That is it a bug if an an IPv4 link local address appears on the eth0 > interface. If there were also a line like: > iface eth0 inet static > *then* an IPv4 link local address should be assigned. Yes - there is an argument for supplying a link local address anyway (by analogy with IPv6) but it's very dodgy. > > interfaces it can (IPv6 address assignment is rather different to IPv4 > > so this is can sensibly be done by the kernel - there's no policy in > > having an ethernet link local address, for example). > Actually the defense mechanism isn't so different -- just the allocation > method. I've been thinking about factoring the link-local defense state > machine out of the IPv6 layer in the kernel and then using it for any > addresses marked as IPv4 link-local. Does it actually use the defense mechanisms for link local addresses allocated on the basis of hardware address (which was the set I was thinking of as most different - as you say the zeroconf algorithm owes something to the way autoconfigured routeable addresses are assigned)? Not that it really matters for this. -- "You grabbed my hand and we fell into it, like a daydream - or a fever." -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
