[pocksuppet]

I think that's actually about avoiding NIC to CPU traffic. NICs have multicast address filters which determine which packets to receive, but they always receive broadcast packets. It would have been more useful in the 90s, when NICs weren't so programmable.

It's pretty silly anyway. NDP is more of a layering violation than ARP, because now IPv6 has a stupid circular dependency on itself. Mapping L3 addresses to L2 is a layer below 3, it is not part of layer 3, it is part of the sub-layer that adapts between 2 and 3. DHCP should be part of that sub-layer, too.

Did you know that for every kind of network that IP can run on top of, there's a whole separate standard specifying how to adapt one to the other? RFC 894 specifies how to run IP over Ethernet networks. RFC 2225 specifies how to run IP over ATM networks.

[holowoodman]

IMHO all this whining about "layering violations" is stupid. One will always need some kind of layer glue, neighbors bordering on each other need to know something about each other, correlate addresses, etc. It is impossible to do anything practical without such violations. And it doesn't really matter if that glue protocol belongs to the below layer, the above layer or is a weird hybrid of both. Because in the end, the glue will necessarily be a hybrid and it will be specific to the combination of both those layers.

The only thing one should really really really avoid is the TCP mistake of not just having some minimally necessary glue, but that tight coupling of TCP connections to IP addresses in the layer below.

[pocksuppet]

It's a layering violation because it makes the inter-layer-glue more messy than it should be. IPv6 having a circular layer dependency on itself is not a good thing. It makes that glue messy. ARP is cleaner glue.