On Tue, Feb 23, 2016 at 10:34 PM, Fernando Gont <fgont@xxxxxxxxxxxxxxx> wrote:
The behaviour that we currently have is due to what implementations chose to do, not to what standards track RFCs say.
This is the first one I found:
Moreover, information
may also be obtained through other dynamic means like DHCPv6. Hosts
accept the union of all received information; the receipt of a Router
Advertisement MUST NOT invalidate all information received in a
previous advertisement or from another source.
This text is also unaffected by the current draft. The draft says that the host should decide whether or not to do DHCPv6 based on the information in Router Advertisements. That does not contradict RFC 4861: since DHCPv6 requires an explicit request from the host before anything happens, choosing *not to start* DHCPv6 does not "invalidate" anything obtained from DHCPv6. If DHCPv6 was not started, then the host will have received nothing from DHCPv6, and there is nothing to invalidate.
Besides, what if say, there's different address information available
via RA vs DHCPv6? --the text I cited above suggests that you accept the
union, but you suggest that you accept only the info provided by the RAs.
The draft doesn't say that the host shouldn't accept information provided by DHCPv6. It says that the host shouldn't request such information if it has all it needs from something else.
Again, I don't disagree (per se) with the proposal to do SLAAC rather
than DHCPv6 . But, I just think the behavior being suggested differs
from what we currently have -- hence the suggested "Update".
The behaviour that we currently have is due to what implementations chose to do, not to what standards track RFCs say.