Hemant Singh (shemant) wrote: > Ole, > > Let's talk specifics, not generics. Of course, we know about section 5.2 > of 2461bis. > > Snipped is following text from Introduction section of our I-D as to > what we think about section 5.2 of 2461bis: > > Sections 5.2 and 7.2.2 imply that the host performs > address resolution before transmitting a packet if the destination of > the packet is on the same link as the host. Some current host > implementations perform address resolution in all cases even when the > destination is not clearly on-link. However, RFC 2461 [ND] section > 6.3.4 implies that hosts must clearly determine that a destination is > on-link before performing address resolution.
Yes. RFC 2461 allowed you to do that with it's "default on-link" assumption. The assumption was that if you did not have an on-link and default routes, they you assumed that the destination was on-link. This assumption has been removed from 2461bis. However, we can't force all existing implementations to change. All we can do is update the spec. Existing implementations can still claim conformance to 2461, but not 2461bis. > > Section 5.2 says "if the destination is on-link, then next-hop address > is the same as the packet's destination address. Otherwise, the sender > selects a router from the default router list." Implementers are not > putting two-and-two together. As described in section 2.1 of our I-D, if > a host has not been sent any PIO in RA, and the host performed DHCPv6, > the host has no way to determine what's on-link for the host. Section > 5.2 just says "if the destination in on-link...". Well, discussion first > has to be made as to how can a host determine what's on-link for it. > These are the clarifications we have made in our I-D. Hm... Searching for "on-link determination" in the text for 2461bis yields 7 instances that describe how it's done and what information is used. > > Proof is also in the pudding. We also say in Abstract of our I-D, if ND > RFC's are almost 10 years old, why are popular hosts still making > mistakes with on-link determination or totally not heeding L bits of RA > ? ND has to be clarified as we have done so in our I-D. That may be... however some of the text in you draft goes too far. I haven't read the last version, but I was in the middle of preparing comments on the last one. I'll re-read the new version and send comments. -vlad -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
