On Wed, 2005-06-01 at 14:11, Soliman, Hesham wrote: > The text does not say that you should do address resolution > before consulting your neighbor cache or forwarding table and do > on-link determination. Although, this can probably be made clearer. > So, I think the clarification needed is in the order of steps. I.e. > The current text seems to skip a couple of steps.
Maybe, but I don't see the need for the text at all. If implementors want to play cute games with ND, then let them. > => You seem to make this assumption because the text does not explicitly > spell out the order of events. Would it be clearer if the text explains > that address resolution is only done if the on-link determination > process indicates that the destination is onlink? If you agree with this > I can send suggested text. It would be better than what's there now, but ideally, the text should just be removed entirely. Doing address resolution on multiple links can lead to ambiguous results. There could be multiple destinations responding to the same address, and you don't know which one you're talking to. This hasn't been studied enough to even be mentioned. If particular implementations want to do this, they can. There's nothing in the spec that prevents them from doing it. Please consider removing the text as Jinmei has already suggested in previous posts (and I apologize for having missed those messages last week). Thanks, -Seb -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
