> > > For each multicast interface: > > > > > > IsRouter A flag indicating whether routing is enabled on > > > this interface. Enabling routing on the > > interface > > > would imply that a router can forward packets > > > to or from the interface. > > > > > > This issue is now resolved > > > > sorry about late comment, but how far would you like to go? > > like interface A forwards to B (and vice versa), and C-D, ... > > how would you describe? > > => I don't understand the comment. The definition implies that you > can't forward from a non-routing interface. Perhaps you mean > that you should be able to forward _to_ a non-routing interface?
no. by making "router" per-interface, you could have multiple set of interfaces which forward packet to/from. for instance, your router has 10 interfaces, and #1 is host mode, #2-#4 exchange traffic as router, #5-#9 exchange traffic as router (it's separate from #2-#4), and #10 is host mode. once you go into "router/host is a per-interface thing" you need to describe such combinaions in full detail. do you see my point? > > i'm for simple "router or host" in document, and leave > > per-interface > > "router" as a exercise for reader ("virtual router" > > concept is not new > > so vendors will make such device anyways). > > => The issue at hand is that the doc is not clear on > nodes that are both hosts and routers. Do you see any > harm in making the definition per interface? yes. i see a big harm and disambiguity introduced by the change. again, keep the document simple, and let vendors do funny/complex things if they want to. itojun -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------