Hi Stan, hi all, Am 31.12.2014 um 20:29 schrieb Stan Ratliff <ratliffstan@xxxxxxxxx>: > If the penalty is "less improvement", and not failure. then I'll push harder > for the text change. Otherwise, we're caught in the procedural issue of a > standards-track doc having a downref to an informational RFC, and the other > problems that Thomas noted - it is legitimate to state that RFC 6621 > specifies multiple schema, and to question whether those schema interoperate > with regard to AODVv2. Replacing it with text saying "It's up to you, oh > mighty implementer." addresses the issue by taking it off the table. > For what it's worth, these arguments seem very convincing to me. However, the information that RFC 6621 describes a suitable way to achieve this may be a helpful hint for someone less experienced than you are (Remembering my struggles with neighbor monitoring mechanisms, I would've found such hints very helpful). Would it be possible and/or useful to downgrade the reference to RFC 6621 from normative to informative by mentioning it in a half-sentence, for example like so: “Implementations are free to choose their own heuristics, for reducing multicast overhead. One example for this is [RFC6621].” ? Regards, Lotte > Regards, > Stan > > > On Wed, Dec 31, 2014 at 2:11 PM, Charlie Perkins > <charles.perkins@xxxxxxxxxxxxx> wrote: > > Hello Stan, > > I am O.K. with that, maybe, especially regarding duplicate transmissions. > However, RFC 6621 specifies ways of implementing reduced relay sets that > can substantially improve performance. > > If some nodes implement RFC 6621, and others do not, the penalty is > not failure, but less improvement. It's always better than brute force, > though, even if some nodes do not implement. > > Regards, > Charlie P. > > > On 12/31/2014 10:57 AM, Stan Ratliff wrote: >> *** This email is about Issue #32 (Multicast transmission) *** >> >> >> ============================================================ >> >> [manet] #32 (aodvv2): Multicast transmission >> >> I updated the issue tracker with some discussion from Abdussalam. >> >> We should ask Adrian if we can close this issue. >> >> ============================================================ >> >> I think the current draft statement "In order to reduce multicast overhead, >> retransmitting multicast packets in MANETs SHOULD be done according to >> methods specified in [RFC6621]" should be changed. I propose we use the >> standard "cop-out" that is in many, many places in DLEP: >> "Implementations are free to choose their own hueristics for reducing >> multicast overhead." >> Provided, of course, that won't cause interoperability issues (e.g two >> different implementations, using different multicast retransmission schemes, >> trying to connect to each other). >> >> Regards, >> Stan >> > >