[Olsr-dev] Seeking comments: OLSR+ETX v/s DSR+ETX

[sebastian sauer]

Hi Hannes,
Hi Henning,

Fri 21 Dec 2007 19:19, Hannes Gredler wrote:
> what about decoupling this:
/me thinks, yes this is indeed a good proposal.

> 1. making the formula as simple as possible:
>     lq = (total - lost) / total
i'm a simple man. i do not like complications. i'm for it, esp. since it's
the only sane thing that i could think of .-) and it's basically fully
compatible to the MIT implementation.
(plz note: the MIT guys added a jitter, and their intervall is a time-intervall. 
*but* OLSR has by spec a packet jitter built-in, to reduce the possibilities of
routing-loops.)

> 2. suppress advertisement of new neighbors (in tc-messages)
>     up until X * hello multiplier.
yes, makes sense. (sure, this would still allow nasty things,
but i do see why it is needed.)

no objections, all for it. :)

cheers,
s.