[net2-wg] Should CTP header have hop count?

[Rodrigo Fonseca]

Hi Phil,
>
> The first step is to see if the byte improves CTP's efficiency. It well
> might, as it is a more accurate indicator of possible loops than ETX.

We are talking about different things, I think. What I proposed is
only in route control packets, and amounts to breaking the cumulative
path ETX into two components: hopcount and cumulative extra expected
transmissions. It doesn't change the forwarding of packets at all, nor
does it need to change the format of the data packets.

You are talking about changing the forwarding using hopcount instead
of ETX for detecting loops.

>
> If it doesn't provide a significant improvement, then we can examine whether
> its presence is needed for other protocols/systems that *might* build on top
> of it.


Adding hopcount to the routing control packets doesn't improve CTP in
any way, just allows the router to tell applications the presumed
number of hops to the root.
I didn't propose a vote as a way of blindly expressing a preference
for one or the other, but as a way to see if people see a benefit.

There have been (albeit limited, and not clear whether not doable
otherwise) some use of the getDepth() interface in the past. There
*might* be uses in the future which we don't know. The cost is not so
high, given that it is only in route formation packets, and this is a
good time to do it.

Rodrigo

>
> Phil
>