Re: [Roll] Intention for draft-thubert-roll-forwarding-frags

[Carsten Bormann <[e-mail]>]

On 05 Feb 2014, at 19:05, Pascal Thubert (pthubert) <[e-mail]> wrote:

> We presented the work at 6lo, but there was a mitigated reception.
> Carsten in particular opposed the work, and complained that it moved places,

Well, my complaint was not just about the fact that the same idea had been discussed in several places, but mainly that it already had received lukewarm to negative feedback in those places, but there had been little progress in the draft apparent when it turned up in the next place.

> plus has doubts about the retry mechanism

That is really the technical part of my and others’ skepticism.
Apart from the apparent redundancy with existing L2 ACK/retransmission mechanisms, one aspect was the unclear interaction of a multi-hop reliability scheme with transport timers.
More generally, there was skepticism that this approach actually yields improvements that are in line with the complexity it creates.
Having some observations from simulations or preferably real networks would be useful to dispel this impression.

But I agree with Ines, sub-IP encapsulation and reliability/performance is really outside ROLL’s scope.  6lo would actually be the right WG, but the technical question marks mentioned above would need to be addressed.

Grüße, Carsten