Re: [Roll] Last Call: <draft-ietf-roll-trickle-mcast-06.txt> (Multicast Protocol for Low power and Lossy Networks (MPL)) to Proposed Standard

["Jonathan Hui (johui)" <[e-mail]>]

Hi Axel, see below:

On Feb 12, 2014, at 1:05 AM, Axel Colin de Verdière <[e-mail]<mailto:[e-mail]>> wrote:

Le 10 févr. 2014 à 15:03, Jonathan Hui (johui) <[e-mail]<mailto:[e-mail]>> a écrit :

On Feb 8, 2014, at 1:26 AM, Axel Colin de Verdière <[e-mail]<mailto:[e-mail]>> wrote:

- In section 9.2, it is unclear when the M flag should be cleared or set: why would it be arbitrarily cleared and not indicate whether or not this message's sequence number is the largest one received?

The M flag helps indicate whether a receiving node should reset its Trickle timer.  If the M flag is not set and the sequence number is lower than what is already known, there is no need to reset the Trickle timer.

I understand that, I just did not fully grasp the formulation in the draft. In clear, I don't get why this text:


When transmitting an MPL Data Message, the MPL Forwarder MUST either
   set the M flag to zero or set it to a level that indicates whether or
   not the message's sequence number is the largest value that has been
   received from the MPL Seed.


Is not simply formulated this way:


When transmitting an MPL Data Message, the MPL Forwarder MUST
   set the M flag to a level that indicates whether or
   not the message's sequence number is the largest value that has been
   received from the MPL Seed.


The text from the draft seems to indicate that the M flag can be arbitrarily set to zero, or set to the proper value indicating the state of the forwarder.

The text is written to allow one to reduce MPL to a simple flood with packet caching to avoid receiving a packet more than once.  In this case, there is no real need to maintain the largest sequence number.

—
Jonathan Hui