On Mar 28, 2012, at 16:08, Angelo P. Castellani wrote:
> http://www.dei.unipd.it/~castellani/coap_m_fsm.pdf
I like that much better.
The RETX ACK is missing its reason (i.e., having got another copy of the CON).
It would also probably be useful to indicate that this is a state per MID, and
that MIDs attach to the *sender* of the NON/CON.
TX-CON allocates a MID.
NONs received get dedupped, too (except see below), and their MIDs time out the
same way.
In my implementation there is also a reply from the RR layer upon a received
CON or NON that the RR layer (or its client) didn't care about duplicate
detection ("stateless"). This MID then goes right back to closed.
Grüße, Carsten
_______________________________________________
Lwip mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lwip
