Hi Alper,
I think in this scenario we may already be treating PANA-Ping in a
special way already since we are not
strictly lock-stepping anymore; i.e., ping processing seems independent
and in parallel of other exchanges.
I don't think it is special. I think it is true for some other messages that
the peer may have an outstanding request while it also has to respond to an
incoming request.
PaC PAA
1<-------- PAR
2--------> PAN
3---> PAR
4 <--- PAR
5 ---> PAR
6<---PAR
7-------> PAN
8<------- PAN
Consider messages 3 and 4 crossing each other on the wire.
In general, I do agree that request crossovers are possible though my
main difficultly is the complexity of the pana statemachine with regards
to handling crossovers specifically with ping (problem of N*2 num of
states mentioned by Yoshi in another thread). Anyway, we should continue
that discussion on that thread to minimize email traffic.
best regards,
victor
_______________________________________________
Pana mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/pana
