On Thursday 18 June 2015 23:49:11 Agrawal, Sachin wrote: > Hi John, > > I guess my language was not explicit :(. > > If we follow the approach suggested by you, an end-point will receive > 'packets' with these possible formats: > > [DTLS Header] [DTLS Payload(which can be encrypted or plaintext...but this > is actually immaterial to our current discussion)] [CoAP Header][CoAp > Payload] > > Since these headers have no correlation, how can an end-point 'reliably' > differentiate between these packets?
By port number. > Although, in above, 'socket F' is being used to send CoAP packets and DTLS > packets too. Will this cause any issues? Yes. Those should be two separate sockets: one for DTLS and one for insecure. But I'd argue that we should remove the the unicast insecure sockets. We should only have secured unicast communication. Whether we can do discovery under group DTLS or not, that's still in discussion. I'd rather we could so we can close all insecure sockets. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center
