On Monday 16 March 2015 23:29:31 Light, John J wrote: > 1. The actual networking code was all written to the Linux socket > layer. I can only trust that the authors of Linux wrote the socket layer > to conform to all of the RFCsf (dozens of them). So I probably conform to > many RFCs, even more than Thiago listed, but if I do, it is to the credit > of the Linux network developers, not to my work on IoTivity.
I listed the ones that you directly used in your code. For example, sockaddr_in6 is defined on RFC 3493 "Basic Socket Interface Extensions for IPv6". You didn't define the structure, but you used it. You also used RFC 4291 "IP Version 6 Addressing Architecture" and RFC 7346 "IPv6 Multicast Address Scopes" when you wrote code that relies on the particular addresses related to multicast (ff02::101, etc.). In your testing, you probably relied on RFC 4862 "IPv6 Stateless Address Autoconfiguration" to configure the link-local addresses. I shouldn't have listed that one, as it is not the only way to configure nor was it required. RFCs 4007 and 6874 are part of the scope ID discussion we had and agreed will be necessary in the near future. RFC 3542 will probably be interesting to get the destination address (IPV6_RECVPKTINFO) so we can tell a unicast packet we received apart from a multicast packet we received on the same socket. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center
