Hi Michael (Tüxen this time, not Welzl :-)), Thanks for the comments too. See inline...
Tom P. [snipped] > > Yes, absolutely, and > I agree with Michael. Not only for NAT traversal, but if you want to run > DCCP on > nodes which do not support it in the OS and you can not open a raw socket. > This is at least an experience from running SCTP on the > iPhone/iPodTouch... > [Tom P.] Yes, this is a useful side-effect of the encapsulation. I know that some people feel that the only place transport protocols should be implemented is in the kernel, but, as I said in a previous meeting, nowadays the net effect of that is to make Microsoft the gatekeeper for deploying new transport protocols. I don't think we want that.
