On Fri, Dec 9, 2011 at 11:28 AM, Leigh L Klotz Jr WA5ZNU <[email protected]>wrote:
> > On Dec 9, 2011 8:56 AM, "Brian Lloyd" <[email protected]> wrote: > > In the case of running where both the TNC and the KISS client are inside > the same machine, I would expect to find a virtual serial port program > connecting the two. OTOH, in unix I would probably use a named pipe. I > would NOT use a socket in the networking stack. It makes no sense to me. > > I don't see why a pipe or a virtual rs232 driver is OK but a TCP > connection isn't. At this point any such abstraction is the same...just a > reliable way to send data bidirectionally. The BSD socket abstraction > makes TCP connections and named pipes look the same anyway. Overhead > difference on the same machine is trivial and possibly the lo interface > gets more attention from developers anyway because X is written to use TCP, > not pipes. > > There is a trend these days to small devices (wrt54g, arduino, quisk on > fpga) communicating via IP. > That is all well and good. If you are going to do that then make your remote device be the router/gateway, not the host computer. In that case encapsulating in IP makes sense. Usually when you are going to the trouble of encapsulating a network or link layer protocol in IP it is to band-aid another problem, usually where someone wasn't thinking and something has been fielded which would be difficult to change. We aren't doing that here. Here we are trying to get an AX.25 network/link layer to talk to a modem. This is pretty much a physical-layer translation function with in-band signaling. If that modem exists inside the same machine with the AX.25 engine, then it makes the most sense to use the most suitable form of internal interprocess communications available. If you are talking about an external device like a real KISS TNC then the right answer is RS-232. If all you have to communicate with your gateway device is IP, then you have no choice. But the selection of proper transport depends on what your topology is and what is available in the hardware. So, even though I am a very strong proponent of IP/TCP/UDP for doing things, I am not in favor of treating them as my favorite hammer and then treating everything else as if it were a nail. -- Brian Lloyd, WB6RQN/J79BPL 3191 Western Dr. Cameron Park, CA 95682 [email protected] +1.767.617.1365 (Dominica) +1.916.877.5067 (USA)
_______________________________________________ fldigi-alpha mailing list [email protected] https://lists.berlios.de/mailman/listinfo/fldigi-alpha
