Re: evolution of nos

[Phil Karn <karn@qualcomm.com>]

>This may appear to be a rather simplistic, perhaps idealistic and
>obviously ignorant observation from a non-programmer... for the
>technically adapt, please go easy on the flames.  My attempt is to
>generate some useful discussion.  Is there any reason
>why we couldn't use existing winsockets or in the <future??> the
>built in hooks to Windows 95, etc for tcpip packet?  I expect that

This is not an ignorant question at all! I originally decided to
promote TCP/IP for amateur packet radio over 10 years ago because I
felt there would eventually be many other implementations available
besides mine, and the architecture of TCP/IP was such that they could
all be used over packet radio even though they weren't specifically
designed for it.

The easiest way to do this is to set up a dedicated machine running NOS
to act as your gateway between an AX.25 channel and a local Ethernet.
Since Ethernet support is just about universal in commercial TCP/IP stacks,
this lets your commercial stack talk to an AX.25 channel via the NOS
box.

Yes, this does require a dedicated computer, but it need not be
expensive.  An old 386 will do just fine, and since it only has to run
DOS you can get away with minimal memory. You don't need a hard disk
either; a floppy for booting is all you need. It's even possible to
dispense with the monitor and keyboard, although that can make
problems hard to debug.

This is how I have my home network of three machines set up. One is
the full-time NOS router, I run BSDI UNIX on the second machine and
DOS, OS/2 and Windows on the third. All can talk TCP/IP over SLIP/PPP
and AX.25 even though the NOS box is the only one with serial ports
and packet radio interfaces.

Phil