From: utzoo!decvax!decwrl!sun!gnu
Newsgroups: net.ham-radio
Title: Re: Packet protocol proposal (long)
Article-I.D.: sun.162
Posted: Sat Dec 11 02:27:34 1982
Received: Sat Dec 11 10:48:56 1982
References: tektronix.842

Sigh.  Yet another ad-hoc protocol invented "because none of the other
protocols were good enough".

I did this once myself (with PCNet) and came out of it with the belief that
in most cases, it doesn't really matter what the protocol is.  It only
matters that there IS a protocol, that is, an agreement among all the
potential users.  The larger the group, the better, since the objective is
easy interchange.

Now that the major packet groups in SF, Tucson, St. Louis, New Jersey,
Washington, Vancouver, and elsewhere seem to finally be agreeing on the need
for protocol standardization (probably due the the prospect of a satellite
mailbox system within a small number of years), out comes yet another
protocol.

Look:  somehow with the horrendous UUCP protocol we've managed to patch
together a network of hundreds of computers and thousands of people.  Surely
with that as a daily example you can live with the problems (and yes, there
are plenty) with the agreed-upon standard link-level amateur packet protocol
(which itself follows several ISO/ANSI standards, making it likely that
cheap available hardware, software, and gatewaying will be available).

Please reconsider?  Yes, it's definitely fun to play around designing and
implementing a protocol for a year or two, but if at the end of it you can
only talk to 4 other people, why not consider doing something with better
odds?

	John Gilmore, Sun Microsystems

PS: Yes, PCNet works, sort of, and yes, I can talk to about 4 people with it.
And yes, it took a year.
PPS: There are cheap ways to calculate CRC in software, like ~10 instructions
per byte as I recall.  It helps a lot to know the math, which I don't.