Subject: Re: IEEE802.3 support ??
To: None <justin@apple.com, pmara@cactus.org>
From: Ross Harvey <ross@ghs.com>
List: tech-net
Date: 02/18/1999 22:35:22
> From: Shashi Mara <pmara@cactus.org>
> Date: 1999-02-18 16:18:59 -0800
> To: tech-net@netbsd.org
> Subject: IEEE802.3 support ??
>
> I am new to netbsd and have some questions for the network gurus.
> How come there is no support for IEEE802.3 in netbsd. I was looking at
> the if_ethersubr.c code. We already have support for 802.2 LLC (used for
> NETATALK, ISO..). I was thinking it should  be quite easy to add the
> 802.3 MAC encapsulation.
>
> 1. Am I missing something ?
> 2. Can Ethernet(V2) and 802.3 coexist on the same subnet ?
> 3. Does any body know approximate % of V2  vs  802.3 users "out there" ?

That's an interesting trivia question, though it was all settled over
a decade ago. It does provide a fascinating insight into the dirty
secrets of standards committees.

RFC1122 requires us to send and receive only ethernet, but says we "should"
also accept 802.3 packets, so that's what we do.  (Gotta luv the way the
IEEE **tried to change the frame format**.) No one actually uses the
IEEE-defined IP encapsulation, BTW, even those few that actually send 802.3
frames. If you watch a network using one of those cool color visualizers
you will see that virtually every frame is ethernet and hardly any packets
are ever sent in 802.3 frames.

It's kind of fun to track down the devices sending 802.3 packets, just to
see who "doesn't quite understand".

This happens more than you would think when a standard gets established
(ethernet) and then _later_ a standards committee (ieee 802.3) meets: the
objective of the vendors that missed the boat (and the academics who didn't
get to fiddle with it) is to _change_ the standard and delay the leaders;
this is the price for their votes.

	Ross.Harvey@Computer.Org