Erik - nice job.   A couple of additions/corrections.

On Wed, Nov 26, 2014 at 1:28 AM, Erik E. Fair <fair-tuhs@netbsd.org> wrote:
        

BerkNET was its own thing:
​... 
It's somewhat similar to UUCP
​Right , like UUCP you could copy files.​

 
because (as I heard it) the Berkeley Computer Center wanted to be paid money
every time an operator had to hang a magtape on a tape drive, and the CS
department was tired of being bled to move small files around.
​Never heard that story.  While there was never a great deal of love between EECS and the BCC, the reason (I believe) was so that folks in Evan's could send email to the ArpaNET.  The only connection to the Arpanet was the Ingress (11/70) in Cory.​   UCB did not have it's own IMP, LBL had one up the hill and a "long host adapter" was used between Cory and one of the ports on the LBL IMP.   It was not until much later, when UCB finally got a C30 in Evans during the CSRG time that there was much more than email support.  Horton lurks on this list and might remember/know for sure.


 

 No redundancy allowed in the network, if I recall correctly.
​I'm not sure what you mean by this.   It was a point to point network like UUCP, but unlike UUCP routing was handled by the protocol (as you said - the tables were compiled into the code).​


 

Pretty slow convergence for changes to the routing table.
​i.e. manually done.   BTW: the Arpanet was not much better at the time​


 

I suspect that BerkNet's colon separator for host:file was how
the rcp command got that syntax, and probably how ssh inherited it.
​Nice idea - but no.  Sam & Eric Cooper modeled rcp from the program of the same name at PARC for the Altos.​  PARC is also what gave us routed BTW.

 

Google turned up the following: http://typewritten.org/Articles/berk-net.html

unless you used a serial interface card that
had some input buffering & DMA I/O like the DH-11.
​Right - the 70s mostly had DH11's which were a full "system unit" on the unibus (they are all SSI/MSI TTL - pretty amazing actually).    Many Vaxen had DZ's which were a mess.   Once the Able DH/DM came out, most of us switched to them and those lines ran at 9.6 or 19.2, but they were short lived, with the entry the 3M ethernet board that linked the CAD group's Vaxen, the Ingress machines and the machine room in Evans.​


 
I insisted that we wait for ARP to be done before we deployed Ethernet &
TCP/IP, because evil old hack of grabbing a Class A IP network number,
pretending that the first three MAC address bytes were always the same (after
all, everyone always uses Ethernet interfaces from the exact same manufacturer
in every host on a given LAN, right?) and mapping the last three MAC bytes
into the host part of the Class A wasn't going to fly in the real world.
​Sigh - that was me in 1979...  when Glaser and I did original IP stack with 3COM (aka UNET) @ Tektronix and since we were not (yet) on the Internet, it did not matter[BTW: look at the HyperChannel code - there are worse addressing hacks in there].  While I knew about MIT's ChaoNet I had never seen the code.   The MIT guys did ARP for ChaosNet which quickly migrated down the street to BBN for the 4.1 IP stack.  Remember that BBN had the contract to do IP for UNIX and that was the stack a number of us ran on our Vaxen for long time [IIRC - Sam actually did the routed and rcp stuff which the BBN stack, before Joy had rewritten it - created the sockets interface etc].


 

The hacks we used to do to make these turkeys fly ...
​Hey it worked just fine at the time.

Clem​