questions about interfaces

[Bill Hacker]

David Cuthbert wrote:

Bill Hacker wrote:

David Cuthbert wrote:

Heh... imagine if the loopback interface actually required hardware.


Imagine?  'Remember when' you mean...

It exists because there was a time when hardware *was* required, and 
one did not always have arms long enough...


I remember using (and still use) this for modems and serial ports (doing 
local and remote loopback tests), but never in the context of an IP 
loopback connection.  This is probably because all of my experience with 
IP/TCP/UDP has been on top of a OS+stack that someone else wrote and 
debugged, running on hardware that wasn't too terribly esoteric.

What eventually became the '7 layer' protocol stack was originally just two when 
I started.

Baudot, BCDIC, Selectric code, or USASCII as the top layer, and either 20 ma 
bipolar or 60ma single-ended telegraph wire ELSE RTTY as the PHY layer. 
'High-speed' was stuff like the Hewlett-Packard parallel current-loop.

DTMF modems were a big advance.  Far more error-free than Bell 103.
And synchronous 9.6 Codex modems were the mainstay of our X.25 and 
point-to-point frame-relay backbone for Novell 2.X and 3.X servers.

Things moved so slowly in those days, that two of us at opposite ends of the 
world could agree to swap cable from 56 Kbps NTE to 9.6 Codex dial-ups on a 
3-count, and not lose in-process IPX/SPX over HDLC / Frame Relay transfers.

OTOH, unlike my older-yet C&W mentors, I didn't have to learn Morse, Wheatstone 
Code, or such...

;-)

PDA's and Mobile phones are now nearly as fast as the TDDL that kept SAGE NORAD 
glued together.

*weird* thing is, I don't remember ever once using the Unix loopback interface 
for testing.  I've always had at least an OS/2 box if not two or more BSD boxen 
or Slackware to test between/among, and tend to run local services on sockets, 
not IP anyway.

Still - glad it is all available.

Bill