Newsgroups: alt.folklore.computers
From: bernie@fantasyfarm.com (Bernie Cosell)
Subject: Re: Who built the Internet? (was: Linux/AXP.. Reliable?)
Date: Sat, 3 Jun 1995 19:14:21 GMT

In article <3qn5b2$h7l@fg70.rz.uni-karlsruhe.de>, Thomas Koenig writes:
} In comp.os.linux.development.system, cmetz@sundance.itd.nrl.navy.mil (Craig Metz) wrote:
} 
} >	It's always nice to know that 4BSD and *only* 4BSD built the
} >Internet. That's right. No IMPs, no Multics, no VMS, no MVS, no research
} >UNIX, no System V, no Cisco/Wellfleet/etc. Yep, *only* 4BSD.
} 
} I thought the first Internet machines were primarily Multics, with a few
} TOPS-10 and TOPS-20 machines thrown in.  BSD came to this relatively
} late.

You're both fairly off the mark, I think, alhtough the quesiton is
a bit underspecified Part of what you get for an answer depends on
what you mean by "built the Internet".  To the extent that one
views the Internet as an evolution from the ARPAnet, then there
were certainly IMPs [indeed, that's all there were].  The initial
hosts on the ARPAnet spanned a huge number of OSs.  The first node
on the net, UCLA, was a SDS Sigma-7.  Rand had some kind of 360.
There was only ever one Multics system on the net --- the one at
MIT.

The most prevalent system early-on was the PDP-10, and in
particular running BBN's "TENEX" OS.  For a while, this was about
the only OS that appeared more than once on the net, and it was
quite good for the net [as, indeed, was the coming of BSD a bunch
of years later].  The reason is that a BIG stumbling block in the
development of the net was the difficulty of getting systems to
support new bits of protocol and new applications,etc.   With
TENEX, though, if the BBN-folk implementing just about anything, it
was *immediately* ready to be tried out in the network [whereas it
would take months, or perhaps years, before other OS's could find
someone to implement whatever-it-was].  That's how email was born:
TENEX had email internally, and Ray Tomlinson put together a hack
that'd allow the email to be sent out over the ARPAnet.  Of course,
at that point the only place you could SEND email was to other
TENEXes [since they were able to run copies of the code].  But it
did make it a LOT easier to try things out, do demonstrations and
evaluations, etc.  [and, of course, email was probably the biggest
success story of the early ARPAnet].

BSD only really becomes a play MUCH later when the network was
going through the incredible pain of moving from NCP to TCP/IP.
What it provided [among other things, of course] was a *standard*
set of protocol managers and applications for dealing with the
network.  It still didn't "do" networking quite as cleanly as
some of the other OSs floating around the net, but it did it *ALL*
and the BSD implementations of the various protocols became the
standards against which other vendors could test their stuff [and,
since it was freely available [courtesy of its having been paid
for by us'uns, the taxpayers, via DARPA], it also provided sample
code.  As far as I know, this is largley still the case, and so
the Berkeley archive version of things like sendmail and BIND and
such are probably _still_ the examplars.

In any event, the multics statement was clearly wrong: there was only
one MUltics among a multitude of other OSs.  But I guess I don't
see 4BSD as being the critical player as Craig does, so perhaps he
would amplify his comments some...

  /Bernie\
-- 
Bernie Cosell                               bernie@fantasyfarm.com
Fantasy Farm Fibers, Pearisburg, VA         (703) 921-2358
    --->>>    Too many people; too few sheep    <<<---