P1: JDW
UNIX WL040/Bidgoli-Vol III-Ch-41 August 13, 2003 17:26 Char Count= 0
HISTORY 497purchase or otherwise obtain a copy of Unix. People
wanted to buy Unix.
But AT&T couldn’t sell it. Not for money, anyway. The
reason for this goes back to 1949, and an antitrust com-
plaint the Truman administration filed against American
Telephone and Telegraph Company (AT&T) and the West-
ern Electric Company. The net result of this was that AT&T
could not do business in anything that didn’t have to do
directly with delivering phone service. Period. Addition-
ally, AT&T had to provide its patents to the public and
could only license (i.e., not sell) their technology to oth-
ers at nominal fees. This basically meant “no business
but phones and telegrams” (Salus, 1994, p. 58). Posing
the question “But can’t we still sell Unix?” to the lawyers
would have been met with even more acrimony than the
question “Can we develop another Multics?”If You Build It, They Will Come
So AT&T decided to license Unix for a nominal charge to
other interested parties. If others, like universities, want to
play with this research, that’s fine, they could do that. Give
it to them, but don’t market it, don’t support it, and don’t
promise to fix any problems with it. By the end of the next
year, the list of “other interested parties” read like a list
of Academic Who’s Who: Harvard, Columbia, Stanford,
Princeton, UC Berkeley, and Johns Hopkins. Universities
liked Unix: There was no warranty to void because there
was no warranty; users got the source code, which made
the system inherently fixable; it came with its own com-
piler, which allowed users to write whatever new utilities
you wanted; it supported multiple users and multiple pro-
grams simultaneously; and it ran on hardware that was
beginning to cost less than $20,000. Unix took off.The Great Schism
In March of 1975, Digital Equipment Corporation intro-
duced the powerful new PDP 11/70, and the Computer Sci-
ence Department at the University of California, Berkeley,
purchased one of the first ones produced. The following
fall, Ken Thompson was invited to Berkeley as a visiting
professor in the Computer Science Department. One of
the first things he did was help Berkeley install Version 6
of AT&T Unix onto their new PDP 11/70. The other thing
he did was meet a new PhD student by the name of Bill Joy.
Version 6 was a popular edition and was widely adop-
ted by university computer science departments across
the world. Operating systems classes began to be taught
using the Unix operating system, and one of these classes
was offered at the University of New South Wales,
Australia, by a professor there named John Lions. He
published a detailed commentary on the source code of
Version 6, which helped countless programmers across
the world repair and improve their installations of
Version 6. Again, the widespread distribution of source
code and now commentary further enhanced the interest
in Unix and the practicability of its successful deployment.
Thompson had written the ed editor, which had been
available since the first edition. It was a primitive line edi-
tor, certainly by today’s WYSIWYG (What You See Is What
You Get) word-processing standards. The ed editor was a
command mode editor that allowed users to open and edita file, make modifications, and save it back to the disk, all
without ever seeing the text they were editing (this actually
has significant practical benefits, because the ed editor is
a filter and can be command-driven from a shell script,
using what is known as a “here document”).
Bill Joy didn’t like ed very much. Most mortals prefer
to be able to see the text they are modifying as they are
making the changes. (Actually, ed did allow you to see the
text, but only line by line and then only on request.) So Bill
Joy wrote a new visual editor called “vi” (for “VIsual”) as
a front-end for ed. Vi also used some of the new terminals
available and allowed the user to see a screen full of the
text that he or she was editing. The source code for vi was
sent to Ken and incorporated into the next AT&T Unix
releases. Bill also didn’t care for the default user shell and
decided he would prefer a shell with a scripting language
that more closely resembled native C syntax than did the
existing shell, so he created the C Shell (csh).The Berkeley Software Distribution
Bill Joy wasn’t content with just writing a new shell and
an editor. He soon turned his sights toward the V6 kernel
itself. His initial focus was on improving the speed of the
kernel. He released a “modified” Unix system in early 1978
through the Computer Systems Research Group (CSRG)
at Berkeley and called it the Berkeley Software Distribu-
tion, or BSD 1.0 for short. BSD 2.0 appeared shortly there-
after, and included Bill’s new vi editor along with his new
C Shell. The first two releases of BSD added only user
programs and shipped with the existing AT&T V6 kernel.
1BSD was sent out in return for a license payment of $50.
In early 1978, a group at BTL in Holmdel, NJ, had just
received a Digital Equipment Corporation (DEC) VAX and
had taken Version 7 and ported it over to the large VAX
instruction set. About the same time, Berkeley received its
own VAX, an 11/780, which was a big minicomputer that
boasted more memory among other advantages. Berke-
ley contacted AT&T and requested that their Holmdel
VAX port of Version 7 be made available to them, and
it was. The Version 7 VAX port became the basis of 3BSD,
which was made available in 1979, and shipped with a new
virtual memory system (supported by the VAX; the “VA”
stands for “Virtual Addressing”) that the BSD group had
implemented for the VAX, which allowed for the support
of many new powerful applications.BSD and the Internet
Unix systems prior to 1976 were pretty much standalone
systems that couldn’t talk to one another over a network.
To address this, Mike Lesk of AT&T created a program
that would allow a file to be copied over a modem from
one Unix computer to another. This program was called
uucp—Unix to Unix Copy. In 1978, Eric Schmidt devel-
oped a small network program for BSD while working at
Berkeley on his master’s thesis (Stevens, 1990, 9). His net-
working program was dubbed “Berknet” and was released
along with 2BSD and provided support for up to 26 dif-
ferent hosts. Berknet provided support for up to 26 hosts
and networking capability for BSD systems and included
capabilities for transferring files and sending e-mail.
By the fall of 1980, Unix was able to run on two hard-
ware platforms: the DEC PDP 11 and VAX 11 platforms.