Paw Prints: Writings of the maddog

My friend Benjamin Scott send some email around today stating that Alcatel/Lucent had published all the old Bell System Technical Journals from 1922 to 1983 online and freely accessible. As Ben said:

Bell Labs practically invented much of our recent civilization (communications theory, transistor, laser, microchip, Unix, the list goes on). The public switched telephone network, before the Internet came along, was probably the most complicated system in human existence. They documented a lot of it in these journals. Making them available like this is a huge boon to technology historians.

My favorite Bell System Technical Journal (BSTJ, for short) was Volume 57.6, published in August of 1978. In it were a series of articles from the people who developed the first UNIX systems. Not only were these articles written “first hand” at a time when people could both formulate clearly the reasons why they created UNIX the way they did, but at the same time it was not so long after they had created the system to have forgotten some of the details.

A small example of this exists in the name UNIX itself. Most people today write “UNIX” as “Unix”, but in these original typeset and scanned PDFs, you can see that the developers consistently spelled it “UNIX”, with all capital letters.

Other gems that are available from these documents include a specification of an early PDP-11/70 computer where Ken Thompson programmed Unix, stating that Ken's system had 768K bytes of core memory, acknowledging that this system was very generously configured, that as a minimal system you could have as little as 90K bytes of core and (as Dennis Ritchie points out) be purchased for as little as 40 thousand dollars (in 1978) of hardware investment. Ken's “very generously configured machine” also had two 200 Mbyte disk drives, 20 dial-up modem lines and 12 hard-wired serial port lines, as well as several interfaces for “machine to machine transfer”, a phototypesetter, a voice synthesizer and a chess machine.

Steve Bourne writes about the original Bourne shell while Dennis Ritchie, Steve Johnson, Brian Kernighan and M.E. Lesk write about the early “C” compiler, and there are some early works about Programmer's Workbench and document preparation. There is even a discussion about the trade-offs of putting UNIX on a “microprocessor” of the day, an LSI-11 computer from Digital that only had 40K bytes of RAM.

Some of the problems they were investigating in 1978 are eerily similar to the problems (and solutions) that we have today. MERT was a real-time and (more-or-less) “virtualized” system, and the papers even address issues in using the computer to control devices.

Take a look at these articles of history, both to find out how UNIX derived and why some things remain the same, even more than thirty years later.

Carpe Diem!

Comments