[TUHS] A Human 'Make' Program (almost)

Forrest, Jon nobozo at gmail.com
Fri Nov 17 03:42:25 AEST 2017


I started with V6 Unix at UC Santa Barbara in 1977. I remember
that when V7 came out, I learned about the 'make' program and
started using it with great success to help me build a large
Fortran package for signal processing.

For its size, there was a lot going on in Santa Barbara at that
time. It was one of the first 4 Arpanet nodes, and there were
a bunch of companies making networking products and doing speech
research as a result.

I was a student at UC Santa Barbara but I started toying with
the idea of finding a real job, mostly to make more money.
I found several possibilities and went to interview at one.

This place had an a need for somebody to, in essence, be a
human 'make' program. The computer they used, some kind of
Data General, was so slow that they couldn't do a build more that
once or twice a day. So, in an attempt to speed up the build,
they wanted to hire somebody who would, by hand, keep track
of the last modification date of all the components in the
package they sold, and do a build that only performed
the necessary steps to generate the package - in other
words a human make program. Apparently they figured that
this would save enough time to justify the $24K salary they
were willing to pay. $24K in 1978 wasn't a bad salary at all.

I didn't take the job, but I've often thought that what I should
have done would have been to take the job under the condition
that I could mostly work remotely. Then, I could have used the
'make' program on our V7 Unix system to generate the optimal
script to build the package, and then taken the script back
to the company to run on the Data General computer. I figure
this would have taken maybe an hour a day. The rest of the time
I could have spent on the beach thinking about ways to spend that
$24K.

Jon Forrest



More information about the TUHS mailing list