[Discuss] JCL in the past, and prognostications for the future

[DR]

Alan W. Irwin wrote:
> On 2008-03-13 16:35-0700 DR wrote:
> 
>> Peter Scott wrote:
>>> [...]Then there are languages like JCL that are obfuscated to begin 
>>> with and were never fun :-)
>>>
>> JCL is not a language it is a mechanism of torture!
> 
> These comments date you both, and I will be dated by my following 
> comment as
> well... :-)
> 
> Actually, I didn't mind JCL at the time because I knew nothing better, U of
> Toronto spent a lot of effort helping us to use it, and with that help it
> enabled me to get my thesis work done on IBM mainframes (which were
> pitifully weak in the early 70's compared to what is available now even for
> entry level PC's -- Moore's law operating for 35 years, and all that.)
> Subsequently, I made the transition to Vax/VMS, Unix, and eventually Linux.
> What helped me to make quick transitions in every case was to completely
> immerse myself in the new paradigm, and never look back to the old paradigm
> so I don't remember JCL at all except I believe "DD" featured large in the
> "language".
> 
> Now to look forward, imagine what your personal computer will be like 35
> years from now if Moore's law continues!  It is hard to imagine what all
> that computer power could be used for, but reliable programming of that
> powerful hardware will be essential.  Thus, let me also predict that by 
> that
> time the freedom of software will make so much economic sense from the
> reliability point of view that the proprietary stuff will be a dim memory
> with developers collectively scratching their head at that time wondering
> what many of today's software companies could possibly have been 
> thinking to
> keep their source code licensed without freedom or perhaps even secret from
> those who have the ability and interest to debug it.
> 
> Alan

I spent the summer of my third year at the U of A, 1967 (I think) 
working for the Alberta Research counsel.  My job was to write a program 
in Fortran (II) for the IBM 360 that took highway cross section data and 
turned it into plots on paper.  The University of Alberta had a flatbed 
Calcomp Plotter with a tape drive hooked to it.

The Fortran programming was pretty straight forward.  I had to learn the 
Calcomp API (we didn't call it that then, we just said how do I talk to 
this thing?) and then no problem.  Read the data off punched cards, 
mulch it around and write Calcomp plot data out.  The problem showed up 
when I had to figure out how to write JCL to write the output to mag. 
tape and prod the operators to mount the mag tape.  Complete with 
instructions to take the resulting mag tape and feed it to the plotter. 
  I think I spent three weeks getting it to work and about 6 weeks for 
the rest of the program.  I still have bad dreams about JCL.

I went on from there to work with an IBM 1800 process control system 
with a whole lot less horsepower and memory than my ASUS EEE PC. But, we 
ran a distillation column, an industrial evaporator, student projects, 
and my coffee pot.  At the same time we could compile and run student 
Fortran programs.  All with 3 megabytes of disk space and about 200kb of 
main core memory.  The most wonderful thing about the 1800 O/S is that 
it did not require JCL.  It had a much simpler and easy to understand 
control language.  Looking back when I left that job after 5 years of 
learning a whole lot about operating systems and writing device drivers 
for the 1800 and patching the OS, I reflected that if I had to do it 
again, I could probably now understand JCL.  I never did have to.

Where will be be in 40 years?  I usually spend a week when I am teaching 
introductory classes talking about the olden days, and probably boring 
my students, but I prefix the discussion with. "I am approaching 
retirement.  The history I am telling you about pretty much spans my 
working career.  What will this field look like when you are close to 
retirement?  How might you affect it?"

Deid