[talk] possible meeting topic: FreeBSD desktop stuff with a side of DJ-BSD redux

[James K. Lowden]

On Tue, 4 Mar 2025 16:03:29 -0500
jpb <jpb at jimby.name> wrote:

> Interesting!  My first computer job (1980) was on a Univac 1100
> series mainframe using a COBOL application.  In fact my Mother was a
> COBOL programmer for a multinational also on a Univac 1100.  She spent
> time in Iran working for a bank.

Here's a sentence you won't see very often: 

	My mother programmed for Univac, too.  

She worked for a heartbeat for IBM, then for Univac outside
Philadelphia in 1968-69.  She worked on the disk access routines, she
told me later, what would be called a "device driver" now.  She wasn't
a great COBOL fan.  She thought it was too clunky.  

It used to be a novelty to say I was a "2nd generation programmer".
Now it happens more and more often.  

It wouldn't surprise me one bit if the application your mother worked
on still runs.  Not for nothing did corporations around the world spend
umptiump millions to have 4-digit years for Y2K.  They sure didn't do it
because the code would be obsolete -- or even easy to replace -- a few
years later.  

My talk at FOSDEM this year was "COBOL is the Original Safe Language".
Someone asked why it's still relevant.  I said there are 3 kinds of
COBOL programs: 

1.  became irrelevant, discarded
2.  became Java because UI 
3.  still run because too hard not to

What I didn't say is that "too hard" refers to programs that were
designed before "design" had its modern meaning, when flow charts were
au courant, and modified over decades by programmers who, on average,
were of average quality.  

It is *always* expensive to re-write an existing application, even when
the language doesn't change.  When the re-write means changing
implementation languages and a new design (that doesn't match the
existing), it gets really hard.  You need experts in both languages
conversant with both paradigms.  And the Tom Jones problem: it's not
unusual for COBOL programs to run 100s of thousands of lines. That's
what decades of programming will do to an application.  

The proof is found in many places that have tried, to the tune of $10
million and more, over a decade and more, and come to tears.  

> Every couple of years you hear about the need for COBOL programmers
> because they are retiring or dying :-(  Sure to be good money for
> people who still know that stuff.

It's true.  When the last IMS admin retires, they give it to the DB2
admin, whose reaction is seldom gratitude for the opportunity.  

>From what I hear, there's bascially no point in advertising for COBOL
positions.  Companies find someone willing to learn, and train them.
And pay them, you bet.  They don't need hordes to write new
applications.  They need a handful who can make adaptations for
regulator requirements or somesuch.  The limelight they save for
someone else. 

--jkl