[TUHS] RIP John Backus

[Steve Johnson]

I had an interesting run-in with FORTRAN's blank treatment very early
in my career.   A couple of weeks after I graduated from college I
had a summer job at Bell Labs.  I was given a job to program a state
minimization algorithm -- they expected it to take me the whole
summer.  A couple of days after arriving, i heard about a new
language, SNOBOL, developed at another location at Bell Labs.  This
sounded like the perfect language to write my program in, so I got a
copy to use (I think I was the first user at Murray Hill).

Now, in those days, there were rooms full of "keypunch girls" (sic)
whose job was to punch up our programs (written on coding sheets) and
verify them and give us the deck back.  The vast majority of jobs
they encountered were FORTRAN, and to avoid ambiguity they simply
skipped all blanks.   (it wasn't quite that easy -- they knew about
column 7 and hollerith strings).  But any blanks that we wanted on
the cards had to be explicitly indicated on the coding sheet.

Of course, SNOBOL had what we would consider now a more modern syntax
with blanks significant and nothing magic about columns 6 or 7...   
So when I gave them my first 2-page SNOBOL program, they typed
everything on each line starting in column 7 and with all blanks
removed.    For some reason, the first couple of cards looked OK to
me, so I submitted the deck, and proceeded to get a thick printout
that I think enumerated every error message the compiler could
produce.

I started indicating my blanks carefully but their habit persisted,
and almost any nontrivial job  I gave them had errors, either because
I hadn't inidicated a blank or they hadn't typed it when I indicated
it.

Since I had been punching cards myself for a couple of years at
college, and when working 2nd shift (when turnaround was much better)
there were no keypunchers available, after a couple of weeks I got
them to agree to let me keypunch my own programs.  A few years later,
the keypunchers were gone, having been rendered obsolete by time
sharing and online editing...

Oh, and I got the job done in 3 weeks once I got SNOBOL to work...  
It really was the right language for the job...

Steve

PS:  For years afterwards, when I punched in FORTRAN programs I left
out all the blanks.  It wasn't until  I worked on a large program
with several other people that I was forced to change this habit, my
coworkers having threatened me with death or dismemberment if I
didn't... 


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20180318/2a26a542/attachment.html>