10 Feb
2011
10 Feb
'11
8:26 p.m.
On 10 Feb 2011 at 18:19, Fred Cisin wrote:
I have to disagree. Although 73-80 were not used by
the COMPILER,
they were "reserved", so that you could put a sequence number on your
cards. That worked a little better for putting them back into order
than the diagonal felt-tip marker lines. Most? FORTRAN coding sheets
explicitly included those columns, so that you could specify to the
keypunchers what you wanted there.
Yes, but they weren't part of FORTRAN, or COBOL, or JOVIAL or GPSS;
they were used for sequence numbers or dirty jokes. I didn't know of
anyone who actually punched sequence numbers by hand.
Remember that COBOL used 73-80, but as "program identification"; 1-6
were used for sequence numbers--and checked by most compilers. So
naturally, if you had half a brain in your head, you left them blank.
Realistically, you put your decks on tape as soon as you could. Most
operating systems had source maintenance utilities (e.g. CDC EDITSYM,
UPDATE or others), some of which extended the card image past 80
columns, so that very detailed sequence information could be added.
If you had to use and keep cards, you punched and interpreted a
duplicate deck using your handy-dandy mainframe or unit record gear
and inserted sequence numbers automatically.
At CDC, we'd use the PSR number as part of the sequence number.
Since UPDATE kept all card images, both active and inactive, you
could always "YANK" a correction ID set to reverse its effect. While
I had several filing cabinets of cards in my office, it was mostly in
case some butterfingers in Test and Integration lost the deck (this
was long before Unix VCS).
...but back to my original point. The 704, on which FORTRAN was
first developed could read only 72 columns of an 80 column card using
the 711 reader. You could program (via plugboard) any 72 columns to
be read, but that was it.
There's a very practical reason why the 72-column FORTRAN statement
came about.
--Chuck