Found it: Clarification of Fortran standards—initial progress | Communications of the ACM
Appendix A-2(2) in the F77 Standard mentions this too.
When I learned Fortran on the IBM 026 keypunch, the keyboard didn’t have parens or an equal sign. You had to use # for =, % for (, and the currency symbol ¤ for ). Printer listings came out the same way. Probably @ for ', too, but I remember more vividly counting characters for Hollerith.
Fun times!
1 Like
LOL! What we had to put up with. My school’s computer center had a row of 029s, with a lone 026 at the end. Anyone in the know would use the 029s. So the poor newbies who didn’t want to wait in line for a 029, and saw that the 026 was free, would soon learn their mistake.
I did this too for a few months before those keypunches were replaced. In my case, the line printer would display the correct characters, it was just the keypunch where you had to do the mental mapping. I also remember a few characters that the keypunch simply did not have, so you had to punch one character to get some of the holes and then backspace and punch another character to get the rest of the necessary holes. The nice thing about standards is that there are so many to choose from.
Or you could hold down the multi-punch key and directly type the positions you wanted in the column.
Of course a good Fortran programmer would also have his own drum card - so that when you pressed the ‘tab’ key, it would space over to column 7.
1 Like
Yes the young folks here don’t know the endless joy they are missing out on with not having to deal with punch cards and keypunch machines. I didn’t remember the 026 not having an = key but I think I was lucky enough to only use an 029. I do remember one keypunch I used had an issue with the T key so all my CONTINUES came out as CONINUES. The Univac compiler I was using at the time was not pleased. In retrospect though I’ve always thought that punch cards had one big advantage over modern interactive input. If you were lucky enough to work somewhere with access to a professional keypunch operator, you could write your code on a coding form (green sheets in the parlance of the day), give it to the keypunch operator and then go off and do some real work like actually thinking about your results and how to implement algorithms. Of course you had to deal with the dark side of punch cards which was having a klutz or meth freak (and I encountered both) operating the card readers drop your foot long box of cards and getting them returned to you in some random order along with a listing with enough compiler errors to break the compiler (again something else I encountered). You quickly learn the value of putting sequence numbers in columns 73-80.
Ugh. I did that exactly once. The data tech who did it made so many typos, that I spent way more time fixing things than if I had just done it myself.
At some point in the early 1980s, IBM refused to renew the maintenance contracts on all their unit record gear. Not only the 029s and 029-like verifiers, but the sorting, interpreting, and other machinery. So they were forced to buy a ‘key to disk’ system. It was nice for the data techs, because some validation (like making sure a date code was within range) could be done before sending the data decks to the mainframes. It was before networking, so the key to disk system would write the completed data decks on 9-track tape.
Of course the initial implementation on the mainframe side simply read the magtapes and routed the data to be punched by the high speed card punch. Then the card decks were re-read via the high speed card readers for processing… (Yes, months later the Admin programmers finally got around to modifying their JCL to simply save the data as permanent files on the mainframe, and have their apps read the data directly from disk. COBOL programmers…)
The Corvette ZR1’s 233-mph run had to start in a virtual world
It turns out it takes a little Fortran to make a faster Corvette
Amusing wording in the description:
“off-the-shelf solutions like CarSim mixed with custom options written in C, Python, and even Fortran, a programming language that dates back to the 1950s and will surely outlive us all”
And, in this forum we converse in a language (English) that dates back to, at least, 1600 AD,and has outlived the Roman Empire!
1 Like
Yes, I was amused by that and have now left a comment, adapting your words about English.
1 Like
The article uses descriptions that aim to impress rather than inform. For example:
According to the simulations, the car should have run faster.
Jake Hedrick, virtual propulsion engineer at GM, started poring over the data. After further investigation, he determined it wasn’t the simulation that was off, but the car itself.
Yes, GM folks, if you wanted the car to run faster, you could have left the car on and turned the simulations off.
They can always do what U.S. manufactuers always do, just put a bigger engine in it. True story. I worked for Lockheed in the 1980’s. Lockheed Georgia Co. had one of the few low speed wind tunnels that was large enough to handle an almost full size car (also could simulate effects of ground plane) so the U.S. auto makers and NASCAR teams would use it to test cars. One automaker brought in a prototype and were talking about all the work they put in to reduce drag but the inital tests were showing a much higher drag than they predicted. As a joke the Lockheed test engineers put the shipping crate that the car came in (it was large enough to hold a full size car) in the tunnel. It had a lower drag coefficient than the car 
.
4 Likes
A whole new meaning to “box cars” and “drag racing” 
2 Likes
Then, the car would have won the next race if they had just left the shipping crate around it?
Chrysler started doing wind tunnel testing on their cars in the 1930s when designing the ground-breaking, but short-lived, Airflow. They discovered that the typical car of the day had less aero drag when driven backward than forward.
I wonder: what was the top speed in reverse gear for the cars of that era? Was that high enough to make drag important? Secondly, doesn’t a test car in a wind tunnel simply sit in an airstream, rather than being “driven”?
Reminded me of visiting the now-closed Living Computer Museum in Seattle some years ago. They had a 029 out in the open loaded with cards for visitors to punch. I sat down and punched a card or two for old times sake. (I couldn’t remember at first how to ‘register’ the card - a docent helped me out.) Then a couple of Gen Z guys walked by and asked how one would ever program a computer with such a device. I punched up a 3-line “HELLO WORLD” program for them - in FORTRAN of course. They were pretty amazed.
They also had a couple of working Xerox Altos nearby. One of my friends had used one ages ago and was entertained by it as well. Knowing him, he probably programmed it in mesa or smalltalk.
Sadly after Paul Allen died, the museum was closed. They recently auctioned off a lot of their historic artifacts.
The Computer History Museum also has a 029 out in the open - as part of their IBM 1401 demonstration. I do some (non-computer related) volunteer work with the fellow who trains new CHM people on how to operate the 1401. He keeps threatening to train me on it as well.
From the U.S. government:
We also have a Social Security benefit estimate program in Fortran, the operation of which is similar to that of the C++ version. The Fortran source code is structured similarly to the C++ version, but has not been updated for some years. The source code for the program, or the program compiled for MS-DOS, is available in a self-extracting zipped file sourcef.exe (384,000 bytes), or in a plain zipped file, sourcef.zip (358,085 bytes).
Compilation instructions are in a file fcompile.txt. @feenberg has maintained the TAXSIM software in Fortran for the U.S. and state tax code for many years.
1 Like
SPIDER (System for Processing Image Data from Electron microscopy and Related fields) is an image processing system for electron microscopy, especially usefull for single-particle reconstruction. SPIDER has been in use since 1978 and contains over 210,000 lines of Fortran code and 7400 files.
The original author of SPIDER is Joachim Frank (winner of the 2017 Nobel Prize in Chemistry).
Source: SPIDER: Overview
1 Like