In high school in Toronto in 1978, we would hand-code our Fortran programs onto optical cards by strategically darkening in circles with pencil, to encode one line of source per card. The teacher would then stack your cards onto a reader, which would scan one card a time (about one or two per minute) and a magical thing called a "modem" would send the program to the University of Toronto mainframe, which would compile and (if you were lucky) run it. The teacher picked up the printouts in the morning on the way to school. Truly an exciting time, but I only saw other lucky students use it; I never got a chance to try.
Then I had a transformative experience through a one-week stay at the University of Waterloo, where we spent all our time in front of teletypes programming in APL. Boy those teletypes were noisy, especially with a room with twenty of them going at it full blast. When I went to bed my ears would still be ringing. Yes, APL was my first programming language, and frankly the trauma hasn't worn off yet.
When I started university in 1980 at McGill in Montreal, there still were punch card machines and punch card readers at the computer center, and its satellite centers. However, pretty much everyone at the school had a "computer code", ie an account on the time sharing system called MUSIC: see wikipedia. It was a lovely system, and you could run commands interactively. It came with Fortran, Snobol, Pascal, Lisp, IBM 360 assembler, and, of course Adventure!
I spent many hours playing Adventure, and of course mapped the whole cave in detail.
At McGill, there were two kinds of terminals: video terminals for the lucky people, and paper terminals from Digital Equipment called DecWriters for everyone else. The former had wisiwig code editors, on which you could prepare your source code and, when ready, "submit" your program to be compiled, run and printed. You would then walk down to the basement of Burnside Hall, where a surly operator would hand you your printout. That's when you learned that you had a typo in your code. Needless to say we were told over and over to WRITE YOUR PROGRAMS OUT before you typed them in, but nobody listened then, and nobody listens now either. In any case, the turn-around time from "submit" to picking out your printout was in the order of 15 minutes, not overnight as in TFA.
The design of Watfiv fortran focused on fast compilation. Your code wouldn't be optimized, but that was OK, because as a student the majority of your runs would have compilation errors. Typically you only ran your program, fully implemented, once or twice, after dozens of botched compile runs.
All in all, it was a great experience. What I appreciate the most, in retrospect, was that CS students were required to learn IBM assembler. Higher level languages don't fully make sense until you know what's going on at the CPU level. I still can't fathom how an intro course in Java can give you any true knowledge of how computers work.