In 1970, I worked for CUC (Computer Usage Company), at that time, the oldest software house in the USA. CUC was contracted by the Viatron Company (Corporation?) to write a full FORTRAN IV compiler.
We wrote two FORTRAN compiler versions, and I don't recall what options or features were in each.
Viatron computers (at that time) came in two flavors (but I never saw a "real" one), a 2K word version and a 4K word. I think the word was 16 bits for the 2k version, 32 bits for the 4k version, making the versions 4k bytes or 16k bytes. I could be wrong on the word sizes. It could've been 8 bits and 16 bits. Also, the compilers were a one-pass compiler. Kingston FORTAN II for the IBM 1620 was also a one-pass compiler, and it worked in about 20K decimal digits, instructions for the IBM 1620 were 7 or 12 decimal digits, so a one-pass FORTRAN compiler is viable (for those of you non-believers).
The "compter(s)" we worked with were only available at the company in a northwest suburb of Boston (Reading?, Woburn?, Concord? ...) and there was no CRT terminal, only a "robot" with many solenoids (one for each each key on the keyboard for an IBM selectric typewriter) which was our console. It supported punched card reader and punching via an IBM 029 (it could've been an IBM 026 though). The robotic-type thingy for the typwriter was a bit slow, so most of the time we used punched cards and had the punched cards interpreted (by some other IBM machine). There was a real printer interface, but it never seemed to be working at the time we were allowed to use the "computer", which was (as I recall) somewhere around 8 pm to 4 am. We wrote the full FORTRAN IV compiler in 4K (no lie) with a pretty secretive macro language that a smart guy at CUC invented. The FORTAN IV compiler had single/double precision, logical, ... I don't remember if it had complex arithmetic or not. It was a fun project, and as I recall, we never got paid for it, even though the FORTRAN compiler passed the Viatron testing. Viatron was having much troubles paying their bills at this time, but we kept working on it as we thought that even if we didn't get paid, we could still sell the FORTAN IV compiler for other "PC computers or home computers". I believe they were also known as hobby computers. The price was to be $99 for the small version, $199 for the bigger version of the computer. The robotic typewriter (pounding away at the keys of the IBM selectric typewriter) was fun to watch. It was like watching an automobile with all the fenders, doors, hood, and body removed, just looking at the working parts. It looked like something a Hollywood studio would've made up for a Sci-Fi movie, but nobody would've believed that it would've worked, let alone be practicable. But there is was. Cassette tapes were used for the FORTRAN IV compiler and libraries, and I don't remember how we updated the source code for the FORTRAN IV compiler. FORTAN source was inputed via the IBM typewriter or via punched cards. For the testing, we used a standard bunch of routines/programs that were on cassette tapes (for speed and durability). The FORTAN IV compiler code was essentially 100%
"assembler" macros, and we had an IBM punched deck of the macros, which was run through the Viatron "operating system" to copy to cards to a cassette tape, and then
assembled" to a punched deck, and the deck written to a cassette tape. The punched cards could've been used as input, but paper cards break down after a while, and the IBM 029 "card reader" was too slow for loading a compiler. I was drooling over the thought of a $200 computer for my home, but alas, Viatron went belly-up.
____________________ Gerard Schildberger