1 Mar
2026
1 Mar
'26
1:53 p.m.
On Mar 1, 2026, at 4:32 PM, Steve Lewis via cctalk
<cctalk(a)classiccmp.org> wrote:
Well, to clarify - my "amazement" was more the idea of developing OS
software using paper (not as much the loading of one from paper, though
that is still a "glad I didn't have to do that" thing :).
There's a nice description, I think in Gauthier van den Hove's thesis, of the
process used for the creation of the world's first full ALGOL 60 compiler, by Dijkstra
and Zonneveld. The two of them did this in about 6 months, which included having to
invent a number of core elements of parsing technology since they did not exist yet.
They would sit at a table, assembly language coding form in hand. One of them would
propose the next line of code, the other would agree or they would discuss it if needed,
and repeat for the next line. Those forms were then handed to punch operators to be
converted into 5-channel tape punched in the odd code that the assembler (also by
Dijkstra, part of his Ph.D. thesis work) could read.
Amazing stuff. Also amazing is that, after several years of production use, only two
wrong-code bugs were discovered, both for highly exotic cases.
So I'd characterize early OS development (meaning
like 1956-1961) as:
developed "in memory" (e.g. CTSS is said to have been written in FAP/MAD),
then the program exported to punch card (or punched tape-- fan tape being a
bit later and fairly exclusive to the "DEC" ecosystem, as it were). Once
verified "yeah this kind of works", maybe that code quickly migrated over
to magtape (bearing it mind this was all pre-ASCII standards). But one
would need some kind of "bootloader" to then initiate it from magtape.
Early PDP-11 software development was done on PDP-10 timesharing systems, including
running it in simulation (MIMIC -- a precursor of SIMH). The resulting bits would then go
to paper tape, I'm pretty sure, to be loaded into the target machine. That might in
fact be a paper tape only system, or if it was RSTS it would have disks but the initial OS
load would be paper tape.
Not all that long after, DECtape would typically be the OS distribution medium. I'm
not sure if those were created on PDP-10 systems; there was FILEX to do that, but I
don't know when that appeared.
By the time RSTS/E V5B arrived, development was on RSTS itself. I don't know about
V4A; system build (SYSGEN) used DOS, but it's hard to imagine people doing software
development on that sorry excuse for an operating system.
paul