22 Jun
2026
22 Jun
'26
1:47 p.m.
On Mon, 2026-06-22 at 16:34 -0400, Paul Koning via cctalk wrote:
Modula-2 was the germ for a lot of good ideas, some of which Wirth
backtracked on with Oberon. The best example is Ada, which has Wirth's
fingerprints all over it, even though he was not an official part of
any of its requirements development or detailed design. The main
failure of Modula-2 was not standardizing I/O, which COBOL, Fortran,
Pascal, and Ada all did. I wrote a lot of Modula code, but it wasn't
transportable because I had to re-write the I/O for every different
compiler + runtime library. I eventually wrote my own I/O layer so that
I could cope with incompatibilities in one place instead of everywhere.
Same as Dave Parnas's 1972 idea for modular programming (but Geschke
and Mitchell had a better idea in MESA — uniform syntax — which Douglas
Ross had earlier described).
Taos, the operating system for the DEC Systems
Research Center
Firefly multiprocessor workstation, was written in Modula-2+, which
added threads, garbage collection, and runtime type dispatch to
Modula-2. History and documentation of Modula-2+ (and Taos) are
available here:
https://softwarepreservation.computerhistory.org/modula2+/
Modula-2-pllus also spread from SRC into the DECnet architecture
group, where we adopted it as the language for writing the pseudocode
for the algoritms in all the Phase V specs.