1972 Unix V2 "Beta" Resurrected

Original Article ↗ Hacker News Discussion ↗

PDP-11 emulation and running V1/V2 Unix

  • People share tools for running early Unix on modern machines, notably SIMH and a PDP-11/20 emulator that currently boots this particular resurrected kernel more reliably than others.
  • That emulator is described as powerful but not user-friendly (no UI); SIMH is recommended for general use, and someone later reports success getting this Unix version to boot under SIMH as well.
  • Several commenters mention playing with other PDP-11 OSes (2.11BSD, RT-11, RSX-11M, VMS) via emulators and even browser-based PDP-11 + VT terminal emulation.

Emulator vs simulator (and compiler vs interpreter)

  • A long subthread debates the difference between “emulator” and “simulator”.
  • One view: simulators aim to reproduce internal behavior and hardware-level details for study; emulators prioritize correct observable results, treating internals as implementation detail.
  • Others argue the distinction is fuzzy and mostly philosophical; implementations and accuracy goals overlap heavily.
  • A parallel is drawn to compiler vs interpreter: in practice many tools do both (compile to bytecode, then interpret; perform constant folding; JIT, etc.), so simplistic definitions break down.

ed and early editing workflows

  • Many are delighted to see “real work” done in ed, with explanations of how its append mode and “dot on a line by itself” termination work.
  • Several reminisce about line editors on early systems (TECO, EDLIN, CP/M’s ED), teletype terminals, and cases where ed was still useful when screen editors failed.
  • There’s historical side-discussion about the “.” end-of-input convention appearing both in ed and ARPANET email protocols, and how these cultures may share ancestry.

Nostalgia vs living in the 1970s

  • Some fantasize about “living in that era”; others quickly point out downsides: leaded gasoline, ubiquitous smoking, asbestos, worse medical care.
  • There’s disagreement on whether quality of life has “skyrocketed”; examples offered include improved conditions for marginalized groups and life-saving modern cancer therapies, contrasted with stagnant wages and complexity-induced anxiety today.

Software archaeology and the SaaS era

  • Commenters praise this recovery as “software archaeology” and lament that similar work will be nearly impossible for modern SaaS: tightly coupled services, single live deployments, no installable artifacts, and deep inter-service dependencies.
  • Others note that much 1970s software was also effectively single-installation and is now lost; what survives did so because it was designed to be installed elsewhere and documented.

Early Unix productivity and minimalism

  • The thread recalls how early Unix pieces (editor, compiler, core I/O) were built in weeks by a tiny team, contrasting that with contemporary development overhead (process, PM, UX, build times).
  • Explanations offered: they were their own users; requirements were clear and small; no heavy process; and they reused experience from earlier systems.
  • Several argue that focused “hermit mode” (few weeks, no internet, minimal dependencies) can still yield substantial systems today, especially for low-level or embedded projects.

V1/V2 Unix file permissions

  • A question about strange ls output (e.g., sdrwrw) is answered using extracted man pages:
    • 1st char: file size class (small/large).
    • 2nd: type/flag (directory, executable, setuid, or none).
    • Remaining chars: read/write bits for owner and non-owner.
  • Commenters note how later Unix evolved these bits into the more familiar modern permission strings.