Open Sourcing DOS 4

Original Article ↗ Hacker News Discussion ↗

Microsoft’s decision to open source MS‑DOS 4, including an experimental multitasking variant, is being treated as a significant win for software preservation and retro‑computing enthusiasts. Commenters delve into the technical and historical quirks of DOS 4.x—its reputation as buggy, its relationship to DOS 3.3 and 4.01, and colorful period comments in the assembly source—while comparing it to modern DOS clones like FreeDOS and emulation solutions. Many hope this release will pave the way for source code from later DOS versions or even early Windows and NT components, though they note legal, licensing, and reputational hurdles that make such moves unlikely in the near term.

Release contents and repo evolution

  • Source for MS-DOS 4.0 plus a separate tree for the early “Multitasking DOS” (MT-DOS) beta was published on GitHub.
  • Some binary/PDF folders briefly appeared under v4.0 then were moved into a v4.0-ozzie subtree for clarity.
  • There was at least one force-push to edit a personal insult in a comment (changing a full name to initials).

Historical and technical significance

  • DOS 4.0 is seen as a buggy, unpopular release; many remember 3.3/3.31 and 5.0/6.22 as the stable “peaks.”
  • The MT-DOS branch is closer to DOS 3.x with experimental multitasking and special disk drivers (bypassing BIOS on XTs, no IDE/ATA yet).
  • Comments point out 32-bit sector support in 4.0 and proprietary large-FAT variants in some OEM 3.x DOSes.
  • Several people actually built and ran MT-DOS on period IBM XT hardware, switching between apps with a hotkey.

Relation to FreeDOS, DR-DOS, and emulation

  • Many say FreeDOS and tools like DOSBox(-X), dosemu2, PCem already run “just about everything,” so this is mainly historically interesting.
  • Some retro users still prefer “period-correct” MS-DOS over FreeDOS.
  • DR-DOS/OpenDOS licensing is debated: some call it “open sourced,” others emphasize it’s only source-available with evaluation/non‑commercial restrictions.

Licensing, openness, and archival quality

  • Release uses a modern permissive license (MIT). Some argue this is unambiguous compared to DR-DOS.
  • Others criticize the “sanitization” (editing comments) and partial nature of the drop as bad archival practice, pointing to external critiques and asking for truly pristine, museum-style releases.
  • Several note legal and third‑party code issues as reasons why fuller DOS/Windows source drops are hard.

Prospects for more Microsoft source

  • Thread repeatedly asks for DOS 3.3, 5.x, 6.x, QBasic, NTVDM, Xenix, and even old Windows versions to be opened.
  • There is speculation and skepticism about ever seeing NT-based Windows or 9x source due to shared lineage with current systems and third‑party components.

Code culture, language, and nostalgia

  • People enjoy spotting in-jokes, profanity, and “brain damaged” comments, and discuss shifting norms around terms like “sanity check.”
  • Many share memories of floppies, CONFIG.SYS/AUTOEXEC.BAT tuning, TSRs, DESQview, and how early systems invited hands-on hacking in ways modern locked-down platforms often don’t.