Forgot your password?
typodupeerror
Open Source Microsoft

Microsoft Open-Sources 'Earliest DOS Source Code Discovered To Date' (arstechnica.com) 44

An anonymous reader quotes a report from Ars Technica: Several times in the last couple of decades, Microsoft has released source code for the original MS-DOS operating system that kicked off its decades-long dominance of consumer PCs. This week, the company has reached further back than ever, releasing "the earliest DOS source code discovered to date" along with other documentation and notes from its developer.

Today's source release is so old that it predates the MS-DOS branding, and it includes "sources to the 86-DOS 1.00 kernel, several development snapshots of the PC-DOS 1.00 kernel, and some well-known utilities such as CHKDSK," write Microsoft's Stacey Haffner and Scott Hanselman in their co-authored post about the release. [...] This source code is old enough that it hadn't been stored digitally. "A dedicated team of historians and preservationists led by Yufeng Gao and Rich Cini," calling itself the "DOS Disassembly Group," painstakingly transcribed and scanned in code from paper printouts provided by Paterson. This process was made even more difficult because modern OCR software struggled with the quality of the decades-old printout.

Microsoft Open-Sources 'Earliest DOS Source Code Discovered To Date'

Comments Filter:
  • by v1 ( 525388 ) on Thursday April 30, 2026 @03:13PM (#66120818) Homepage Journal

    modern OCR software struggled with the quality of the decades-old printout.

    I'm amazed we can OCR in Egyptian scrolls on papyrus, but struggle with 30 year old green-bar printouts?

    • Re:OCR struggled? (Score:5, Informative)

      by LordHighExecutioner ( 4245243 ) on Thursday April 30, 2026 @03:42PM (#66120878)
      Don't be too surprised. I am struggling since a few months with the rebuild of an old FORTRAN program. It is about 15,000 lines of code, only a printout was available, and the OCR replaced randomly '0' with 'O', '1' with 'l' an so on. Multiple pass with compiler and ftnchek solved most problems, but still something is not OK. Luckily, in the printout there are some test examples...
      • Re:OCR struggled? (Score:5, Interesting)

        by 0123456 ( 636235 ) on Thursday April 30, 2026 @04:47PM (#66121014)

        The PGP encryption source code was printed in a loose-leaf book with checksums on each line to make it easy to OCR.

        It was still a huge project because they forgot to convert tabs to spaces (or vice-versa) before printing so software had to be written to try all possible combinations of tabs and spaces on lines where the checksum check failed.

        For the Apollo Guidance Computer code they got lucky and had a binary dump of the compiled executable at the end of the listing so they could run the OCR-ed code through the compiler and check for mismatches in the compiled binary to find the OCR errors.

        It's definitely non-trivial and can be even if the developers went out of their way to try to make it easy.

        • Compute magazine used to print program listings (as did a lot of the old computer magazines) and they had a program called MLX for entering Commodore 64 programs.

          You entered the numbers from the magazine line by line and the last number on each line was a checksum for that line.

          It worked really well. I remember typing in pages and pages of numbers and eventually ending up with the Speedscript word processor.

          • I'm still keying in programs for HP calculators. The 12c has a Moon Lander script running, and I suck as a pilot. Still.

          • by v1 ( 525388 )

            Back in the day most of the computer magazines had one of those. I think it was Nibble magazine that published several programs for the Apple ][.

            IIRC you'd start the program in the background and it would watch what line your cursor was on and display a two digit checksum in the upper right corner of the screen that would update as you typed. Just make sure that number matched the check on the end of the line in the magazine and you were clear to hit Return to save the line.

            A different magazine had a simi

            • by kenh ( 9056 )

              I have been involved in the personal computer world since the late 70s, reading about S100 bus computers and CP/M, but I don't remember these "checksum" tools for entering programs, but I was much more in the TRS-80 world, not Commodore or Apple, maybe that's why? Anyway, what I do remember is one magazine (Kilobaud?) that put flexible 45 RPM vinyl records inside the magazine, and you would "play" the record on your turntable, connected to your cassette interface.

              I also remember in the TRS-80 world there wa

          • When I got my first computer it was a Commodore, and I had a ton of those magazines. Eventually my disk drive broke, so if I wanted to play one of those little games or things they had, I literally had to type the whole program in each time.

            I would leave the computer powered on as long as possible so that the RAM didn't clear :).

        • One PGP edition had a lot of problems being OCR-ed. The one after that fixed it, where it was using a better font, checksums, standard spacing, etc.

          I may not be remembering this right though.

      • by antdude ( 79039 )

        Did they not use a slash in zeros to avoid confusions?

    • Not suprised - ribbon ink can fade like a mofo.
    • by Tablizer ( 95088 )

      I'm amazed we can OCR in Egyptian scrolls on papyrus, but struggle with 30 year old green-bar printouts?

      Scrolls don't have to compile.

  • ... that all internal self-references say "86-DOS", and it displays "Seattle Computer Products" on the boot screen.

    But yeah, let's pretend Microsoft created it.

    • Re:It's so old... (Score:5, Informative)

      by tlhIngan ( 30335 ) <slashdot@wo[ ]net ['rf.' in gap]> on Thursday April 30, 2026 @03:52PM (#66120910)

      Well;, it's well known that after IBM failed to get an NDA with Digital Research for CP/M-86, they went to Microsoft and asked if they could supply the operating system. Bill Gates agreed and then they purchased a full license of 86-DOS from Seattle Computer Products.

      They mildly patched it to get it working on the IBM PC (it was originally designed for SCP's 8086-based computer).

      Note the source code actually existed - the Computer History Museum actually has it as a digital artifact. The only problem was it wasn't open source - it's was until now only available as a source-available license for studying and curiosity. What Microsoft did now was put it under MIT license so it's under a fully open open-source license that lets you compile and build it.

      Also, Microsoft paid for a per-customer license for 86-DOS, paying $90,000 for it. The did this knowing they only had one customer - IBM. Eventually they hired the programmer of 86-DOS.

      MS-DOS 1.0 wasn't particularly interesting other than appearing like an independently created version of CP/M. MS-DOS 2.0 added additional services that made MS-DOS look a lot more like an operating system - instead of CP/M opening the files for you (and passing their handle in your process control block), MS-DOS 2 let you actually open a file by calling an open function. (MS-DOS 2.0 inherited a lot of semantics from Xenix).

      • The NDA thing is a bit embellished. Yes, Gary Kildall of CP/M was out flying the day IBM showed up unannounced, and his wife and business partner did not want to sign anything without him around, but the issue was resolved quickly once Gary got back into town and they had a deal with IBM. The problem occurred when Digital Research started getting way behind in their benchmarks and IBM was getting nervous that they would miss their release date. They mentioned it to Bioll Gates whi decided to pounce on a pot

        • by kenh ( 9056 )

          Tandy negotiated a one-time fee for Level II Basic for the TRS-80 Model 1, as I understand the price was based on selling about 10,000 units - RS eventually solde a quarter-million TRS-80s. Quite a good deal for Radio Shack.

        • In one version of the story there's an additional factor, namely that Microsoft wanted to pay a fixed, one-time, fee for the license whereas Kildall wanted to follow his usual practice of getting a royalty on each copy.
    • 86-DOS? Be careful, that sounds threatening! Somebody might sue!

    • by kenh ( 9056 )

      When has Microsoft denied buying 86-DOS from Seattle Computer Products?

  • And here I thought we were getting the code for the oldest denial of service attack known. I'm not sure anyone really cares about old versions of DOS.

  • This isn't the Kernighan and Ritchie we're looking for; move along

  • by JThundley ( 631154 ) on Thursday April 30, 2026 @03:50PM (#66120906)

    Does this signal a change in Microsoft's strategy of releasing bad operating systems?

  • by rsilvergun ( 571051 ) on Thursday April 30, 2026 @04:00PM (#66120924)
    The earliest basic code but somebody had already cleaned out the MIT trash cans
  • by Elektroschock ( 659467 ) on Thursday April 30, 2026 @04:01PM (#66120926)

    I wonder why Microsoft does not release more relevant code as Open Source such as Windows XP

    • Maybe too embarassing?

    • Re: Historical (Score:5, Insightful)

      by Malc ( 1751 ) on Thursday April 30, 2026 @04:18PM (#66120958)

      Probably contains bugs for things theyâ(TM)re still including in recent versions of Windows.

      • That and as scary as it sounds there's probably still critical infrastructure out there running on it. Likely would be a national security nightmare to release it so someone could run an AI though it and find every possible vulnerability
      • by AmiMoJo ( 196126 )

        Probably still have customers using it. There is an "embedded" version of XP that is used on things like ATM machines.

    • I would rather see win95 win98se get open sourced, just think if you can pull the windows kernel out from under it and build a custom Linux kernel for it, so if and when it crashes it don't crash the kernel too and only userland restarts, oh heck I'll just install iceWM and use the win95 theme
      • It would be nice for creating a win98 subsystem on modern versions of windows, to ensure that things still work for preservation purposes

      • Even something like MS-DOS 6.22 + Windows 3.1 wouldn't be bad. The one thing I'd like to see open-sourced is the Stacker code, as that did some pretty surprising reliable voodoo for its time.

    • Because all versions of Windows after XP are just slopped on top of it. Release XP's code and the slop house falls.

  • it hadn't been stored digitally.

    ... that there's a Zip disc [slashdot.org] sitting around somewhere.

  • ... will this be a compulsory upgrade from Windows 11 ?
  • by Reygle ( 5392954 )
    I look forward to when they open source modern Winblows in 2785 so everyone can laugh themselves to death.
    • Nah, everybody will be too busy watching the latest hit movie, "Ass"

      • by kenh ( 9056 )

        Nah, everybody will be too busy watching the latest hit movie, "Ass"

        And still waiting for "The Year of the Linux Desktop"

  • They couldn't type it in and build it as they go thru the code? It can't be that big.
    Using OCR seems half ass and lazy,
    as if the motivation for those in charge was only to generate some executive PR hype.

A boss with no humor is like a job that's no fun.

Working...