‹ BackHN Continuity

Thread

Ask HN: Who's still keeping a DOS machine up because the business depends on it?

297 points · 304 comments · mlaux

  1. smackeyacky · · focus · HN ↗
    I know of a DOS machine at a wool testing place that uses an ISA card based imager for measuring the micron and other things. They also have a supporting DOS machine with a dBase looking database on it. When I spoke to the owner he didn’t really understand the risks involved in running ancient hardware and in fact had a storage room with period PCs in it if the machines failed. He didn’t have much of a backup plan for the imaging hardware though.

    My opinion is that if these guys are still running this gear, they already have a strategy in place (no matter how flawed) and would likely retire before throwing money at something they consider a non problem.

    Good luck with the idea. It likely is a tiny niche market with customisation for every customer and given they are normally hardware constrained AI may not be much of a help.

    All of them have no money as well.

    1. Dwedit · · focus · HN ↗
      You can still get ISA slots on systems as new as Broadwell with an adapter.
    2. mlaux · · focus · HN ↗
      My idea is basically a DOS runtime with hooks on interrupts and I/O that would allow for either migrating databases to modern formats, or converting I/O accesses for modern USB peripherals. Customization for every customer and AI not being much of a help are pluses for me... lack of money not so much.
      1. Froedlich · · focus · HN ↗
        I have been porting sections of the business software I mentioned earlier from QuickBASIC on DOS (nice to have the source code!) to Tcl on Linux using ChatGPT to automate the conversion. It gets it right the first try, almost every time.

        The BASIC code is very simple by modern standards, and Tcl overlaps BASIC neatly, making conversion easy.

        Unfortunately it's just a short-term stopgap to add some features management suddenly can't live without, that weren't practical to implement in BASIC/DOS. They'll replace the whole stack with a commercial product and support contract when they find something they like.

      2. jasomill · · focus · HN ↗
        For anything business-critical, I'd tend to want to rely as much original code as possible, running actual MS-DOS on emulated hardware that could be tailored to suit your I/O needs.

        Lots of forkable code exists to get started with this, QEMU, 86Box, DOSBox, PCjs, MAME, …

        And if you do want to go the route of implementing DOS and BIOS interrupts yourself, DOSBox does this.

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.