For example, the "Domain_OS SR10.4 - 01 VUE desktop" is a bit confusing, and may cause people to miss actual DomainOS.
Apollo DomainOS (or Domain/IX, or simply Domain) had many unique and interesting things about it, but disappeared soon after being acquired by HP. It looked more like it might look if you took a programmer who had mostly only seen text terminals, and gave them a megapixel display with pixel framebuffer, a mouse, and the freedom to design the keyboard hardware, and told them to make what they would want to use.
VUE (around when the Unix workstation vendors collaborated on standarding on a common desktop environment) was for HP-UX , which was a very different operating system, and entirely different user experience. More of an early attempt at let's give non-power-users an accessible computer with virtual desktops and everything.
Similarly, Solaris had innovative OpenWindows (including but not limited to a networkable display system based on PostScript) before they got the common desktop environment.
SunOS 4.x (retronym "Solaris 1.x") and earlier could run the earlier SunView environment, which was more like monochrome early Mac than the later Open Look look and feel of OpenWindows.
Rather than just another name for Domain/OS, Domain/IX was actually a Unix compatibility layer that was an add-on product for pre-SR10 AEGIS versions, with SR10 merging it into the base OS (pre-SR10 had no built-in Unix compatibility).
AFAIK even though it's usually associated more with HP-UX, VUE actually originated at Apollo before HP bought them, although I'm not sure if they ever actually released it before the acquisition.
https://en.wikipedia.org/wiki/Pick_operating_system
My first actual job was working for a local health authority here in the UK, and they had a Pick computer running some database application thing, I think to do with accounting. I had to run the backups. Sorry to be a whinger, I don't mean to belittle the monumental amount of work.
I loved working in Pick BASIC on those systems. So much you could do with "dict items"
[1]https://www.facebook.com/groups/retrocomputers/posts/7062462...
I’ve also found source for an AEGIS menu system (mouse, hotkeys) written in Forth.
Not all gear got junked. When I was a teen intern, I got some obsolete Apollos (and 2 logic analyzers and a terminal) from my employer, and other people were also bringing home gear the company "sold" them.
Somewhere, there might well be an industry or university sysadmin or programmer who brought home a box of old QIC tapes, and one of them says "AEGIS" on the label, and it's in a garage/attic.
Also, rumor has it that at one point Boeing physically archived at least one Apollo network, because they apparently take documentation integrity extremely seriously. If that's true, they might have an engineering librarian or someone who could take an interest in making sure any versions of Aegis/Domain they need (and have preserved media for) can run on emulators or something?
Currently the terminal doesn't really process input itself, it just gives the program running the "raw" fd.
If instead the terminal gave the processes a pipe (for instance) and consumed all the pty input itself (and its end of the pipe being a buffer of that content), why wouldn't it be the same?
- Convergent Technologies CTOS
- Whatever the Rational R1000 ran (https://en.wikipedia.org/wiki/Rational_R1000)
- OS/400 (emulating an AS/400 or IBMi is problematic - not enough information available)
- Tandem's NonStop
- Stratus' VOS
- Nixdorf's NIROS/TAMOS
- Data General's RDOS and AOS. DG/UX is also kind of rare (the 88000 was a flop, but it ran on the Eclipse platform)
I hadn't included the R1000 emulator because originally they emulated it at the logic gate level and it literally took a week to boot. It looks like they now have a more traditional instruction-level emulator that is supposedly comparable to the real hardware in speed, so I guess I should try to install that.
I think a major reason that no AS/400 emulator exists is because people are afraid of IBM sending them a C&D, although I'm not sure if IBM would care all that much if somebody wrote an emulator for IMPI AS/400s. Those haven't been supported for well over 2 decades AFAIK.
I do know somebody is slowly working on S/32 and S/34 emulators, but they haven't shared either because both apparently are still somewhat flaky. Hopefully they release them at some point. The entire midrange/office computer category is very under-preserved. The only such machines represented in the OS museum currently are HP3000 and S/3.
I've never seen images anywhere for NonStop, VOS, or NIROS, nor do emulators exist for the original hardware they ran on. Later versions of NonStop and VOS run on x86, but I'm not sure if they run under virtualization.
I'd second this. Of course, there would need to be 3 emulators - the original proprietary CISC systems, the tightly-lockstepped MIPS systems, and the loosely-lockstepped Itanium systems.
HPE still sells NonStop on X86-64 systems now, including virtualized NonStop as a Service "cloud" offerings.
I'm also wondering whether/how they include OSes from devices that VICE already emulates, since that could save some work if they want to include OSes of Commodore devices.
1) I run my own systems in emulation and its always educational to see how other people handle configuration and sysadmin type problems. Much like programmers reading other programmer's code for educational purposes.
2) I have a genuine philosophical question which it appears I cannot answer by any means simpler than running it and trying it. Similar to the halting problem LOL. I wonder how the project handles operating systems like MVS/360 where there exists a perfectly good 1960s installation (which I have installed by hand from tape for the experience) however no one uses that IRL because the various MVS Turnkey projects provide seemingly infinite debugged and dependency organized patch sets. There's quite a difference between trying to white knuckle a homemade bare basic MVS/360 from the 1960s vs "MVS Turnkey 4" which basically just works out of the box.
Another example of #2 above is there's DEC PDP-8 OS-8 which technically boots... but the most common distro had a non-working but trivially fixable FORTRAN compiler (IIRC the runtime package filename was wrong or something similar). There's a lot of fun customization.
Another example of #2 above is I wonder how the author handles RSX-11M, distribute the ancient unpatched unmodified OS from DEC or ship something like the Billquist distro, or does the author ship the PiDP-11 RSX-11M (or is PiDP-11 shipping the Billquist RSX-11 distro now?)
I guess for people not into retrocomputing it would be like claiming some rando RedHat .iso from the 90s is "The" Linux operating system. Well, its "a" linux from one instant in time... Likewise there seems to be no "The" MVS/360 operating system there's a zillion possible local installs of all capability levels and eras, all very different and fun.
VM and MVS back then were interesting beasts. The source was available and many people customised them extensively.
I have been playing with VM/370 Community Edition for some time (https://rbanffy.github.io/fun-with-big-computers/fun-with-vm...), and it's an interesting environment. The other day I decided to make plain VM/370r6 available as a docker image (alongside my other images) and it is a very bare operating system, much less comfortable to play with than CE. The same applies to MVS 3.8j and the various community enhanced "Turnkey" editions.
You are basically expanding the zip file, and you can pick and choose.
I think I got it on an early Packard Bell Pentium system in 1994. I remember I used it even though it sucked because it seemed a little better than Windows 3.1 mostly due to the fact it didn't try to look like a functional windowing operating system. Once I got my hands on Win95 beta, I never ran it again. Of course, early Win95 also sucked as a real OS but it was enough better than Win 3.1 that I could slowly begin to transition off my beloved Amiga 2500.
Huh. I thought the term "shell" generally referred to command line interfaces to the OS and that Unity's way of describing itself as a graphical shell was some "new" (2011) generalization of the term, but I guess there is at least this precedent for using the word like that.
In Windows 3.1, the SYSTEM.INI had a setting called "shell" for overriding the default program started after Windows had loaded. Use of the word "shell" in this sense to describe a graphical interface dates back at least this far.
There was also a crude character-mode graphical interface called MS-DOS Shell, in 1988.
"shell" might be best generalised as "the first interactive program to run after boot" rather than "command line interface".
Lol
Past me's head would have exploded just to have seen modern Wikipedia and GitHub back then.
This is "hunter gatherer experience": click an element, see if it does something (opens a link, displays a menu), see if it brings any value. Such a waste of human life.
Just look at what happened to McDonalds.
It's one of those strange memories from my youth that I've been unable to confirm as an adult.
macOS sucks, but it's pretty
This preservation of old OS is important.
Spread the word, this needs to reach anyone who's interested in it.
Oberon?
Oberon, Oberon System 3, Oberon-2, Active Oberon (AOS/BlueBottle), Component Pascal, Zonnon, Oberon-07
one that i noticed missing: Novell Netware, I spent several years in de 90s developing software for it. It was the main office network server software on those days.
3.x, 4.x ran on relatively regular 32-bit PC server hardware. 2.x ran on the 80286 in protected mode, the only OS I know which did that.
Copies can be found at archive.org.
And even though there weren't very many 286 protected-mode OSes there were still several of them, with the OS museum including:
1B/V3 (a Japanese OS with an object-oriented desktop and extensive compound document support, part of the TRON project) Microport SysV/AT Prologue TwinServer (an obscure French OS that originated on 8080/Z80) Multiple versions of OS/2 1.x QNX 2.21 QNX 4.0 IBM PC XENIX
1B and TwinServer are especially notable since they were maintained as 286 OSes long after x86-32 machines had made 286 machines completely obsolete; the last versions apparently being in 1997 for 1B and 2002 for TwinServer (although the last version of TwinServer has some limited support for 32-bit code, it can still run on a 286)
But, originally wasn't it mostly a network system to support network printers and file systems?
BTRIEVE would run on top of that. But, as I understand it, Netware wasn't required. They just went together really well.
Finally, especially with Netware 386, they supported "NLMs". "Netware Loadable Modules". This was what let you deploy applications to the network server. Some databases ported to that I believe. I think Informix had a NLM version of Informix OnLine.
So, to me, early Netware seemed more an interesting network utility more so than what I, at least, would consider an "OS". Perhaps it was an OS, but just sealed off. At least until NLMs arrived, making the system more extensible.
I have no idea what facilities were available to NLMs, or how they were developed.
I think they were usually developed in C. Metrowerks had a compiler that could build them, and Open Watcom can still do so as well.
My friends father worked for a shipping company and their office ran off a 286 Netware server until the early 2000's. It was a big white label tower with classic orange monochrome monitor and large Epson dot matrix printer with tractor feed paper.
SCO Xenix.
Concurrent DOS/286.
DR FlexOS.
OS/2 1.x.
Coherent 2 (IIRC).
0: create an empty VM called osmuseum with 4 cores, 8192 MB of memory, no drives, OS type Linux 6.x - 2.6 kernel using whatever vmid fits your schedule
1: unzip the relevant part of the distribution:
unzip virtual_os_museum-2026.05.19-lite.zip virtual_os_museum-2026.05.19-lite/VirtualOSMuseum.utm/*
qemu-img convert Data/host_x86.vdi Data/host_x86.raw
qemu-img convert Data/guest_images.vdi Data/guest_images.raw
qm importdisk 700 Data/host_x86.raw ext-lvm
qm importdisk 700 Data/guest_images.raw ext-lvm
5: set the scsi0 drive as bootable in the vm Options settings page and move it up in the boot order (e.g. ide2, scsi0, net0)
6: boot the vm and change the network configuration in /etc/network/interfaces to fit your needs. In my case I changes the address for br0 to a free address on a local network segment, corrected the broadcast, network mask, gateway and dns-nameservers parameters to fit my network and restarted the networking service (service networking restart)
7: Things should now work, you may want to reboot the vm but it should not be necessary.
Run System 7 in an emulator and the menus look right, but the input feels wrong. What we're really preserving in these collections is the screen output, not the interaction. Which is fine for an archive — just worth being honest it's a museum of appearances, not of use.
It "feels right".
I wish we could have USB versions of period appropriate keyboards such as the VT-100/200/300, the ADM-3A, Apollo, Xerox Star, Symbolics, Apollo Domain etc, as I believe a lot of the realism is the physical parts of the machine you actually touch.
The screen is important, but if you have a fast-refresh HDR monitor, you'll be close enough. Many more recent machines had keyboards that were very close to modern ones and didn't any have special keys absent from a PC-104/105 key one. Mice are also relatively easy, as few machines had very different kinds of mice.
I remember the contrast between a SPARCStation and an SGI O2 and how the O2 mouse was smoother than the Sun's. Same between Windows and Macs at the time.
Design priorities...
What's strange, is that SunOS, Solaris, and even NextStep all supported higher baud serial mice. If you look at the mouse driver on SunOS for example, you'll see the logic which loops over baud rates until it detects valid mouse data.
And Sun did ship a mouse with a higher polling rate/baud. One. The wired ball mouse for the SPARCstation Voyager.
NetBSD doesn't have the baud detection loop, so there, for this single mouse, you have to change the kernel to make it work: https://www.netbsd.org/ports/sparc/faq.html#voyager-mouse
I don't know if it includes "every operating system I can think of". I can think of some things: TempleOS, BTRON (there might be more than one implementation; I know of an (apparently) abandoned FOSS implementation), Serenity OS, and some others that I do not remember what they are called.
Also, what might be useful for preservation is, in addition to the files and emulation, also the documentation for programming those operating systems. There would also be such a thing of consideration as documentation of old computers (including their instruction sets), which might be a separate project but potentially might be useful in combination with this.
Another thing would be somehow you can download individual systems together with information about the emulation, in case you want to use your own emulators for it instead of installing an existing collection with its own installers and launchers etc.
Some people mention uncommon features (and features that work in an unusual way). I think that would be worth making a article about too, and just because a feature is common does not necessarily make it good.
All of those OSes you mentioned are included. BTRON isn't a single OS, but a small family of OSes based on a common specification (just like Unix is); the OS museum includes the demo 1B/V3 and Chokanji 4. The FOSS BTRON implementation you're thinking of is almost certainly B-free/EOTA, which is also included. EOTA never actually implemented BTRON proper before it got abandoned. It basically just ended up being like a Unix based on an ITRON kernel.
Documentation for some OSes is included, although I've focused more on user/administrator documentation over developer documentation. It would probably be a good idea to include developer documentation though.
I've thought about making individual images available for download, but many of them are dependent on particular emulator versions and/or the common launch scripts so it isn't quite that simple.
I ran it very early in its life, got it to boot, then couldn't do much else with it :)
... and you may consider DESQview . Not quite an OS, but a precursor to Windows.
Apollo had a "UX" which didn't get in the way of the user, customizable ( i know, not in fashion anymore). Apollo had a multitasking, multiuser network distributed OS which ran on 4MB of RAM, with 3 personalities (AEGIS, SYS V and BSD) and with a versioning filesystem.
The download is a Linux VM, gotcha.
Are other OS-s nested virtual machines inside that Linux VM, or emulators (in which case, holly mackerel, that is even more impressive :O... and also why??).
Readme seems to imply it's emulators, but it also uses the words "virtual/virtualization" or "VM images" liberally sprinkled.
I have a container that runs a 4.3 BSD userland using opensimh; it's not super hard to set up, just takes a bit of patience and willingness to learn how opensimh works.
Nested virtualization for certain x86 OSes running in QEMU is supported, although you will have to enable it manually (VirtualBox has a checkbox for this in its settings). For VMs that support it, the QEMU launch scripts will automatically use KVM if available and fall back to TCG if nested virtualization isn't enabled.
So far on retry/resume #12, 97.3/120GB done (i am live updating this comment as long as i can)
seeding now. please seed too :)
and it is not resuming ...
2026-05-19 16:23:03 ERROR 522: <none>.
I'll dig through my collection of "abandoned" OS distros to see if I have something that could make an addition to your museum.
I have long held anxiety that many of these would vanish as certain university archives disappeared. It is nice to see them protected.
I am always on the hunt for AST, which was like, a vendors custom shell for Windows 95 but sold\included as if it was an OS in its own right. Its been eaten by history I think.
Are there any any operating systems that you'd like to add to the collection but haven't been able to find?
Maybe someone here at HN could help with that.
It sucks that there's no good way to port Linux directly to WASM UML-style, since WASM insists on implementing memory safety at the bytecode level with no way to bypass it. There is a very limited port, but it doesn't support paging. Not all the emulators would run on a full-featured WASM port if one existed, but that could be dealt with by just using user-mode QEMU to run whichever ones are x86-only.
It definitely needs to be extracted to the C: drive to work
Also, make sure you update your copy of Oracle Virtualbox
Once it's working, change the input from PS-2 mouse to USB tablet (even if you have a mouse), otherwise your pointer goes invisible, but is still active.
I found TempleOS 5.03 in there, by the way.
I'm looking forward to playing with the various DEC PDP-11 and VAX/VMS options, reliving my days at Rose-Hulman and Purdue Calumet.
The HP-41 calculator is a nice touch!
[edit] No, found it!
This is a treasure trove. And glad you made the whole museum downloadable, so this treasure does not get lost.
Oh, and did I say it was also a threaded-mail reader? A threaded-news reader?
Oh, and lastly, Emacs is a torture device.
org-tables would disagree.
https://www.gnu.org/software/emacs/manual/html_mono/ses.html
Also:
I'm not aware of any fully working TENEX images unfortunately. There are partial images, but last time I checked they weren't in a state that was even remotely usable.
“Microsoft Bob was a Microsoft software product intended to provide a more user-friendly interface for the Windows 3.1, Windows 95 and Windows NT operating systems, supplanting the Windows Program Manager.“
I'm also planning to add earlier versions as well as the later forks at some point.
https://os.ryo.lu/ Quite cool.Clicking on "download" though, I get this:
> Full edition (121G zipped, 174G unzipped): From Internet Archive
Not to be picky about free stuff offered by others, but I'd be more happy to download a non-zipped torrent, ready for use, where I can contribute BW back to the project itself as a means of gratitude.
https://www.atarimagazines.com/startv4n4/transputer.html
They were pretty much ahead of time with multiprocessing.
https://github.com/nanochess/transputer
First article in the series: