Ask HN: What is the oldest, still supported OS?
I recently discovered that TSOS, an old Univac OS that I used (and loved!) in the mid 1970's and first released in 1968 by RCA, is still supported (although the name has changed) as Fujitsu's BS2000 OS. Unix was released a year after that (1969). Is there something that beats these?
236 comments
[ 0.23 ms ] story [ 278 ms ] thread[1] Easy to observe when running z/OS under Hercules, as one of the console modes shows the CPU address size mode. A lot of the time during bootup of basic ADCD you'll see 24, a bit less 31, and rarely 64
as someone to young to have seen punch cards live this sounds really strange.
Note that bcd, ppt, and morse are all in the same binary. ppt simulates punched paper tape and morse will encode/decode International Morse Code.
I actually ran these yesterday when I was explaining some computer history to my 19 year old son.
You can still create virtual punched cards on Linux by using the virtual punched card driver in the Linux kernel (which only works on IBM mainframes). Arguably a more realistic virtualisation than bsdgames since rather than just being a novelty it can actually be used to do something useful (such as integrate Linux applications with legacy IBM mainframe software when both are running in separate VMs under the IBM z/VM hypervisor - you can connect two VMs such that cards sent to one VM’s card punch appear in the other VM’s card reader.)
[1] https://www.theregister.co.uk/2013/06/19/nuke_plants_to_keep...
The oldest (known) surviving code in FreeBSD dates back to November 1974, committed by Ken Thompson.
https://www.researchgate.net/figure/The-oldest-surviving-cod...
Because you can still buy a new PDP-10 (well, PDP-10 compatible, iirc a lot of I/O is different, but it can run TOPS-20 V7 with patches)
Of course S/360 has longer continuity of hw/sw combination.
Who is still selling new PDP-10-compatibles? Systems Concepts? XKL?
So various signs on heaven and earth point to still made equipment using TOAD-2s.
I don't think that proves they still include TOAD-2 in new systems. It is equally compatible with all TOAD-2-based systems going EOL: likely they had some new boards set aside as spare parts for the production systems, once all customers stop using the old TOAD-2-based systems in production, the spare parts are no longer required, and so XKL may have decided to quietly give away (or sell) those spare parts to well-connected 36-bit enthusiasts.
Furthermore, there is a good reason to believe they don't use TOAD-2 any more: if you look at the manuals at XKL.com, you'll find recent products are running an operating system called "DXMOS". It appears to be a custom operating system based on *BSD but with a UI inspired by Cisco IOS (so similar in architectural concept to Juniper JunOS, but aiming to be closer in CLI syntax to IOS than JunOS is). The various copyright acknowledgements in their manual (all three of FreeBSD, NetBSD and OpenBSD) support that conclusion (most likely they have picked one of the three as a base, but borrowed some code from the other two as well). It seems rather unlikely a *BSD is running on a 36-bit CPU such as TOAD-2, it seems much more likely it is running on some industry standard 32/64-bit CPU such as x86 (or maybe ARM or POWER or MIPS). Furthermore, the copyright acknowledgements and release notes also show dependence on a number of other software libraries (SSLeay, OpenSSL, OpenSSH, Dropbear SSH – not all at the same time, of course, in recent versions they switched from OpenSSH to Dropbear) which seem rather unlikely to have been ported to run on a 36-bit platform.
I believe my employer's PDP-11 ran RSTS/E or some variant, and the software was written in BASIC by my current boss. There were a lot of operating systems available for the PDP-11 for a lot of different contexts (including UNIX).
Z80 also dates back a really long time, and there is a ton of stuff that still uses Z80-family processors in many embedded applications (microwaves, calculators, etc).
I realize neither of those are probably the absolute oldest, but this is going to hinge on what each commenter's exact definition of "families" and "still in use" are. Active development/new chips being released? Still in widespread legacy use? One-off legacy systems still hanging on in an industrial installation somewhere?
In terms of what is the absolute oldest - again depending on your exact definition of "family", z/System is arguably the oldest computer family, you can trace a direct lineage back to punchcard tabulation programs, which date back into the 1880s. z/OS systems can run programs originally compiled for System/360, which was designed to be directly ported from plugboard programs for programmable tabulators, so in a "ship of theseus" sense that is likely the oldest computer family.
https://en.wikipedia.org/wiki/Datapoint_2200
Anyway, the PDP-11 wasn't really an OS but an architecture (like ARM or x86). The robotics software in question probably runs bare metal. It was very, very popular and will indeed remain around in pockets (in emulation) probably until the 22nd century. Ain't broke? Don't fix.
Not to be confused with Encom MCP[1], which was defeated by Flynn and Tron in 1982.
[0]https://en.wikipedia.org/wiki/Burroughs_MCP [1]https://www.imdb.com/title/tt0084827/
MCP is the most user hostile OS I've ever seen. It may be less intelligent, but shows the same contempt for users as its fictional counterpart.
Besides... I'd expect something emulating S/3x.
You could alter the system message catalog however you liked... Sounds like something that would happen to Whitman College's system :)
https://en.wikipedia.org/wiki/FBI%E2%80%93King_suicide_lette...
https://www.latimes.com/business/hiltzik/la-fi-hiltzik-email...
i.e. someone did sue about email. it's relevant.
https://youtu.be/1Lb3CNe6k4Q
Who would be using Burroughs MCP now and…why?
Just for fun?
Dysfunction is unbounded.
That is also the marketing messages when you check ClearPath MCP marketing materials.
[1] https://en.wikipedia.org/wiki/OS_2200
z/OS was released in 01966. BOS/360 made it out the door earlier, in 01965, thanks to the disastrous delays in z/OS, but it's no longer supported; DOS/360 (z/VSE) also beat z/OS out, is still supported, and is arguably the continuation of BOS. https://en.wikipedia.org/wiki/DOS/360_and_successors
Unix wasn't released in 01969. I think it wasn't released until Fifth Edition in 01974, though Thompson and Ritchie described the Fourth Edition in CACM in 01973. Fourth Edition had "over 20" installations, but I think all within AT&T. https://en.wikipedia.org/wiki/Research_Unix
...I think this might have been meant for another comment/thread anyway.
I think it's plausible that things we say or do now will matter 7900 years from now. The languages we speak today, and the beef, pork, and wheat we eat, are to a significant extent the result of the Bug-Dniester culture 7900 years ago; and Voyager 2 will very likely still exist 7900 years from now unless we catch up with it. And the advent of writing has made it much easier for direct cultural influence to span vast gaps of time.
I think it's very likely that the human species will become extinct within a century, but quite possible that something will survive because of what we say and do today.
Basically, Excel has a numbering system that uses A-Z for 1 to 26, but no 0. But the system still works just fine.
The book 'Purely Functional Data Structures' has lots more obscure numbering systems, and uses them to construct data structures.
To give a trivial example: unary numbers are basically equivalent to linked lists.
Skew binary number are equivalent to something more interesting!
[1] https://longnow.org/
For an extreme version of this, 1961 might be written in scientific notation: 0.196100e4
This is also future-proof as we will always have a built-in 0 at the beginning of the number implying that we never get to "1" and there will always be more time to solve the problems we can't even begin to understand that may exist 10k years from now.
Sorry for the sarcasm, it was the easiest way to make my point.
(To be fair to the Long Now, not that they necessarily need fairness, and I'm not a member I just read about them here on HN mostly, adding the initial insignificant zero was intentionally an easy "hack" to keep things recognizable and not get into quagmires like changing the epoch date.)
In the case of the Holocene Calendar the bugs would just be a question of negative dates; in the case of UTC-12:00 it would be a nonnegative epoch time that is nevertheless in 01969 instead of 01970, which will only happen if someone adopts a more westerly timezone like UTC-13:00, pretty much guaranteeing you won't test that case beforehand. (Samoa did something similar in 02011, but it was in the other direction.)
Of course, as other people have pointed out, you shouldn't use fixed-width date fields in your software.
(Hint: look at the publication date.)
128 bits isn't even enough to get to the putative half-life of proton decay.
There's some truly mind-boggling numbers in https://en.wikipedia.org/wiki/Future_of_an_expanding_univers... "the non-zero probability of producing a new Big Bang in roughly 10^10^10^56 years"(!).
These numbers are far beyond https://datatracker.ietf.org/doc/html/rfc2550#section-2.4.1 , which is only 10^14; about the time star formation ceases.
Time for an updated RFC!
Does preparation become pointless sometime after 8,000 years from now, but before 92,000 have elapsed?
Is the hypothetical problem so bad that solving it is worth the ambiguity between octal numbers and the extra effort for readers to parse 02022 as year?
Instead of just sitting around discussing vague ideas they thought it would be interesting to design and build something to showcase their ideas, so the "ten thousand year clock" was proposed. It is amazing how many problems you have to solve to build a clock that has even the slightest chance of still be running ten thousand years from now.
A side effect of the clock project was that some of the dates involved would have five digit years, so they thought using five digit years for current dates would be a good way of shocking people and exposing them to their ideas.
sprintf for a better tomorrow.
It's also annoying to read. It costs a half-second to think every time we read a year. That's rather reader-hostile.
Also, surprisingly, it flushes out the people who would have supported Turing's ultimately fatal prosecution because OMG he's such a weirdo, before their support is required for anything important. It makes super clear the distinction between who's here to have interesting discussions and who's here to attack other users.
See eg http://liyang.hu/church.xhtml
Sounds like a good thing to me.
It quickly identifies you as someone who prioritizes personal idiosyncrasies over communicating whatever it is you want to communicate - since pretty much every reader will do a double-take on the date (or is it a date?) and be distracted from reading the content.
That's helpful, since one can generally assume that anyone with that attitude isn't communicating anything useful.
Please don't do that.
But, obviously enough, I already know I'm violating the date formatting convention, and I could hardly be more aware of those people. My theory is that, like Carmen Ortiz, they like attacking people because it makes them feel powerful, and violations of arbitrary conventions give them the opportunity.
Most people will interpret this sort of thing as extremely pretentious and also assume that you're fishing for "compliments" and trying to look cooler than everyone else. Like an inside joke only you know about, so will someone please ask you about it?
This type of behavior is typical in a classroom but becomes a lot less culturally acceptable the older you get.
Who cares what most people think?
And, yeah, it's true that most human cultures are very conformist, although I'm not sure I agree with your assertion that social groups full of old people are the most conformist. That conformism is why, for example, it took the humans 2000 years from recognizing that light darkens silver chloride until they invented photography, and they're constantly killing each other en masse in warfare, and 50 years after the first Moon landing there are still no cities off Earth.
So, lighten up. Play a little. Violate arbitrary conventions. The mind you save could be your own.
How do they write BCE years?
I commonly say things like "Troy fell around 01100 BCE". Except that usually I look up the date so I don't get it wrong like that. I don't know what the Long Now folks do.
Except conventions derive their value from the proportion of people using it. If a convention is stupid for other reasons, it may be worth trying to convert people. But writing numbers without a leading 0 is not stupid; in fact the opposite is. Why not write the current year as 0000000000002022 to preemptively protect against the Y(10^16)k problem? Why not write your age as 027 so that you're prepared when you hit 100?
> The mind you save could be your own.
You telling people you're liberating their sheeple mind, like literally every conspiracy theorist says, is not helping your case.
False; see US vs ISO date formatting.
You're welcome to use the date formats you've proposed if you think they're better.
On what account is the Earth disappearing in a mere 8000 years?
It's superior in every way: minimal, unconstrained, universal.
Why write such a lengthy comment telling someone you're dismissing them rather than just dismissing them? Is it to discourage their behavior, or to warn other readers? Or simply to share how you feel?
"Anyone who shows noticeable personal idiosyncrasies in their writing must not have anything useful to say" strikes me as a heuristic of exceedingly little value.
I was familiar with the general Long Now ethos, but seeing prefixed years in the wild and going down this HN rabbit hole has been worthwhile.
Also seeing how politely & reasonably you folks debated back and forth on this usage makes me love the HN community (if we can even call it that?) even more.
I'm glad you enjoyed the conversation, though, and found it worthwhile.
One of the best ways to be polite is Principle of Charity and that's mostly what I saw here.
2022 says "I'm an abstract integer expressed in decimal: two-thousand and twenty-two", whereas 02022 says: "Likewise, but I'm a five-digit data type requiring a padding zero for some reason that will likely cause a problem when I'm eventually replaced by 99999, and that then needs to be replaced by it successor."
https://twitter.com/longnow/status/1436364868131586054
(Edit: fortunately no longer the top-voted reply.)
If you'd your conversations not be derailed by discussion of esoteric date formats, that can achieved very simply: use the typical date formats everyone is familiar with.
The problem is that there are bullies posting personal attacks, not that I'm the one being attacked. It's true that when griefers and bullies are around, it's the tall poppies that get cut down. But locating the problem in the poppies rather than the scythe-wielder is obvious nonsense. Someone looking hard enough for an excuse to attack others, Twitter-style, can always find a victim.
But you also brought this on by using their propaganda, perhaps in an attempt to promote them or perhaps not.
Knock it off.
Long Now doing NFTs is bullshit. You can choose to show your love for them and people should absolutely call you on that.
1. Your intent was indeed to attack me.
2. Your attack is motivated by my perceived affiliation with a group to which you consider yourself superior; indeed, a group you despise.
3. If this angers me, you consider that a very desirable result.
4. You intend to keep attacking me every time you see me posting a 5-digit year, because you interpret that as a sign of affiliation with the group toward which your antipathy is directed.
5. There is nothing I, or anybody else, can do to change your mind.
Am I interpreting you correctly? It's possible there's something I've misunderstood, so please correct me if that is the case.
If that is correct, you are expressing an intention to convert HN into an open sewer. I will say again: please immediately desist from this contemptible misconduct.
If it makes you mad, so be it. I don’t care about how it makes you feel. You do seem incredibly mad about it that you’re still replying to the thread days later, so I’ll assume you’re annoyed by the fact someone replied to you with something you don’t like, so yes I will continue to reply to anyone using 5 digit years pointing out the Long Now bullshit.
If you feel attacked by this, don’t. It’s not about you. You’re not something I need to worry about. We know some common people.
I will say this and please take this with jest and not any illwill, but stop fucking replying to this thread.
It sounds like you're saying I interpreted you correctly on points 2, 4, and 5.
I don't understand the contradiction between "I don’t care about how it makes you feel.", and your previous comment, "I love how mad you’re getting". That seems like a direct contradiction to me. Are you saying your previous comment was not an accurate description of your motivations, or that you were previously motivated by angering me, but have now ceased to care how I feel?
I have considered your request for me to shut the fuck up ("stop fucking replying"), and I decline it.
Mythology and literature is full of really bad things that happen to people when they claim to be gods.
If you assume that the convention is to just always add exactly one leading zero, it works out.
(It's still a bit silly overall, but there's nothing special about five digits then.)
(Just an interesting point of reference.)
DEC Alpha has extremely weak memory ordering. [1] In fact, it's the weakest ordering of any arch supported by linux, which includes extra fence instructions to support it. The memory model is crazy weak, but it apparently allows for extra speculative execution parallelism.
[1] Awesome Raymond Chen post, totally worth a read: https://devblogs.microsoft.com/oldnewthing/20170817-00/?p=96...
Is this still used en masse somewhere and needs updates? Consider me 100% ignorant.
Why care about the instruction set? It a very good regression test for the kernel locking primitives and data structures. Something "boring" like x86 with its TSO memory model will hide bugs.
Such a thing actually exists as part of VMS! DEC was one of the first to use binary translation commercially like that. First to run VAX code on Alpha, and then Alpha code on Itanium. And now Alpha and VAX code on x86. (Doesn't look like VSI will bother with Itanium to x86 on the assumption customers will have source.)
I don't miss having to think about architecture and different Unix variants all the time, but I do miss the variety.
Apparently the L1 cache was implemented with two blocks handling even and odd lines, and they would process coherency operations out of order with respect to one another which is why dependent loads can see data out of order. So not additional speculative execution parallelism but probably a simpler physical design for the cache.
http://www.theos-software.com/
Edit: Looks like it died in 2014 or 2016 or, at least, that’s the latest update I see.
Update: Yup, it’s now listed as suspended. Looks like it was started in 1983.
In my mind supporting an old OS means e.g. that someone is still running old terminals that still run actual Windows 95 (emulator or not), DOS etc.
It wouldn't be meaningful to say "this runs on an OS from the mid-80s!". Only to find that it's all C# code targeting last year's release of Microsoft Windows (which has its eventual origins in the mid-80s).
[0] https://www.technologyreview.com/2015/08/06/166822/what-is-t...
A near-guaranteed contract renewal as long as the system wasn't replaced. If it was to be replaced, you'd likely have a good amount of warning. You'd be one of a very small handful of people in the world able to support and maintain the system. You could become a deeply knowledgeable expert about the system over dozens of years likely without having to learn a new language, framework or system more than a few times in your career.
Of course, the downsides are huge, especially if you limited yourself to that one system. If the contract didn't renew, finding a new job would likely be incredibly difficult. I could see it getting mind-numbingly boring, I guess that might be a plus for some people, but would likely drive me crazy.
https://web.archive.org/web/20041204030934/http://www.blackb...