Today, it hit me that iOS is already ten years old. I consider iOS a relatively new and fresh operating system, but can we really say that at ten years old? In order to figure that out, I quickly threw together a little graph to visualise the age of both current and deprecated operating systems to get a better look at the age of operating systems.
It counts operating system age in terms of years from initial public release (excluding beta or preview releases) to the last release (in case of deprecated operating systems) or until today (in case of operating systems still in active development). I’ve included mainly popular, successful, consumer-oriented operating systems, leaving out more server or embedded oriented operating systems (such as UNIX and QNX), which tend to have vastly different needs and development cycles.
As far as the nomenclature goes, Windows 9x includes everything from Windows 1.0 to Windows ME, and Mac OS covers System 1 through Mac OS 9.2.2. Windows CE is currently called Windows Embedded Compact, but its line also includes Windows Phone 7, Windows Mobile, and Windows PocketPC.
Red indicates the operating system is no longer being developed, whereas green means it’s still under active development. The only question mark in this regard is Windows CE; its latest release is Embedded Compact 2013 in 2013, and while I think it’s still in development, I’m not entirely sure.
This graph isn’t a scientifically accurate, well-researched, quotable piece of information – it takes many shortcuts and brushes several questions aside for brevity’s sake. For instance, looking at the last official release doesn’t always make sense, such as with Windows Service Packs or Mac OS X point releases, and I haven’t even been entirely consistent with these anyway.
On top of that, the graph doesn’t take months or weeks into account, and just counts everything in terms of years. Linux shouldn’t technically be included at all (since it’s just a kernel), and you can conceivably argue that, for instance, Mac OS X is older than its initial release in the form of 10.0 since it’s so heavily based on NEXTSTEP. Amiga OS is also a bit of a stretch, since its development pace is slow and has even died down completely on several occasions. You could maybe possibly argue that BeOS is still in active development in the form of Haiku, but I consider Haiku a reimplementation, and not a continuation.
In any event, I originally wasn’t planning on doing anything with this, but I figured I might as well publish it here since it’s an interesting overview.
The granddaddy OSs are Unix and IBM mainframes.
Given that FreeBSD arguably has a direct lineage back to the earliest Unix (it’s directly descended from the original *BSD’s which were effectively patches to the original Unix), it’s not unreasonable to squint a bit and call it 45 (or 47) years old.
While FreeBSD has roots in ye olde Berkeley Standard Distribution, really, it started as 386BSD in ’92, became FreeBSD in ’93, and was heavily rewritten and released in ’94 as FreeBSD 2.0 (with no AT&T code).
Since X11 was available from day 1, it’s easy to say FreeBSD as a desktop OS has been around for 25+ years.
The lack of OS/2 on the graph is a bit peculiar, though. If you count eComStation and ArcaOS (makes as much sense as counting AmigaOS), then you’re looking at 30 years of OS/2.
FreeDOS should definitely be included, since it recently turned 23 and is still under active development.
https://sourceforge.net/p/freedos/news/2017/06/freedos-is-23-years-o…
Agreed, but going down that road one should also include ReactOS, and if we’re including that one (a re-implemented Windows) we should also include Haiku. For that matter, where is OS/2?
This could end up being a huge chart if we really wanted to get pedantic with it.
I’d not include ReactOS or Haiku because they are both pre-releases… However yes.. OS/2 forsure esp. since there was ecomstation and now Arca Noae.
What about SkyOS? Lol.
FreeDOS is widely used, though.
The same can’t be said about ReactOS or Haiku
We can’t really include OS/2, as it technically isn’t being developed anymore. The current incarnation is still the same WSeB core from 1999 w/ updates that go to 2002 and changes to USB stacks and such. All IBM does is keep milking that kernel so others can bundle and (re)sell it to others (like me).
Several of the OSes in the chart itself also are no longer being developed. That wasn’t a metric for inclusion.
DOS in general (not necessarily or just freedos) is very much alive and well today and in constant use. So the age is incorrect on the chart. Just about 40 years depending on when you start counting (1977 or 1980 or 81, etc)
You generally touch or are with someone who touches a device/computer one to a many times per day that are either running DOS or DOS was used in their production.
I’m sorry, but what??? Did you use Google Translate to translate this post to English? Because I don’t understand what you were trying to say.
Maybe it’s a bit convoluted, but actually understandable if you take the time.
I understood it just fine. But, just because the OS is still being used does not mean it is still being developed. Thom’s chart was only intended to include active development of the OS.
This is incorrect. In fact, he specifically said, Red indicates the operating system is no longer being developed, whereas green means it’s still under active development. The chart clearly has OSes that are actively in development as well ass OSes that are no longer in development.
DOS, as developed as a clone of CP|M in QDOS, then evolved into MS-DOS, is only actively developed in the FreeDOS project. There were, and I’m sure still are, many other DOSes which have absolutely no relation to CP|M/QDOS/86-DOS/PC-DOS/MS-DOS/DR-DOS/FreeDOS.
Just because it’s called DOS, doesn’t mean it’s what we’re talking about.
IBM was the earliest with DOS/360, released in 1966, it was a subset of OS/360 for the System/360 (I’m seeing a trend here) mainframe. DOS/360 was used on lower end S/360s, which had a disk drive. There was also TOS/360, for tape drives, and BOS/360 for even lower spec S/360s with no storage at all.
Apple had AppleDOS and ProDOS underneath Apple// Basic.
Commodore had CBM-DOS, which was baked into their floppy drives, and talked to the main system’s Basic in a networking fashion.
AtariDOS was a software layer to provide high level access to diskettes, but it was not the primary OS.
Amstrad used AMSDOS.
The ZX Spectrum used GDOS and G+DOS.
None of these are compatible with the DRI/MS/IBM DOS ancestral line you’re familiar with.
You may run into Datalight ROM-DOS, which is specifically designed to be version compatible with particular versions of MS-DOS, but it’s not developed beyond maybe fitting it onto newer EEPROMs.
If you really want to be completionist, CP|M (Control Program|Monitor or Control Program for Microcomputers) was “inspired” by DEC TOPS-10 (Total Operating System-10), and written in Gary Kildall’s own PL|M (Programming Language for Microcomputers). I don’t know how deep that “inspiration” went, I’ve never used either.
If NT includes windows phone, then the same logic would follow that iOS is Osx, ie both run off the same kernel and subsystem (Darwin)
It doesn’t matter in either of those cases. Those mobile OSes fit inside the timeframe of their ‘grandparent’ OS and so wouldn’t change the total timeframes.
Edited 2017-08-02 01:04 UTC
Windows Phone is included in the WinCE graph, not the NT one.
But that makes little sense for newer phone os versions that have the same NT kernel, but lots of sense for the older ones that have the Win CE descendants.
He never said it would make perfect sense, and that he took a lot of shortcuts and glossed over a lot of things.
Perhaps a better name would be Mobile Windows, to include everything from the phone side of things?
ProDOS 8 for the Apple II was originally released in January 1983 (34 years ago) and the latest update was last year: http://www.callapple.org/vintage-apple-computers/apple-ii/announcin…
Here are some more: https://retrocomputing.stackexchange.com/a/2380/75
Except that release isn’t really a release in the sense of it being under active development. ProDOS is done, though the hobbyist market keeps using it. The release you’re talking about was made by patching the last release from Apple and contributions from the hobbyist community, but probably more important for this chart is that it wasn’t an official release from Apple in any sort of supported capacity.
How do you get 5 years for BeOS???
BeOS was first released around 1996, we’re in 2017, so I count 21 years there. MacOS first came out in 1984, that’s 33 years there.
Please explain your methodology Thom
Was going to ask this myself, well the BeOS part anyway.
It counts operating system age in terms of years from initial public release (excluding beta or preview releases) to the last release (in case of deprecated operating systems) or until today (in case of operating systems still in active development).
Hi,
I think this has confused a lot of people.
It should have been “time from first release to now” instead, or called “time in active development” (and not “age”), or possibly combined both (e.g. a line for “first release until now” that is twice as thick during the period of active development).
– Brendan
Except if you equate ‘last release’ to ‘death’ as Thom appears to be in this case, the use of the term ‘age’ makes perfect sense. The problem is that most people don’t think like this.. As an example, I and most other sysadmins I know consider a piece of software ‘dead’ when it is both no longer actively developed, and is not used anywhere in production, which would equate to the present day with nothing on the list being dead for all the listed operating systems (although quite a few not on this list (such as Multics) are functionally ‘dead’ by this measure).
MacOS hasn’t been under active development since OS X was released in the early 2000s. Seventeen years is correct. The five years for BeOS is also correct. It started in 1991 but wasn’t publicly available until 1995. Like MacOS it lasted until about the time OS X shipped. I think R5 released in 2000.
Edited 2017-08-02 01:44 UTC
Yep, August 2000, though it was still sold commercially until late 2001. I bought my copy of R5 in the summer of 2001. A few months later Be, Inc. folded.
Thom, have I gone mad? I was using MS-DOS waaaay before Linux, and I started using Linux the year Linus released it to try to get myself off of minix.
I thought MS-DOS was released in 1982 (as in, I was there, man). That would make it 36 years old.
If memory serves, and sometimes it does not, BeOS was released in 1995 as a friend purchased a BeBox in 1995 and made a big deal about the OS…
I think some of these years in the graph are off…
Technically the initial release of MS-DOS is just a renamed 86-DOS by Tim Patterson, whom cont9inued his work at microsoft after the sale of the software by SCP.
So by that metric DOS in it’s unbroken chain would have to be counted from late summer 1980 until MS-DOS8 (realeased in 16 September 2000) so roughly 20 years, give or take a month or two.
I bet that it is listed as 19 years, since there is a month left for it to reach 20 full years. But one might consider Q-DOS as well, and then it reaches above the 20 year mark.
Edited 2017-08-02 01:58 UTC
MS-DOS 8 wasn’t released as a seperate product, and seperately installable, was it?
Soo not the point of this timeline.
From the article:
So 86-DOS definitely doesn’t count, since it never had a public release. With MS-DOS it’s less straight forward, since MS-DOS 7 and 8 have never had a seperate release from Windows (95 and ME respectively). MS-DOS 6.22 is the last independent version of MS-DOS released, in 1994.
Why does it not count? The list does NOT specify criteria that woul disqualify it. I think you are just Adam Cutlering out of your behind.
I quoted the criteria Thom used. It doesn’t count because Thom says it doesn’t count. God, either you are a troll or really thick…
QDOS (Quick and Dirty OS) was the original name. It was changed to reflect the stodgy nature of IBM.
this isn’t a list of how old each OS is, it’s a list of how long each OS was developed. MS-DOS was born in 1981 with MS/PC-DOS 1.0 and ended development in 2000 with Windows ME. 19 years.
2000? You’re forgetting embedded MS-DOS. For example, my Windows phone has a copy of embedded MS-DOS mobile.
MS-DOS Mobile Version 1.0
Copyright (C) Microsoft Mobile 2015
All rights reserved.
C:\>
Admittedly, this was released on 1st April 2015.
Nothing byt the version nuber was changed from DOS8.
For example: If i release BeOS 5.1 claiming it to be a new OS and make sure that it has a f–king stupid GUI call to break compatibillity.
Oh no that was Bernd Korz.
Edit: And i like the guy. I wish that it would have been legal.
Edited 2017-08-03 23:47 UTC
https://blogs.windows.com/devices/2015/04/10/our-guide-to-the-new-ms…
It was an April Fools prank.
While I understand some of the shortcuts and choices, I think some are not really easy to justify: Windows9x for example includes versions that where no real operating system (from 1 to 3.11) as they ran on top of versions of DOS, while 95 to Millenium stayed on its own without the dependency. So Windows9x should be cut down to just 5-6 years (from 93-94 to 2000)
Edited 2017-08-02 01:47 UTC
Win95 to Millennium bundled their own versions of DOS. It was always there.
All of these systems are hard to categorize, but this particular one seems the hardest. Consider the timeline: Windows 1 & 2 are real mode graphical interfaces on DOS; Windows 386 uses a protected mode memory manager and has to thunk back to DOS when its services are needed; Windows 3.1 uses a protected mode disk driver to do pagefile IO without DOS; Windows 3.11 uses a protected mode file system so it could do file IO without DOS; Windows 95 used the protected mode file system almost exclusively except where TSRs are present, and implemented long names there; Windows Me removed the UI to exit back to DOS. All had DOS underneath, it just transitioned from implementing half of the functions applications needed to being completely bypassed.
(Random trivia: if you shut down Windows 95 and are left with the “It’s now safe to turn off your computer” screen, it’s really a DOS prompt with a bitmap. Run “mode co80” to see for yourself.)
I’m aware that 95 to me had stripped down version of DOS underneath, the difference with the previous one is that the effort required to make these versions to run on a non MS-DOS DOS version was massive (and as of 98Se, impossible) so while it was there, it was merely reduced to a bootstrap mechanism and a subsystem… not a full OS as it was still within 3.11 for example
No it was full dos. To acces DOS8 in ME was nothing more than a bin hack with a simple hex editor to the boot loader.
Correct, unless you executed win.com. Then Windows took over and virtually nothing of DOS was left (see the link in my other post).
Thank you sir. That is and was my point.
Not that i need anyone agreeing with me for it to be correct, as Sagan said “It just is”
It was not a stripped down version of DOS. Windows 95 through ME had absolutely full versions of DOS, and could not function without them.
They could, and did function without them just fine.
No. They. Didn’t.
Win9x/ME had to have DOS boot the machine for them, they could not boot independently of DOS.
Everything needs a loader.
It would take only minimal work to make Windows 9x boot without DOS.
Besides, there is a HUGE difference between “Windows uses DOS to boot,” and “Windows is running on top of DOS”
Once Windows loads, DOS code is ignored, since it is just a bootloader.
The only time DOS is accessed by Windows is when you are using 16-bit DOS TSRs, as a compatibility layer, and even then, DOS is only used just enough to make convince the TSR that DOS is indeed running, even though it isn’t.
Based on DOS, though, Windows 9x is not. It certainly doesn’t run on top of DOS either.
Wow! I guess people back then(including me as a child) had lots of patience. I don’t know how I never noticed that screen allowed for input!
I think it’s not fair to call Win16 a DOS program. It was an operating system, more than DOS ever was. It was no less of an operating system than Linux, which you could call a BIOS or UEFI program.
Windows 1.* and 2* was a dos executive and not a gui. (especially true with 1.0) as you could not move the windows. It was more like a skinned dos prompt.
What? A GUI isn’t defined by movable windows…
Windows provided a GUI and it provided more advanced services than DOS did. Stop trying to revise history!
Windows 95 and 98 could be run on other versions of DOS, like DR-DOS, or PC-DOS. I ran then on DR-DOS for awhile at home. Other than using more 32-bit stuff, there wasn’t really that much difference between 9x and 3.x.
ME made it impossible to run on other versions of DOS, but MS-DOS most certainly was there (it’s how the PC booted, for example).
technically, 95 through millennium still ran on top of MS-DOS
Technically, no, they did not.
Technically speaking they did. Also win9x came with a full dos and not a limited dos. Everytime an application was launched in win9x dos was invoked to allocate a small memory block to said application. And win.com was itself running as a dos application, using a memory extender to bypass the 640 kB barrier.
Novell Netware otoh, only used dos during boot after which dos was terminated and removed entirely from the system, unlike win9x/me which merely partially bypasses dos.
The notion win9x/me did not depend on dos is a relatively new idea, and outright silly. Dos was not merely bundled with Windows9x. It was Windows 9x.
Well, Windows 95 disabled virtually all of DOS, unless you had some 16-bit legacy device driver loaded:
https://blogs.msdn.microsoft.com/oldnewthing/20071224-00/?p=24063
Technically you are just plain wrong. See comment by dpJudas above for explanation.
Windows 9x did not run “on top” of DOS. They had DOS as a component for backwards compatibility, but did not run “on top” of it. At best, you could say they ran “side by side”. And in ideal circumstances Windows 9x could do away without any DOS at all.
Edited 2017-08-02 12:45 UTC
Windows 9x does mostly the same thing as Netware. The only significant difference, though, is that Windows also absorbs the DOS environment, since it still needs to provide DOS-compatible virtual machines. That’s what the DOS prompt is in 9x, afterall – a DOS virtual machine. (Actually 16-bit Windows apps ran in a virtual machine, too, but it was one virtual machine for all applications)
The only other function DOS had in Windows 95 was to provide the 16-bit driver layer. If you weren’t using DOS drivers for your hardware – say, you had a Windows95 sound driver instead of a Windows 3.1 or DOS driver, then DOS wasn’t used at all.
In fact, even applications that used software interrupts for DOS system functions rather than the Windows API – say, int 21h for filesystem functions – those were still handled by Windows, unless there were any DOS-specific TSRs loaded that needed to work.
Edited 2017-08-02 21:08 UTC
Win95 through WinME could not boot the computer without DOS there doing so for them. whether they knocked DOS out of memory once they loaded or not is inconsequential. without DOS, none of them functioned. DOS is essentially their bootloader.
Windows wasn’t a self contained OS until Windows NT (in workstations and servers), and XP (on home PCs).
on a side note, I’ve never tried this, but really should; I’m wondering if Windows 95 will run from FreeDOS.
By that standard, Linux runs on top of DOS, when you boot via LOADLIN.EXE
I mean, the Linux kernel knocks DOS out of memory, but as you said, it’s inconsequential.
Linux does not NEED DOS to load, it’s there as an option. Windows9x/ME NEEDS DOS to run, it cannot boot on it’s own.
Linux still needs something to load it into memory. If Windows is a DOS-based operating system simply because DOS acts as a bootloader for it, why isn’t Linux a GRUB based operating system, or an EFI based operating system? After all, the only thing DOS does is load the Windows kernel into memory and start executing it. When that happens, Windows nukes DOS and takes control of the system.
In other words, why does the bootloader define the whole operating system in one case, but not in others?
DOS only loads Windows if it’s told to by either AUTOEXEC.BAT, or from the DOS prompt. Otherwise DOS only loads itself, in a fully ready and usable state.
Linux is the same in that respect. Linux only loads itself, unless it’s told to load a CLI or GUI. The only real difference is, if you don’t tell it to load a CLI or GUI, you can’t really do much.
Edited 2017-08-05 11:34 UTC
Not with 9x, or at least specifically, Windows 98, where a fresh install has empty AUTOEXEC.BAT and CONFIG.SYS files.
It boots straight into Windows
By that logic, Linux is just a GUI on top of BIOS and not a full OS. So is Windows 7. Or almost any other OS.
OK, so now we have the “Alternative Facts” camp chiming in.
Strictly speaking, Linux is only the kernel, it is not the GUI. OR the command line. OR the OS as a whole. Most people only refer to Linux as the OS for brevity, or misunderstanding of the ecosystem.
The BIOS/UEFI starts the system and preps it to hand over to the bootloader.
For Linux, GRUB, LILO, BURG, or SysLinux (among many others), which then loads the Linux kernel, which then loads the drivers and other system processes, configuration files, and any software configured to load on boot. You can use Linux just fine without using one of the many Linux based GUIs, but you still have to use one of the many CLIs to do so.
MS-DOS’s is built into the MBR of every disk it formats, but is only there to make sure IO.SYS and MSDOS.SYS are also on the disk, and then hands the boot process over to IO.SYS. IO.SYS then loads MSDOS.SYS which then loads COMMAND.COM. After that, if you have a CONFIG.SYS and/or AUTOEXEC.BAT, those are processed, and depending what’s in AUTOEXEC.BAT, you’re either dumped to the DOS prompt, or loaded into Windows(v1 — ME) via WIN.COM. No DOS -> no way to load WIN.COM -> no Windows. Once AUTOEXEC.BAT is done processing, whether Windows loads or not, the system is done booting.
While Windows 3.x to ME are certainly quite alot more than just a GUI, they are ALL still dependant on DOS to set the system up to run.
Edited 2017-08-04 15:22 UTC
The Linux kernel is still dependent on the bootloaders to run.
Once the Windows kernel is loaded, there is no DOS code being executed anymore, in the same way as if you use LILO – once the Linux kernel is executed, there is no DOS code being executed anymore.
Weird, that you’d be so nit-picky with Linux being only the kernel, but completely ignorant about DOS being only the bootloader.
Edited 2017-08-04 18:03 UTC
I’m sorry, I was over-generalizing when I said that, to be fair, I said DOS acts like Windows’ bootloader, not that it strictly was. The DOS bootloader is built into the FAT12/16/32 MBR.
IO.SYS and MSDOS.SYS comprise the kernel, and COMMAND.COM is the CLI.
from Wikipedia:
“In versions of MS-DOS from 1.1x through 6.22, the MSDOS.SYS file comprised the MS-DOS kernel and is responsible for file access and program management. MSDOS.SYS is loaded by the DOS BIOS interface, IO.SYS, as part of the boot procedure.
“In Windows 95 (MS-DOS 7.0) through Windows ME (MS-DOS 8.0), the DOS kernel has been combined with the DOS BIOS into a single file, IO.SYS (aka WINBOOT.SYS), while MSDOS.SYS became a plain text file containing boot configuration directives instead.
“With the release of Windows 95 (and continuing in the Windows 9x product line through to Windows ME), an integrated version of MS-DOS was used for bootstrapping, troubleshooting, and backwards-compatibility with old DOS software, particularly games, and no longer released as a standalone product.
“In Windows 95, MS-DOS 7 can be booted separately, without the Windows GUI; this capability was retained through Windows 98 Second Edition. Windows ME removed the capability to boot its underlying MS-DOS 8.0 alone from a hard disk, but retained the ability to make a DOS boot floppy disk (called an “Emergency Boot Disk”) and can be hacked to restore full access to the underlying DOS.
“In contrast to the Windows 9x series, the Windows NT-derived 32-bit operating systems developed alongside the 9x series (Windows NT, 2000, XP and newer) do not contain MS-DOS as part of the operating system, but provide a subset of DOS emulation to run DOS applications and provide DOS-like command prompt windows. 64-bit versions of Windows NT line do not provide DOS emulation and cannot run DOS applications natively.
“Windows 9x used the DOS boot process to launch into protected mode.”
Are Android (And more so iOS) even separate operating systems since they rely so much on Linux and Java in the case of Android and BSD and OSX in the case of iOS?
Android is a forked Linux (and iOS a forked BSD), but I think both have changed enough to warrent status as a seperate OS. Java is just a software layer on top, and not integral part of the OS. (Note this is all simplified, no time for the details…)
iOS and OS X are not BSDs
Let’s make a simple rule: if the ABI & API are incompatible, it’s a different OS.
Problem is, this would either make all OS releases a different OS, or it would make Windows 95 and Windows NT the same, depending on how strict you dictate the backwards compatibility.
Let’s make a simple rule: if it has it’s own kernel, it’s a different OS. If it uses same kernel as something else, it’s NOT a different OS.
Edited 2017-08-02 11:20 UTC
That would make Server/Desktop Linux and Android the same OS.
Which, in essence, it is.
No, it’s not. An OS is the kernel AND its userland. Linux is only a kernel. Server/Desktop versions of Linux wrap a completely different userland around the Linux kernel than Android does. They’re not the same OS, they’re merely the same kernel.
So if I put Audi rims on my Honda my Honda is now an Audi? Seriously?
Simplifying logic down to retarded.
An OS is more than just a kernel.
Audi rims won’t fit (5×112 vs 5×114.5), however most Ford rims will fit Hondas.
So I guess Hondas are Fords.
I don’t know what “rims” are, but if it means “engine”, then yes. The very essence of OS is it’s kernel.
Edited 2017-08-03 06:31 UTC
and to extrapolate the car analogy a bit more..
A. is the essence of a car really more the Chassis+bodywork combo – or the engine..?
It is of course both. But I think most folk would agree the chassis+bodywork is (generally) the larger component, but it won’t “drive the same” without the original/correct engine
B. Similarly when you get Microsoft reconstructing “linux” with Ubuntu userland, utiliities, libraries etc, adding their own translation layer and (of course) using their own NT kernal still ….it might “run” linux programs…..but it won’t drive like it any longer.
And only SOURCE CODE APPROVED mega-partners can really tinker with the engine etc. You or your local garage are SOL
So I just looked up what “rims” mean. It seems you are trying to sound retarded on purpose: comparing rims to OS kernel, seriously??? Rims would be better compared to window decorations in theming software such as Stardock WindowBlinds. Kernel is much better compared to car engine. And yes, if you put Audi engine in your Honda, it’s no longer Honda.
Edited 2017-08-03 08:51 UTC
Actually, no. According to the DMV, it’s still a Honda.
What’s a DMV? And why would I care what they think?
Putting Audi rims on Honda is like applying Windows XP theme to Gnome running on Debian. Don’t act stupid comparing rims to kernel.
Then don’t make retard level comments in the first place.
Well, you were the first one to pretend to be stupid and make ignorant comments. I just followed up to show you how absurd you sound.
Aside from the number of years and questions posted by others, I’d like to see some additional entries. Focusing on consumer-grade operating systems, I’d say OS/2 is missing. OS/2 is arguably still maintained and just made a release this year.
I question the validity of 31 years of AmigaOS.
Some point out 19 years of MS-DOS, but I’d say that’s technically correct given it’s original release in 1981 and final release in 2000. I think it’s too easy to conflate “supported by the original developer” vs. “still in active use in some capacity”.
FreeDOS is worthy of listing as well. It’s certainly been around for a long time and sees use in many applications.
If you think OS/2 is missing, than AmigaOS is definitively valid.
There is even a new bugfix release recompiled from original source for the 68k branch by Hyperion from last year.
(If you don’t want to consider the PPC reimplementation as the real real)
It is easy to establish a timeline for some operating systems: they were released on a particular date and development (more or less) ended on a particular date.
Yet Amiga OS, and many operating systems that were not mentioned, raise red flags. Did the development of Amiga OS ever stop? Are long periods with only minor updates considered active development? I am not terribly familiar with the story of Amiga OS, but what little I know sounds remarkably similar to the story of RISC OS: it was under active development for a time, the operating system switched hands with different parties developing different branches, and there were long periods with very minor updates. The one thing that I do not know about Amiga OS is whether development was halted for periods of time. (In the case of RISC OS, it was.)
As for Windows 9x, my main quibble is nomenclature. At a technical level, Windows 9x was quite different from Windows 3.x. In terms of development, it was the direct descendant of Windows 3.x. I am fine with using the latter for categorization, but labeling it as Windows 9x makes it sound like a technical distinction. Perhaps calling it Windows, or Windows (non-NT) would be better.
I don’t know why he lists MacOS X vs MacOS 9, but lumps all of Amiga OS into one thing. It definitely stopped being developed in 1999(I think that’s when BB2 was released) for 3.x. 4.x, while supposedly based on the 3.x code, has most everything rewritten for PPC.
Different CPU ≠Different OS
Android runs on ARM, MIPS, and x86, but they’re all Android.
Linux runs on many more, but they’re all still Linux.
MacOS and OS X both started on one CPU architecture, and ended up on another, MacOS was still MacOS, and OS X is still OS X. and yes, I agree with this particular division, as OS X is more of a descendant of BSD and NeXT than it is of MacOS9.
AmigaOS3 and AmigaOS4 are very different. I guess technically MorphOS is out there too. Most view them as Amiga-NG systems. When you have to emulate to run the old software, there should be a division, agreed?
that i will agree to, yes.
Not as different as you seem to think. The emulation argument breaks down too when compared to Mac OS X – when that switched to Intel, it also needed to use emulation to run older PowerPC software (Rosetta). Amiga OS4 is no different. The API is compatible, just the CPU is emulated so to the eye there’s no difference between software written for OS3 and that written for OS4.
Edited 2017-08-03 07:48 UTC
Pretty sure there were the two jumps for Mac though, from 68k to PPC to Intel.
I know about Rosetta, but I think there was still a compatibility layer between OS 10 and OS 9, right? And it would have been very similar to AmigaOS4 and OS3.
All I’m saying is if this list feels like it should differentiate between OS 9 and OS 10, it should between OS3 and OS4. Hell, AmigaOS4 is even developed by an entirely different company, so should be split even more so.
Amiga OS is of the “I’m not quite dead!” “Oh, shut up, you’ll be stone dead in a moment.” “I’m getting better.. I think I’ll go for a walk now..”
Yep, there were two CPU jumps on the Mac, but each time the same OS was on either side of it – OS8 for 68k to PPC and OSX for PPC to Intel.
The jump from OS9 to OS10 (X) was far more substantial, so much so that it’s an entirely different OS, and there wasn’t really compatibility between the two. Software had to be recompiled to run on OSX. On the other hand, OS4 is a port of the OS3.1 code to PPC and maintains full, transparent compatibility with it, just like a 68k app on an OS9 Mac or a PPC app on an OSX Leopard Intel Mac. There’s no compatibility layer other than CPU instruction translation; 68k apps use the PPC APIs natively, and it’s impossible to tell whether an app is 68k or PPC when you run it.
What you might be thinking of is that there is also an emulation of the classic Amiga chipset, however that’s a machine emulator (UAE) and not the same thing at all. It’s needed for software that *didn’t* use the OS but banged the hardware directly. That’s not really incompatibility between OS3 and OS4 though since such software doesn’t bother using either.
March 2001 was the last Haage release, 2006 was the first hyperion release, however shitload of patches and fixes was released in between. For example the ctdl lock bug was fixed and Hyperion had the source all this time, and was just not allowed to release a full OS because of that asshole Bill McEwen, as he retained the name and the rights to the 3.1 source code.
Bill Mac Ewen held back amiga development several years, for what? An UAE emulator for phones that worked less well than e-uae (amiga anywhere) and fucking amiga shirts that was never delivered.
He sold over 1000 shirts with the checkmark logo that was never delivered, and when people wanted a refund he was gone.
Maybe someone should hire a bounty hunter to track down him down and let all the people who got ripped off for that throw Red/White boing balls at him.
I was going to say something about lumping in Windows 9x with everything from 1.0 to ME, but damn, Windows underwent some massive changes during that process.
Really, 9x should be separate from 3.x. But, then again, Windows 3.x should also be separate from what came before, as well.
It’s sort of amazing how much Windows has changed while remaining compatible with previous versions.
Indeed, and I’d go further: Windows 1 and 2 aren’t OS’es at all, just graphical shells around DOS.
Not really. They took over and had more advanced operaing system functions so parts of DOS and part of Windows were what was running. Gray zone, sure but is wasn’t just a shell.
This Unix chart has been always fascinating to me.
https://upload.wikimedia.org/wikipedia/commons/7/77/Unix_history-sim…
And here is the extended version of it
https://www.levenez.com/unix/
(mouse over the “white strip in the upper part of the page or look at it as a pdf e. g. https://www.levenez.com/unix/unix_a4.pdf)
Atari TOS is still in development and it is from 1985. (TOS is based on GEMDOS): http://myaes.lutece.net
I was going to say that, but is it really? At least from the original Atari TOS. We have EmuTOS + FreeMiNT. Now with FreeMiNT I would agree, that’s what became of TOS and is indeed still in active development, with the tooling and build process somewhat recently moved into github.
Shouldn’t that be included with Be?
Surely the BSDs such as OpenBSD, NetBSD and FreeBSD should be on the graph? these are important, widely used operating system, certainly more than BeOS (which is admittedly very cool).
In other words fewer and fewer teens will have any clue about other operating systems besides iOs and Android.
As many people already pointed out, there are sooo many problems with this “data” and your approach to interpreting said data, that entire thing is completely moot/irrelevant.
Based on this graph, and accepting its methods, it looks like FreeDOS would be one of the longest lived operating systems. I think it’s been in development for around 25 years now, longer than the life of the system it was based on. I guess the same applies to Haiku, which has had a longer life span than BeOS by a long shot.
Solaris (OpenSolaris still uses Sun code)
Symbian (since PalmOS and WinCE are in the list)
JavaME (since Android is in the list)
VMS (!!!)
CP/M
Edited 2017-08-02 13:18 UTC
O yes! Solaris, Symbian and VMS are absolutely necessary in such list. How can one include Android and iOS but skip Symbian is beyond me… It wasn’t some kind of niche or little known OS — it was The King (at it’s time, of course).
I’d make reimplementations part of the chart, starting from when they first appeared. For example, AROS (reimplementing AmigaOS) first appeared in 1995, and is worked on today, making it 22 years old. Haiku, ReactOS, and FreeDOS would all be similarly listed.
I’d like to add RiscOS to the list of pet OSes that could have been included in the chart. Not sure if you could count the whole line from 1.0 to the latest as one OS, but there has been at least some development fairly recently: it was ported to run on the Raspberry Pi.
You should consider posting high resolution/DPI images next time.
I like the idea of comparing the lifeline of operating systems implementations.
Maybe we could create a collaborative spreadsheet to enter the initial and latest release of operating systems and use this to generate the graph?
Congratulations Thom, your little inconsequential graph is stirring up quite the debate, haha
Here’s an idea, maybe osnews could make up a more official chart for operating systems, and keep it maintained. It really should be upgraded to a “gantt chart”. Maybe osnews members could maintain it?
Also, as other’s have already noted, your choice of what to include and exclude seems personally biased, it ought to be more objective. Solaris, although not as popular as windows, it was big on desktop computing especially in industry and academia. Omitting it (and others like it) doesn’t do justice to the amount of competition we had in the earlier years.
Yes, one thing he has always been good at is stirring up debate.
I’d love to have seen the BSDs in there.
> You could maybe possibly argue that BeOS is still in
> active development in the form of Haiku, but I
> consider Haiku a reimplementation, and not a continuation.
Either BeOS is dead after 5 years but Haiku is alive since 10+ years, or BeOS is still alive thru Haiku since 20+ years.
I fail to see how it could be both dead and alive as a reimplementation.
Edited 2017-08-03 17:48 UTC
I did not realize the original source base was still under active development or support. I knew of clones and other OS’s that were tributes to AmigaOS…
I guess technically the source was released for 3.1… which arguably complicates things in the Amiga community. People don’t want to look at the code who work on AROS, because then some of it might bleed in, and then some jerk could potentially close the project down, or cause massive re-writes…
Sadly, so few turds in that bowl try to stink it up for the people who just want to breathe life into it.
As frequently happens, your pragmatic choices seem to be nitpicked with narrow definitions. You never claimed it’s perfect; it’s a fascinating survey.
Somehow the old systems seem shorter-lived and the newer ones longer than I’d have intuitively thought. Maybe that’s just a matter of how we experience time.
What, no love for RISC OS, released in 1987 and in continuous development in both private and open source up to today? That’s a good thirty years!
Wow, no IRIX love here, not even on the diagram.
entire chart is so flawed that you only cherry picking the OS..you didn’t put entire OS when you want to post your articule on the site why waste people time to read it at first..tried rewrite the articuled better for our understand and don’t leave out the important details ..its very rude to leave all UNIX flavors and apple os, all the BSD…if you going start mention the age of os..why stop at few favorites ..your flawed as editoria on your site. suppoese reasearch your work first and combine into great story about our OS history..start from the grandfather pdp’s and UNIXes..and wish you mention the fate of MacDOS and MACBSD..even microsoft unix version..even BeOS i was fond at one point in my childhood but LINUX still is my favorites and current os i used now for WIndows layer os
Edited 2017-08-04 14:27 UTC
VMS went public with v1 in ’78.
Still developed (kinda) by HP. Still being actively developed by VSI. Being ported to x86 by VSI. <3
I am thinking to buy Operating System PLR Articles from here:
http://www.easybuycontent.com