Putting aside all the drama surrounding the contested ownership of the rights to AmigaOS, I think it's amazing that I could install new 3.2 ROMs in an Amiga 3000, in 2022, install AmigaOS 3.2.1, install a TCP/IP stack, then install a web browser, with working SSL / TLS, that works with contemporary web sites (which I did). This is on a 25 MHz m68030 with 16 megs of memory.
I noticed Google centric .webp images didn't load. I checked Aminet, and, of course, someone has uploaded a datatype for webp images. Downloaded and installed it, and the browser (IBrowse), is loading .webp files.
Web browsing is slow on this machine, but for most other uses it's quite snappy. I can't help but think we've all gone so wrong because we have machines which can quite literally do tens of billions of instructions per second, yet rendering a web page can take half a gig and multiple seconds on what's an insane amount of computing resources.
We can learn lots from Amiga. The fact that it's usable in 2023 already tells us we haven't been looking far enough ahead.
Other than gaming and a few specialized workloads[0], many people would do just fine with such an Amiga today, which is something most OEMs don't want to hear about, naturally.
[0]- Naturally there is the issue of memory protection, but that could be solvable.
Very late models did have an MMU, but it was only partially used due its history.
Naturally I am not advocating zero changes, also compare with similar SGI devices of the time.
Amiga 3000 was released in 1990, so that would be SGI IRIS 4D, IRIS Indigo, and IRIS Crimson timeframe.
Also keep in mind that Amiga designers were UNIX fans, and Amiga only turned out that way because of constraints that made them pivot their original plan, so an Amiga with MMU could be something into that direction.
The MC68030 also has an MMU, although some Amiga models used the MMU-less 68EC030. It's interesting that it can apparently do modern SSL/TLS at native clock speed, since that's what tends to limit the usability of older machines on the modern Internet.
I remember an old HP iLO that just didn't do TLS >1.0. Which for such a tiny OS really shouldn't be a problem. But they rather have you spend money on a new server.
I have this theory that the various new AI "Chat" services will bring new life for old machines, since their interface is mostly just plain text, they should work just fine even on the dumbest of terminals.
> I have this theory that the various new AI "Chat" services will bring new life for old machines, since their interface is mostly just plain text, they should work just fine even on the dumbest of terminals.
Except old machines tend to generate a lot of waste heat. Even if all you need is a text interface, you'll be better served by something modern sipping the milliwatts.
>Except old machines tend to generate a lot of waste heat. Even if all you need is a text interface, you'll be better served by something modern sipping the milliwatts.
My accelerated Amiga 1200 draws some 15w.
In contrast, my Ryzen 5800x3d based machine idles at some 75w.
If it's about interacting with an instance of chatGPT that's running in the cloud, the Amiga would be far more efficient.
If you compare to Ryzen or GeForce, then sure. But those devices are not optimized for power usage, but for performance.
There are also modern low-power devices, like Raspberry Pi, that offer a better power/speed/price ratio (i.e. it's probably less wasteful to run a C64 emulator on an RPI than to run the original C64)
Yes, but I'm not saying that you should use these old machines just for that purpose. I'm saying that if you're already using these old machines, as a hobby, it's much easier to make an AI Chat client than it is to make a web browser.
Amazing how much drama that can happen around failed operating systems that no-one cares about beyond a few diehard fans.
Similar thing happened with RiscOS, where weird small companies fought over scraps rather than open-sourcing and benefiting everyone. AmigaOS (and RiscOS) was amazing for its time and still has a lot to teach, so for it to be locked behind such nonsense is a pity.
Am just glad that the various people behind RiscOS did actually manage to get it together and open it properly [0].
(Still sad that Magic Cap got swallowed by IV, never to be seen again.)
Haiku is an amazing piece of work, but it would have saved everyone a lot of time and it’d be a lot further advanced if the original code was available, rather than having to implement it from scratch.
(Haiku developer here.) The correction is wrong; while some select portions of BeOS were released as open-source (most notably Tracker, Deskbar), the OS as a whole was not, and Haiku is thus largely a clean-room reimplementation.
Just about everything in this situation is disputed, but one theory I've heard is that they primarily publish these updates to establish active business use of the related IP rights (and hence standing/damages in a lawsuit).
I fail to understand what's the point in keeping closed source an operating system that runs only on a platform that ceased to be manufactured 30 years ago, besides maybe the usual patent nightmare, but anyway... if you want to get an idea of how the Amiga worked, the AROS operating system is an Open Source recreation of the old AmigaOS that runs on different architectures, including x86 and ARM, so you can run it on a PC or a *Pi board, either native or hosted as a process.
Even if there was a patent nightmare at some point, like you said the hardware ceased to be manufactured 30 years ago. I doubt AmigaOS is still covered under any patents.
I futzed around with AROS a year or so ago, and even played with it in 2005 on a hosted Slackware, if memory serves. The installation process was highly confusing. No doubt it works for the developers, but for those that just want to kick the tyres, it could be a lot friendlier.
Small tweaks to their website would help. In the main menu they have "Aros-Exec". The should call it "Forums", because that's what they are. "Archives" would better be called "Software", or "Packages". Sometimes tiny things can mean a lot.
There seems to be an active community for software uploads, which is nice to see.
One thing that is interesting to think about, what if they had gone OpenSource. They were a full on vertically integrated hardware manufacturer. Maybe the could have gotten more technical people, programmer, nerds, even universities on their computers, specially the Amiga.
Open source, as we understand it today, barely even existed when Commodore collapsed. If you did find someone who was familiar with the concept, they were likely to view it as a radical new idea. Those who understood the history of computers would likely see it as a quaint reflection of computing's roots in academic environments.
For most people free software meant freeware, software that was both free of cost and free to distribute but rarely included the source code. When source code was shared, it was usually for the benefit of other developers (e.g. examples and libraries). Those who did may have put it into the public domain, requested money for its use, or left it in legal limbo by saying nothing at all.
Yet the idea of distributing the source code as a means of collaborative development or to extend the life of software, few outside of the academic world would have accepted such notions. Such notions had been purged from the commercial world at least 20 years prior.
And that's even before one considers the personalities of the people running Commodore.
Yes, if you look at the programs I submitted to BeBits you would see I included the source files, but I was the exception. Too many times I would see an interesting program but no source file so I could modify it to my needs.
And I was more talking about early on, not late in the game.
> When source code was shared, it was usually for the benefit of other developers (e.g. examples and libraries).
Yeah that is the main benefit you try to get.
If you are launching a computer, like the Amiga, in crowded market you need adoption by developers.
From everything I read about Commodore their metal model of the computer market was that they sold computers and they just needed a new model of computer every year. They didn't really have a longer plan then what new computer are we releasing next year.
But as far as I know, they were pretty open in terms of releasing data sheets and so on.
If you consider these releases "official", maybe, at least as far as "desktop operating systems" go. It's older than Windows; Mac OS is older, but you could argue that modern-day Mac OS isn't really descended from classic Mac OS, so it doesn't count.
Sibling comment points out that various still-alive Unixes are, in a sense, still the same operating system as the original Unix.
I suspect various mainframe operating systems are older and still maintained, but that's a little out of my wheelhouse.
System V Unix was released in 1983 and I'm pretty sure Solaris, AIX, and maybe some other still-maintained Unix variants are still based on that same source code.
EmuTOS is older I believe. Or at least just as old. GPL'd version of the Atari ST operating system. It's directly descended from the GEM/GEMDOS sources from Digital Research dating back to 1983 if not earlier. And I believe there's pieces in there that go back further -- stuff that came from e.g. CP/M 68k.
2.11BSD still gets maintenance fixes when problems are found and of course you can trace a chain of descent for it all the way back to the first versions of Unix which were installed at Berkeley.
In terms of operating systems that mere mortals would have access too, quite possibly, though IBM may beg to differ (along with other companies that have inherited mainframe clients and IP).
I noticed Google centric .webp images didn't load. I checked Aminet, and, of course, someone has uploaded a datatype for webp images. Downloaded and installed it, and the browser (IBrowse), is loading .webp files.
Web browsing is slow on this machine, but for most other uses it's quite snappy. I can't help but think we've all gone so wrong because we have machines which can quite literally do tens of billions of instructions per second, yet rendering a web page can take half a gig and multiple seconds on what's an insane amount of computing resources.
We can learn lots from Amiga. The fact that it's usable in 2023 already tells us we haven't been looking far enough ahead.