Bringing the first native OS for Arm back from the brink
Steve Revill of RISC OS Open chats to us about taking the project into the future
Can't watch the video? Here's a transcript
Below is a machine-generated transcript of the video for those who can't watch right now. Blame any typos and whatnot on the AI...
Liam Proven (LP): So hello, and welcome to Retro Tech Week on The Register. I'm Liam Proven. I'm the FOSS and open source and public cloud correspondent for The Register. And joining me today is Steve Revill of RISC OS Open. Steve, thank you very much for coming. Would you like to introduce yourself?
Steve Revill (SR): Yes, sure. So I am the director of a company called RISC OS Open. We were formed in about 2006. We all came from a background in a company that was British-based called Acorn Computers. And we're very passionate about open source. So I'm very pleased to be sat here talking to you today.
LP: So um, yeah, thank you very much because I do have and use RISC OS myself. I have both Pi 400, and because I'm moving house quite soon, in its box, a 4 and a 400 sitting here, that has RISC OS on it.
So RISC OS was the original native operating system for the Arm chip, and came out of Acorn Computers who invented the Arm. How come RISC OS is open source these days?
SR: Well, back in the its origins, RISC OS was a proprietary operating system owned by Acorn Computers – developed by, as you said, the same people who developed the Arm chip. The Arm architecture... went off on its own path with involvement from various companies, notably Apple, but eventually it was spun out on its own to dominate the world. And RISC OS stayed as an internal project within Acorn Computers, and around the end of the '90s Acorn Computers was taken over.
And RISC OS largely became a forgotten project. But there were a lot of people who worked on RISC OS in the past, who felt passionately that this was a significant milestone in British computing history, its origins being tied so closely to the Arm architecture, and so we believed that it needed to have some sort of preservation.
And at the time, late '90s, early noughties, there was a lot of growing exposure of open source around the world, and especially Linux was certainly taking off and starting to take over. And we believe that open source was the way forward for the preservation, at the very least, of this piece of computing history.
And we negotiated with a company that took over RISC OS from Acorn called Castle Technology. We negotiated quite hard with them for many years, and eventually around 2006 We agreed with them that we would form a company called RISC OS Open and they would publish the source code for RISC OS under a permissive license that still not closed source, but it's not open source. It's somewhere in the middle. And that was called a shared source license, which was written from scratch, with lawyers, very expensively, when when I think we could have just picked an open source license off the shelf.
But one thing they were very concerned about was making sure that any work that was done on RISC OS they could stand to benefit from it financially because they'd invested quite a lot in buying the technology from its owners.
And our job as RISC OS Open was to do the grunt work and help Castle to relicense all of the source code under their shared source license. And that was an entertaining process. It was a part archaeology, finding out who owns the rights to various bits, are they still alive? Do they still remember anything about it?
Maybe it's companies that have been taken over several times and trying to agree sensible licensing terms for everything that we are publishing. And then it's trawling through the code itself and saying, is this suitable for release to the world?
Because you can imagine a closed source private piece of software. There's a lot of developers and there are a lot of pressures sometimes. They might write things in their code that you perhaps don't want publishing.
LP: Take out all the swear words!
SR: Yeah, that's correct. Any people insulting other people? So we even, we wrote custom code just to scan the code base for things that might be considered offensive, stuff like that to us. And it wasn't it was a very long multi year project with a small group of about four people really, all either ex-Acorn, or from the Acorn sphere of influence. And we're all software engineers. And we, piece by piece, published the source code and the Castle shared source license. And that was really that was the first step of a much more ambitious and longer journey.
LP: So now correct me if I'm wrong, I've been following the world of Acorn and Arm for... well, since since since, you know before the Arm was launched, I could never afford a BBC Micro in the day. But the thing that came out of Pace and getting the code from Pace was the Iyonix. That was kind of the first new product from Castle, wasn't it? Because they continued making some of the original Acorn hardware like the RISC PC, but the Iyonix was a whole new computer with an Intel-made Arm chip that didn't have the 24-bit mode, that Acorn used?
SR: The 26 bit. That's right. So there was a small group of us who many went on to become RISC OS Open in the demise of Acorn, and we eventually formed a small company called... oh, what was it called? I forget what it was even called now. Somatic, that's the one. We offered professional services and Castle who had bought RISC OS, needed a company to help them develop the next generation of Acorn, RISC OS computer, because they didn't just want to keep selling legacy hardware until eventually the whole market disappeared.
They wanted to spur ongoing development and ongoing growth in the market. So the Iyonix was born from a collaboration between Somatic and Castle technology. And as you said, architecturally, the chipset was passing an important milestone in Arm's history, whereas a lot of its legacy addressing mode – 26 bit mode – was thrown in the waste bin and everything was fully 32-bit. And for RISC OS it was a big problem. Because a lot of RISC OS and the applications that run on RISC OS were written in 26 bit mode, and they just wouldn't run on this chip. Now, before Castle took over acorn and pace had been aware of this change coming and a lot of work had been happening but it had never been finished.
And so there was a rush to get all of that work finished and alongside the launch of the Iyonix but it was an important step because that's opened the doors to everything that's that's happened subsequently.
LP: Okay. Okay. And, and so, that led, I think it was called, that was the sort of dawn of RISC OS five, is that right? That was like the 32 bit version.
SR: That was the beginning of RISC OS five, yes.
LP: Yeah. And, and that's the form that's now the now is open source. So now is is is rule risk is that there is another one, there is this other fork, but now I've written a bit about RISC OS on the rage recently, and I talked to the guy behind risk averse development, Andrew Ronsley. And he told me that as far as he knows that that sort of fork of it is pretty much dead now that there hasn't been any new development on it in
SR: That's right.
LP: That was mainly aimed at selling upgrades to existing acorn machine owners back in the day. And now it's owned by this company that sell an acorn emulator. So it's sort of aimed at the legacy side. And really RISC OS five is is the only version that's kind of going forward and developing. You run on the on the Raspberry Pi. But that is not by any means the only platform. It obviously only runs on ARM processors, but you run on various other machines as well. Isn't that right?
SR: That's correct. So as well as the legacy platforms that came out of the acorn era, and the iyonix that came from Castle they've been subsequent devices built from the ground up for RISC OS, such as the titanium, which is from a company called LSR. And that uses a much more modern chipset.
And then there's a lot of essentially open hardware platforms or close to open hardware platforms such as the OMAP range. And that's like the BeagleBoard the PandaBoard that I get five. So there's those platforms that we've ported to. There's the i MX six board, the wand board, and the pinebook. And we also have some ports to some more esoteric platforms that are a bit rough and ready, like Psion notebooks and various other products. I do they're probably not feature complete. Probably not as polished as some of the other platforms but it has been ported to anywhere we can. You will find a popular, easily accessible ARM based platform. That usually would be a very good target for porting because what we really want to do is lower the barriers to entry for anyone wanting to have a play with risk is an experience for the first time. And from that perspective, the advent of the Raspberry Pi was wonderful.
LP: Yeah, yeah, I mean, you know, they are remarkable little machines for the price. And for me personally, I think it was a lovely bit of a coincidence, really, I I am on a an internet bikers mailing list that's been around for years called Excel. And there was a guy on there saying, so I've got a Raspberry Pi. In fact, I've got two or three of them. What can I do with the thing you know, I don't want to build a weather station and I don't want a smart doorbell. And you know, I don't want to learn to write see what what can I actually do with these things? And I said, Well, there's this really cool little operating system called RISC OS that I used to use on Acorn Archimedes.
It's still around and it is like nothing else you've ever seen. It's it's very, very different. You can get it for free and and he got it and he played with it.
But another chap piped up and said, "Oh a mate of mine wrote that." Yeah, makes mine wrote that. And that ended up with me getting in touch with Paul fellows and him giving a talk at Google about about 10 years ago. So that was that was great fun and unlikely sort of circle. So um, these these days, it's a 32 bit operating system. But there is a tool to me like 26 bit for for some old acorn software. I believe I don't I don't know if it's included, but there was Gemini to allow dual head I think and you know, the PI four has to monitor outputs. It doesn't get working. But at the moment, it only supports Ethernet wired networking. And I understand that there are sort of some efforts afoot. So what are you guys doing to to modernize the networking because I recently discovered this might be a bit more complicated than it looks on the surface.
SR: It's quite complicated. And there is a wireless for the Raspberry Pi, there is a wireless option. So you can get a third party hat, and there's drivers for that. So that's great. But it's a bit of a halfway house, because from the perspective of the operating system, it looks like a wired network interface. It's cleverly written code. Really what's needed under the hood is for the network software on in risk is to understand the concept of wireless networks, provide all the user interface around managing passwords and all of that good stuff. Now, there's a lot more to that than meets the eye. And it's a bigger picture of the network stack being based on a quite an old version of BSD's network stack. There's a modernization program that needs to come first. And then there's additional stretch goals that come out of that modernization program, one of which would be supporting wireless interfaces, okay. So that's a program that is underway. It's just not a complete program of work.
LP: Okay, but the existing IP stack is actually already BSD code, is it?
SR: It's ported from BSD.
LP: Oh, I didn't know that. Okay.
SR: Yes. And it's essentially net being a very old net BSD at its core.
LP: Now I gather one of the other things is that some new generation arm chips and this doesn't apply to the, to the PI it only support 64 bit mode. And I gather there's a there's already a plan in hand for that. How far has that got along so far to this 32 bit ARM OS on a 64 bit only?
SR: There are various bits of experiments going on in various places. And there's certainly a clear awareness that there's clouds on the horizon very much like the transition from 26 to 32 bit we're going to see the same transition. We're seeing arm chips now that have ditched the legacy 32 bit mode altogether. Because every piece of the architecture, the ARM architecture you include in your core is just more die space, and it's more cost.
So if you only need the 64 bit node, then that's all you'll pass on. So, we know of the various routes forwards one of them is just a pure emulation routes, and there is a free risk is emulator called RPC MU, and that looks like a 26 bit ARM from RISC OS's perspective. So we can run RISC OS on that at the moment on for example, Windows or Mac or even Linux. There may be a similar step that's required to write an emulator that runs on a 64 bit pure Arm that makes it look to the operating system. Like it's a 32 bit is a bit sad, because realistically, you're not getting any of the benefits of the 64 bit architecture or bringing you.
Another route forward is to look at the core of RISC OS essentially as kernel and you look at all the for example, the privileged parts of it that what the called privileged parts, the bits that are really talking to the bare metal of the chip. At the very least, those would need converting, so that that's quite a long process. It's probably going to take many man-years, but that's something that we're looking at at the moment.
LP: I'm glad to hear it. I wish you luck because I think it's an amazing little survivor from the '80s. Really there's I think RISC OS and NeXTSTEP, are roughly contemporaneous they both sort of reached something like their modern form in '87, '88. And they're the only two that haven't been sort of completely rewritten in that time. So, um, where do you think RISC OS might go in the future then if apart from just trying to adapt to, you know, still work? Is there anything you'd really like to see that you'd love to see?
SR: Yeah, absolutely. So just one thing to be clear about is it is now true open source, it's under the Apache two license, and that was an important next milestone. So the dream going forwards. I think the my dream personally, is that we get good commercial sponsorship of some of the big ticket items like 64-bitting, or multiprocessor, support all of those things. They're quite big pieces of work and although RISC OS Open, we are a not for profit organization, it's very part time. And as you say, when you've got families when you've got other commitments in your life, it's hard to commit to something that is a many man-year effort. So what I hope is we can attract some commercial sponsors who see some value in using risk assessment, perhaps vertical markets. And can pay some wages of people. We've got our black book we know who the good people are. We just need to be able to afford to engage them for a few years to do some of the fundamental work. That would be my dream for the next steps.
LP: That's a that's a noble goal. I wish you luck. Maybe if I can, you know, bring it maybe to a few more people's attention via The Reg and so on that might that might help a little bit because I think it still has quite a lot to give and to teach. Right. I think we are heading up towards our time limit now. So thank you very very much for joining me today. Um, if it's okay, I will I will continue this conversation by email and maybe we can we can get some future article or something out of this. But thank you very much for all the work that you have done so far and for for leading and spearheading this effort, because you've saved one of my favorite bits of 80s computing and it's a delight to me that I can I can still use it today. Somewhere on my Twitter history, you'll find my first tweet: Free Pi on RISC OS which made me very happy indeed. But thank you very much it's been a pleasure. Thank you very much indeed. And we'll talk again soon. Thank you.
SR: Thanks, Liam. Thank you.