No more big 32-bit cores for RISC OS from 2022

plenty of Acorn Shall Rise Again fanatics

I have this horrible suspicion that at 70 I may be not very much older than the median age of this rabble.

Not sure that’s quite the right question though. Building the OS might be easier second time round for two reasons: better tools, and hindsight – we’ve got an (implicit) spec already, whereas the original team were developing the spec as the thing grew.

You, sir would never be considered for a management post – you’re far too quick on the uptake.
There is of course less pressure to invent new rather than “borrow” features and of course the look and feel needs to be largely the same.

However, any techie knows that estimating a larger value and coming in under budget is useful because you can spend the excess on a few extras…

the only thing their 100 metre mega-yacht is missing.

Acorn fanatics have yachts? What am I doing wrong…?

or would you get people insisting that such a thing only justified a minor version number?

;-) It’s definitely 7. Or maybe SEVEN. Anything else would be wrong.

That said, if it’s flexible enough to be easily ported to x86, I’m sure we can have long and tedious arguments as to why it would be called RISC OS on a CISC processor…

would never be considered for a management post

Hmm. I was deputy manager of an ITeC for a whole year (of a two year contract), depute (!) co-ordinator of a TVEI centre for two years (of a three year contract), and Director of Information Technology in a public school for a year. You’re right it’s not my métier, but it was my decision to leave each of these posts.

The most recent time I was offered such a job though was at ARM, and I wrote this little story immediately after turning it down:

https://www.deviantart.com/coshipi/art/The-project-leader-28849467

To the boss in question’s credit, he took it very well, and we’re still friends, albeit rarely in touch.

why it would be called RISC OS on a CISC processor…

I think the meaning of RISC must have shifted quite a bit for ARMv7 to be called RISC, never mind ARMv8.

Boss: You are, but you’re the project leader.

Hmmm, “you’re the project leader”

“Oh, sorry I missed that email and HR clearly didn’t get the promotion details”¹

Sometimes you can’t help saying things – in my case most of the time.
I’m quite surprised my job title actually includes the word “manager”.

¹ I think I have the wording roughly correct.

No, the wording in my story is closer to what actually transpired – probably, I don’t remember exactly, fourteen years later. But the boss and I were actually pretty good friends. He wasn’t really any more management material than I am – nor for that matter was his boss, also a good friend. Last time I saw middle boss was at big boss’s untimely funeral. Untimely: this was a few years ago now, and he was a fair bit younger than me.

Just a friendly reminder that conversations not related to the thread subject header should be moved to the Aldershot forum.

I’m quite surprised my job title actually includes the word “manager”.

I had the opposite issue – I was working as the “System Manager” of VAX/VMS systems. The term System Manager was used within DEC and all the manuals as meaning the person doing the job I was doing. However the HR department of the company I was in didn’t like the use of “Manager” for the job title since it didn’t involve supervising people. For most of the time the job title was “Senior Scientist”, which didn’t start to describe the job.

Building the OS might be easier second time round for two reasons: better tools, and hindsight – we’ve got an (implicit) spec already, whereas the original team were developing the spec as the thing grew.

It might also benefit from being able to port a lot of functionality¹ from other free open-source software, instead of having to build everything from the ground up.

¹ Once again, assuming the existence of a multimillionaire to pay the team of programmers who would be doing said porting.

Alas, while there are plenty of Acorn Shall Rise Again fanatics, I don’t know of any who feel that a computer running RISC OS 71 is the only thing their 100 metre mega-yacht is missing.

I think real Acorn’s fanatics are all probably busy debating how to best use SWP/whatever ARM instruction and how someone could re-implement MEMC in FPGA and break the 4MB RAM limit looool… Acorn was expensive yes, but it was for computer/hardware geeks and hackers and RISC OS 5 is the testimony of what Acorn was about.

Maybe Acorn’s management or Acorn’s employee wanted something different (I have heard many times the story of “competing with Apple”). But if they really wanted to compete with Apple then they did it all wrong ;)

To be able to judge the amount of resources it would take to update RISC OS, you need to say what it is about RISC OS you would keep and what you would change. For example, some people would like features that would make applications more robust, but other people would disapprove of changes that restrict low-level hardware access. For any big change you make there would always be people who would say that the result is not RISC OS.

Maybe someone needs to draw a line under “classic” RISC OS and just tell users to run it in an emulator or in a virtual environment. Either take what they like about the original system and build something new to sit underneath, or just use an existing kernel with good hardware support. Then work on ways to integrate the “classic” environment with the new system.

Nice – if there was a ARM 64bit diss mode for Zap :-)

Once again, assuming the existence of a multimillionaire to pay the team of programmers who would be doing said porting.

Linux was written by a single guy, it sucked at the beginning, but his ideals and ideas convinced others to join for free and now that OS dominates the world… So if we are talking about multimillionaires then we are talking about Windows… which in so many years of development / re-factoring and re-development still struggles to deliver the original idea…

Same for macOS until Apple embraced Open Source and restarted from there, yes then they added a lot to it, no doubt, but it still started from a free available source…

RISC OS doesn’t necessarily needs investors, it simply need ideas and ideals:

• What exactly is it for?
  • Retrocomputing? It’s already perfect for it, you can develop on it like it’s a C64 or a BBC Micro (RISC OS Pico is THE best retrocomputing platform EVER)
  • Modern Desktop? Then it lacks a lot of features and architectural choices and needs to be changed
  • Nostalgia nerds? RISC OS 3.20 is coming up, get on stardot for updates.
  • Hardware Hacking? It’s on the right path, but it needs multi-core support, WIFI and make sure it can use ALL the underlining HW included EMMC, USB3 etc…
  • Compete with macOS and Windows? Good luck with that, it’ll need all the changes of the Modern Desktop point + a proper Application Shop that offers a market to bedroom coders and start ups
  • Modern Coding? well read up…
  • Future proof? Then it has to be re-written in C, compiled for 64bit ARM (and possibly other RISC CPU, you know just in case), add a Universal Binary format for which compilers have to compile code for, add support for modern programming languages and be ready to change as the times changes
  • Hackable OS? It’s perfect for this, you can do whatever the heck you want on it already :)
  • IoT? It has a lot of potentials, but still needs full HW support and a task scheduler that works also without the WIMP. It should also be headless, but this can be done already.
  • Embedded Applications? same as IoT pretty much

• What’s the ideals behind it?
  • Protect user privacy? It had a good start, doesn’t spy on the user, but then it needs strong security and support for Secure OS enclaves etc.
  • Offer a safe Desktop? Think again, loads of changes ahead
  • Open Operating System for learning? Great start, but not as good and portable as Minix
  • OS for schools? That train has passed long ago, and modern school OS requires security as well as warrantee a future to students, in other words they have to be Industry Standards etc.
  • OS to have fun with? THE best OS for the job already, simple, small and offers most of what people may want to hack around and have fun with.

• Which Ideas does it offers?
  • Reviving CMT? This has potentials, but the OS and the APIs have to be built to be coherent and easy to use
  • New coding approaches? Not really, but people may come up with new ideas
  • New ways of processing data? No, AI/ML works better on Linux for example (right now), and when they will work better on RISC OS it’s not a new approach anymore.

Regardless where each of our vote may go in the list above (or adding more options to it) RISC OS won’t be able to cover all the points, so I think at some point people will have to make some decisions and this is so also for ROOL/RISC OS Developments/Cloverleaf etc.

However, with good ideas and good commitments, it is possible to re-write RISC OS so that it will be able to support the past and the future and, through time, also full-fill a lot of the points above as long as the Architecture of the OS will be that flexible and portable and as long as people are willing to make compromises.

Future proof? Then it has to be re-written in C, compiled for 64bit ARM (and possibly other RISC CPU, you know just in case), add a Universal Binary format for which compilers have to compile code for, add support for modern programming languages and be ready to change as the times changes

And even with that you’ll not be future proof at all. From a back-end point of view, future is ML/cloud platforms. And for user interaction, it’s web and mobile platforms. Oups, sorry, it’s not future ; it’s happening today.

The big web browsers are not and will not be compatible with RISC OS.
But we have still room.

1/ as back-end on IoT platforms
2/ as front-end on small HID devices

And so the question is not 64bit or not. But…
1/ MPU support or not? (to open RISC OS to Cortex-M cores)
2/ Good support of tactile mode + Weboob like platform + PinePhone port?

Linux was written by a single guy,

No, the kernel was originally written by a single guy. All that is Linux (or GNU Linux to the RMS fanboys) is a huge collaborative effort.

which in so many years of development / re-factoring and re-development still struggles to deliver the original idea…

I think that’s because Windows doesn’t actually know what it wants to be. Look at the abortive attempt to shoehorn tactile paradigms into a UI based around keyboards and mice.
This is perhaps one of the more interesting ideas in the world of Linux, that there are multiple window managers to choose from, each with their own strengths and weaknesses.

but it still started from a free available source…

Darwin, as in the underlying kernel, is based upon open source projects, yes. Because all that low level nitty gritty stuff is hard, so probably best to build upon existing code. The UI isn’t open, however.

RISC OS doesn’t necessarily needs investors

We desperately need investors, because ideas don’t get stuff done.

Compete with macOS and Windows?

Never. Anybody who thinks that is delusional. It’s a cut throat enough world that compared to the many Android clones and the iPhone, Microsoft largely failed to do anything at all in the mobile arena. Microsoft!

Granted, they had no idea what they were going to offer, as a desktop UI simply doesn’t scale to a mobile device (you’ll notice most Android apps are full screen and those that can split screen do so with restrictions; and can iOS multitask properly yet?), however Microsoft had a lot of know-how and a lot of money and still failed epically.
Us? Compete with that? Don’t even waste your time imagining…

Future proof?

There’s no such thing. “Future proof” is just marketing speak for “not immediately obsolete”.

Hackable OS? It’s perfect for this, you can do whatever the heck you want on it already :)

Yes, indeed. And on future devices?

It has a lot of potentials, but still needs full HW support

Yes, I think it really ought to be a priority to get onboard WiFi and Bluetooth working. I think somebody is fiddling around with the maze of twisty passages that is Bluetooth, and the WiFi part might be slightly hobbled by our prehistoric network stack. Hell, we don’t even do IPv6 yet.

a task scheduler that works also without the WIMP.

This is, actually, a rather interesting idea. I wonder if it might be possible to have a module that can register and run tasks in a polling loop similar to the Wimp, but with no UI components? With the caveat that programs are executed in-situ (none of this &8000 malarkey as it would clash with everything else) and these programs get forcibly time sliced if they don’t hand of back to the coordinator in time. That way, they can sit in the background and just “do stuff”.

Protect user privacy?

Depends upon what you mean by user privacy. The OS doesn’t spy on its users (something I find rather abhorrent), however there is nothing to stop me adding a few lines of code to Manga to quietly rummage around your harddisc looking for things of interest (PhotoDesk or ArtWorks registration key?) to add as a payload during the next check.
I wouldn’t, because pulling that kind of crap is even more abhorrent, however the OS offers no barrier between the applications and anything on the filesystem. Of course, doing that in either an Android way (ask permission) or an iOS way (flat no, you’ll just get a little sandbox that nothing else can see) would break this lovely hackability, but do note that freedom and security are mutually exclusive.

Open Operating System for learning?

No, not until it’s written in a language that people can read.
I think part of the overall lack of enthusiasm with hacking and tweaking bits of RISC OS was never the licence, and isn’t really the DDE being non-free. It’s more simply a dwindling number of people capable of actually having a clue that the hell all that gibberish means.
Kids in the ‘80s learned assembler because it was kind of necessary to get useful things happening in sub-10MHz processors. Few people these days delve that low, there’s no reason when the stuff we counted cycles over can be done just as easily in Scratch, Lua, php, and a dozen other interpreted languages.

OS for schools? That train has passed long ago

I concur. The easiest way to watch a system break is to let a bunch of schoolkids loose on it. If there’s an obvious flaw, it will be found.

Case in point. My shiny A3000 plugged into the school Econet with a FileStore and a bunch of Beebs/Masters.
My 8MHz RISC processor more than capable of keeping up with the really slow network clocking that the BBCs needed (and the FileStore, that was bloody slow as well). A network where all of the passwords were sent over the wire in the clear, using a topology where every station saw the data every other station was sending.
Join the dots.
It took minutes to grab the SYST password right off the wire, just by hijacking the ADLCs FIQ and dumping everything received to a big circular buffer in memory.
Sad I don’t have that code any more, it was the only time I’ve fiddled that low, and it was…fun…getting it to work. Many complete machine freezes due to random stupidity. ;-)
I did a version that simply used OS_WriteC for everything received, because as slow as the vdu drivers were, Beeb era Econet was even slower. But obviously that couldn’t be used with a teacher around.

OS to have fun with? THE best OS for the job already,

Depends upon your definition of fun. I’m quite liking Android since it allows me to watch The Bureau Of Magical Things (that’s a pleasant way to end a day that doesn’t require too much thinking nor runtimes measured in hours).

Reviving CMT?

Personally I think a hybrid approach would be best. CMT by default (there’s nothing wrong with CMT done correctly) with the system able to seize control and poll over to another task if an app stalls without specifically telling the system “this will take time”.

future is ML/cloud platforms

They’re all the rage today. But will they be in the future when people understand better that their data is no longer their own, and their access to it depends upon the whims and management of somebody else?

Cloud is easy and cheap. But if your business depends upon cloud, you’re effectively dropping all of that stuff in the hands of somebody else.

MPU support or not? (to open RISC OS to Cortex-M cores)

Not. Too much of how RISC OS currently works depends upon shuffling pages of memory around. Always has.
The lack of MMU also means that an errant program may also totally trash the kernel (no memory management implies no access control).
There’s µCLinux that can run on no-MMU systems, but it’s rather different to Linux; it also needs a Cortex-M3, though I don’t know if it needs the M3 or if you just don’t get enough memory on M0 class devices?

You might be able to make a hack of something that feels a little like RISC OS that has stuff in fixed (restricted) areas and can only run one “app” (at a time). Might work for embedded devices, I suppose? But you’d probably need to lose 99% of what makes RISC OS what it is.

Personally, I’d wonder if making something akin to a BBC Master on acid (or, you know, Arthur) might be a better option here? A separate creation from the RISC OS stable, but not RISC OS but something with similarities that can run on the more restricted hardware?

Good support of tactile mode

Yes. In this day and age, an “X, Y, button” is inadequate. But since there was never any actual proper mechanism for adding “pressure” (like tablet stylus tips), I won’t hold my breath for anything exciting like multitouch. ;-)

But will they be in the future when people understand better that their data is no longer their own, and their access to it depends upon the whims and management of somebody else?

That’s completely the case today.

Not. Too much of how RISC OS currently works depends upon shuffling pages of memory around. Always has.

I know. But it’s perhaps an easier task than a full 64bit OS :)

The lack of MMU also means that an errant program may also totally trash the kernel (no memory management implies no access control).

MPU does that.

separate creation from the RISC OS stable, but not RISC OS but something with similarities that can run on the more restricted hardware?

Why not. A limited RISC OS for microcontrollers would be great.
And we could reuse some RISC OS compenent and tools: Basic, DDE…

“Future proof” is just marketing speak for “not immediately obsolete”.

I took the view, a number of years ago, that anything I implemented at work was going to last a lifetime¹ so I didn’t have to revisit the job. Sometimes that meant producing script/code that was so heavily commented that it was obvious to a five-year-old what needed altering to fit some new circumstance.
Variations to fit, but the general idea is that nothing is allowed to fail or need renewing before I retire.

The easiest way to watch a system break is to let a bunch of schoolkids loose on it. If there’s an obvious flaw, it will be found.

I note this: “Raspberry Pi OS (previously called Raspbian)” and wonder is that Linux flavour diverging even more from the Debian? If so is that to aid kiddies getting down into it or to stop them fiddling?

¹ For “lifetime” read my working lifetime. You see, the future is someone else so when my stuff finally breaks or can’t cope – N.M.P/S.E.P. I did my bit.

No, the kernel was originally written by a single guy.

Actually Linux IS the kernel… the entire distribution is Ubuntu or Debian or Fedora… Stallman insists it should be called GNU Linux because the tools used to build Linux and a Linux Distro are GNU.

We desperately need investors, because ideas don’t get stuff done.

Investing on WHAT exactly? Apologies, not trying to be mean or something, I am simply struggling to understand.

Investors traditionally wants their money back + interests produced by a product/idea/solution built using their investments and that has sold/created a market/produced more money…

So, my question to you is: Did you mean we need someone to pay for all the work that has to be done regardless the outcome (for example charity donations) or did you really mean investors in the traditional sense?

And on future devices?

That’s exactly my point, it can be kept a hackable OS on future devices, but hackable will unavoidably result into a non stable system for a user that is not careful, hence is it a goal or a side effect? my point was this is a requirement that would collide with the Daily Driver OS requirement for example… makes sense?

This is, actually, a rather interesting idea. I wonder if it might be possible to have a module that can register and run tasks in a polling loop similar to the Wimp, but with no UI components?

Yes of course it is possible! Do you want to start a thread on here or on chatcube and put some ideas together and starting some code for fun?

Depends upon your definition of fun.

True, my definition of fun is hacking some code to do something creative, play music and recording it, playing games that I like and that do not cause me stress and infinite learning curves lol have low level access to interface some electronics and do something creative with it, design new programming languages or just play porting old ones… so all stuff RISC OS can do, but you are right this one is subjective.

future is ML/cloud platforms

Sure, but you still need an edge device to access that (or to get your own data stollen lol).

I think RISC OS should be perceived as edge computing platform, not cloud. We have cURL now, and HTTPLib, embedTLS and AcornSSL, box access and others, so pretty much getting close to access a lot of cloud stuff and sure we can add more and more protocols for this, but I would recommend we get the multi-core sorted first and the modern browser too.

Why not. A limited RISC OS for microcontrollers would be great. And we could reuse some RISC OS compenent and tools: Basic, DDE…

Agreed and not just that. IoT is an actual market, in other words Start Ups and Companies willing to invest to have a better platform for their own products and ideas.

As always just my 0.5c…

Stallman insists it should be called GNU Linux because the tools used to build Linux and a Linux Distro are GNU.

Stallman is <Insert insult to taste> for many reasons.
This is another example – if I put together an Ikea flatpack drawer unit with a Stanley screwdriver set it does not make the Ikea drawer unit a Stanley unit.
My wife could buy a Gucci handbag and fill it with Avon cosmetics but it doesn’t make it an Avon handbag.

My take on this question:

Did you mean we need someone to pay for all the work that has to be done regardless the outcome (for example charity donations) or did you really mean investors in the traditional sense?

Simply put, I think investing in RO should be eligible for tax relief like other charitable donations.

I think RISC OS should be perceived as edge computing platform, not cloud. We have cURL now, and HTTPLib, embedTLS and AcornSSL, box access and others

Correct. That’s why I suggest a Weboob like platform.

Agreed and not just that. IoT is an actual market, in other words Start Ups and Companies willing to invest to have a better platform for their own products and ideas.

Absolutely. There is (a lot of) money here. And it’s a better target than to believe that RISC OS could compete with Android or Linux on multicore 64bit processors.

That does not mean that we should not evolve. I dream of a Titanium like board with a connector for compute module, heavy thermal solution (2.147 GHz) and perfect (PCIe) SATA support :)

1/ MPU support or not? (to open RISC OS to Cortex-M cores)

Cortex-M cores only support the Thumb1 or Thumb2 instruction set, not the arm32 instruction set. (this would be as big a port as the 64 bit port, but with the added addition of trying to bodge a flat memory map onto risc os, no mmu)

this would be as big a port as the 64 bit port

Please tell me that we’ve learned the lesson about not writing it all in assembler…

but with the added addition of trying to bodge a flat memory map

Just treat it like a supercharged Beeb with some RISC OS like additions. ;-)

Stallman insists

Nothing unusual there, then.

because the tools used to build Linux and a Linux Distro are GNU.

What relevance is that? Is everything built using those tools supposed to have a GNU prefix attached?

The better explanation is that RMS jumped onto the Linux kernel after seeing what one guy did himself (given the fact that GNU’s own kernel never went anywhere). Thankfully Linus was savvy enough to adopt the current version of GPL but only that version, so the kernel isn’t tainted by the restrictive crap that has found its way into GPL 3.

Investing on WHAT exactly?

That’s the question. Though you’d likely be better off directing that sort of question towards people who know a little more about it than, well, me.

but hackable will unavoidably result into a non stable system for a user that is not careful,

Kind of like how shoving a metal spoon into the top of an operating toaster will likely result in fire/electrocution/death. One needs to accept a certain amount of responsibility. If they aren’t up for that, point them towards an iPad… ;-)

in other words Start Ups and Companies willing to invest to have a better platform for their own products and ideas.

What planet are you writing from? IoT is a valid thing, but it’s start ups happy to try to coin it by flogging whatever barely functional “just about does what it says on the box” crap that they can make the cheapest.

Allow me to demonstrate: https://heyrick.eu/blog/index.php?diary=20170617

That’s why I suggest a Weboob like platform.

It’s probably supposed to be said like “web oob”, but…

Allow me to demonstrate: https://heyrick.eu/blog/index.php?diary=20170617

So, that’s three years now? The reply should be with you any day now. Yes, Steven, take your medication now…

It’s probably supposed to be said like “web oob”, but…

Yeah, I even tried typing it upside down/back-to-front on a calculator but that W / M has me stumped.

What relevance is that? Is everything built using those tools supposed to have a GNU prefix attached?

not my problem lol, given that you were trying to be precise I just brought your attention on the fact that I have been extremely precise in the wording chosen on the original comment ;)

For me they can debate as much as they want on GNU yes/no, what matters to me is what it can do for me at work or in life… There are hundreds of Open Source OS and/or Kernels out there, Linux (+Kernel, for who still thinks Linux is an entire OS distro) dominates the world because:

• it actually works
• does what the majority of people needs
• it’s portable
• it’s debuggable/fixable

Please tell me that we’ve learned the lesson about not writing it all in assembler…

Amen on that!

What planet are you writing from?

Earth, you?

IoT is a valid thing, but it’s start ups happy to try to coin it by flogging whatever barely functional “just about does what it says on the box” crap that they can make the cheapest.

Isn’t RISC OS already free? Not sure what point you’re trying to make, dare to clarify please?

Also a little help for you:

Start Ups would:

• Pay N developer(s) salar(y|ies) to work on RISC OS if it does what they want/need.
• Many would also submit fixes and patches for the kernel and filesystems, network stack etc…
• By using it ti would promote it more than a campaign on FB to try to gain new desktop users when RISC OS cannot compete with other well established OS out there
• By spreading it they would create more job requests where they requite a developer to know how to code on RISC OS

The above is already a lot compared to what RISC OS has now, don’t you agree?