25 years since Black Thursday

This week marks a work anniversary for me. 25 years ago this week, I started my first post-graduation job at Acorn User magazine, as ‘Expansion Manager’ and webmaster for the magazine’s website.

Editor Steve Turnbull had purchased the magazine from IDG Media earlier that year, so Tau Press was born, with the intention of publishing Acorn User. We rented some office space within Europa House near Macclesfield, which for many years had been the home of not only this magazine, but other titles such as The Micro User (later Acorn Computing), way before IDG took over Database/Europress Publications.

IDG had just published the 200th edition of the magazine, which I had curated the cover CD for during that summer. The 201st issue would have been Tau Press’ first.

However, my work anniversary is somewhat tainted by the memory that, days before I officially started, Steve received an anonymous phone call. The call came from someone within Acorn. That day was Thursday 17th September. What came to be known as Black Thursday. Steve took the call, which lasted 1, maybe 2 minutes. He calmly put the phone down, and proceeded to inform me, and our colleagues, that Acorn was shuttering the Workstations division, making at least 70 employees redundant, and cancelling the Phoebe computer.

For someone just starting their first job, this was pretty awful news to contemplate. And I was getting married in a few weeks!

I appreciate that for some, September 1998 was effectively the end of an era, and the start of a really painful period for the whole Acorn and RISC OS community. Things that happened in those first few years continue to be felt to this day.

However, I’d like to think that the platform is now in a far better place than those darker days of the late 90s and early 00s.

But more to the point, where have the last 25 years gone?!!!

for some, September 1998 was effectively the end of an era

Yes and no. There was some amount of chaos and pain (though thankfully being unplugged from the world between 2002 and 2009(ish) I only got a gist of what happened by virtue of some really gonzo messages on forums and mailing lists.

All this time later? Well, despite it all, RISC OS isn’t dead, people are still using it, and (shock!) you can set up a pretty good system for under a hundred quid if you recycle some bits from older machines.

One could almost say we’ve never had it so good, though the decade of lack of development and current development through volunteer efforts does rather show in how far behind RISC OS is compared to elsewhere (though, arguably, the rot may well have been what caused Acorn to throw in the towel).

In the end, what happened happened. And, yeah, I’ve been over here in France for coming on to 21½ years. WTaF? I’ll be a corpse before long at this rate!

It was the end of an era. Back then we were using RISC OS computers in large industrial control applications, and were keeping Windows PCs at bay via ease of use of the software and rock solid reliability compared to anything running 9x or NT. We had planned a new generation based on Phoebe, but we got the call from Dave Walker himself on that fateful day, and it was all over pretty much over night for general purpose RISC OS machines being used commercially.

Yes we all got together and rescued RISC OS from the remains of Acorn for use by enthusiasts, and small cheap Pi hardware allowed us to carry on using it natively long after Acorn era hardware stopped being used, but I would not say we’ve never had it so good, there are less and less of us every year, and software development is a trickle compared to quarter of a century ago.

However, I’d like to think that the platform is now in a far better place than those darker days of the late 90s and early 00s.

One could almost say we’ve never had it so good

Can I please have some of what you’re smoking?

quarter of a decade ago

I suspect you mean quarter of a century… 8~)

One could almost say we’ve never had it so good

I hope you mean here “comparing to the absurd disaster left by Acorn” and not like something positive…

RISC OS is in a state of extremely low maintenance, after looking at all the sources shared on the ROOL’s GitLab it’s pretty clear, if someone has some understanding of ARM assembly and modern Operating Systems concepts, there is no way this can be denied. (read full stop, aka fact, aka not for debate, sorry)

RISC OS was a nice thing done wrong for most part, and there are good reasons why at the origins it was made they way it was. The problems started to arise after RISC OS 2, when the OS needed constant and consistent improvements (which never materialized), so, for me, the beginning of the end it’s after the release of RISC OS 3.10, when (according to some of the comments here) it was still alive, but it was clearly dying inside (there you have it! I have been somewhat poetical here, instead of being the usual direct).

With the lack of 32bit on RO 3.50, the never completed full memory protection, the never happened introduction of preemptive multitasking (even in an hybrid form) and the never fully extended WIMP, network stack to include WiFi, never ported to C, never introduced elements of security in its architecture, only confirm what Peter has just asked, so, in a more polite form:

RISC OS is a relic of an history that never happened, because it couldn’t happen the way folks at Acorn managed things and took decisions. It was an illusion that lasted until competition started to take their “s..t” together, and when that happened there was no history, RO belonged to the past.

What it is now, it’s a fun toy to play with, for curiosity, intellectual challenge and nostalgia. If one wants to believe it’s more than that, feel free to delude yourself, reality will always be out of the door waiting.

Can I please have some of what you’re smoking?

I’m smoking roasted nuts. How ’bout you?

Can I please have some of what you’re smoking?

Just trying to be positive, even though it’s increasingly harder these days (just ask any young girl).

“comparing to the absurd disaster left by Acorn”

Not so absurd. Okay, it wasn’t good – but an operating system written as a quick’n’dirty port of the MOS in assembler in 1987 wasn’t exactly up to scratch a decade later.

Look at our contemporaries… Microsoft went from DOS to 16 bit Windows to 32 bit Windows to the better protected NT architecture to 64 bit Windows.
Apple has changed architecture four times (68000, PowerPC, x86, ARM) and – oh the irony – they’re making their own custom ARM silicon.
Android didn’t even exist then, neither did the Linux that underpins it.

Now, shall we look at the turnover of these companies? They have huge budgets at their disposal, armies of programmers, and can make decisions like “no more headphone socket” and others follow.

Acorn? Had what? Maybe a half dozen people tops working on their core OS? And no budget to allow for better.

I believe that the biggest positive development of RISC OS happened between 2 and 3. Following that, it seems to me that the core OS only really changed as was necessary for differences in hardware. And even then without necessarily much coherent thought ¹.

a state of extremely low maintenance

Yes. I think the number of people who want to deal with massive wodges of assembler are rapidly diminishing.

RISC OS was a nice thing done wrong for most part

Well, it wasn’t that unusual to write stuff in assembler in 1987. And, remember, RISC OS is the runner up as Acorn were putting their cards into the ARX hat.
The big problem is that having arrived at an OS, I think they were sort of stuck with keeping it around.
I recall at the time that Computer Concepts really didn’t like it, and were going to make something better. Of course, we only see little bits of that lurking within Impression (like Impulse) because, well, writing operating systems is hard.

With the lack of 32bit on RO 3.50

Indeed. It was a big missed opportunity, as the platform was vibrant enough at the time that they could have brought RISC OS into a 32 bit world.

the never completed full memory protection,

Not really possible. Too damn many things handing out pointers to blocks of memory in the RMA, so you couldn’t flag system areas like the RMA as inaccessible to user mode.
Plus damn near everything added to the system is modules stuffed into the RMA, code intermingled with data, all with kernel level privs. Want a bit of memory that won’t get swapped out on a task change? Yup, the RMA…

It shouldn’t be hard to throw together a build of RISC OS with the RMA blocked off as inaccessible to user mode code. I wonder if such a machine would make it to the desktop. ;)

Of course, these days we have extra modes (like SYS) that would be useful for stuff like modules, but then we also have three decades of legacy meaning nothing’s going to change.

the never happened introduction of preemptive multitasking (even in an hybrid form)

As I keep saying, this is liable to completely break how wimp messaging works (not to mention stuff like window redraws). It’s all based upon the idea of “switch in a task and get it to do stuff now”. Wimp2/Tornado had to do various tricks to defer some actions until a later time. So some degree of pre-emption is possible, but would it be enough to satisfy people?
Certainly, it won’t help in porting across stuff from a world where this is just expected behaviour.

network stack to include WiFi,

Uh… The oldest version of WiFi that anything actually supports is 802.11b which was introduced in 1999. Acorn went tits up in 1998…
(even back then, WiFi was such a novel concept that when Apple introduced it in 1999, they called it AirPort)

never ported to C,

I think the problem that we’re suffering now, as well as potentially back then, is that C is really just an extension to the core OS rather than an integral part of it. We can see this in the use of veneers to bash the assembler-style API into something that works with C code.

Systems such as Linux, Minix, and I assume Windows, get the C runtime up as soon as possible with a scant few lines of assembler, and from then on it’s a C world.

never introduced elements of security in its architecture,

Again, too much baggage and not enough budget or power to say “this is what we’re doing now”. I’ll refer you again to Apple that has changed processor architecture four times now.

RISC OS is a relic of an history that never happened

Whoa, that’s deep. So my entire life is a fabrication? Hmmm…

You know, if I was to imagine a world I could do a better job than the world we have. For starters, there would be actual dragons in Wales.

and when that happened there was no history, RO belonged to the past.

Ah, I see. Maybe it’s a different proposition in your language.

There’s always a history. From British colonial nonsense to Acorn’s decisions, history is how one got from hunter gatherers to sitting hunch-backed over the keyboard screaming “NO! YOU’RE WRONG!” at people one has never met. History doesn’t cease to be, unless you’re the loser in a war and the other side massages your culture out of existence.

What you meant, in English, is: and when that happened there was no future, RO belonged to the past.

it’s a fun toy to play with, for curiosity, intellectual challenge and nostalgia

Yes, yup, uhu, yeah.

If one wants to believe it’s more than that, feel free to delude yourself,

Depends on what you’re looking for. RISC OS’ simplicity means that it’s a useful way to get things done without distractions, telemetry, spyware, reporting everything back to the mothership, and constant incessant advertising for either Temu or that stupid game where you have to rescue a king…

Computer devices are cheap enough that one no longer has a computer like in the old days. I have a phone/tablet for watching Netflix and most of my internet needs (though I’m using NetSurf right now). RISC OS doesn’t have to be and do everything as there are other things that are more appropriate. But, then, that’s always been the case hasn’t it? A Renault 5 isn’t going to be the right vehicle to shift two tonnes of gravel…

reality will always be out of the door waiting.

I opened the door, looked into the abyss, and then slammed the door shut again.

Note: my most powerful desktop computer runs XP; pretty much everything I regularly use and have done for a decade runs Android; doing stuff like this is a hobby and a way to pass more hours of my life than I’d ever want to count, it’s not a job, it doesn’t pay the bills.

¹ ABC still doesn’t support the r,g,b forms of COLOUR/GCOL despite the manual implying the new forms of these commands are supported as of v4.29; and this situation has come about because various bits got bolted on to the BASIC interpreter meaning there are maybe a dozen versions of BASIC with subtly different sets of keywords/behaviour…

A Renault 5 isn’t going to be the right vehicle to shift two tonnes of gravel…

But at need it can do it, albeit taking several trips for the job…

A Renault 5 isn’t going to be the right vehicle to shift two tonnes of gravel…

Oh, I don’t know about that…

Cut the top open, and run two chains, one frontish and one rear through the passenger compartment and through the floor, underneath and back up on the other side.
Fill with gravel.
Lift with a truck mounted crane onto the back of the flatbed.
Drive to destination.

Where there’s a will (and enough dislike of a Renault 5) there’s a way :)

NB. Cheaper proto-skips exist.

Even our Honda Accord Tourer would take four trips – or more if the route’s at all rough or you think the police might be watching…

Cut the top open, and run two chains

Not quite what I had in mind. ;)

I’d say “thinking outside the box” but the director at work claims that people talking about that confuses me as I don’t even recognise the box.

How about describing it as “considering all the options”? :)

as I don’t even recognise the box.

👍 There is no box.

There is a box, but it’s inside the director’s head, nowhere else.

@ Rick,

Now, shall we look at the turnover of these companies?

No, let’s have a look at a one man project like:

https://www.redox-os.org/

Easy, I answered all your points in one single comment, that’s handy ;)

But wait, what if you comment next: “but but but back then it was the 80s”, well no, I specifically stated: all good till 3.10. Issues started from 3.50… but still let’s assume (by absurd) that 3.50 was coded in the 80s… then let’s have a look at a very small and very well designed OS in the 80s: QNX (to name one)

“But no, really RISC OS was still revolutionary! So, no it wasn’t an illusion as you say!”, OK let’s analyze the facts once again:

a) 1987 Archimedes gets released with Arthur a port of the Acorn MOS to ARM, this is btw why I’ve got so excited about the Archimedes, it was everything I could possible dream of when I was a kid, RISC CPU (so much hype on Italian magazines about the RISC architecture back then) and my beloved BBC Master Compact OS on it and running way faster than I could possibly imagine and with more colours I could count on both my hands! It was like wooooow and I still love it for this, even in this day an age, where voluteers and ROOL made the miracle of having my Master Compact running native on a Raspberry Pi and I can enjoy it at performances no one could ever possibly have imagined the BBC Micro would have reached (please focus on the BBC Micro/Master term here, this is key)

b) 1989 (focus on the year) RISC OS 2 gets released, now this is the beginning of the RISC OS with “multitasking” and an actually watchable desktop. Was this the beginning of an innovative OS or is the innovative the illusion I have mentioned?

Answer: it was an illusion. As I have mentioned already many many (way too many) times, Steve Job released NextSTEP in 1988 <—— 1988 is before 1989 (last time I checked ;) )

But even if one really wants to play the silly guy and ignore even this fact, then ok 1985 Amiga was released and it came with preemptive multi-tasking and HW acceleration, features these that were truly innovatives for the time especially in the personal computing world.

On top of all that, Paul Fellow and others clearly stated all they did was bringing the Acorn MOS to ARM, so it wasn’t meant to be innovative or anything. Indeed it was very cool and these guys did an amazing job for the time they had, so again problems started LATER (read 3.50).

Acorn? Had what? Maybe a half dozen people tops working on their core OS? And no budget to allow for better.

History Mr. Murray, history please ;)

Windows 3.1 and all the subsequent 3.x releases were built and designed by a team of 11 people including Cuttler (the manager). But, beside this, again read up QNX, or now Redox OS or even early Linux and more. And even if you want to close your hears and start singing, then just to be clear, in software engineering, small teams perform generally better than large teams, this is a fact.

So, again, was it an illusion? (again the innovative OS part) YES, and it was fueled partially by Acorn original effort on ARX and later on by following every potential opportunity (although if in a bit of a crazy way if you want my opinion).

“But RISC OS was fast!”, Yes it was and DTP and MIDI music making were unbeatable on RISC OS in terms of performance, but Windows wasn’t slow, it had a different goal in mind:

Making coding on Windows easier than everything else. Which had Microsoft making choices like having virtual memory (that is what has slowed down Windows for decades).

Why Virtual Memory was central to Windows and what does it has to do with making coding easier? Simple, making coding easier is generally achieved through a set of stategies of which an important one is the use of libraries, libraries that are not meant to solve a single problem and so they normally can add a lot of volume to an executable and even more to a process, so virtual memory allowed folks to develop full software solutions on Windows writing relatively little code using tons of libraries, this reduced what is called time to market which, combined with the lower price of PCs that happened in the 90s, did the trick of having a real revolution.

RISC OS was fast, but it also required longer to write big applications for it. This, combined with the lack of certain assumptions on the matter of operating systems (by that point), together with the high price of the HW platforms, did exactly the opposite of the Wintel effort and created an involution effect (or users migrating away from the RO platform).

The history that put the lot of us here together today, isn’t really the history of an innovative OS (that has never been), is the history and evolution of Acorn MOS that we all grew up with and love, but that has nothing to do with an (illusion) OS that some folks try to pretend can be used instead of Linux, Windows and/or macOS.

I still love RISC OS and use it a lot, because it makes me happy, but I don’t use it for anything that is important in my life.

On a side note: The “Illusion OS” could have become reality, if people would stop spending too much time imagining it and, instead, would sit, learn how to code and start helping to build it ;)

No, let’s have a look at a one man project like:
https://www.redox-os.org/

Not a great example. According to Wikipedia it currently has 79 people working on it (I can’t be bothered to go to GitHub to find out how many), and the project’s website says:
Redox OS as of this release generally does not support Wi-Fi, USB, and any other hardware not listed above.

Even RISC OS supports USB! 😋

assume (by absurd) that 3.50 was coded in the 80s

Well, technically much of it was. Or did you not notice the source files referring to Arthur?

So, no it wasn’t an illusion as you say!

It existed, therefore it wasn’t an illusion.
I apologise if that word means something entirely different in your language, but in English, it means:

• a wrong or misplaced sensory experience (like a fast moving wheel slowly rotating backwards)
• a deception

small teams perform generally better than large teams, this is a fact.

Up to a point. Do you think 11 people work on Win10 or whatever?

“But RISC OS was fast!”

Was. Was. I dunno about you, but I find file operations to be painfully slow.
As long as it takes SparkFS to index a moderately large archive, is about twice as long as it takes 7zip to actually unpack it all (2.4GHz P4 running XP).
It’s not just a SparkFS problem. I dunno if it’s FileCore or loads of service calls or what, but it’s… just not particularly fast.
For a more ARM to ARM equivalent, look how long it takes to rotate a full size image by an arbitrary angle (not a right angle as there’s a flag in the header that can say that, by using an arbitrary angle each pixel has to be calculated).

Now try that on your phone.

Why Virtual Memory was central to Windows

Because memory was still kind of expensive and bloated applications needed a lot of it?

a lot of volume to an executable

Uh, isn’t that was DLLs (OCXs, etc etc) are supposed to be for?

that some folks try to pretend can be used instead of Linux, Windows and/or macOS.

Again, depends what you want to do. I’m using RISC OS to write this. Wanna tell me I’m wrong or imagining it?

for anything that is important in my life.

The only things that are important are going to work to make money to pay taxes and buy tea bags. Everything else is a choice or the consequence of a choice.

Not a great example. According to Wikipedia it currently has 79 people working on it (I can’t be bothered to go to GitHub to find out how many),

That is easy:

https://gitlab.redox-os.org/redox-os/redox/-/commits/master

The main author is still the main contributor, Wikipedia lists everyone who did at least one contribution, so we shall list ALL the people who worked on RO at Acorn, RO Ltd, Castle, ROOL and now ROD (even a single contribution) to compare ;)

Redox OS as of this release generally does not support Wi-Fi, USB, and any other hardware not listed above.

That is why I used it as a comparison with RO, similar lack of HW support except Redox started in 20 April 2015 <—- another element to focus on vs RISC OS back in the 80s, this was the message and in topic with the original post part “where 25 years are gone?”
Answer: they are gone on people talking about RISC OS, but not improving it enough to become the “illusion OS” that has been created by comments over 25 and more years of forum/chats/mailing lists.

Now, if you take all the text (I call it from “the cult”, which I consider to be a set of “overly enthusiasts” that tend to describe RISC OS above and beyond what is and was meant be by the original authors and never brought to be by Acorn and who followed, either for good or bad reasons) and combine it together, you will notice these people have produced (together) far more text than it’s needed to transform the actual RISC OS code base into the “dream OS” they talk about.

Hence the solution: Stop imagining it and start writing it. I know, I sound a bit like Nauls of the old days, but in this case it’s true, people have been writing a lot about this “legendary OS” that was never coded the way they describe it.

Again, depends what you want to do. I’m using RISC OS to write this.

Indeed, if one wants to use 80s / early 90s approaches and resources, well RISC OS is ok for it, I agree XD

Wanna tell me I’m wrong or imagining it?

Tsk Tsk… try to use gmail on NetSurf then, go on :)

So, let me correct your English here (sorry): If one needs 80s / early 90s approach and only uses stuff that still behave like in the 80s and / early 90s, then yes RISC OS is ok, except for the security part.

Was. Was. I dunno about you, but I find file operations to be painfully slow.

Indeed, it’s extremely slow for what it should be and it’s ARM assembly code is no longer optimized for modern ARMs either, so very slow, which is part of my original point

Uh, isn’t that was DLLs (OCXs, etc etc) are supposed to be for?

DLLs still sucks memory both for code and data, less than static linked libraries, but still they need memory to exists and to process data that belong to a process they are being mapped in. The issue isn’t DLL vs Static, the issue is very large libraries (and that use a lot of memory) vs specialized libraires written specifically to solve an issue, title “The age of frameworks and how to grow your simple 100 lines program into a monster multi-megabytes code”.

But still it did the trick of giving people what they wanted, cheaper and more available solutions and it made quite a few software house happyer that needed to code and debug less (at least in theory) which resulted in lower costs for them (again at least in theory). The whole shenanigans got uncovered when IBM made a study on the maintenance costs related to using Windows based laptop vs Apple based ones and that was painful for the Windows market, but it’s another story for another time.

The only things that are important are going to work to make money to pay taxes and buy tea bags. Everything else is a choice or the consequence of a choice.

None of which is safe on RISC OS ;)

I think it’s a shame that we didn’t get the Arthur project started a tad earlier than August 1986. Might have had something rather better than Arthur 0.2 in your A310 on release.

There is no box

There is a box, but it’s inside the director’s head, nowhere else.

I thank you both for the confirmation that the limiting item is a figment of the limited thinking of a small set of people.
As you both probably realise, if there was a box it would have a cat in it, because boxes need a cat to sit in them.

The main author is still the main contributor,

Could say the same for the Linux kernel. ;)

so we shall list ALL the people who worked on RO at Acorn, RO Ltd, Castle, ROOL and now ROD (even a single contribution) to compare ;)

Sure. Think we’d make it to 79 people in three decades and five companies?

talking about RISC OS, but not improving it enough

There’s a vast difference between “improving” an OS written mostly in assembler, and one written in something like Rust (or, heck, even C).

For starters, the underlying API of RISC OS is tied extremely tightly to how the ARM processor actually works. The only comparison I’ve come across in my time are the INT calls on DOS where you’d poke stuff into AX, BX, etc and call an INT and hope that you got something sensible in return.
But, hey, DOS was an assembler OS too. They pretty much ditched that stuff when they went to an OS written in something a little more maintainable.

transform the actual RISC OS code base into the “dream OS” they talk about.

My “dream” version of RISC OS wouldn’t run most RISC OS software – dumping crap into the RMA and handing over a pointer to it would be nuked from orbit.

Ideally, it would not even support the existant API, it would do something similar but not in a way that requires processor-specific features. I’d like to know how Linux(/Windows/etc) actually handles OS calls, because it would be great to have RISC OS “just work” but not really caring much about what it is actually working on.

So… a ground up rewrite that isn’t remotely compatible with the current software base. Yeah, I can see people queueing up for a piece of that.

Stop imagining it and start writing it.

I don’t consider myself sufficiently competent to write an OS. Sure as hell not a filing system.

… try to use gmail on NetSurf then, go on :)

I don’t use GMail period. Don’t like its layout, and I’m not giving all my email login details to GMail to “manage”.
On devices that don’t have a stock email client, I use K9. A real email solution. Not whatever the hell the GMail app thinks it is.

“The age of frameworks and how to grow your simple 100 lines program into a monster multi-megabytes code”

Ah, yes, so you’ve met Visual Basic.

It was brilliant. It was a game changer ¹. It had many megabytes of dependencies.
Didn’t you mention something about a need for virtual memory? Ahem… ;)

because boxes need a cat to sit in them.

Just so long as the box isn’t the possession of a quantum physicist.

I think it’s a shame that we didn’t get the Arthur project started a tad earlier than August 1986. Might have had something rather better than Arthur 0.2 in your A310 on release.

Indeed, I know you guys would have built something even more exciting! We’ll never know, but I loved my Archie and I still do (and still own it btw). It was my dream machine of the time and did NOT disapoint me one bit.

The younger me would have wanted more games lol But hey, without them it pushed me to learn more about coding ’cause of the need. So that, in the end, turned out a great thing, I had an amazing career as Software Engineer and worked in many different countries and in places as legendary as Silicon Valley and close to Cupertino, I worked and developed a lot of sucessful software (and also a lot of way less sucessful! ahaha), all thanks to the work people did back then and the passion it has given me for software development, so I do not have a single complain, as always thank you very much!

Ideally, it would not even support the existant API, it would do something similar but not in a way that requires processor-specific features.

That’s what I meant with “it a nice thing done wrong” in my first comment, I am glad we agree in the end ;)

And again “done wrong” is not a judgement on the original work on Arthur and RO 2, it’s in reference to the “growing up” process that should have happened from 3.50 onward.

There’s a vast difference between “improving” an OS written mostly in assembler, and one written in something like Rust (or, heck, even C).

???? MS-DOS was written in Assembler, Windows was NOT, again moving from the old RO architecture to a new one was on Acorn, it never happened because there are no vision, no direction beyond selling the next computer. You can se this even in the architecture of the RiscPC, in a moment where it was clear that insinsting on putting everything on the same motherboard was NOT a winning strategy, what Acorn did? Well put everything (again) on the same motherboard. That is, BTW, what killed many non PCs of the time, together with the OS in ROM. These were approaches that made it very expensive for a user to upgrade, and, believe it or not, even for Acorn, Commodore and Atari themselves.

I don’t use GMail period. Don’t like its layout,

So that makes RISC OS amazing? lol just messing with you man, I understand I have touched some feelign there, typical, but these comments won’t change the reality. Take another dynamic layout website of your choice and crash and burn your eyes seeing how it gets rendered ;) – Go on (again) after you’ve finished your tea, just to avoid you to spit it out while commenting (in proper Rick style) WTH!!!

dumping crap into the RMA and handing over a pointer to it would be nuked from orbit.

And yes you can abstract the RO API on top of a better kernel and fix all of this. RMA hell can be fixed by intecepting the passage of the pointer and by implementing what Linux (as an example) already has, functions like copy_from_user and copy_to_user as well as memap for when you need performance. It’s already been invented and can work on RISC OS if done right. Why? because the OS is essecially single task, so the App will naturally freeze when the OS will need to copy. The problem is more related to the absence of a Process concept which makes things more complicated than they should be.

I gave up on that and just wrote a VM that abstracts the whole thing and software appears to be perfectly happy (at least at this stage). It allowed me also to be able to start experimenting with preemptive multi-tasking which was trivial for non WIMP apps within that level of containerization, but that will be painful when I’ll have to abstract the WIMP, given all the messy things the WIMP carries on. However internal priority queues for the events and then a good dose of patience to handle the redraw requirements and a typical long session of swearing (lol sorry, but yeah it does happen) and many coffee will help me to get through that long storm. Stay tuned for updates if interested… oh an it does already have a good security model implemented and allows even user app data encryption in RAM with the byte code havign no idea about the encryption/decryption keys. So, thing scan be done on RO, but again it needs people to spend less time imagining a “dream OS” that it is not and actually sitting down and dealing with the “devil” to tame it into something that, at least, resembles a “good approach”.

P.S. comments here are always in reference of 25 years of time, not few months, just to be clear. It’s a unit of time large enough to rewrite the OS from scratch OR to add an abstraction layer in between to start separating Apps from the actual OS implementation, to then have ground to re-implement the OS itself. In other words, “it’s a doable in 25 years for whatever team size and with whatever amount of available time”. Also, please note, I am not pointing fingers at people, if they can’t do it, they can’t do it, it’s not a reason to attack people.

the “growing up” process that should have happened from 3.50 onward.

Problem was, no staff no money.

Even ROLtd, who did a lot of work with the code, only really nibbled at the edges in terms of stuff written in C.

no direction beyond selling the next computer

Yeah, that’s kind of the part that puts bread on the table.

That is, BTW, what killed many non PCs of the time, together with the OS in ROM.

The OS in ROM does have some benefits, but certainly it’s a bit of a surprise in a world with CGA, VGA, Hercules, and some other oddball video cards, plus easily expandable memory, that Acorn didn’t at least bung the video chip on a slot in card.
The problem, though, might be that the VIDC looks after more than just the screen?

They were actually slightly ahead with the slot in processor cards. Typical PC motherboards aren’t really designed for processor upgrades (it can be done, but it’s tricky).

insinsting on putting everything on the same motherboard was NOT a winning strategy

That piece of 💩 memory bus speed, podules (and thus expansion I/O) still chundering along at 8MHz, and a really lame IDE bus implementation. Oh and RISC OS 3.10 with some pretty coloured chrome effect stickers added on. ;)

So that makes RISC OS amazing?

No, I’m well aware of the shortcomings of the system. RISC OS does still have its uses. Maybe not for everybody, but then there are plenty of other options to choose.

and crash and burn your eyes seeing how it gets rendered ;)

“It doesn’t” is the usual result.
Don’t have to wander far, just pop over to the source on GitLab with NetSurf and, oh, right…

RMA hell can be fixed by intecepting the passage of the pointer

Meh. How about a magical brick in a sock that will come swinging out of the screen to smack anybody that does such a thing?

All these interceptions, that’s how you end up with a twisted pile of legacy noodles all wiggling around in a crazy mess that few people dare to look at, never mind touch.

comments here are always in reference of 25 years of time

Things move slowly.

Once upon a time my hair had colour.

“it’s a doable in 25 years for whatever team size and with whatever amount of available time”

Doable, but doable correctly?

There are loads of hobby OSs that come and eventually fall by the wayside. A part of me suspects that there’s a big difference between “making an OS because that’s a cool thing to do” and actually designing it well in the beginning to avoid pain later on. Obviously you can’t forsee everything, and there is always the danger of overplanning (so your little toy OS has a feature set to rival Windows), but some basic questions thought about in the beginning would be better than just hacking out some code.
Thing is, knowing the right sorts of questions to answer.

Here’s one. Machine boots. It beeps. You’re looking at the command line.

You enter commands [A], which are typed in at the keyboard [B]. Character by character ©

A: Like RISC OS? Like Linux?

B: Serial port? SSH? Can one machine support multiple connections? Can they be different users? Do you even have a concept of users?

C: First, define “character”.

See? Plenty to think about before touching that editor window. ;)

I’d concur with Rick on his point that we are arguably in a much better place than previously. As a recent returnee, I note:

• The availability of RISC OS on fast, cheap hardware
• The comparative lack of crashing compared to the RISC OS 4 ere (on a new RiscPC, I might add)
• The lack of cost to obtain RISC OS
• The availability of some excellent software, a lot of which is now free
• Availability of newer technology like up-to-date Python
• The RISC OS users making videos and writing articles teaching people how to program on RISC OS
• The people dedicated to RISC OS having a viable future and either giving their time and expertise to achieve that end, or running businesses also to achieve that end

As per RISC OS tradition, some people derive notable satisfaction from negativity, and focus on what they see as failings in the system or availability of software. Luckily there are others who code away developing RISC OS or the programs we still use so the rest of us can enjoy using RISC OS and at least have the confidence there are people doing their level best to take RISC OS into the future.

I have an Apple Mac, and it’s great being able to just install the latest stuff and use it. But everything is bloated, overloaded with unneeded features (the latest being AI) and, my pet hate, software puts stuff all over the place. I have never liked not knowing what’s going where, and knowing that when I uninstall something it’s probably left bits of itself in various folders I don’t know about. With RISC OS I feel in control and although I need to brush up after a few years away, I know what’s going on with RISC OS as a user. When I turn a RISC OS computer on, I also don’t get loads of notifications telling me this and that program has an update to download, or that I can now pay for a new version. I can just get on with whatever I turned the computer on to do.