RISC OS is alive and well and running on the fastest hardware it's ever been on – and the kit only costs £120. But "kit" is the operative word... Acorn may sadly be no more than a brand name attached to fairly generic netbooks now, but Acorn's products are thriving. The Acorn RISC Machine, later renamed the Advanced RISC Machine …
This is all well and good, but when is someone going to get me a port of RISC OS for my iPad? After all, it's not so different to the Archimedes A420 I have kicking around somewhere...
Zarch/Lander would be fantastic with the iPad's accelerometers!
Not to mention Archimedes Elite...
But, more importantly, where's the Econet port?
"ARM is the world's most successful processor – ARM's partners had shipped over ten billion units by the start of 2008, dwarfing the x86 market."
I'm a fan of the RISC chip and had an Archimeded way back when, but when it comes to measuring markets then it's traditionally done by value and not by units shipped. It's a bit like claiming the worlwide bicycle market is bigger than that for cars. I think it very unlikely that the value of the ARM processor market is anything approaching that for Intel and AMD x86/x64 processors, simply because the unit costs of the latter are anything from fifty to several hundred times higher.
Lies, damned lies, and statistics...
@"it's traditionally done by value and not by units shipped"
No, its done by whoever wants to bias the statistics in their favour. Companies do this all the time. Evidently you are biasing the figures in the favour of Intel.
Plus phrases like "most successful" are all subjective and often incite fan base rivalry (which is a good way for news sites to stir up forum discussions. ;) ... the point being, you should be mindful of such Jedi mind tricks ;)
"Most successful processor" is subjective and debatable, but ARM based processors are the more numerous processors which is more specific. (“More” as in more than x86 based or any other processor family). As for comparative value of the ARM vs X86 markets, that's very debatable and it varies based on so many criteria. For example, historically valuable market?, growth of the high power computing marking, growth of the mobile computing markets etc.. Plus does valuable mean cost per unit value or valuable as in part of a value chain of products producing valuable markets from this underlying technology. Sales people exploit the complexity of the subject to push their over simplified biased version of it, to highlight only the aspects they wish.
Its all lies, damned lies, and statistics. ;)
Yeah, but ...
x86 hardware blows goats.
The fact that ARM has basically taken over all the other embedded device markets is something that is indeed impressive. The day someone comes out with an ARM-based "PC" and actually gets it to take a bite from the Wintel majority, it will be as revolutionary as the first PCs themselves.
Meanwhile, I'll have to keep on suffering the almost 30 year old x86 arch that has been patched over to look like it's actually good.
Oranges and Apples
The mistake you are making is, as wonderfully demonstrated by your bicycle/car example is you are not comparing like for like. Sure, there are many more ARMs shipped than x86 silicon. But then, there are probably more self tapping screws shipped than x86 silicon. And how can you speak of value when the ARM is sold as licences and then resold as fairly inexpensive units? How many Beagle boards can you buy for the price of a lump of (expensive) x86?
Does my PVR run an N270? Does my Zen run x86? No, both run ARM. Because ARM is better targetted for small embedded devices than fully blown computer systems.
What is nice to see, irrespective of market size, is that it is shipping loads. It is a veritable success. Yay.
"world's most successful processor"
Taking simply units shipped of ARM family processors have you checked statistics on the MSC 8051 family and derivatives - you might be surprised how few devices ARM processors are in.
And the bad thing is Intel are trying to supplant ARM with low power X86 chips. They think Atom2 will be suitable for smartphones.
...is apparently a glorified C64-style OS from the 80s. Serious Operating Systems need protected address space and non-cooperative multitasking.
AFAIK QNX and Symbian would fit that bill. Certainly, ARM would be a fresh start in computing and looks promising, but RiscOS will not be part of this.
<pedant mode>It's RISC OS not RiscOS.</pedant mode>
RISC OS not RiscOS
There are two completely different operating systems with similar names. RISC/os and RISC OS.
RISC OS could be thought of as a glorified version of the BBC micro operating system, in the same way windows could be thought of as a glorified version of CP/M.
Novel Netware obviously wasn't a serious operating system, by your criteria.
I would be curious to see a GUI on symbian (or even Windows mobile) that was as good as the versions of RISC OS from the early 90s, let alone its current UI.
I was going to say that serious opererating systems need good GUIs, but then I remember the state of GUIs on other systems.
Wash your mouth out!
"glorified C64-style OS"
If you had said glorified BBC Micro OS I might have let you get away with it! The C64 OS was primitive compared even to that.
But neither had the fluid, consistent, well designed GUI of RISC OS. Sadly, nor does anything else :-(
I keep waiting for someone to catch up with the RISC OS GUI, but after 20 years it doesn't look like happening. It's not going to be Windows with its history of appalling GUI design, Mac OS X did get closer, and while you can change the Linux desktop it doesn't change the apps which all behave like WIndows apps.
Touch sensitive devices or gesture recognition will probably completely change the way we interact with computers, so the WIMP desktop metaphor will be replaced by something "better".
This new ARM hardware will hopefully mean I can continue using RISC OS until that happens :-)
*Officially* it is RISC OS
However you'll find numerous instances of "RiscOS" in the source code.
@jlocke. Such is the percieved wisdom.
But I believe that ARM has always had a protected mode. It was certainly enough to have BSD 4.3 ported onto it in the A440/R140 days. It was called RISC iX.
And I can't see that it would be impossible to add interrupt driven multi-tasking to RISC OS. It's a programmed interrupt timer that will effectively call the same context switch code in the OS as is used in the co-operative system. May have to move some of the context save function into the interrupt handler, but that should be trivial.
The main drawback was that although RISC OS was intended from the start to be multi-tasking, I don't believe that it was ever meant to be a multi-user OS. This means that some of the protections you may expect in a modern OS could be missing. As far as I can remember, RISC OS programs are relocatable, and run in a shared address space, but this should not be a barrier to making it fully protected. If the current implementations have the requisite hardware, it should be perfectly possible to set up a Virtual Address space that will allow existing applications to run, just protected from each other.
The people who designed ARM initially were clever bunnies, and would not have missed so obvious a trick.
It may be interesting to look at ROOL to see how difficult it would be to retrofit.
Co-operative though it was, it was at the time leap years ahead of Windows 3.1 and then Windows 95 came along stealing the task bar concept (in fact the 'Icon Bar' pre-dated Windows task bar and Mac's dock. In general UI experience, RISC OS was speedy compared to clunky Win 95 and its pre-emptive multitasking (and whilst were on it, why is it that Windows even today can still take down the entire computer if a pre-emptive multitasking process locks up?).
C64 OS... not even remotely in the same ball park. Didn't even have a windowed interface. C64 was equivalent to the BBC Micro in terms of OS, with barely even any form of multitasking at all, and even then MS DOS was way more advanced, and RISC OS was way beyond that. RISC OS was near equivalent of Windows 95, just without the pre-emptive multitasking and the hideous registry system.
It's long in the tooth now, but for embedded systems it would be pretty good. Pre-emptive Multitasking just bloats Symbian, but otherwise the two have similar footprints. Cut out the desktop and you could build an embedded RISC OS that would be pretty rock solid.
Best thing about RISC OS was installation though.
Forget registries and installers. Just click on the app, drag to folder, job done.
Oh, and let's not forget the super speed of loading the OS thanks to putting it into ROM, and the inherent security this provides. Again, ideal for embedded systems, and if there's no registry/internal database concept (which Symbian does have), then if anything goes wrong, just power cycle and the device is back to factory default. Dodgy app? Just click, delete, gone.
Off and On again
Turning it off and on again actually worked and was a totally valid solution!
I admit that I never actually used it, but I used MacOS 9 and win95 and both suffered from lacking memory protection and noncooperative process switching. Only NT and MacOS X brought this "mainframe" feature to the commercial PCs (forget OS/2).
Probably a major redesign of Risc OS could do that, too, but they haven't done it yet, right ?
GUIs are certainly important, but first the tech below needs to be rock-solid. I know because I developed on MacOS 9 and it was a real pain compared to Solaris and NT.
@ Peter Gathercole
The ARM has "supervisor" (SVC) mode and user (USR) mode. User mode can't fiddle with hardware, SVC mode can. Traditionally the call "OS_EnterOS" gets you into SVC mode, though there's no reason that this could not be altered to have things a little more discriminating (like "does this app have the right to do this?").
Early RISC OS used to have fairly poor memory protection (!8=0 would overwrite the SWI vector and stuff the machine), though again there's no reason - especially on later ARMs with better MMU hardware, it can't be locked down to be read only or perhaps even access denied.
Perhaps one of the best things about RISC OS was also its biggest weakness. Like the old Beeb MOS, it positively invited you to fiddle with it. You could throw together a bit of assembler to bit-bash IIC on the parallel port, and what did you need? Some data sheets. That's all. No fancy DDK, no expensive compilers, and to boot pretty much all the important stuff in the PRMs was available as StrongHelp files (think HTML-like in the days before HTML).
There was a working implementation of preemptive multitasking of a cooperative application in a cooperative OS. It was called "Tornado" by Niall Douglas. It worked surprisingly well, actually. Perhaps this method is something that could be integrated into the ROOL version?
I'm not aware of *any* multi-user feaures in RISC OS. I think it just wasn't a design consideration.
Niall Douglas actually did Wimp2, as well, although it basically needed a whitelist of software that it would work with, and it worked far better if stuff was coded to use it.
That said, I do wonder if adopting Wimp2, and porting as much as possible to natively use Wimp2 may be the best bet - it'll allow poorly behaved apps to run (although they'll be in the CMT environment - Wimp2 runs as a process within that CMT environment, essentially, IIRC,) while making (hopefully) the majority of apps in the future able to run in a true PMT environment.
(Also, extending Wimp2 to natively handle SMP wouldn't be a bad idea - seeing as ARM chips are going multicore, SMP is going to be useful.)
But, none of this is trivial, of course.
...the best OS in the world, ever.
But still it's pretty much dead. I saw a picture on the IconBar once of an RiscOS user show... gone are the day of huge conference halls in london. It was a room about 50' sqare with about 10 desks. You could walk round it in 5 minutes... that's if you could squeeze past the overweight guts of the 10 or so men who turned up to it.
Sad... but RISC OS will never be anything anymore.
Here is an example of what I mean:
Man, it's sad, and this is the big RISC OS event of the year... it looks like something from 1982... and looking at the people attending, they have been timewraped in from then too.
...I remember driving down to London in my Dad's car to attend the Acorn User show when it was held at Alexandra Palace! Good old days...
"...and this is the big RISC OS event of the year..."
No, this http://www.riscoslondonshow.co.uk/ is the big RISC OS event of the year :-)
from the look of the website...
... I'm not that hopeful that the size of the hall or the people attending will be much different :-(
hmm.. all this talk of RISC OS makes me miss Chocks Away.
Chocks Away was awesome.
I still wake up sweating thinking about attacking one of those Gotha bombers...
Can that be ported to the iPad as well?
And the game with tanks (Conqueror, possibly?) That was fun. Ahhh, happy days.
Who says pictures don't lie?
Well, they don't - but they can be misinterpreted.
For one thing, you're looking at pictures which are mostly taken of individual stands, rather than the room(s) in general. Since that seems to have been the goal of the photographer, it seems reasonable to take pictures of the stands when there are fewer people obscuring them.
For another thing, some of the individual presentations, such as the one from RISC OS Open Ltd, were very popular and had a very noticable emptying effect on the room as the crowd all tried to cram into the room used for those. It wouldn't surprise me in the least if those pictures were taken while one such presentation was going on.
FWIW, my stand was next to the RISC OS Open Ltd one, and on more than one occasion during the day it was drowning in a crowd of visitors to their stand.
RE: timewraped in from then too
Not all of them!
"I am 9 years old and am VERY good at computing."
RE: people attending
There's been some talk of seeing whether Jason Kridner of TI can vist this show:
If this comes to anything, there may be an announcement or simply just a surprise visit!
In any case, you might as well attend yourself for old times' sake!
A handheld Achie would be great. Using new-fangled accelerometer-thingies we could have a BRILLIANT remake of David Braben's Zarch!
Paris, 'cos I'm all excited now!!
How about a port to the OpenRD boards? They have a 1.2GHz ARM cpu coupled with 512Mb of memory, and come as a ready to use system in a case... You can even fit a 2.5" SATA drive inside.
I'll tell you what's amusing...
...is that RISC OS apparenly has a thriving and popular warez scene!
Hilarious stuff for an OS that's been effectively dead for a decade or more.
Thriving? A few Acorn User cover discs? They don't even have my port of Wolf 3D up there ;)
Actually, it is in there
Check the RISC User CD...
@ Eddie Edwards
Looks like you caught somebody's attention... http://thepiratebay.org/torrent/5663951
I remember having to collect my jaw up off the flaw the first time I got to play with one of the original A440s at college back in '88, when I was able to write stuff in BBC Basic that ran faster than things I had written in assembler on a BBC Model B.
@RichyS Zarch/Lander!!! My god, yes!
@jlocke um... I have to wonder if you have the first clue of what you're talking about...
"Glorified C64-style OS from the 80s"..??!!
" ARM would be a fresh start in computing"... erm, hello? the ARM chip is at the heart of almost every major mobile computing platform... Nokia N900? Palm Pre? Blackberry Bold? HTC Hero? iPhone?? iPad??? Have you heard of them? Jesus...
"@jlocke um... I have to wonder if you have the first clue of what you're talking about..."
To be fair, jlocke only needs some clue. RISC OS did have memory protection, but for a long time the tasks (not really like proper processes in, say, a Unix sense) were protected from each other but not prevented from arbitrarily reading and writing kernel memory. Thus it was quite possible for a null pointer access to stiff the machine.
To the delight of people who thought RISC OS was the absolute best, Acorn never delivered a grown-up non-"C64-style" OS of their own for the hardware, and although this let them compete for a few years (just like the Mac with its even more perverse architectural restrictions, before the days of OS X), such skimping on proper operating system technology (along with other things) did it for the company in the end.
Perversely, some of the RISC OS technology would have been pretty good on top of a GNU/Linux system, but by the time Linux was becoming unstoppable, Acorn were being wound up.
Calm down, dear!
It was only a troll! ;)
Actually, I'm rather glad to see this as my pair of Risc PCs are certainly dog eared now - one won't stay up for more than half an hour and the other, an Acorn original from the mid-1990s, is still on RISC OS 3.7 and needs a new drive. I suspect that I may be looking at a new system here since second hand Iyonixen are like rocking horse manure!
As for being dead, I'm not so sure. I still like using Impression and still prefer Pluto to anything I can find elsewhere, but that's my preference. As long as users still get use out of their RISC OS machines, it's Not Quite Dead.
"skimping on proper operating system technology (along with other things) did it for the company in the end."
I don't think the OS was an issue there. If you compare to the competition at the time, RISC OS was not outdated by any measure. What killed Acorn was partly a heavy investment in net computers tat never really panned out in combination with over-priced hardware and lack of an up-to-date laptop (the only they ever made was the A4, which was long outdated by the time Acorn folded).
Other factors was the public impression that Acorn computers were only good for school use and the lack of games (compared to Windows PCs and the Amiga).
"I don't think the OS was an issue there. If you compare to the competition at the time, RISC OS was not outdated by any measure."
In microcomputer operating systems from 1989 and into the early 1990s, RISC OS was competitive, but its user interface advantages were eroded by actual improvements in the competition. Even though Windows 3.x sucked, those editions of Windows brought various development environment and feature enhancements that Acorn couldn't be bothered to bring to RISC OS.
Development of major applications was a lot easier on the competitors' platforms, and when Windows 95 finally came out, in the few areas where Windows 3.x had only had feature parity (competitive multitasking), Windows 95 had feature superiority. Although we can argue that the RISC OS Desktop was still nicer, there were still lots of inadequacies that probably still exist in the RISC OS variants around today.
"What killed Acorn was partly a heavy investment in net computers tat never really panned out in combination with over-priced hardware and lack of an up-to-date laptop (the only they ever made was the A4, which was long outdated by the time Acorn folded)."
Quite a few people would have bought a laptop, and Acorn's excuses never did seem that convincing. If you're a company of a certain size, getting the case design done (excuse #1) is just not a major problem.
But again, the operating system rears its ugly head. Acorn saw that the future was network computing and yet the Internet support was mediocre. TCP/IP support was an expensive add-on until people did just enough work to make a free alternative, and the fundamental deficiencies of RISC OS didn't make it a good platform for solutions that were completely oriented around networking.
In short, network computing was possibly one of the least suited areas to pursue if RISC OS was going to be the vehicle. That's why the RiscBSD people got Oracle's business in the end before Larry lost interest in the whole thing.
The value of the shares Acorn owned in ARM exceeded the value of Acorn at the time. Some people might think that the real reason Acorn was shut down was to free the ARM shares.
The root of all ...
Vague Incendiary FUD
there were still lots of inadequacies:
Any chance that you might back up this extraordinarily weird statement with anything even approaching an explanation or example?
Re: Vague Incendiary FUD
"Any chance that you might back up this extraordinarily weird statement with anything even approaching an explanation or example?"
Sure. You had a desktop environment which was built on a bunch of BBC Micro-era graphics primitives - not such a bad thing, given that they sort of did the job, doing stuff like clipping graphical output to regions of the screen, albeit slowly - but then you also had the basis of the graphical user interface being provided by a bunch of effectively kernel-level routines offering only the simplest user interface controls.
On top of this, there was no decent official high-level user interface library, meaning that a lot of people ended up using an unofficial one apparently written by an Acorn engineer in his own time, and all that did was to wrap these register-level calls with C functions. Consequently, there wasn't even a standard multi-line text editing widget in widespread use, at least during the entire RISC OS 3.x series. (Acorn offered a colour chooser in RISC OS 3.5: big whoop, there.)
The font manager produced very good visual results. However, the API specified in the reference manual was register-level and mostly sucked. (See a pattern forming yet?) As a result, applications were split into two camps: those which went the whole hog and implemented a complete text editing solution, and everything else which offered really primitive controls. And I don't know when Unicode got supported in RISC OS, but I imagine it was sometime this century, not in the 1990s.
There are people out there who have spent a lot of time rubbing up against the inadequacies of RISC OS. That they don't parade around in the RISC OS scene still pretending how wonderful it all is doesn't make their frank assessment of the platform's failures "FUD". Try a bit harder with the brush-off next time: in the catalogue of inadequacies, we're barely done with the first chapter here.
Ye Authore Himsselfe responds
I must admit, even as a regular ROUGOL attendee - 'cos they're a good bunch - I do not actually ever use RISC OS any more myself. I ought to dig my A5000 out of the attic and find out if it still works.
But plain old nostalgia aside, it did have some really good aspects to it. It gave the world the icon bar, which begat the Windows 95 taskbar - indeed, Windows 7's taskbar has gone back to something considerably closer than Acorn's 1987 original. The system-wide font anti-aliasing and full-window-drag are now ubiquitous, but RISC OS did it first. And personally I feel that the RISC OS desktop was one of the most elegant and efficient GUIs ever, even if it did (and to some extent still does) lack some amazingly basic features, such as cursor-key folder navigation.
No, I wouldn't recommend RISC OS to anyone today as a mainstream OS - but it's an interesting thing to play with. It may have been largely forgotten but it was tremendously influential. And on the gripping hand, I'd give good money for one of the tiny ARM-powered netbooks running RISC OS instead, if only as a portable writing tool.
RE: Which pretty much mandates that you have to have a working Acorn box (with USB) to hand
Fret not, this *isn't* the case!
RISC OS can read FAT formatted sticks fine. Just follow the instructions :-)
@Bod re: Mac Dock
The NeXTSTEP Dock preceded the Acorn Icon Bar. Guess what they call NeXTSTEP these days? That's right, Mac OS X...
Would this be the operating system? "Nextstep 1.0 was released on September 18, 1989"
RISC OS 2 was released in April 1988
Arthur (the previous version) was released in 1987
Wikipedia actually says...
"Nextstep 1.0 was released on September 18, 1989, after several previews starting in 1986"
Then it says "citation needed".
I went Win95 because MS swore it was pre-emptive multi-tasking
Not like that unsafe RISC/OS stuff.
I wrote *fairly* simple code. It crashed and took the PC with it.
Rather like the Transputer certain Windows messages transferred control to the core OS. They *could* be part process by an app, but *must* be passed on to Windows so it can service other app.
Swallow the message whole (or forget to code the pass on function) and bye bye PC.
Not impressed. No one running anything below Windows NT got multi-tasking the way a Unix person would recognise the term.
Thumbs up to keeping this OS alive BTW it dates from roughly 1986, Windows from 1983 and Unix from 1969 to keep things in perspective.
Windows in 1983?
Wiki's article says November '85. God, that predates OS/2! But the first worthwhile version of Windows (3.0) turned up in 1990.
RISC OS is from 1988, but before that came Arthur in June 1987. It is a bit bogus to throw in a comparison with *Unix*, so at least be honest and say X came about in 1984, while the current X11 was released in September 1987 (nerrr! after Arthur :-) ).
I wonder - if Win3 was co-operative multitasking, and Win32 (95, etc) was pre-emptive, I wonder how much was 'kludged' so a PMT OS could handle CMT applications? Let's not forget that Win32 was [supposed to be...] 32 bit, while many of the older applications were 16 bit. I can still run Win3.1 apps on XP. Useful, but also a bit disturbing. My main gripe with W95 was how often the damn thing used to bluescreen. Long gone days, it takes a *lot* to get XP to bluescreen. I think I've had my eeePC for nearly a year now, hasn't BSOD'd on me once. [this, of course, is tempting fate, isn't it?]
But, then, having said all of this, there are good reasons to want to keep RISC OS alive. It is a simple OS, simple to understand, simple to use. It is highly modular, I mean, hell, the entirety of the OS beyond the core kernel is a bunch of "modules". It is powerful and feature-laden for its size. RISC OS fits on a 4Mb ROM (later versions use a bigger flash because I think it is circa 5Mb). How big was Windows 3.11? How big was Windows 95? My Neuros OSD PVR came with a version of Debian installed in a 16Mb flash. If you update to the latest firmware, it resides (permenantly) on a CF card because it outgrew the flash space. Of course, it makes it easier to run a live filesystem than all the read-only flash stuff, but still... If I ever succeed in porting any version of RISC OS to the unit, I fully anticipate being able to softload it into the OSD's memory (32Mb, I think) and running with it there in-situ. Pare it down a little bit (as Neuros did to the Debian), I don't see why it wouldn't fit in a teeny-tiny 4Mb.
And finally (yes, I am known for long rambling posts, so there's light at the end of the tunnel now!), substantial parts of the core OS were written in assembler by the very people involved in the design of the processor it is running on. Howzat for speed freak satisfaction? RISC OS demands little but gives a lot. It may be lacking in modern codec support, but I think that is only due to it never really being used in a CPU/DSP system. There doesn't seem to be any technical reason why you couldn't design some sort of media player running on RISC OS - you'd only ever see the media player interface anyway. Perhaps why Linux is used more is because Linux is truly open in the OPEN sense of the word. RISC OS is only partially open and... is it ALL available yet, or are there still bits missing? They've only just published FPE's sources...
was fine as long as you did a clean install, (not over 3.x) and stuch to DOS and Win 32 programs .
The irony is that many schools moved from early RISC IS to Windows 3.1. The WIndows UI since windows 95 was far more like RISC OS than it was like windows 3.x, so by adopting "industry standards" they actually made it harder for the pupils who would mostly encounter Win 95 by the time they finished school.