RISC OS 5 modernisation to cost millions - RISC OS News, Software and Information

RISC OS 5 modernisation to cost millions

Published: 22nd Jun 2004, 07:40:34GMT Source: drobe.co.uk
By Chris Williams

Page 1 of 1

Going all the way, baby

How much will it cost to modernise RISC OS 5? Give it your best guess. Well, yesterday Castle estimated that, for them, it'll probably cost between eight and ten million quid. And therefore, it's going to be a very gradual process.

Tematic logo

As part of Monday's press conference with Castle, COO Peter Wild gave a brief introduction to Castle's future plans for RISC OS 5, which partly explain why Castle are so eager to press ahead with issues like the RISCOS Ltd. dispute - Castle say they want to push the market forward but want everything "in order" before they do so.

Also, copies of the Tematic presentation, that was given at last month's RISCOS Ltd. shareholder meeting, were passed onto RISC OS media outlets to report on. As you may know, Tematic is the design arm of Castle, and has been since the two companies merged at the start of the year, although Tematic were behind the design of the Castle Iyonix.

The game plan
We know they've been saying this for years, but Castle really want to get as big a slice as possible from the huge worldwide market of ARM processors. In 2001, over 400 million ARM based processors were shipped worldwide, so if Castle can push RISC OS onto even a tiny percentage of this number, the stage will be set for bigger things. Or at least, that's been the dream.

Not every ARM processor will be suitable, as every chip considered will need to meet certain requirements for RISC OS (notably the presence of some kind of MMU amongst other things), but their sights are definitely set on modern, 32bit ARM 9, 10 and 11 processor cores. Nevertheless, and (again) we know they've been saying this for years, Castle want to really move into the embedded arena, and hopefully take on some 'virginal' markets. Think PDAs, Internet appliances, in car entertainment systems and other embedded devices that Acorn hinted at too. The concept of people using RISC OS without knowing, in their ethernet enabled fridges or organising their lives with a RISC OS powered PDA, does tickle us.

In terms of recent efforts to develop the embedded side of RISC OS 5, according to the Tematic presentation, Castle have worked on enhancing their C/C++ compiler package, implementing USB 2.0 and IPv6, and working on new video codec support (including WMP), new HALs for ARM 9 devices and also device drivers - Castle claim to have ported the RISC OS 5 kernel to a new processor and system-on-a-chip device in two weeks. In the future, Castle want to focus on even more ARM processor cores, with "emphasis on low power portable and hand held devices" and wireless support.

Right place at the right time
In order to squeeze their way into whatever market space they'll eventually target (or have found, but aren't telling us), Castle realise that RISC OS needs to be 'modernised', and that this will come at a price. So their plan is to approach things gradually and build the OS up to the point where they say they can afford to give RISC OS features that its competitors have. Peter Wild said he'd eventually like to see PMT (that old chestnut) and real time processing in the OS, which was an unexpected comment given (what we imagine to be) the sheer amount of work required to plough these features, sorry, OS architectures into RISC OS 5.

Castle also want to use the same RISC OS kernel in both their embedded systems and in desktop products, like the XScale powered Iyonix. You would be forgiven in thinking that it may be a good idea to have a particular kernel tuned for low foot print, reliable embedded products, and another kernel tuned for feature rich desktop users - but it's Castle's party and they'll build their OS the way they want to. Castle's justification is that, by focussing development on honing an embedded kernel, desktop users will benefit from using a reliable kernel that's undergone the necessary software quality assurance (SQA) tests required for an embedded kernel.

Castle CEO Jack Lillingston was keen to stress that they're not leaving the desktop market behind because they need it to provide a development platform for their clients' engineers, and they also need a desktop market to kindle future generations of programmers. Peter Wild also added that the embedded and desktop markets need not be mutually exclusive.

For instance, Castle launched in May the Merlin project, which will release in phases new features for RISC OS 5. Castle hope to review the feedback they've received from that, allocate resources to features that can be implemented and then inform users of the features they'll be adding. The Tematic presentation also mentions the "possible incorporation" of some RISC OS Select features: merging the RISC OS 4 and 5 kernels is out of the question (leaving Castle to play catch up with the kernel bugs that RISCOS Ltd. have spent the past 4 years fixing), but suggests that RISC OS Select desktop features could work with the RISC OS 5 kernel if RISCOS Ltd. co-operate - which beautifully echoes what RISCOS Ltd. has been asking for the past four months, except RISCOS Ltd. want co-operation with Castle.

Arguing that they are "committed to supporting loyal users and giving them what they deserve" with "plans to support legacy users" despite claiming that 32bit RISC OS 5 is "the only way forward", the highly detailed Castle road map (sans time scales) looks like this:

Iyonix
Iyonix successors
Other ARM based platforms
The rest is secret

Castle also want to support VirtualAcorn, "in the right commercial context".

Another step in Castle's plans is the issue of licensing RISC OS 5 to companies outside our desktop arena, and this bit is straight forward in Castle's view: they'll license a lot. By getting RISC OS 5 onto as many hardware platforms as possible, Castle hope to secure shipments with unit volumes that out number the total number of RISC OS computers sold over the past 5 years - which isn't too difficult, given the oh so super success of their desktop competitors RiscStation and MicroDigital.

And the rest is history?
It was pointed out to Castle that it appears that they are following Acorn's footsteps rather closely, perhaps too close for comfort and biting off more than they can chew by pushing RISC OS on the desktop and embedded fronts and, of course, we all know what happened to Acorn. There were tears. However, Castle quickly pointed out that Acorn was a completely different company in different circumstances: it employed "geeks" who were "managed by accountants" and marketed by "school teachers", and was a weak company sitting on a big pile of ARM shares - it didn't stand a chance in the wake of anxious suits, hence its break up.

So, in summary, are Castle pre-announcing? Are they whipping us into a furious state of zealotry the likes the platform has not seen since 1991, or is this talk of future success bringing back that bitter after taste of computers cased in yellow and promises of FPGA based graphics systems inspired by electrical storms? Hopefully Castle (and other remaining players in the market) are aware that nowadays, as far as the RISC OS platform is concerned, talk is cheap, and that it's their realised ambitions that we're most interested in.

Normally, by playing their cards close to their chest and not revealing anything until a product's finished, Castle have avoided the usual abuse levelled at other market players, who have sometimes failed to keep their hyped up, pre-announced promises. Now, having outlined their intentions, Castle have goals to publically commit themselves to, and as time passes, goals by which we can later measure them against.

Castle, though, haven't quite discarded their no-pre-announcement policy: they refuse to comment on whether or not there'll be a native, ARM based RISC OS laptop or a 'budget Iyonix', teasing us with the usual, "we'd like to engineer a product first, then tell everyone about it."

When they do, you'll be the first to know.

Links
Castle website

This article has been linked to, or is available in the following formats:

[Printable] [Digg this] [Blog search]

harmsy 22/6/04 8:14AM	I like the "geeks, accountants and schoolteachers" bit. That's so true.
zito 22/6/04 9:02AM	Cant help but think that having a central OS focus will be a good thing and that CTL are about the only company that can pull it off.
em2ac 22/6/04 9:21AM	my god, it seems that castle have been thinking about what the users want (or need if they havnt realised it yet). I think this will move me closer to a PDA, Iyonix, and VA for my laptop! Nice Job Castle! Knew they were the team to do it
fylfot 22/6/04 9:28AM	Another excellent article. But it looks like some more long waiting before I get me Ix lapdog.
vshears 22/6/04 9:38AM	This all sounds very promising. It is good to see the future direction of RISC OS some focus.
Sawadee 22/6/04 10:04AM	People point out the weak marketing of Acorn as the blame to it's company sell out, but wasn't it these "geeks, accountants and schoolteachers" that were responsible for the foundation in the making of this Acorn machine? If not who was it then? :acorn: Although the Acorn RISC OS computer lacks so many features and years of development compared to the Multiple Sclerosis (MS) computers, we still have a great RISC OS system. Acorn /RISC OS are "so different" to MS, I think that Castle will eventually succeed if their plan develops a new RISC OS that's even more "so different" to MS. I use Acorn RISC OS machines and I don't have an Iyonixs (because of lack of money for now!) , but what ever RISC OS develops into (and eventually leaves behind), now is the warning to plan, save and change for the future pathway. Like many of you, I really don't know the answers but at the moment I'm beginning to think maybe we better have more trust and faith in Castle. Why trust Castle? Well they could shut the door on us completely and walk away from all of this, they do say that they need a desktop for their client's engineers and future programmers (that doesn't sound like us at the forefront!) :blush: Cheers, Steve.
Fuzzy 22/6/04 10:07AM	Castle obviously seem driven to move RISC OS into the imbedded market, an area that is ideally suited to our nice and trim OS. I particulary like the line "and they also need a desktop market to kindle future generations of programmers", yah! Maybe 10 years from now when RISC OS is featured in every handheld, fridge and toaster we can all look back and be glad we stuck with our RISC OS machines even after Acorn expired. Of course for this dream to succeed Castle, RISCOS Ltd, etc need to sort their differences out right now. Plus an affordable Iyonix for us people with tighter budgets, after all the machine is a few years old now. I shall cut and paste once more: "and they also need a desktop market to kindle future generations of programmers"
nijinsky (-0.1) 22/6/04 11:10AM	why do people think pda or washing machines when someone mentiones embedded???? The OS is probably not ideal for a pda now palm is multitasking. Also the PDA sales have dropped and more focus is being put on smartphones (wher palm has the ideal OS for that, and the BEOS programmers). THings that it would eb good for are modular systems like protein purification units, sequencers, unltrsond machines and othe medical equipment. Cheers Bob
em2ac (-0.1) 22/6/04 11:23AM	ah but not forget most mobiles, now have ARMs in them! thats where we need to be (and the java might filter back from java for mobiles!
mrchocky 22/6/04 11:26AM	In any case, the figures quoted sound extravagant to say the least. Even given the enormous difficulties in implementing PMT, and perhaps resolving some other very hard OS issues. Some back of an envelope calcuations suggest to me that a mere cool million should get you pretty much everything that you want. I guess Castle and I are working from different assumptions about OS development, however.
rod 22/6/04 11:55AM	So where does this leave the likes of Microdigital and RiscStation? What have these companies got to say about all this, given that they rely on RISCOS Ltd's RISC OS 4 product? And where does this leave support for customers who have Omegas, Micos and R7500's? The silence from certain corners of the RISC OS market is worrying...
guestx 22/6/04 12:11PM	"We've learned from Acorn," claim Castle, whilst embarking on the same old long march to nowhere. Eight to ten million quid plus a fair amount of time spent fixing up RISC OS means that by the time Acorn, erm, Castle finally reach their destination (if they ever do), they might as well ready the big sign with large letters formally inviting a thorough slapping from embedded Linux and its established competitors (or rather those still in business by the time all this comes about). But before exploring the fantasy, Castle and RISCOS Ltd. have to get out of their mutual death grip. Even having Nokia continuously sniffing around Symbian is less intimidating for the average embedded systems developer looking for a platform.
AMS 22/6/04 12:17PM	I suspect that some of the cost may be licensing (probably things like Flash or WMP - which does* get a mention). Otherwise Peter's Million would sound realistic. In reply to guestx:* Does sound that they're serious enough to me, remember inspite of Linux relative success in the embedded space they aren't the only option on handheld devices (WinCE (god) also figures). I guess ROS can beat WinCE and ROS was designed for the ARM whereas Linux wasn't
guestx 22/6/04 12:41PM	*In reply to AMS:* "I guess ROS can beat WinCE and ROS was designed for the ARM whereas Linux wasn't" Being designed for a particular CPU architecture ceased to be seen as a virtue for operating systems in the early 1970s. One can argue the usual about the compactness and efficiency of RISC OS, but a number of factors are at work here. Like Linux, RISC OS has grown in size somewhat, and if RISC OS had the wealth of software available for it that Linux has, combined with the liberal licensing of Linux, you'd have some pretty big RISC OS distributions out there. Moreover, the resources available on embedded devices is increasing all the time - expect the embedded variant of Mozilla to debut on mobile 'phones at some point. Meanwhile, the really small embedded device scene doesn't appear to have graduated to CPUs like ARMs, although one might argue that the big bucks aren't going to be made in that scene anyway. One can't blame Castle for going for a slice of the bigger pie, but the question is whether their credentials earn them a seat at the table, especially when RISCOS Ltd. and other forces keep snatching their napkin.
ninja 22/6/04 1:02PM	To get the required PMT and RTOS capabilities will take so long that the embedded market will have moved on by then, so in my opinion that speculation is just pie in the sky. On the other hand, Castle are sensibly keeping exact details of their more immediate plans under wraps till they're ready. I'll just file this in the "Woo yeah!" folder along with all of Steve Jobs' speeches then.
jess 22/6/04 1:13PM	Why should PMT be difficult for the embedded market? The difficulty is PMT and backwards compatibility. For embedded solutions, you are in control of all software that will be used. (Assuming that the apps you want to use are easily updated)
ninja 22/6/04 1:25PM	*In reply to jess:* Is that in reply to me? PMT will not {IMO} be especially difficult to do for the embedded market. It will {IMO} be difficult to do full stop. I /think/ that PMT will need to be implemented in the kernel {certainly will if they're looking an an RTOS}, and that's a major change for any OS, let alone RISC OS.
mrchocky 22/6/04 1:35PM	*In reply to ninja:* The "embedded market" is enormous. There is plenty of potential for RISC OS to be used without PMT, i.e. before it "moves on". And even if it does "move on", precisely what direction do you think it will take - furthermore, do you not think that this is the same direction that RISC OS development would take. What's perhaps more interesting is that the areas which make Linux particularly useful in the embedded arena, are some areas RISC OS is pretty poor at, or simply don't exist. That is, extensive driver support, low-level protocol support (IP, etc), and the multitude of tools Linux has for doing, well, just about anything in an environment where you may well not have a GUI at all.
ninja 22/6/04 1:43PM	I don't predict what the embedded market (if you can treat is as one market even) will do in the next 12 months, let alone however long it'll take Castle to produce PMT. I just reckon that it'll look significantly different. You're probably right that there's enough of an embedded market for RISC OS to be sold to make a profit, especially for a small company like Castle, though. Some of the tool work and work in porting the OS to other hardware that Castle announced alongside Merlin probably does look more immediately useful, considering where the OS currently is. And I was surprised about their comment on IPv6, but maybe I've not been following developments closely enough.
egel 22/6/04 1:50PM	Is PMT not already done? http://www.nedprod.com/programs/RISC-OS/Wimp2/
ninja 22/6/04 1:58PM	True, but compatability with existing software wasn't stellar, and I wasn't personally too impressed by the improvments in responsiveness. I think I still run it when my machine boots, but I only apply it to a few known good programs.
em2ac 22/6/04 2:18PM	if you look on the wimp 2 site there is a patch for red squrillil (and presumably VA) to use win 32 calls one advantage is wheel mouse scrolling under RISC OS 3.7 [Link: www.nedprod.com]
Smiler 22/6/04 2:58PM	This looks like a great way foward for RISC OS, and so much for my previous comment on yesterday's article! One advantage RISC OS has over Linux is size - RISC OS can fit a fully working OS with GUI into a 4MB ROM - Linux will NEVER be this small with a GUI. Also, someone mentioned the embedded form of Mozilla. I'm using Firefox (a 'cut down' version of Mozilla) to write this now and it's using 21,792KB of memory - 22MB! In the ebedded arena, this much space is expensive. Sure, memory prices are falling but more PDAs and smart phones only have 32MB of FlashROM storage space, let alone RAM! I do like where this is going - Castle certainly seem to be whipping RISC OS into shape. I'm interested to see what happened when their modernisation reaches the GUI - to get ahead of the market they'll need Mac OS X type features; alpha channels, shadows, slick and fast animations. That's what will apeal to people when they're buying their new smart phone - they don't care whether it can multitask, or what goes on underneath, as long as it works and look pretty! Well, here's to the furute of RISC OS 5!
arenaman 22/6/04 3:22PM	*In reply to Smiler:* I would object to too much alteration of the RISC OS GUI, which is surely it's selling point (I mean for the desktop). Like you say, users don't care how it works under the bonnet, so one assumes the GUI is why people bought Acorns. Personally, I think a great selling point is that the GUI is very responsive and easy to use, plus it's very attractive - without having to resort to animation and drop-shadows. RISC OS looks as good as MacOS X - MacOS X is more in-your-face, RISC OS is much more subtle.
guestx 22/6/04 3:27PM	*In reply to Smiler:* "Also, someone mentioned the embedded form of Mozilla." Yes, precisely to elicit this kind of reaction. No, I will never be able to run Mozilla or Minimo on my current mobile 'phone, but future generations of 'phones will have the memory to do it. It'll be easier to add a stack of memory to the device than to get RISC OS into the functional ballpark of doing all the tasks that an established and well-supported embedded operating system can do. Meanwhile, on the "modernisation" front, I want my mobile 'phone to reliably send messages, handle voice calls, communicate with other things and generally just work. It's good to have nice visualisation of the goings-on, but as 'phones have become more powerful, they've also become more unstable, needing power cycles and other things done to reset them from dodgy states. When your 'phone reportedly needs a reboot because you've roamed between networks, who cares about alpha blended animations?
mrchocky 22/6/04 3:27PM	*In reply to Smiler:* please try and be a bit more informed before you post. FireFox is a fully featured browser - to say it is "cut down" is not accurate, and not the browser in question. That is Minimo. Secondly - no, you can't fit a Linux desktop into 4MB. Nor can you with RISC OS, if start counting applications. But these sorts of numbers aren't actually that important - Linux systems can run on 32MB PDAs, but newer PDAs are 64MB or 128MB.
4 comment(s) are below your moderation threshold. Login to view them.
Use the forum for more comments on this article