Spending a Million on RISC OS

My New Year’s Resolution is to win big on the lottery (I have as much chance of keeping that one as I do any other I’ve ever made, although I suppose my chances are slightly reduced by not actually playing it!), and by the time I’ve done everything I want with the money, I reckon I’ll have a million left to “invest” in RISC OS.

So the big question is… Where should the million be spent? Would it be worth buying the sources to RISC OS 4/6 and giving them to ROOL? Should I pay for some enhanced development of Netsurf? Should I invest in multicore support? What about wifi support?

What’s the consensus if I actually manage to keep my resolution?

Are you looking to advance RISC OS technically, or bring it “up to scratch”?

If it was me, I would like to purchase the sources to RISC OS 4/6 and the services of Justin Fletcher (who has the most experience of the codebase (note that numerous changes have been made compared to the Castle codebase)) to merge everything together so that – going forward – there can be a uniform “this is RISC OS” instead of a simpler version on new modern hardware and a fancier version for those clinging to ancient hardware.
Given the <cough> history </cough> I wonder if a one followed by six zeroes would even be sufficient…

Assuming we could get this done for a quarter mil (let’s not p*** away money, there comes a point where doing it again would be cheaper), I think the next work should be on WiFi and multicore.

Let’s face it. Unless somebody pulls some sort of Linux compatibility layer out of a magic hat, we ain’t never gonna have a capable browser or anything that resembles video playback that does justice to the capabilities of the hardware. It would not necessarily make sense to spend a lot of money trying to compete with what is already both extremely capable and free. Want videos? Load RaspBMC… So it might be better to think about what RISC OS can offer and what makes it different. Personally, I think that with some stability enhancements (could be hard – think of trashing R14 by accident when in SVC mode) RISC OS would make a good embedded OS for when you really don’t need the great hulking monster that is Linux but you’d like something a little better than some code to run on a PIC. ARMs are fairly inexpensive, RISC OS is lightweight and fast. It’s a good match.

I think this is a great question, not just for wishful thinking, but to get a good strategic roadmap for RISC OS.
From a technical point, I am not qualified to give an answer, but I would appreciate others to add their view on this.

Steve Revill made a very succient statement at the ROOL presentation in London recently (I watched the vid on Youtube), when he said that “Software is just not worth anything”.
So pumping $$$‘s into RISC OS or indeed any OS is probably not a wise investment to make if you want to see any return on it. That’s why I think Castle should just Open up RISC OS 100% and give it the best chance to flourish. If Castle, as the current owners of RISC OS are not willing to invest in it, why should anyone else?

That’s why I think Castle should just Open up RISC OS 100% and give it the best chance to flourish.

I thought people understood that the bits that aren’t already source available to public were restricted because of a mixed copyright.

What would be handy is a concise list of the bits that are closed – something a bit easier to read than trawling the sources for closed references. Those are the bits people need to do something about.

That’s why I think Castle should just Open up RISC OS 100% and give it the best chance to flourish.

If exciting things aren’t happening now, I’m not convinced a simple “change of licence” will make any difference.

If Castle, as the current owners of RISC OS are not willing to invest in it, why should anyone else?

I suspect you very much underestimate the effort that had gone into getting RISC OS this far. Sure, Castle may not have spent much (if anything) on RISC OS last year. But making the code available in the first place wasn’t free. Making a complete 32 bit conversion wasn’t free. And giving the code to us to play with, learn from, expand, and help bring to newer hardware…was a generous thing. Once the Iyonix ceased to be, this version of RISC OS could have ceased with it (and the non-free branch has cobwebs hanging from it, following a period of subscriptions without obvious return – would that be a better model?). But it did not die, it is here with out today.
I expect Castle are playing this as a long term gamble – we get RISC OS, the system stays alive in some manner, and they will get a small fee if somebody decides to use it commercially. That’s a big IF in a world where every other embedded device runs some hacked about clone of Linux (and, for what it is worth, totally ignores the GPL – try finding the source to build the firmware in your router, for instance…).

For noncommercial use, the only restrictions are (and we went through this in the GPL thread):

1. you can’t use it commercially (need an OEM licence for that)
2. you can’t port it to non-ARM platforms (which would be non-trivial given that huge amounts are “scary assembler”!)

There is no reason other than political why the platform cannot flourish until the licence is changed. I suspect if a person’s only objection is to Castle getting a small return for commercial use, then that’s just….rude.

I looked, largely out of interest at getting a commercial OEM licence.
I was expecting something in the hundreds of pounds.
Anyone who wants to moan about the Castle Licence, go to the castle website and see what the cost of the licence actually is.
I’m not disagreeing with any of the points, I kind of agree with everything here.
I would be suprised if Castle’s return on the licencing is enough to keep the website running to be honest.

I had similar thoughts about a lottery win, my thoughts were paying 5 people full time for 10 years up front to work on RISC OS and three application developers.
A million wouldn’t be enough :-(

Anyone who wants to moan about the Castle Licence, go to the castle website and see what the cost of the licence actually is.

Looked way back, which is why the GPL thread was basically tosh in my opinion¹
Cost? cost? cost?cost?

The text doesn’t go small enough to express it.² It’s not pocket money, it’s pocket lint without the fluffy spangle.

¹ Quite apart from the global domination element

² It might but my eyesight probably wouldn’t cope with what the screen displayed

Here’s what I’d do with my million (I’d probably burn through it pretty quickly – not sure if we’d manage all of this):

• Buy RISC OS from Castle – not worth a lot in all honesty, I’d guess low five figures
• Buy RO4/6 from whoever has it now (Advantage Six?) – worth even less than RISC OS from Castle
• Make all the Castle and ROL bits BSD
• Revamp the ROOL website and server infrastructure/facilities
• Pull together a developer team to work on key headline tasks:
  • Replace Wimp with more modern one supporting pre-emptive multitasking
  • Update Kernel to support pre-emptive multitasking and multiprocessors
  • Finish all the file system development
  • Add native support for latest USB, WiFi and Bluetooth
  • Improve the debug and development tools available
  • Finish and fully integrate the package management stuff
  • Get multi-monitor support working
  • Port some key OSS applications (e.g. Libre Office, Chrome)
  • Better memory management and (dare I say it) security features
  • Lots of other tweaks, refactors and tidy-ups
• See if we can finally get the DDE made into a free download
• Get the PRMs fully updated
• Pay people to work full-time running ROOL (dealing with email, forums, admin, etc.)
• Get out looking for niche markets that RISC OS could apply to
• Drum up income from industry sponsors before we run out of cash!

I suspect the RISC OS that we’d have at the end of all that might be quite different (at its core) to the one today, with lots of things (like pre-existing applications) running in some sort of compatibility mode, but it’d be a damn sight more attractive to commercial uses and far more able to take fuller advantage of current and future hardware (e.g. use more than just one processor core).

Here’s what I’d do with my million (I’d probably burn through it pretty quickly – not sure if we’d manage all of this)…

Brilliant! I shall take my quid to the shop now, just on the off-chance!

Get multi-monitor support working

Anything that could be done here with Geminus, with slightly less than £1m?

Port some key OSS applications (e.g. Libre Office, Chrome)

Ooh, do I have to fund the porting of the really not very nice Libre Office? Would much prefer AbiWord and Gnumeric. Chrome I can go for, though…

Buy RISC OS from Castle – not worth a lot in all honesty, I’d guess low five figures
Buy RO4/6 from whoever has it now (Advantage Six?) – worth even less than RISC OS from Castle

Let’s hope for five numbers and the bonus ball then!

What would be handy is a concise list of the bits that are closed – something a bit easier to read than trawling the sources for closed references. Those are the bits people need to do something about.

Off the top of my head, I think ShareFS, MbufManager and the DDE are the only closed-source components remaining (apart from crusty old STB-era components which are either closed due to licensing issues or because they’re not important enough for ROOL to have processed them for release).

Get multi-monitor support working

Anything that could be done here with Geminus, with slightly less than £1m?

I did recently half-jokingly suggest that getting mutli-monitor support working should be one of my new year’s resolutions. Maybe it’s time I made it official?

Ooh, do I have to fund the porting of the really not very nice Libre Office

Given that there are one or two OSS programs to pick from, I’m pretty sure ROOL would involve the RO community as a whole before selecting which one(s) to attempt to port.

mutli-monitor support working should be one of my new year’s resolutions

Resolutions. I see what you did there… ;)

Off the top of my head, I think ShareFS, MbufManager and the DDE are the only closed-source components remaining (apart from crusty old STB-era components which are either closed due to licensing issues or because they’re not important enough for ROOL to have processed them for release).

I can’t find the source to Resolver, or does that one count as super-crusty¹ pre-STB-era component?

¹ Crusty to the point of fossilisation?

Well I did say “off the top of my head” :)

It does look like Resolver is one of the closed-source components.

Well I did say “off the top of my head” :)

Perhaps you have bigger things occupying your attention.

It does look like Resolver is one of the closed-source components.

Desperately needs replacing. I recall looking at what it did¹ some years ago while trying to persuade Dave/Frank to include a blacklist facility in AntiSpam. Part way through I discovered the nastyness inside Resolver.² Frank did a partial job to cover the reverse lookup AntiSpam needs for that facility.

¹ What it does is make some really horrible assumptions for things like reverse lookup.

² A short while later I shifted to spending a lot of time on work related stuff and largely left RO alone. Must try and dig out the bits and pieces to refresh my memory

Steve wrote:

Buy RISC OS from Castle – not worth a lot in all honesty, I’d guess low five figures

Buy RO4/6 from whoever has it now (Advantage Six?) – worth even less than RISC OS from Castle

So, for low five figures, you could buy and unite the codebases. That’s very nearly crowdfunding territory these days.

So, for low five figures,

We don’t know that. Prices quoted here are speculation.

you could buy and unite the codebases.

Finally.

That’s very nearly crowdfunding territory these days.

Indeed. Any brave volunteers to ask for a guideline price? If we have a target, we’ll have an idea if the plan is realistic…

That’s very nearly crowdfunding territory these days.

Indeed. Any brave volunteers to ask for a guideline price? If we have a target, we’ll have an idea if the plan is realistic…

I guess ROOL are best placed to do this, given their track record of actually getting us this far. Could be very interesting…

So, for low five figures, you could buy and unite the codebases. That’s very nearly crowdfunding territory these days.

Nowhere near enough money around.
Steve R is wrong about the ownership of 4/6 – it’s 3QD that bought the ROL remnants.
So basically you need to get Aaron to talk nicely about uniting the two branches. I think you may have better results with matter and anti-matter.

Buy RO4/6 from whoever has it now (Advantage Six?)

3QD Developments Ltd – i.e. Aaron Timbrell’s company. (Advantage Six Ltd is Stuart Tyrell Developments).

I wasn’t sure of the correct spelling of Stuart’s name, so I did a search to make sure I had it right, and this led to two quite interesting factoids:

1. The STD website is displaying a banner for Wakefield 2015.

2. We therefore now have a date for Wakefield 2015 – 25th April

(That’s the date I predicted in the 2014 show report – along with a note that it is one day before the 30th anniversary of the first piece of code being run on an ARM1 – 26th April, 1984 1985.)

Ah, that STD’s website is showing the banner isn’t that interesting after all – it’s just lifting it from the show website, so it’s always showing the latest banner. The site itself refers to the 2012 show.

Steve R is wrong about the ownership of 4/6 – it’s 3QD that bought the ROL remnants.
So basically you need to get Aaron to talk nicely about uniting the two branches.

and
bq. 3QD Developments Ltd – i.e. Aaron Timbrell’s company.

There’s an echo, echo,ech…

I also suggested experiments with matter and anti-matter might give better results.

There’s an echo, echo,ech…

I hadn’t read to the bottom of the thread when I wrote that¹; I saw after that you’d already pointed out it was 3QD.

I also suggested experiments with matter and anti-matter might give better results.

Quite!

I suspect the value Steve suggests, and the price Aaron would want (if he was even willing) might differ somewhat.

¹ I use the RSS feed, not the website; I read Steve’s comment, clicked through to the site, logged in, then headed straight to the comment form.

I suspect the value Steve suggests

Price based on talking with someone thinking business.

and the price Aaron would want (if he was even willing)

Billions, plus the ritual sacrifice of certain individuals?

So basically you need to get Aaron to talk nicely about uniting the two branches.

And don’t forget to spin it that so long as the emulator is capable (can VRPC boot RO5? anybody tried?), his product will stand to benefit from the use of the versions of RISC OS based upon such merged sources. It’s a win for everybody, right?
The alternative, he can continue selling his emulator with an increasingly ancient version of RISC OS.

Put it like this. I do not have RISC OS 4 (or 6), and as far as I can determine, it is not being developed further. It certainly doesn’t run on new-era hardware. Thus, when I develop my software, I concentrate on RISC OS 5.xx and – where there is no technical reason otherwise – I check the software works on RISC OS 3.70. If it also works on 4/6/Select/Adjust, then good. If not, too bad.

I think you may have better results with matter and anti-matter.

That’s why I asked if there were any brave volunteers. I’ve read some of the… uh… interesting… perspectives expressed in the comments on Drobe.

And don’t forget to spin it that so long as the emulator is capable (can VRPC boot RO5? anybody tried?),

I believe VRPC detects the version of OS and behaves differently for each version.

his product will stand to benefit from the use of the versions of RISC OS based upon such merged sources. It’s a win for everybody, right?

You’re talking sense, I don’t think that fits the situation.