what do people think about having to pay around 100ukp for RISC OS 4.5?

http://www.riscos.com/news/news_items/PR180501.htm
Cheers,

Andrew (in Cambridge still!)

So...

what do people think about having to pay around 100ukp for RISC OS 4.5?

http://www.riscos.com/news/news_items/PR180501.htm
Cheers,

Andrew (in Cambridge still!)

I would have been willing to pay the full UKP 175 for the entire package but the announcement that work on the 32 bit OS has been "put on hold" has just about told me to get out and get out now!

First off people have been prepared to pay deposits in advance for non-existant hardware like Omega. Perhaps ROL figured people might be prepared to stumpf up for RO's future development as well. After all even if Omega/Imago and something new from Castle were available NOW without an OS it wouldn't be worth a whole lot would it ?

As to 32 bit or not that is as much a hardware issue as a software one if RISC OS was 32 bit NOW there would be NO hardware able to run it to any great effect. I suspect the hardware is at least 6-12 months off (if Omega is out by Autumn 2001 I'd be surprised (I hope that's famous last words on my part!!!)).

If we want RISC OS we need to pay for it - full stop end of story. I, for one, am prepared to pay - simply because the alternatives (Windows and Linux) just don't appeal.

One thing for certain ROL would not be adopting this approach unless they had to, the way it was announced I think is where the real problem lay. The second issue was not turning up at Wakefield, this pill has to be sugar coated - simply ignoring the users and your potential customers is not good politics.

I strongly suggest ROL talk to people and do so soon, I accept that what they've done is necessary, but that having been said many don't and they need and MUST be talked to.

RISCOS Ltd. have shown a certain amount of disregard to both developers - many of whom are shareholders, and who ROL was set up to serve - and the general public in recent and not-so-recent actions, and it'd be nice to hear some positive news from that direction for once. Going into competition with software developers (Castle/R-Comp with the Ant thing), charging punters almost twice as much for the same service (shouldn't we expect a certain amount of upgrades/bug fixes to RISC OS to be included in the asking price?) and telling their hardware-making customers that they're not bothering with RISC OS 5 (but still went ahead with RON when no-one wanted it) isn't going to win them any friends, and that's what you need most of all in a market like this.

Or maybe I'm just tired and emotional after a long day, so please feel free to slag off any of the above diatribe.

Looks like lots of people are going *bye

Not I, £104 is about one pint of beer a week I can spare that to help keep my favourite OS going :-).

Looks like lots of people are going *bye

Not I, £104 is about one pint of beer a week I can spare that to help keep my favourite OS going :-).

The price is not the issue on my behalf either, development of an OS costs money and £104 is not much for an OS, how much does Windows cost? How much for a MacOS upgrade?
My problem is that no work will be done on making RISC OS 32bit in the near future, making the platform come to a complete standstill in terms of hardware. There's been no real advances with hardware in several years and when we could finally see the light at the end of the tunnel someone shut the door!

I strongly suggest ROL talk to people and do so soon, I accept that what they've done is necessary, but that having been said many don't and they need and MUST be talked to.

The announcement suggests that they aren't willing to talk to people - no phone support, no guarantee that emails will be read...

My vote is to get rid of ROL.

The Free/Open Source movements show what a fundamental mistake this is. I want to contribute to my community in the best way I see fit, i will only embrace "our" OS, never "theirs". Look how Linux is set to benefit as punters turn away from the new Microshaft licensing deals. What's happening to RiscOS? Wakefield turnout disappointing, that's what!

Maybe Pace or whoever pulls ROL's strings needs to read this too.

There is a real need for clarity here. Just what kind of a show are we all trying to put on?
The RiscOS community just won't support outdated commercial software much longer. What is there to lose by seeing if it can support serious Open software?

That's two diatribes in one thread.
Hohum, sorry folks.

On a lighter note, I don't see the death of 32-bit as fatal. However, it does restrict RiscOS to the low-performance portable/embedded market. Quick, before EPOC32 aka Symbian gets there!

The user base, however, is not a big one and they (ROL) have to make money someway. This, sadly, means we all tend to wind up paying more. It's the approach that's taken to the user group needs to be polished somewhat.

Regarding 32 bit I would suggest that ROL/Pace should release (under NDA) source for key elements of the OS and have the more experienced programmers from the user group work on it to make it 32 bit. After all if ROL don't release the whole thing, no harm done, they'll have some work done so there is LESS work to do later.

Perhaps they're worried if they announce a 32 bit RISC OS for this time next year people won't buy 4.5 when it comes out this year. It worryingly means we also will have to wait quite sometime for xScale to run on Omega or anything else.

[Edited by ams at 18:41, 22/5/2001]

I think the real issue is RISC OS Ltd (like Acorn before them) have a tendency to take their user base for granted.

The user base, however, is not a big one and they (ROL) have to make money someway. This, sadly, means we all tend to wind up paying more. It's the approach that's taken to the user group needs to be polished somewhat.

I personally wouldn't mind so much if they at least told us what we were getting for our money.

Regarding 32 bit I would suggest that ROL/Pace should release (under NDA) source for key elements of the OS and have the more experienced programmers from the user group work on it to make it 32 bit. After all if ROL don't release the whole thing, no harm done, they'll have some work done so there is LESS work to do later.

I personally think this is a far better suggestion than the cries to make it open source. Firstly RISC OS is Paces Intellectual Property. Secondly open source isn't the big miracle cure people seem to think it is.

Perhaps they're worried if they announce a 32 bit RISC OS for this time next year people won't buy 4.5 when it comes out this year. It worryingly means we also will have to wait quite sometime for xScale to run on Omega or anything else.

If the announcement hadn't been made I would have bought a RiscStation because there will apparently be some sort of upgrade path to Evolution. I would probably have brought 4.5 when it appeared because certain features (like DHCP) would still prove extremely useful.

It's a matter of confidence. Judging by the newgroups my recent comments would be branded as hysterical and I'd be blamed for aiding the fall of RISC OS. I have no object to spending money (recently I've upgraded / bought new software to aid my writeup, I've just renewed my foundation membership and I still subscribe to AU). Of course I have to be shown a product that I believe is worth the cash and one that isn't going to be a white elephant.

This is all very important to me. Think about it. I've stated on several occasions that I think the Kinetic is flogging a dead horse, I'm not willing to put a deposit down on a machine that isn't available and Imago has gone quiet. The prospect of Evolution would have enticed me to live with a less than top spec machine (although considerably faster than my current box) knowing that, in the long run, I'd get to where I want to be. Of course now this is looking some way off I'm back to square one.

i completely agree!

...eric

Regarding 32 bit I would suggest that ROL/Pace should release (under NDA) source for key elements of the OS and have the more experienced programmers from the user group work on it to make it 32 bit. After all if ROL don't release the whole thing, no harm done, they'll have some work done so there is LESS work to do later.

I personally think this is a far better suggestion than the cries to make it open source. Firstly RISC OS is Paces Intellectual Property. Secondly open source isn't the big miracle cure people seem to think it is.

The trouble with NDA is that ownership, and hence direction, is on ROL's terms. This makes it harder to keep voluntary programmers motivated. Yes, Pace have the final say, which is why they need to be brought in to the discussion. IMHO they will eventually have to go 32-bit too, and need to choose now between an open 32-bit OS and no 32-bit OS. OK maybe a kind of NDA solution might work, but it's a pretty high-risk strategy. I don't say opening up would be trouble-free (in fact i think most ROL staff would need to find a day job), but i see it as more fail-safe in the long term.

Here's a big idea. If m$ can port CE to the ARM, why can't all those swarms of x86 C codies out there port RiscOS to x86? They ported Unix...

If the announcement hadn't been made I would have bought a RiscStation because there will apparently be some sort of upgrade path to Evolution.... Of course now this is looking some way off I'm back to square one.

H'mmm... i'm getting fed up with waiting... must check the RiscStation harware spec again...

The trouble with NDA is that ownership, and hence direction, is on ROL's terms. This makes it harder to keep voluntary programmers motivated. Yes, Pace have the final say, which is why they need to be brought in to the discussion. IMHO they will eventually have to go 32-bit too, and need to choose now between an open 32-bit OS and no 32-bit OS. OK maybe a kind of NDA solution might work, but it's a pretty high-risk strategy. I don't say opening up would be trouble-free (in fact i think most ROL staff would need to find a day job), but i see it as more fail-safe in the long term.

Why do Pace have to choose between an open 32-bit OS and no 32-bit OS? You're making the assumption that Pace can't do the work themselves. This seems somewhat unlikely given that most of their engineers working on RISC OS are probably ex-Acorn engineers.

Pace have absolutely no reason to give up their IP on RISC OS. To think overwise is ridiculous.

Here's a big idea. If m$ can port CE to the ARM, why can't all those swarms of x86 C codies out there port RiscOS to x86? They ported Unix...

Because both CE and Unix are written in portable C. RISC OS is 99% pure ARM code.

H'mmm... i'm getting fed up with waiting... must check the RiscStation harware spec again...

It's not fantastic but it was incredibly responsive in the desktop for the few quick tests I tried.

Pace have absolutely no reason to give up their IP on RISC OS. To think overwise is ridiculous.

Here's a big idea. If m$ can port CE to the ARM, why can't all those swarms of x86 C codies out there port RiscOS to x86? They ported Unix...

Because both CE and Unix are written in portable C. RISC OS is 99% pure ARM code.

ROL aren't giving up - merely acknowledging that they can't do 32bit right now. They shouldn't have to - Pace should (and have). However the announcement made it sound like they weren't bothered about it.

I'll say it again - open source is not a miracle cure. Personally I think having Pace do the work is best (assuming we can get our mits on it) because they have a vested interest in keeping it up to date.

The GNU community haven't gone through any pain... All that has really happened is that they've perhaps learnt how to use revision control in the most effective manner possible.

Frankly, to me the real problem is the lousy track record in commercial management over the years. I just hate seeing ROS die because ROL still can't get a simple public announcement right.

well, if the problem really is with the announcement and not with ROL I will be a bit happier, but maybe they need to recruit Peter Mandelson? And *I*'ll say again, i'm not suggesting open source as a miracle cure, just that it is better than a death sentence. Lets hope Pace have something better up their sleeves.

I personally think (and I stress this is MHO) that open source would be the death knell of RISC OS. Without the resources of a company like Pace, or a large, willing developer community, I don't see that RISC OS would progress.

They have also learned how to manage open projects, e.g. to cope with errant five-day wonders who fire up a great idea and then wander off. Something even the current limited plans to open up the devloper base would need to do.

It's relatively easy to

a) ignore someone until they write some code
b) rollback code in CVS to remove that persons code from the source tree.

Frankly, to me the real problem is the lousy track record in commercial management over the years. I just hate seeing ROS die because ROL still can't get a simple public announcement right.

Quite. They really should've run it past a few people to gauge reaction and they certainly should've waited until after Wakefield to release it!!

How come its taken nearly 2 years to tell us it is too much effort to do?

How's this for another poll.
1) Would you buy RiscOS 4.5 if it supported 32bit hardware
2) Would you buy RiscOS 4.5 if supported all the feature in listed in the Select Scheme
3) Only if it does both 1 and 2
4) Wont buy

I am not saying that the new features are not needed, its just that the features listed will be only useful to a few of the users in the RiscOS community. I think the majority of users bought RiscOS 3.7 to support StrongArm not for new features.

How's this for another poll.

I think you're wasting your breath suggesting a new Poll here on Iconbar, no one seems to be willing to change that darn JavaScript Poll! IB staff - do you need a Poll administrator?

I think you're wasting your breath suggesting a new Poll here on Iconbar, no one seems to be willing to change that darn JavaScript Poll! IB staff - do you need a Poll administrator?

That's because no-one's made a sensible suggestion in ages! The only other poll request since the JavaScript one would have required a complete re-write of the poll code!

We are talking about maybe sometime going 32-bit and MS Windows are moving to 64-bit

http://www.microsoft.com/PressPass/press/2000/Feb00/MSWin64Pr.asp

Peter.

What is going on here.

We are talking about maybe sometime going 32-bit and MS Windows are moving to 64-bit

http://www.microsoft.com/PressPass/press/2000/Feb00/MSWin64Pr.asp

Peter.

An increase in instruction size does not necessarily mean that the performance will be much faster at all. Intel's Itanium (one of the processors that will run 64bit windows) is not even clocked as fast as the current Pentia. It's hope that its VLIW architecture will allow a sufficient degree of paralellism to allow it to make up the difference (we will see).

Windows has (in actual fact) being getting SLOWER over the years so more and more powerful processors have been required to keep it running adequately. What performance increases Intel/AMD give in one hand I am sure Microsoft will take away in the other.

I would also point out that there are (and have been) 64 bit Unix OS implementations for sometime, so why the fuss over Windows only NOW approaching a point where it can run 64bit code. No one clamoured for 64 bit windows as the 32 bit one was crap and unstable - now it will simply become crap unstable and 64 bit.

The only benefit of 64 bit is being able to process large volumes of data in memory (servers and large DBMS engines benefit from this). Those sort of users value stability and so far based on the 32 bit Windows experience that sort of stability has proven somewhat lacking.

Ok, here's one:
What RISC OS programming languages do you know?
- Assembler
- Basic
- C/C++
- 2 or more of the above
- None

I think the problem with this is that you'll get a lot of 2 or more, a lot of None's and few of the others.

If you know assembler it's pretty guaranteed that you'll also know either BASIC or C (or both). I suspect if people know C they'll know some assembler and the same will probably go for a chunk of BASIC programmers as well.

The other problem is that the poll misses languages like Ada which have been used to develop significant products (IIRC one of pieces of CD Writing software is written in Ada).

The other problem is that the poll misses languages like Ada which have been used to develop significant products (IIRC one of pieces of CD Writing software is written in Ada).

Yes. !CDBurn from Steffen Huber/Warm Silence is written in Ada95.

I personally think (and I stress this is MHO) that open source would be the death knell of RISC OS.

at least we still disagree on something, then.

sigh. hey, maybe they would be willing to sell us some

That would be the general idea.

and a right pig it is too. ptui!

Only because the GUIs are pretty naff. As a development environment it's fantastic. Then again most of what I do on it can be done on RISC OS with a linux file / CVS server.

I'm thinking more of the x86 free software community, most of whom shun VB and are more into C.

It's still a long way from writing C code to writing OS code. Besides most of RISC OS isn't "C compatible". A lot of the features (ie callbacks and the like) require the C tools to generate veneers (helper code) to enable the C code to work.

The source code would be pretty useless, too embedded in the wrong hardware. The OS would have to be reinvented from the PRM and the functionality, which information is already available.

No, because without the source code you cannot guarantee that the resulting behaviour will be the same. If you can't guarantee this this

a) it's not a port
b) you can't guarantee that applications will run

The PRMs are incredibly vague in some areas.

Apple have rewritten MacOS,

No, Apple wrote their OS on top of a Unix style kernel. I can't remember which one but MacOS X does not represent a ground up rewrite by Apple.

DR rewrote MSDOS,

MSDOS was even less sophisticated that the BBC operating system. Hardly fair to compare it against RISC OS.

iirc Psion wrote EPOC32 from scratch (and got it more stable than RO3, first time round)

EPOC had the advantage of nearly a decades worth of new experience in OO design / implementation which wasn't available when RISC OS was developed. In addition the target processor was faster so higher level languages became practical.

and Amiga/Atari did something similar,

They did? The Amiga gained a better GUI - things like pre-emptive multitasking were already there (the first home computer to get it).

MGT rewrote Sinaclair BASIC,

A programming language has to run on the services provided by the OS. Can you honestly say a language compares to an OS? I rather suspect the people who write compilers / interpreters would strongly disagree. My old university used to make coursework out of writing parsers, assemblers etc, you wouldn't ask a student to write an OS though.

and then there are all those emulators out there.

Again an emulator can call upon all the services of the OS it is running on. An emulator isn't an operating system - it doesn't control the hardware or provide an API which abstracts away form the hardware. At their most simple level an emulator just converts commands into the equivalent on the machine it is running on.
However emulators can be quite complex.

MS only took so long because of backwards compatibility and a thousand active nuisances, and Win3.11, NT and ME are all actually pretty different. You can't really compare the manhours spent on 3.11/95/98/NT/2000/CE/ME bloatware, web browser integration and active paperclips with one tightly-coded OS.

Win 3.11 - effectively a GUI on DOS

Win95 - implementation of 32bit API on 16bit DOS (ie slightly updated Win3.11 DOS).

Win98 - Win95 with GUI changes and some other, minor enhancements to the Win32 API.

WinME - first break with DOS, although DOS is emulated for compatibility. Most of the code base is probably still from Win98 - it represents the last of the 9x kernels.

NT - completely different product with a completely different code base aimed at a completely different market. The only similarity is the Win32 API allowing apps to run on both 9x and NT kernels (although it's not quite that straightforward)

Win2K - enhanced version of the NT kernel, major updates include allowing the use of DirectX.

WinXP - AIUI consumer version of the Win2K code optimised for home use.

Therefore MS have effectively only ever written two OS' - the original DOS and NT. Everything else has built on those bases. You're right, you can't compare NT with RISC OS - NT is in a completely different league. You can compare the 9x OS' with it though.

Bear in mind that MS have a hell of a lot more going on that just updating their OS'. For example IE would not have been written by the OS team. However because the app developers wrote it as a COM object it was easy to integrate into the OS. As a COM object it would be usable from any app in the way that a module is on RISC OS, hence why the filer and many apps can display webpages - it takes about 5 commands to do it.

ROL are in the same boat, and seem to be giving up. I am comparing closed-source and open-source development models, to which this is largely irrelevant. At least opening up allows bright sparks to come up with what they can.

ROL aren't giving up - merely acknowledging that they can't do 32bit right now. They shouldn't have to - Pace should (and have). However the announcement made it sound like they weren't bothered about it.

As for opening up to open source and see what bright sparks can come up with - when you get a few hours take a look at www.sourceforge.org and see how many inactive open source projects there are. Now convince me that a minority OS won't go the same way. - Look at how many RISC OS apps / libraries have already gone this way - a few include

Popcorn
Gamesuite
Warp
Mesa (ported ages ago, no one has offered David much help)
Anything on the coding vault (run by Nathan)

I'll say it again - open source is not a miracle cure. Personally I think having Pace do the work is best (assuming we can get our mits on it) because they have a vested interest in keeping it up to date.

The point is, we already know what we want ROS to do, and the GNU community have gone through the pain for us of finding out how to do it. we don't have to reinvent their wheels, and it would run much smoother for us.

The GNU community haven't gone through any pain. Think about it - revision control systems were available before open source was even dreamt up and all the theory for an OS like linux has been known for decades. All that has really happened is that they've perhaps learnt how to use revision control in the most effective manner possible.

...but I'm willing to bet that Pace have a hell of a lot of goodies that we haven't seen.

Firstly why would the x86 developer community have any interest in RISC OS? Those who are interested in OS development already have linux.

Besides the vast majority of the x86 development community are VB/Java/Delphi developers, not OS developers.

You're also assuming that Pace would allow all those x86 developers access to the ROS source code. Hardly a realistic proposition ... Besides it'll still break all those ARM code applications.

how many OS' have been rewritten? You'll find the number incredibly small (if any). Look at how long it took MS to ditch DOS - it's only in WinME that it's finally gone (but still emulated) yet it was supposed to disappear with win95 (originall win93 IIRC) and they've got more resources at their disposal than the entire ROS market.

You cannot develop a better OS if the underlying architecture is the same as before. "Improvements" such as pre-emptive multitasking, threading, interrupted I/O, full memroy protection etc can't just be bolted onto RISC OS as they're fundamental to the architecture of the OS.

It's also easy to forget that Unix was a multi-flavoured mess, so GNU/Linux had to be reinvented as well as just rewritten. And the whole methodology of developing free software was being developed too. This all led to some nasty code forks and duplication of effort - gnome/kde, .tar.gz/.rpm/.deb and SuSE/RedHat filesystems spring to mind. ROS would be spared all this.

Not if it was open source it wouldn't. True Unix was (is) a multi-flavoured mess, it's inherent in the design. Linux goes a fair way to address this but it's not perfect. The closest thing there is to a truly ubiqutuous platform is Java, and even that isn't quite "write once, run everywhere".

You seem to be suggesting that ROL and Pace would have developed 32-bit independently. On the face of it this would have been absurd and it was quite right to knock the idea on the head. Such an absurd code fork that I wonder if I am missing something?

No I'm not suggesting that - merely pointing out that ROL do not have any rights to RISC OS other than what Pace give them. AIUI 2 years ago ROL bought what was effectively a CVS snapshot of the source code from Pace and both parties have effectively gone their separate ways ever since. This does not mean that from time to time they haven't collaborated but I'm willing to bet that Pace have a hell of a lot of goodies that we haven't seen.

I said it was a big idea

Understatement of the year

That's why we'd need to tap into the huge x86 developer community. A lot of work has to be done anyway, simply to convert the original code to 32-bit. In my experience, rewriting from scratch is a lot more efficient than picking over and updating old stuff, so the job doesn't scale in proportion to the changes. And let's face it, every OS was written from scratch once.

Firstly why would the x86 developer community have any interest in RISC OS? Those who are interested in OS development already have linux. Besides the vast majority of the x86 development community are VB/Java/Delphi developers, not OS developers. You're also assuming that Pace would allow all those x86 developers access to the ROS source code. Hardly a realistic proposition.

As for starting from scratch, the only reason to do so is if significant gains can be made. Merely porting ROS isn't enough to justify it. Besides it'll still break all those ARM code applications.

Finally yes every OS (like every piece of software) has been written from scratch. However how many OS' have been rewritten? You'll find the number incredibly small (if any). Look at how long it took MS to ditch DOS - it's only in WinME that it's finally gone (but still emulated) yet it was supposed to disappear with win95 (originall win93 IIRC) and they've got more resources at their disposal than the entire ROS market.

I disagree, at lest initially. Alright users wouldn't gain much at first, just save a few quid and get better support, but in the longer term we would gain the independence and freedom to develop just exactly that better OS which ROL can't resource.

You cannot develop a better OS if the underlying architecture is the same as before. "Improvements" such as pre-emptive multitasking, threading, interrupted I/O, full memroy protection etc can't just be bolted onto RISC OS as they're fundamental to the architecture of the OS. You only have to look at how many abortive attempts at multithreading are out there to see that this is the case. (Dreamscape, pThreads from Rozilla, Simtecs module, WIMP2 - all require specific compilers / libraries to work).

It's also easy to forget that Unix was a multi-flavoured mess, so GNU/Linux had to be reinvented as well as just rewritten. And the whole methodology of developing free software was being developed too. This all led to some nasty code forks and duplication of effort - gnome/kde, .tar.gz/.rpm/.deb and SuSE/RedHat filesystems spring to mind. ROS would be spared all this.

Not if it was open source it wouldn't. True Unix was (is) a multi-flavoured mess, it's inherent in the design. Linux goes a fair way to address this but it's not perfect. The closest thing there is to a truly ubiqutuous platform is Java, and even that isn't quite "write once, run everywhere".

That's because no-one's made a sensible suggestion in ages! The only other poll request since the JavaScript one would have required a complete re-write of the poll code!

Ok, the poll system is probably very simple - only one option can be selected and no other user input. What are the maximum no. of options for the poll? If we know this, maybe we can suggest more sensible polls