markee174 (+3.5)
18/10/07 7:21PM |
"He also claimed that there would never be another native RISC OS hardware platform, "
But we could well see RISCOS on ARM hardware given the number of devices with ARM chips in them. |
rmac (+2.1)
19/10/07 4:54AM |
"He also claimed that there would never be another native RISC OS hardware platform ... "
Without wishing to seem unnecessarily pessimistic this would seem to explain the lack of Select for the Iyonix. The ROL's business plan appears to be focussed on the past - ie the large pool of RiscPC and A?000 machines out there.
In comparison the number of Iyonix is too small for them and with a belief that no new machines will eventuate they are looking backwards rather than forward.
I hope that I'm wrong! |
stevek (+2.0)
19/10/07 7:16AM |
I disagree with the statement that ROL are looking backwards but I do wonder what the strategy is (or if they actually have one).
Once upon a time ROL looked at RISC OS on Netbook, a project that never eventuated for many reasons. Since then a huge amount of work has been done to make RISC OS more portable buy abstracting hardware dependences. They also have a much better understanding of what is required now than they did back in the RON days.
As I see it the only real options for ROL are:
- continue to sell to the shrinking RiscPC / A7000 user base
- produce versions for the A9home (easier) and then Iyonix (harder)
- hope people continue to buy emulators
- target ARM hardware from other manufactures
The major issue with the last option is that they need to pick a device that will stay in production long enough for them to port RISC OS to the hardware and will appeal to people who are likely to buy one so they can run RISC OS on it. Obviously this costs money and is probably quite risky.
It looks like ROL have put many of the things in place that are required to run RISC OS on other hardware but resourcing, slow development, finding appropriate hardware and alternative operating systems are obstacles that ROL must deal with.
|
jc (+1.0)
19/10/07 12:00PM |
ROL have provided a means to a future that gives space to partners (at present known or unknown). I don't know who might take up those options but they are the only ones available for our area of interest. It was necessary for ROL to be inclusive (old hardware & virtual) and a pity Castle/Iyonix haven't made it easy for their product to be included too. I'd be surprised if the comment from Paul wasn't taken out of context though. |
Leo (+1.0)
19/10/07 12:34PM |
In reply to stevek:
"I disagree with the statement that ROL are looking backwards but I do wonder what the strategy is (or if they actually have one). "
Someone asked what the 'wishlist' for Select was for the next year. Paul stated that everyone at ROL had their own ideas, but he wanted to concentrate on fixing bugs that have been present in RISC OS for ages (You know, the ones nobody notices).
In follow up to the questions about possible new ARM hardware, Paul said that he considered an emulator running on a minimal Linux distribution was his ideal solution. Where Linux can handle dealing with all the various devices (graphics, audio, network etc.) and the emulator would then just access them via common Linux APIs. |
jc (+1.0)
19/10/07 12:44PM |
In reply to leo:
Paul is obviously considering the success of the VA on Windows when he talks about his ideal solution being on a basic Linux foundation - and there is a lot of sense in that. My very strong preference is for RISC OS on ARM - though a solution that included Linux somewhere along the line might (at some point in the future) solve a number of problems. Developing an OS for all hardware developers is critical though. |
cables (+1.0)
19/10/07 12:49PM |
Beta3 of VRPC for the Mac? Damn this post strike! I can't go to the show tomorrow, so I'll just have to wait for the post strike to end before I can have my copy.
A new release of Techwriter is always welcome as well. |
CKH2 (+1.0)
19/10/07 3:05PM |
Will the guys from Advantage6 be there to talk about the A9home? And about when flash 3 will arrive? Or should ROL have something to say on this? Is there a theatre slot for them? |
Leo (+1.0)
19/10/07 3:09PM |
In reply to jc:
"Paul is obviously considering the success of the VA on Windows when he talks about his ideal solution being on a basic Linux foundation - and there is a lot of sense in that."
I'm sure he is, and of course that's probably his personal opinion and not necessarily ROL's official position.
The bonus of someone producing an Emulator for Linux is that it opens up another market for ROL to sell RO Select to, without them having to do any work...
" Developing an OS for all hardware developers is critical though."
It may be, but its not what ROL is doing (Paul stated that its just not cost effective to produce a version of Select for the Iyonix). Maybe ROOL will have better success at fulfilling this critical task?
|
diomus (+1.0)
19/10/07 3:13PM |
In reply to CKH2:
Going by the show website and what I've been told, Ad6 will not be present. I assume CJE will be the people to speak to about Flash3; Chris Evans mentioned its progress on the A9home list (quoted on drobe). |
CKH2
19/10/07 3:27PM |
In reply to diomus:
Shame Adv6 won't be there and their web site (experienced user support) is just never updated. Good news is that the guarantee period for my A9home hasn't started yet and doesn't start until it's upgraded to the retail spec. - overall therefore quite a cheap (i.e. zero cost) form of extended guarantee...
|
dgs
19/10/07 11:18PM |
In reply to jc:
"I'd be surprised if the comment from Paul wasn't taken out of context though."
Almost impossible - Paul talked about it at length, clarified it in response to questions, and re-iterated (part of) it after the meeting ended.
(I'm not the eyewitness quoted in the article, though, I didn't have time to email anything. Someone else was taking notes for Archive).
dgs |
hzn
20/10/07 8:22AM |
As for "Paul suggested there won't be a Select 4i3 out for the South East show, but there might be one for the Christmas shows" ... well well, I didn't expect any new Select anyhow since I assume(d) that ROL is very, very busy tackling to get RO6 done for A9home - I do remember that Paul told me at the RISC OS Roadshow in the Netherlands (oops, that was well over a year ago) that the primary goal was to get RO for the A9home done.
As for Pauls "He also claimed that there would never be another native RISC OS hardware platform...": Well, I dare say that I'd be indeed very surprised if any other company starts with native hardware based on ROL's RO6 considering the track record for A9home - you know, that really nice, small, silent RISC OS box for which ROL claimed to be working with high priority to get the new Select done for... On the other hand I wouldn't be surprised if CTL suddenly has something like IYONIX2 on sale.
As for "VirtualAcorn on Linux": If I remember correctly RISC OS 5 licenses shall be pretty cheap now so perhaps a VA for Linux (or any other OS) using RO5 might be interesting since it could perhaps be much cheaper... On the other hand isn't VA the only future market for ROL since I have the impression that they're not into IYONIX Select and progress for A9home doesn't suggest much interst either and Paul stated that he doesn't expect new hardware anyhow 
|
jess
20/10/07 9:54AM |
I would hope that as the RISC OS kernel is broken into smaller modules, it is still following a plan to allow it to work on all sorts of hardware, rather than just relying on emulation. Hopefully to the extent that if new hardware came along, RISC OS itself would not be need modifying, just the support modules/drivers would need to be created for the hardware.
This would allow small developers to make their own RISC OS systems far more easily.
They write all the drivers in house, rather than having to pay ROL to port it.
I would hope that both versions of RISC OS remain compatible, and the kernels are developed to a similar plan. |
sa110
20/10/07 10:01AM |
In reply to jess:
An email I had from Advantage 6 seems to suggest what you wrote is true. AD6 wrote they were in the process or porting their drivers to be RO6 compatiable. Although personally I would not have thought that process would have taken them this long when you consider they already had drivers working with RO. Or is RO6 really that much different from RO4.42? |
jess
20/10/07 11:04AM |
One would hope that ultimately, once drivers were written they wouldn't be tied to one version of the OS. (And hopefully not one fork.) However the OS may not be sufficiently modularized yet. Perhaps RO 6 is improved in that respect, and that is why a rewrite is needed. |
AMS (+1.0)
20/10/07 4:32PM |
hzn wrote>"Well, I dare say that I'd be indeed very surprised if any other company starts with native hardware based on ROL's RO6 considering the track record for A9home "
Yes it's unlikely with ROL's offering but let's not forget ROOL/Castle's RO5.xx - so long as ROL don't succeed in "tying up" a sufficient number of developers to "closed" contracts then the option of porting the more open RO5.xx model to other hardware will be available.
and jc wrote>"It was necessary for ROL to be inclusive (old hardware & virtual) and a pity Castle/Iyonix haven't made it easy for their product to be included too."
The sad reality appears to be that ROL *can't* actually build an OS for any platform that *isn't* simply a RiscPC or something emulated to look like a RiscPC. It appears that even with sufficient technical information (a comprehensive published TRM, a CD full of datasheets, a swodge of RO5 source and a site full of API details) that ROL can't seem to deliver an Iyonix Select port. Even with a business partner (Ad6) that is committed to help ROL there seems to be issues getting RO finished for A9Home. As it's not just a Castle/Iyonix issue the problem must lie elsewhere (nice try JC - but no cigar  )
The path of least resistance open to ROL would be a purely emulated one (at the moment the most viable option being VA on Windows - and I don't see that changing anytime soon - I'd view a Linux port a long way off).
So if Paul is saying emulation is the way to go - yes for a purely software oriented company with little low level hardware programming capacity emulation may be the only route. So long as the emulator vendor can present a platform that looks to RO like a RiscPC then ROL will be able to use (without worrying about the low level hardware niceities
Is this a good thing for the platform?, probably not. Any loss of hardware diversity and capacity to support it is, I would argue a bad thing. But at least the ROOL option offers an alternative. |
sa110
20/10/07 10:06PM |
In reply to AMS:
ROL have 32bitted their RISC OS. They have provided both 4.40 and 4.42 for the A9home, AD6 have written their drivers for both 4.40 and 4.42 (therefore it does work on non RPC hardware).It would appear the issue with RO6 on the A9home is down to drivers. Drivers that AD6 wrote and I'm led to believe are either rewriting or patching to work with RO6. Last I heard, it was thought a release of RO6 for the A9home before the end of this year was said to be not unrelealistic. However only time will tell. |
AMS
21/10/07 11:51AM |
Drivers require the " low level hardware programming capacity" I was referring to. Yes ROL have undoubtedly done much good work on 32bitting and producing a product that is a bit more portable than what they originally received - but if the problem is producing drivers being a difficult and time consuming task then emulation may appear to be an easier route as Microsoft (or other hardware vendors) have handled all the driver stuff for you and the emulator vendor has produced a product that to all intents and purposes looks, feels and moves like a RiscPC.
I don't doubt that ROL will finish the OS for the A9 (as to when who knows) the issue is more would ROL ever take on another ARM native hardware project (in that I'd include porting Select to Iyonix) and I suspect unless someone dumped a suitcase full of cash on their desk I would think not.
It really is down to persuading people to part with money for something that doesn't already exist and then with limited resources trying to implement it. Doing that with "new" hardware requires more resource up front - it also can have unseen difficulties and can take a long time (Omega and A9 being cases in point). In comparison porting to a "known" platform (like an emulated one) is less "greedy" of these resources and makes it more feasible to satisfy delivery in a reasonable time. A9, unfortunately, was just a tad too different to be easily accomodated. You'll note that updates for RPC's and VARPC are more frequent and take less time to arrive.
As A9 becomes more of a "known" quantity to ROL it'll become like the RPC and VA and get more frequent updates I suspect. Problem is if the quote from Paul Middleton is accurate it very much sounds like they'll never put themselves through all that hassle again - and that the relative ease of supporting an emulated ARM environment will win out.
Future native hardware will, therefore, need OS support from elsewhere (i.e., independant developers updating the ROOL/Castle sources) rather than from ROL. |
sa110
21/10/07 12:13PM |
In reply to AMS:
You stated "I don't doubt that ROL will finish the OS for the A9". People keep saying the OS is not finished. However the A9home I use certainly seems to have a stable and finished OS. An OS which is later than Adjust but prior to RO6.
What people are waiting for it would appear is not ROL, but AD6. AD6 have recently made available via CJE a USB update (an update that I have had for sometime). I expect the next update will be to the sounddma module, this fixes the problem with the sound stopping. An A9home compatible version of RO6 is the next thing we A9homers are holding out for. Not having RO6 on the A9home is not in anyway stopping the A9home being fully used. |
AMS
21/10/07 12:36PM |
In reply to sa110:
That is because in the old days Acorn did all the OS - hardware drivers included. ROL don't (which is what I said earlier).
They may feel confortable about delegating someone else to do the drivers - but this takes the OS development out of their full control. When all the drivers are present and everything works then the OS is complete - not before. The fact that a part of that work has to be done by an organisation other than ROL is neither here nor there.
From ROL's perspective having an emulated platform only would eliminate all the hanging round on others to complete drivers. They could just "ship" the RO product the moment their internal and beta testing had finished. |
druck
22/10/07 8:51AM |
If the OS is structured properly, there is no reason for the core OS and drivers to have to be shipped in lock step. With a well defined interface and separation of the binary image, development and release could occur independently. I hate to say it, but that is a key advantage of the RO5 HAL model, which should have been at least replicated in the RO6 work instead of sticking with the single ROM image and module chain.
As for emulation only, that assumes that the RISC OS environment will never be able to interface to any new peripherals, reducing it to a dead OS and mere toy. If emulation has any future, RISC OS must make direct use of equipment such as scanners, webcams and other USB kit in the same way it can currently use the hosts networking, without having to drop back out to the native environment and use non RISC OS software - otherwise that's where they'll stay. |
AMS
22/10/07 11:19PM |
In reply to druck:
Agreed.
I am not actually advocating emulation as a route to preserve RISC OS - just trying to see it from ROL's perspective - where emulation (from their view point) may throw up fewer development problems than trying to support the native ARM platform. |
druck (+3.2)
23/10/07 12:09AM |
If the developer of an ARM based operating system, can't support ARM based platforms, its time to pack up and go home. |
jess
23/10/07 9:37AM |
Since I'm normally using machines via remoter access, an emulator or virtual machine on a decent host is a very suitable solution for me.
However I don't think it is partically ideal for non IT people. It is certainly unlikely to attract new users who aren't very IT literate already.
I think the best way forward with emulation would an x86 system with an arm emulator at a very low level in the system, so that in effect RISC OS would be the only system.
(Apps would be ARM only, but presumably modules could be written in x86 to take advantage of the native hardware speed) |
AMS
23/10/07 1:15PM |
In reply to druck:
Yes if ROL were the only option and if they couldn't implement ARM based versions of the OS I'd agree. Fortunately the ROOL option is an alternative. That having been said I'd be saddened to see ROL unable to continue supporting ARM.
In reply to jess:
Yes if there were absolutely no alternative - BUT if ROL preferred to use emulation so as to avoid developing drivers for low level ARM use they'd be even less inclined I'd think to try make all the x86 ones that would be required to get RISC OS running "natively" on x86.
I'd also point out the ROOL license precludes use of its sources on emulated ARM platforms (which is fair enough it's bad enough one RISC OS developer supporting emulation having both support emulation would - in effect - truly mean the end of the RISC OS as a computing platform IMHO). |
dms
23/10/07 1:57PM |
"I'd also point out the ROOL license precludes use of its sources on emulated ARM platforms (which is fair enough it's bad enough one RISC OS developer supporting emulation having both support emulation would - in effect - truly mean the end of the RISC OS as a computing platform IMHO)."
No it doesn't. The licence states "... PROVIDED THAT such work is only intended to be used in conjunction with an embodiment (whether physical or emulated) of one or more versions of the ARM processor architecture." |
AMS
23/10/07 7:24PM |
In reply to dms:
My sincerest appologies... (dagnabbit my brain is becoming more like swiss cheese everyday!).
Yes you're right and I got it wrong. I wouldn't mind but after a brief trawl I found me asking the question does that mean a JIT (Just in time recompilation of ARM code) would be allowed. If not you could emulate an ARM to run RISC OS - just not translate the code to x86 to speed things up (your emulation would be a bit tardy but would work). Don't know if that ever got an answer.
So yes it does seem (I'd agree with your reading of it) that you can emulate ARM and use ROOL source in doing it.
Appologies to all (again...) |
fantasian2
23/10/07 8:29PM |
Can anyone enlighten me for what exactly purpose has Nokia been using RISC OS? Please? |
jess
24/10/07 12:12AM |
In reply to ams:
The whole point of separating the hardware drivers from the kernel is that the OS writers don't need to write drivers, they just create a good interface. The drivers can be written by others. |
| Use the forum for more comments on this article |
The RISC OS South East show, taking place this Saturday in Guildford, Surrey, will see a number of new products unveiled - from databases to graphics. Here's a round up of what you can expect to see revealed this weekend.
