Please Stop the MadnessBy Peter Naulls. Published: 10th Nov 2003, 18:02:37 | Permalink | Printable
Oh how we laughedAs you may know, it is drobe.co.uk policy that all correspondence is for publication, unless otherwise stated. With this in mind, and after our recent articles on emulation we received the following flame that we're happy to share with our readers:
Date: Mon, 10 Nov 2003 16:45:37 +0100
From: Axel van Almsick
Subject: RiscOS for everyone
Please stop the bullshit.
After RiscOS 32 bit I told ROL in Dutch that it is now Time for RiscOS
to go on every platform.
Why HAL and Emus, when it is possible to make out RiscOS for ARM,
RiscOS for PowerChip, Intel, etc.
ARM and RiscOS is not together anymore.
It is bullshit to wait that one company is bring out a ARM-Chip
(Intel, Samsung?, etc). It will be too expensive.
But to buy RiscOS for Intel, PowerChip, etc. like Linux or BSD,
without Emu, will bring more fun and sense.
But ROL is like Acorn, far away to be clever and see and accept the
Please ask ROL when it will bring out RiscOS for everyone platform.
Make a petition.
Or better, ask Castle.
Maybe Castle have brains.
Oh, Axel, we're touched; really. We like your mail so much we're thought it would be nice to do an article based around it to clarify some of the issues about emulation, new machines and so forth. So, without further ado, let's see what Axel really thinks he's saying:
"After RiscOS 32 bit I told ROL in Dutch that it is now Time for RiscOS
to go on every platform."
A noble cause, I'm sure, but you've not bothered to enlighten us on what they said on the matter. But perhaps I can suggest why they might have said by examing what else you've got to say:
"Why HAL and Emus,"
Oh dear. HAL stands for "Hardware Abstraction Layer". The whole point of it is precisely to allow RISC OS to be moved to other machines, just what Axel seems to be advocating. That is why the HAL is so important. However, the RISC OS 5 HAL is really ARM based, it was designed to allow moving RISC OS to different ARM platforms. In principle you could have a new abstraction layer if you wanted RISC OS to run on multiple processor architectures, but it would probably need to be quite different and larger. Multi procesor OSes such as Linux and the BSDs have various layers of abstraction that perform this function.
As for the RISC OS 5 HAL itself, I have examined its sources in some detail, and although such things often have be compromises in how closely they're tied to hardware, there isn't too much I'd change if I were redoing it myself. Some parties in the past have attempted to critisize the HAL, but they've not bothered to follow it up with any kind of justification.
"when it is possible to make out RiscOS for ARM, RiscOS for PowerChip, Intel, etc."
RISC OS (which is, incidentally, the spelling) runs on Intel and PowerPC processors? Gosh, why didn't someone tell us sooner? I'm interested as to how you came to such a conclusion? Perhaps what Axel really means is that it would be straight forward to convert RISC OS to run on another processor.
Well, let's assume for a minute that you've rewritten the approximately 40% of RISC OS that is in ARM assembler (I'm sure that would be no problem for a technocrat like you now, would it Axel?), and let's assume you've fixed all the Basic and C in RISC OS that assumes it's running on an ARM processor. In short, we're talking about a comparison with the 32-bit conversion, but on a much larger scale. So, assuming you thought it was appropriate to invest massive amounts of money to make this happen (instead of for example adding PMT to RISC OS), then what?
Well, you'd be able to run your Basic programs, but you'd have to recompile all your C RISC OS applications, and rewrite from scratch all your assembler ones in a high level language - and then you'd have to make sure they worked on all the processors you were targetting. Doesn't sound like much fun, does it? Sure, this is what happens on Unix systems, but there are millions of users to deal with such things.
But there is one more option, we could do what Apple did when they converted from 68K machines to PowerPC - emulation. The ARM instructions could be emulated (JIT or otherwise) to the host instruction set and then excuted - naturally at a performance hit. But wait, wasn't emulation the basis of Axel's objection? What a conumdrum.
"ARM and RiscOS is not together anymore."
We're not entirely sure what Axel means here. It's true that it's been many many years since RISC OS drove ARM development. But it's not true that RISC OS doesn't rely on an a processor with an ARM instruction set (whether that's an emulated processor or otherwise).
"It is bullshit to wait that one company is bring out a ARM-Chip
(Intel, Samsung?, etc). It will be too expensive."
Again, we're not entirely sure what he's getting at. Sure, Samsung have dragged their heels a bit bringing out their ARM chips, but Intel haven't - just recently we covered Intel's 800Mhz 80331 XScale processor - a later version of the 80321 found in Castle's Iyonix. Perhaps Axel's referring to the cost of developing a new motherboard around such a chip - but I'd be interested to see his comparison against converting RISC OS to another processor. If his argument is about price, then surely he should be advocating, again, emulation - VirtualRPC on a PC involves no further development effort.
"But ROL is like Acorn, far away to be clever and see and accept the
If you say so, Axel. Be sure to tell us when you have some facts and we'll pass them on.
"Please ask ROL when it will bring out RiscOS for everyone platform.
Make a petition.
Or better, ask Castle.
Maybe Castle have brains."
We'll be sure to pass that on too. Honest.
At drobe.co.uk, we're well aware of the debate that emulation has caused; we've tried not to express any particular opinion on it, and made sure everyone has equal coverage. But for us to stick our heads in the sand and ignore emulation would certainly be wrong - it's now part of the RISC OS landscape, like it or not.
What we do think is great is that, even without VirtualRPC and friends, that there's more choice than ever of machines to run RISC OS on - even if it's still just a souped up, and entirely usable RiscPC.
News? Comments? Flames?
Previous: Drobe price comparison chart
Next: Castle bids farewell to RiscPC
DiscussionViewing threaded comments | View comments unthreaded, listed by date | Skip to the end
Please login before posting a comment. Use the form on the right to do so or create a free account.
Search the archives
Today's featured article
Of subnets and routers
Part two of IP networking and RISC OS
2 comments, latest by AMS on 7/9/04 8:11PM. Published: 7 Sep 2004
It's all over before it'd begun!: Omicron pull out after bankruptcy
Discuss this. Published: 1 Apr 2001
News and media:
RISCOS Ltd •
RISC OS Open •
MW Software •
Advantage Six •
CJE Micros •
Liquid Silicon •
Chris Why's Acorn/RISC OS collection •
The Register •
The Inquirer •
Apple Insider •
BBC News •
Sky News •
Google News •