Reviving ROLF or ROX

Having recently discovered Timothy Baldwin’s astounding looking port of RISC OS to run on Linux, I’ve been wondering if it would be possible to use it to launch RISC OS applications in individual windows on the Linux desktop without running the full RISC OS desktop. Then all we’d need is an updated version of either ROLF or ROX in order to put a RISC OS-style GUI over the top of Linux and we’d have a set up that can run the full range of RISC OS software alongside modern software like Chrome, Google Apps etc. The whole shebang could be made available as a distro.

We’d be able to run RISC OS software at near native speed using one of the cores on any 32-bit capable ARM Chromebook. At the same time Linux would be making full use of the rest of the hardware. It’s pretty unlikely that any single core consumer devices will be released in the future, so the idea of porting RO to run natively on any modern hardware feels a bit wasteful.

I’m far from the most technically astute person here, but the above looks to me to be the simplest, most maintainable way to preserve everything that people like about RISC OS (UI, software, BBC Basic etc.) while giving access to the modern niceties that computer users now take for granted. It’s also pretty close to what Apple did when transitioning from OS 9 to OS X.

Would anyone else be interested in something like this?

keep the RISC OS more pure.

Purity is a pointless illusion.

Would you like “impure” ports of stuff from elsewhere that works and helps to plug some of the obvious gaps from two decades of little development? Or would you rather retain this noble sense of purity like an electronic chastity ring and have…next to nothing?

We need modern useful software that does the sorts of things that are common place elsewhere. It’s utterly inadequate to say that HTML is an overcomplicated mess and modern video formats use too much processor power ¹, because these things are just normal on pretty much every system that matters. Mac? Windows? Android? People just expect it to work, so we have a lot of catching up to do and can’t afford to be picky and consider things as nebulous as “purity”.

The language used to write something is not relevant.
The library linkage method is not relevant.
That we might burn a couple of hundred extra cycles in doing something ² is not relevant.
That we can’t Zoom directly from RISC OS is relevant.

¹ Completely bogus argument. Any sane system will toss that to the GPU or some sort of DSP co-pro that has stuff intended to make video decoding quick and efficient and take very little power. It probably ought to say something that a Pi1 running OSMC wasn’t fazed by 1080p H.264 video. The same device running MPlayer under RISC OS could just about manage a 320×240 DivX single tasking. You just don’t do video decoding in software unless you have no other choice, because it’s slow and extremely inefficient.

² Before you complain about what ELF’s linking does, I might suggest you rummage around the CLib sources to see how swix() is implemented, and that happens for every single SWI call made. I avoid swix() like the plague Covid because it’s enough of a hit taking the SWI trap, I can’t justify an additional hundred instructions just to work out how…
Perhaps the best approach to these sorts of things is the Wizard of Oz approach. Appreciate something for what it is and does, but don’t try looking behind the curtain to see how it does it.

fixed for you

Not really. Rick mentioned Zoom because it’s one of the tidier examples of a conferencing setup.

Cisco bought Webex, it does OK but requires more resource and ports open.
GoTo Hell?¹ have yourself booked into a mental facility.
MS Teams? Was it Mr Huber who described it as bloatware? He’s right.

¹ Well, it’s goto Meeting and goto-this and goto-that only two things are consistent: The Goto prefix and the level of adverts embedded.
Go to their test instance and test, then block all adverts and trackers and test again. It will probably timeout because of the delays caused by it waiting for the blocked ads to render.

Also would make it possible to run RISC OS simi-natively on hardware not natively supported by RISC OS proper (the best point of the RISC OS On Linux port).

This is what interests me about the proposed set up above. There are so many ARM-based computers that are being released that are powerful enough to be used as someone’s sole computer. If you had a RISC OS-like distro of linux that could run RISC OS software at close to native speed then you’d open up that software to a much wider audience. There are applications on RISC OS that would add interesting options to the apps available for linux e.g in DTP. It’d also allow old hands to use RISC OS software more frequently if it would run on their main system without the need to switch desktops.

I’m not terribly attached to maintaining the purity of the OS, but it seems to me that attempting to port Zoom (and all the other applications you need to interact with the modern world) is a bit like trying to port them to Mac OS 9. There’s a reason why people don’t do that. RISC OS works really well as a hobbyist OS on a Raspberry Pi 1. It’s barely using the Pi 4’s hardware and would ridiculously under-utilise the fast ARM Chromebooks etc. slated for release next year. RISC OS on Linux seems like the obvious way to make use of all the exciting ARM-based kit coming onto the market at the moment.

To return a bit to the subject, I’ve used the ROX filer a bit on my Linux-system and I like it a lot, but it handles mime-types very poorly, so I haven’t gotten around to replacing my regular filer with it. If it could be brought up to speed to use, or at least be compatible with, modern alternatives for how such things are handled, I’d be very excited!

Why not get RISC OS itself in whole to run directly on the Linux Kernel?

Already done. That’s what RISC OS on Linux is.

… also with ROX GUI.
Debian+ROX+Linuxport on ARM.
Worum geht’s? ;-)

A linux kernel and a few tools yes. Why would you need the full distribution?

:) Sorry different world. My most powerful computer at any time in the last 30 years has always been ARM based and had RISC OS as the primary OS.

I think even for members of this forum that puts you in a pretty exceptional position! Most of us have ended up with a computer running a more up to date OS on x86 in order to interact with the rest of the world. I don’t doubt that there are more efficient approaches to video conferencing than Zoom, but if your work demands that you use it (mine does) then you have to run an OS that can handle Zoom.

The point I was trying to make about people not backporting modern apps from Mac OS X to OS 9 is that end users had a number of years running OS X with up to date software with a familiar UI while still being able to run older OS 9 software via a compatibility layer. They didn’t back port things because they could just use OS X.

To return a bit to the subject, I’ve used the ROX filer a bit on my Linux-system and I like it a lot, but it handles mime-types very poorly, so I haven’t gotten around to replacing my regular filer with it.

Very interesting! I’m going to have a play around with ROX and possibly ROLF (anyone used this?) on my Raspberry Pi. There must be few things that need to be updated – it’s been 6 or 7 years since either were changed by their authors.

Why not get RISC OS itself in whole to run directly on the Linux Kernel? That is no Init, no Unix environment, just the Linux Kernel, its HW support modules, and RISC OS in whole on that.

The issue with this is that, as I understand it, RISC OS would still be limited to using only one of the cores on a multicore system. Buying an all singing, all dancing ARM laptop just to leave most of its hardware untouched seems pretty unattractive to me. I suspect the vast majority of other people will feel similarly. Surely it’s better to have a full Linux installation that can run modern apps and utilise the hardware + an updated RISC OS-like GUI and RISC OS on Linux for backwards compatibility? Ideally, people should be able to pause whatever they’re watching on Netflix and then switch across to work on an Impression document without having to move to a different desktop or switch computers.

I am still looking forward to seeing the (partially there already) AMP multicore support by native RISC OS. Seems a better way to add multiple processor/core support to me.

Over the course of the next year huge numbers of ARM Chromebooks, Windows laptops and Macs are going to be sold. The approach in this thread is the only one that will give the majority of them the option of a RISC OS compatible system that can run all of the software they need day to day. It’s going to take ages for RISC OS proper to be made 64 bit and multicore compatible and for software to be ported across from Linux to fill current gaps. By then we’ll have new, bigger gaps in functionality and fewer users and developers to fill them.

If I was running a business focussed on RISC OS software and hardware then I’d be screaming for a ROX & ROL set up to be put together. Upgrades to RISC OS itself can carry on regardless. All I’m saying is that we shouldn’t ignore the opportunity that has opened up to bring on board lots of new users, especially amongst Chromebook owners.

I am not talking about and a few tools, those few tools are the difference between just a kernel, and running on a unix like system. Remember Linux is a kernel, not an OS, the RISC OS on Linux runs on an OS,

You’ll need a few tools to make something of the Linux kernel. Networking configuration, for example, or disk management (format, check). These tools are perhaps the difference, but they don’t make a difference. And they are needed anyway: the Linux kernel just can’t work alone.

I am talking about the OS being RISC OS, with only the kernel under/within it, that is NOT EVEN INIT outside the kernel, just bare kernel then RISC OS ROM Image, nothing else, no shell, nothing.

IMHO, a lot of efforts just to save 1 MB of disk space.
Nobody cares of these tools: you’ll never see them.

You would expect that RISC OS would be the first group to want to try the rout of custom HW designed as a hobby

No, because we don’t need too. We are lucky: we have hardware. Semi custom one (Pi), or really custom one (Titanium). We even have an Open Source CPU projet: Amber.

are those that look for G3 HW to keep going

G3? You mean the iMac? I have one of those in the garage. I think I’ve come across an iso image of MacOS9 in English, so maybe if I’m really bored…
https://heyrick.eu/blog/index.php?diary=20131027

Yet the current talk gives the appearence that the RISC OS users are not as dedicated.

We don’t have to be as dedicated as we have contemporary ARM hardware to play with. Indeed, the platform itself isn’t stretching the abilities of what we have so there’s room for better things right there.
The problem is, the modern world has shone a light on some of the architectural deficiencies of the OS, which is part of the reason why the Wimp doesn’t just fire up the other three cores and spread out the task load.
If we’re to move forward in any great capacity, something first has to change with the OS itself, not the platform it is running on. The use of RISC OS on the various ARM 32 boards has more or less been keeping the OS exactly as it was only running on a more powerful machine. We have benefitted from the machines being broadly compatible in the instruction set (differing in platform specific ways like I/O, video, etc). Well, we’re about to run into a pretty large obstacle, and while I can see the purpose of having third party FPGA style ARM cores, I think it’s going to be quite a while before we have anything that can match the higher end devices already available to us.
In other words, do we try to devise a plan to embrace the future, or do we stay with current hardware (and emulations of) in a semi-stagnant manner?

It doesn’t mean we aren’t dedicated, it just means we’re looking in a different direction.

If we’re to move forward in any great capacity, something first has to change with the OS itself, not the platform it is running on.

I think the platform that the OS is running on is actually pretty important. The Raspberry Pi gave RISC OS an influx of new and returning users who stimulated further development. We’re about to see millions of people shift onto ARM-based systems as their main computer. If we can offer them an alternative Linux distribution with ROX + ROL then we have an opportunity to get thousands of extra people using RISC OS.

So your argument is what?

Simply that getting RISC OS on Linux, Debian & ROX working together seamlessly should be ROOL, R-Comp, CJE Micro and anyone else’s top priority in the short term. Grab the quick win then do the hard stuff when more people are around to help out.

So why are so many people running to change the platform it runs on, instead of continue the work to the OS, or at least be supportive of the ongoing work to the ARM OS. No everyone talks about running to a different platform.

Citation needed.

It does not help to progress in this discussion if you keep inventing stuff all the time. And insist on your own strange wording, like “only AArch32 is ARM, AArch64 is not ARM”. Or that the G3 iMac is in some way not a G3 Mac. Or coining your own terms that nobody but you has ever used before (“26bit R15” or “DosEmu”) for those things. And of course making up stuff out of the blue (“There is a reason almost no one has used M$ Windows in the last couple decades” or “most people use the indirect strings almost exclusively”).

How is your BASIC compiler coming along, by the way? And your fast FPGA implementation of ARMv2? And the 100% pure assembler USB stack you apparently had ready some years ago?

Arguing this point of view is taking time that could be better used by me getting my CPU documents retyped and up on my site. A better break time is to work on my software instead of the restating my point ten-thousand times in response to strange re-twistings of the argument.

There’s no reason to argue with those of us who’d like to see RISC OS make use of 64 bit ARM hardware. I think the arguments for doing so are compelling. You don’t have to agree. I’m sure there will be others who want to stick with ARM32, just as there are people still on PowerPC and 68K.

I’m quite happy for anyone to continue to work on Aarch32 and have nothing to do with Arach64 developments, but please stop spamming ever other topic that trying to move RISC OS forward.

Spam, as in every topic I read, there are runs of half a dozen consecutive posts by the same person. There is no need to separately respond to every single post by other people, try to make one concise post taking the debate forward. There is no need to repeat the same point ad-infinitum, I think everyone now knows where you stand on Aarch32 and the use of assembler.

Possibly relevant to the original topic;

https://www.youtube.com/watch?v=LHsHU-dDwAk

Possibly relevant to the original topic; https://www.youtube.com/watch?v=LHsHU-dDwAk

That’s a very nice piece of work. Is it (hopes rising) the reason why you now own a company named “Acorn Computers”?

@ Peter

that’s a very nice piece of work, and the obvious question follows: Where could I get it? :D

@ Peter that’s a very nice piece of work, and the obvious question follows: Where could I get it? :D

Educated guess: http://www.marutan.net/themes.php

Charlotte wrote:

That’s a very nice piece of work. Is it (hopes rising) the reason why you now own a company named “Acorn Computers”?

Completely unrelated. I mainly registered Acorn Computers to stop other nutters getting it and using the name for releasing unrelated things … also I like sometimes introducing myself as the Managing Director of Acorn Computers :)

In response to one of your queries in another post. I neither own, or have licensed any IP, copyright, trademark or patent with regards to the classic company. As such my Acorn Computers will likely have to remain ‘dormant’ (non-trading), at a cost of 10 quid a year to me to keep it registered.

Paolo wrote:

that’s a very nice piece of work, and the obvious question follows: Where could I get it? :D

You don’t want it yet :D

It is very easy to make a video that shows all the working bits, and skipping all the non working bits. There is also some hardcoded paths in the config that would make it tricky to get going if you’re not called ‘peter’ (/home/peter/sprites :D )

Though if you would like to follow the changes, its repository is here;

http://home.marutan.net/hg/riscwm/

I have yet to commit the most recent menu related changes.

Steffen wrote:

Educated guess: http://www.marutan.net/themes.php

That is where I had reached about 10-15 years ago, but my realisation was that just theming existing window managers did not allow for some of the nuances of a RISC OS desktop. As such I’d have to make code changes to support things.

I searched the massive list of existing X11 window managers looking for something to base it on, and settled for JWM, a very lightweight WM, written in C, with anti-aliased fonts and a windows 95 style interface.

After that it has been very occasionally working on it for 7 years or so, slowly battering it into a shape where it can fool some people, some of the time.

I mainly registered Acorn Computers to stop other nutters getting it and using the name for releasing unrelated things…

So, I guessed right. Seems like a neat way of keeping the dodgier elements of the business world away.

@Peter I’d like to thank you for holding the Acorn Computers company name, as I was so annoyed back in 2006 when someone started selling cheap and nasty Windows PCs under the Acorn name.