No more big 32-bit cores for RISC OS from 2022

Simply put, I think investing in RO should be eligible for tax relief like other charitable donations.

LOL brilliant! :D

Then ROOL needs to set up a charitable foundation.

Well, Paolo picked up on the irony anyway…

I’m missing the daily (Mon-Fri) wit and repartee at the office.
WfH (because Covid) is fine on travel time (none) but the interplay of 8 techies in one barely big enough office is fun. I’ve trained them all to be almost as acidic.

Paolo and at least one other. But irony is harder to spot in plaintext.

But irony is harder to spot in plaintext.

True, some expect it though.
Some at work think it’s related to taking creases out of clothes. :)

Cortex-M cores only support the Thumb1 or Thumb2 instruction set

Correct, so no use here.

dominates the world because

Because it is free. Don’t underestimate the aversion to paying licences.

You’ll notice that in the great majority of devices running a Linux kernel, the owner probably doesn’t realise it. Webcams, routers, smart printers, blah blah. All can be built, have Linux and busybox and some crappy custom software shoved in, no licence payments, and to hell with the GPL in far too many cases.

Isn’t RISC OS already free? Not sure what point you’re trying to make, dare to clarify please?

There’s more to a device than just the underlying OS. And if you could with generic branded hardware (IPcams, media sharers, etc) you’ll soon come to note that there is a cheap “made in China” clone of something known, where the vendor company has done little more than stick their own branding on some barely functional hardware and firmware.

In other words, IoT is a cut-throat race to the bottom.

David:

1/ MPU support or not? (to open RISC OS to Cortex-M cores)

And a few days later:

Correct, so no use here.

Care to justify this?

Because if we do it properly (defined as don’t write it all in assembler!) then whether it’s ARM32, ARM64, or ThumbX shouldn’t be particularly relevant. Okay, it will need thinking about how to handle stuff like SWI calls, but it isn’t RISC OS and it won’t run RISC OS software (unless BASIC gets ported ¹) so it doesn’t have to behave in exactly the same way. Just similar enough to be familiar.

¹ Doesn’t the new Micro:Bit support BASIC? Is it onboard? Is it the same or a lesser BASIC?

it won’t run RISC OS software (unless BASIC gets ported) so it doesn’t have to behave in exactly the same way

It has to work in a very similar way, even if BASIC gets ported. Even RISCOS software that’s written in BASIC has to drive the WIMP… BBC BASICs on other platforms are pretty pointless really.

@ Rick

Because it is free. Don’t underestimate the aversion to paying licences.

Apologies, but this is getting boring now…

1. FreeBSD is free
2. OpenBSD is free
3. Haiku is free
4. RISC OS is free
5. FreeDOS is free
6. ReactOS is free
7. IllumOS is free
8. AROS is free
9. DexOS is free
10. AtheOS is free
11. MenuetOS is free
12. Mach is free

Should I keep going?

The difference between Linux and all the above is that for many people Linux does more and better in many situations (not all situations), not that it’s the only one free.

Some practical examples for you:

• Linux got multi-core support before BSD did
• Linux got ported to certain architectures before other OS did
• Linux is MODULAR enough to be able to be built as an IoT cooperative multitask, small OS from the same Kernel that can also be compiled to be full featured Desktop System or HPC.

There’s more to a device than just the underlying OS.

And who said that the OS is the only thing? The reason for focusing on RISC OS is because of trying to find a real market for it. A market where it could be used with the “smallest” amount of changes and with the highest ROI possible given the situation.

Modify to see a causal thread

Linux is free, the alternative on PC (Windows) is not – that’s interesting to the rebellious geeks

Linux got ported to certain architectures before other OS did

The rebellious geeks want other architectures “because”

Linux is MODULAR enough to be able to be built as an IoT cooperative multitask,

Extension of the previous, but also Linux is modular because different people wrote different bits in a slightly anarchic way, but mostly in bits.

Thus, a free, modular OS that had already been ported to multiple architectures was then targetted at a set of cheap, mostly ARM based, boards usually labelled as for IoT.

In that timeline RO was not free and only relatively recently became open source (with licence) and then very recently open source “do what you like”.

RO? Possibly a bit late to the party?
However, unlike Linux it doesn’t need slimming down to fit on the development boards.
Pi4 has what Linux fans would see as an adequate amount of RAM and storage available.
That same board has what RO fans regard as a bewilderingly massive amount of RAM and storage available.

If there isn’t some mileage in that I’d be surprised. Whether there is the developer resource to take advantage is probably the real question.

@ Steve

Linux is free, the alternative on PC (Windows) is not – that’s interesting to the rebellious geeks

BSD was always free, AROS came out of people being frustrated with AmigaOS, MiNT came out from people who wanted multi-task on Atari TOS, WIMP2 came out from who wanted to get as close as possible to preemptive multitask on RISC OS.

Yes, innovation commonly come from who wants more or something different :)

Extension of the previous, but also Linux is modular because different people wrote different bits in a slightly anarchic way, but mostly in bits.

Said this way sounds like “Linux got lucky” which is a bit diminishing the effort of so many people around the world, me included at work (just finished a big port of a complex kernel framework we created from Kernel 4.9 to kernel 5.4).

IMHO Linux, despite what some untrained people may say, it’s really well designed in many aspects, not all aspects. It does indeed solves a lot of technical issues and it provides many protocols that behaves in a standard way. The modularity offers another advantage: Because it can be used in multiple situations “I” don’t need to learn how to code on another platform. This has been a huge point of traction for many things in IT.

In that timeline RO was not free and only relatively recently became open source (with licence) and then very recently open source “do what you like”. RO? Possibly a bit late to the party?

Sure and let’s also add to your list:

• RO had its own quirkiness
• RO was forced/confined to ROM only. This is one of the big reasons of slow update process in the past for both manufacturers and users (all ROM based mjor player of the 80s died also for this reason). The average Jo couldn’t update his/her system easily (some updates even required extra PCB to be fitted and soldering). This was in a time where MS-DOS was on a single floppy disk and also on the HD, so easy to install/update and if the newer release was not compatible with Application X you could still run an older version from floppy and run application X just fine.
• RO ran on systems (late Acorns) that struggled to be competitive with PCs when PCs started to get their stuff together
• RO never really evolved looking at the future as other OS did, this can be seen as a good or a bad thing depending on the subject, but it’s certainly another big difference for RO compared to many other OS.

However, unlike Linux it doesn’t need slimming down to fit on the development boards.

Absolutely true, however it needs to be “fatten up” in order to provide a modern Desktop experience, run AAA games, run on big systems etc… so doesn’t sounds to me an advantage according to what people generally want. But yes it’s an advantage in markets like the IoT.

Pi4 has what Linux fans would see as an adequate amount of RAM and storage available.

True again, but that is because of the pure RPi “philosophy”, in other words: What type of computer can we build in 2020 for just 35 GBP price??? in 2020 you can have 4GB RAM and 4 cores ARM for that price etc…

RPi IMHO is the BEST possible ideas from the old British Computer companies put together: Low price of a Sinclair ZX Spectrum, expandability ala BBC Micro and processing unit ala Acorn Archimedes all put together in one single and very cool device.

Linux on the Pi is just another OS that can run on it. However it’s also an OS that can use all the features offered by the hardware ;)

If there isn’t some mileage in that I’d be surprised. Whether there is the developer resource to take advantage is probably the real question.

That IMHO is a really really good point.

You got me thinking, so this one is a bit Rick Murray. :)
Hope people don’t get bored before the end.

Said this way sounds like “Linux got lucky"

Yes.

1. Lucky on timing and lack of coherent competition from the “industry standard”¹ Windows
2. Lucky to have so many rebellious geeks with no other easy option to play with.

That’s not diminishing the effort of those people, but it’s a lucky circumstance they all stayed playing with the same stuff.

RO had its own quirkiness
RO was forced/confined to ROM only

Yes, definitely.

Absolutely true, however it needs to be “fatten up” in order to provide a modern Desktop experience, run AAA games, run on big systems etc… so doesn’t sounds to me an advantage according to what people generally want. But yes it’s an advantage in markets like the IoT.

You’re seeing something where the OS overall has to centrally provide those and “fatten up”.
I’m seeing something where I think the OS needs some house-mates, where the guys in the other rooms (cores) do parts of the household chores.
IoT – if the guy at the front of the house does the “window dressing” and takes the orders (RO as now plus a bit) the guy in the backroom(IoT device) can be set a task and left to it. (I think some aspects of Dave Higtons home control stuff are on those lines.)

True again, but that is because of the pure RPi “philosophy”, in other words: What type of computer can we build in 2020 for just 35 GBP price??? in 2020 you can have 4GB RAM and 4 cores ARM for that price etc…

I think the driver for most of the extra speed and memory is the “fat” nature of Linux as used on the Pi. The Raspbian build was only really there to provide a programming environment for the kids in schools that were supposed to be the target market. When the Linux fans got hold of it and discovered it could do, slowly, what they wanted: they asked for faster/‘bigger’ versions of the hardware.
Think x86 PC development (speed/RAM etc) driven by MS code bloat.

In ARM terms Linux hit the platform pre-bloated and the platform is expanding to accommodate. The question is: can RO expand into that space?
Some people on this forum have got RPi400 units. It was launched, what – 3 days ago? Today the tweak to make RO work looks to be complete. That’s about the same time taken to get most of Retrokit working on it.
There’s an increasing number of interested people with a cheap unit.

RO has things like network drivers operating an I/O in a fashion that co-operates with the rest of the OS in use of the same core, while typically (RPi4/400) has three cores sat idle.
How many other CPU intensive or blocking operations does RO have that could massively improve performance of the OS in general, and their own, if they were bounced off into another core?

Linux on the Pi is just another OS that can run on it. However it’s also an OS that can use all the features offered by the hardware ;)

Including all the cores – but it’s still slower on various aspects than RO running on one. Definitely a development opportunity there.

¹ How many times did I hear that bull in the 1990’s?…
Industry standard was big IBM stuff, so the PC fans were talking sh**

Care to justify this? Because if we do it properly (defined as don’t write it all in assembler!) then whether it’s ARM32, ARM64, or ThumbX shouldn’t be particularly relevant.

Correct too. In fact, a full 64bit port of RISC OS is probably impossible. But a very small core with a SDK on RISC OS and a BBC Basic port could be quite easy to adapt on different ARM cores. A bit like the excellent PICSuite.

BSD was always free,

No it wasn’t. It was always “source available”, but the original BSD used bits of AT&T code, for which a licence was required. And given that Unix was used in heavy iron (think Vaxen and such), the licences were expensive.
So you could look at how it worked, but you needed to cough up cash to actually use it.

It wasn’t until the Tahoa (sp?) release that the offending AT&T code was replaced, and the infamous dead-easy licence applied. From there spun off a number of different versions (FreeBSD and NetBSD to name two) as groups of people differed on what the path forward should be. A bit like RISC OS 4/6 and 5, then. ;-)

WIMP2 came out from who wanted to get as close as possible to preemptive multitask on RISC OS.

My understanding was that Niall wanted to talk with Acorn about having some measure of pre-emptive support in the Wimp. Acorn, unwisely, said that it wasn’t possible. And Wimp2 (originally called Tornado) was Niall basically saying “bollocks, it can be done”.
It’s a very hacky implementation (for a start, it degrades everything to be a RISC OS 2 era task with no message filtering ¹, and working around numerous special cases) but it does rather prove the point that something is possible.

Said this way sounds like “Linux got lucky”

In a way, it did. Please don’t underestimate the political angle, especially those adhering to the GPL movement.
Before you respond – ask yourself one question. Why Linux? It wasn’t the first free incarnation of Unix around. Hell, Linus was inspired by what was happening with Minix, so really it’s a bit of a latecomer. So why Linux? Don’t underestimate the political angle.

Linux got {blah blah} support before {other OS} did

Yes. Because it has a veritable army of developers. There were fewer in the beginning, but now many. It’s called traction. Do you want to code on a minority OS, or the big one everybody else favours?
It’s perhaps worth noting that Linux alone probably has more different distros than there are alternative free Unixen.

The latecomer is now top dog.

And who said that the OS is the only thing?

We are talking about IoT devices. While many of them use Linux, it is partly because Linux is free, but as has been discussed, partly because it has the ecosystem. Can you point me to a little IPCam that runs NetBSD? I can point you at two (the two I own) running Linux.
This isn’t to say that they (the device creators) actually give a crap about Linux at all. They’re basically just parasites cashing in on other people’s work knocking out cheap FOSCAM clones and the like. But they use Linux because it’s there, it’s free, and it does what they need it to do.

unlike Linux it doesn’t need slimming down to fit on the development boards.

Oh, yes.
My original PVR had a little custom build of Debian with some tools and the codecs/UI to do its job.
An upgrade brought enhancements, one of which was a snazzier UI based upon Qt4 (I think, it’s been a while). Unfortunately, the firmware needed to hijack the CF slot because the on-board 16MiB Flash was too small.
Turns out this was useful later on, as a hack of the firmware called OSDng made much of the filesystem read/write so stuff could be modified as desired. Sadly, despite being GPL, the author of OSDng never made the mods available.

16MiB. I think you could fit a build of RISC OS and a fake filing system containing quite a lot of the !Boot resources in order to be able to start up a fully featured system.

A market where it could be used with the “smallest” amount of changes

That would be nice, but we’re always going to run into two issues. Security and stability. Security on RISC OS is pretty much a lost cause. Thus it may be better in a domestic embedded sector where it can quietly do its thing with nobody any the wiser as to what’s actually running the device.
But, stability. It’s still extremely easy for a fairly trivial mistake to take down the entire machine.

My build of RISC OS (5.23 ²) quite often crashes when I try to shut it down. It’s one of those annoying repeats-forever ones (the the desktop never fully shuts down). A mistake in one of my modules? Perhaps. But it does demonstrate the inherent fragility of things.

It does indeed solves a lot of technical issues and it provides many protocols that behaves in a standard way.

The basic point of an OS these days. ;-)

As an aside – I am firmly of the opinion that in a few years time nobody will much care what an OS is. Windows, MacOS, Linux, Android… they will eventually cease to be relevant, with the importance placed upon what a device can do (as in video calls, social media, streaming…) rather than what it is running to make that happen.

So while companies like Apple are going out of their way to make themselves incompatible with other things in favour of a private protocol just for themselves, this mindset might one day bite them in the ass. As it is, I can fling files around using Bluetooth to everything except my iPad and RISC OS. Newer devices can support WiFi Direct for a quick private transfer. The iPad can to, but AirDrop. Not anything the rest of the world speaks.

all ROM based mjor player of the 80s died also for this reason

I think one of the main reasons of death wasn’t that the OS was in ROM, it was because when the eight bit home computer boom fizzled out, nobody really knew where to go from there. The primary next generation machines were the Archimedes family (for the rich!), the Atari TOS machines, and the extremely popular Amiga family.
Archie was expensive, TOS was a bit odd, and I think most others died a death knowing that they just couldn’t compete with the Amiga.
Plus, knocking out 6502 or Z80 based machines was a bit of a bandwagon. It stood to reason that most of them would fizzle out and eventually be remembered only by die-hard enthusiasts. Hello Oric! :-)

This was in a time where MS-DOS was on a single floppy disk and also on the HD, so easy to install/update

Essential given that the damn thing had an extremely shonky filesystem and was ridiculously prone to virus infections, especially clever ones that could wiggle themselves into pre-boot due to the peculiar way the BIOS starts up PCs.

RO never really evolved looking at the future as other OS did,

Yes. RISC OS was well ahead of the game and quite ass-kicking, in the late ’80s. Unfortunately it sort of sat there staring at the clouds failing to notice the rest of the world not only passing it by, but lapping it a couple of hundred times.

Mom used to know when I was using RISC OS… because it was the ‘dated looking one’.

What type of computer can we build in 2020 for just 35 GBP price??? in 2020 you can have 4GB RAM and 4 cores ARM for that price etc…

Double that, you can have a basic LattePanda. Intel Cherry Trail Atom (quad core) at something like 1.92GHz, 2GB RAM, 32GB eMMC, GPU/HDMI, 1xUSB3, 2xUSB2, WiFi/Bluetooth, can run Windows 10. Has a built in Arduino ATmega32u4 for extra hacking fun.

Yes, a proper Windows x86 SBC for twice the price of a Pi.

Oh, and note that the Pi4 with 4GB is closer to £45 than £35 (price: Farnell).

However it’s also an OS that can use all the features offered by the hardware ;)

Pow!

FX: Blows smoke from barrel.

I think the driver for most of the extra speed and memory is the “fat” nature of Linux as used on the Pi.

This. It’s only just recently that people have been saying that using the Pi as a desktop device didn’t suck. Maybe that’s why the fancy keyboard has only just appeared now, now that there’s a Pi with enough grunt to do the heavy things that people want?

There’s an increasing number of interested people with a cheap unit.

Yes. And the in-keyboard form factor is a lot safer than having stuff plugged into bare boards (though it might be ‘interesting’ getting an RTC in there – really they should have tracked out IIC to an internal header so people can bung in their own). It’s tidy, it’s compact, it only really suffers from sticking with the crappy micro HDMI sockets.

if they were bounced off into another core?

In theory, many.

In practice, it wouldn’t make any difference. The design and behaviour of RISC OS doesn’t understand blocking. So if core #1 was to handle the file ops, core #0 would sit there patiently waiting for core #1 to do its thing. Because RISC OS is, at heart, a single process single state OS with smoke and mirrors to make it look like more.

Industry standard was

…a buzzphrase loosely translated as “what everybody else is using”. Hence why it invariably meant Word, Access, Excel, Powerpoint, or some unholy combination of the aforementioned.

For a while, when you could be assured that people’s laser printers were HP compatible and hooked to a parallel port, I used to troll people that sent me WORD files. I’d usually read it in Edit or Zap, then I’d write my reply in Ovation, print it to file, and email back the saved printer dump.
I’d tell them:
Windows probably won’t understand this, but your PC itself will. Just save this file to disc, it’s called blahblah.dat, save it in C:\. Then go to the DOS command line and type in COPY /B C:\BLAHBLAH.DAT LPT1: and it’ll be interpreted by your PC and printed for you.
More than once I got back “oh my GOD, it worked!” and messages to that effect.

¹ This is part political – Niall didn’t like the inclusion of C stuff in RISC OS 3 because it was getting away from the purity of pure ARM code.

² I’ve noticed that SerialUSB 0.09 (I think?) sometimes fails to open the file handle on 5.25 and later. My weather station software doesn’t cope with that, as it shouldn’t really be happening. It’s definitely a change in the OS as 5.23 works fine. As I can’t be bothered fiddling with my software to make it more bulletproof, I’m just staying with the one that works…for now.

So if core #1 was to handle the file ops, core #0 would sit there patiently waiting for core #1 to do its thing.

So you don’t think your desktop would be snappier response if a different core was handling a file transfer from one location to another?
Or that the transfer itself would be faster if the process to move files and the update of on screen info about the transfer weren’t competing for CPU time?
Or that network I/O done by a different CPU would be faster?

No it wasn’t. It was always “source available”, but the original BSD used bits of AT&T code, for which a licence was required. And given that Unix was used in heavy iron (think Vaxen and such), the licences were expensive.

Rick, look I think you’re getting the dates a bit wrong…

Linux first official release 0.01 was September 17 1991, that release was unusable (talking by direct experience back then) and was just the beginning of the development releases.

FreeBSD first release was 1993 and was fully usable. FreeBSD is a descended of 386BSD which was also available for free and was officially released in 1992 and it was usable on many 386 PC compatibles.

Linux, until 1994, was only a development project and not much more, the first usable release (1.0) was done in 1994. After that there have been few years of problems because of the initial drivers API etc…, so technically some major issue continued after 94 if we really want to be pedantic.

So, with this clarified, I am sorry but your answer is not correct, BSD was freely available BEFORE Linux reached the state of usable.

What caused problems to business use of FreeBSD in the early years was a fear (yes just a fear) caused by the initial legal issues. The actual legal action against BSDI company in early 1992 resulted in the dismissing of most of the AT&T claims in 1993 and the legal “drama” continued until beginning of 1994 but the part 2 was Berkley Uni. counterattacking. Anyway at the beginning of 1994 the whole legal dispute was fully resolved and Novel acquired the rights on UNIX and FreeBSD had all the claimed copyright parts replaced (they were like 2 files or so IIRC).

You can comment on this as much as you want, but please provide proofs, not just opinions, peace out!

You’ve made two different arguments (‘BSD was always free’) and (‘BSD was freely available BEFORE Linux reached the state of usable’). Rick’s comments appear to be in response to the first one.

Even RISCOS software that’s written in BASIC has to drive the WIMP… BBC BASICs on other platforms are pretty pointless really.

Um. You’d better tell that to the people who write Windows applications in BB4W¹, then.

¹ Not that I’m advocating doing this, you understand. I find hacking about with the raw Windows SYS calls in BB4W a reminder of why BASIC should really be put out to pasture and used for “turbo-charged scripting duties”.

@ Chris Mahoney

You’ve made two different arguments (‘BSD was always free’) and (‘BSD was freely available BEFORE Linux reached the state of usable’). Rick’s comments appear to be in response to the first one.

Is this some sort of “twist and relaunch” game? Anyway…

First:
I provided the right context by listing FreeBSD and OpenBSD as FREE look at my original list and Rick said no… The 2nd time I shortened the list of BSDs available for free as BSD only, my bad I should have said xBSD I think I made a typo and meant to write BSDs.

Second:
If now we are getting into the mood of “Nah I don’t care about the right context and typo, I just want to be right!” fine let’s dig a little more in the history then:

In 1991 Berkley released Net/2 BSD which was also shortened as BSD and it was FREELY redistributable, so even in the case of the typo above the concept actually still holds truth.

Net/2 BSD (or just BSD) is what started the legal battle, why? Guess what: because it was freely available and redistributable (at the time this was defined as a “permissive license”) and because also granted access to source code.

Also another note Net/1 BSD was not a complete Unix, while Net/2 was…

So, because Net/1 was not a full OS and Net/2 was and was also the first free available Unix clone, I guess we can all be finally happy with my point? If not then bummer, can’t make everyone happy lol :D

Peace out guys and hopefully we can re-focus on RISC OS now ;)

Peace out guys and hopefully we can re-focus on RISC OS now ;)

This thread reminded me of this article in a MacOSX book I picked up from the recycle, not that I have Macs, but the book mostly covers bsd and shell utilities.
gnu/bsd page
Regarding the gpl3, how is it that ExFat is described as gpl3, yet it has been in the linux kernel for a year?

@ Ronald

This thread reminded me of this article in a MacOSX book I picked up from the recycle, not that I have Macs, but the book mostly covers bsd and shell utilities. gnu/bsd page

Actually you make a great point there, and probably I should have mentioned the historical point on Free as freedom and also the fact that, even if xBSD has always been free, it is entirely possible that someone worked in a company who paid licenses for it because being Free means someone may also charge for it and that’s fine.

Regarding the gpl3, how is it that ExFat is described as gpl3, yet it has been in the linux kernel for a year?

The history of exFat doesn’t starts with GPL actually, the original exFat developed by Microsoft was proprietary back then. In 2019 Microsoft has released the patents (together with a lot of other patents) and published all the specifications.

Linux introduced native support for it from Kernel 5.4 (I am just out of a lot of work on K5.4 at my job).

However there was also previous support for it via FUSE (user-space FS, which allows Linux to be more elastic on licenses given that the code in FUSE do not link directly with the kernel etc.) and also I think Samsung previously released a GPL licensed driver for it (but I never used it).

Anyway the reason of the adoption since Microsoft has allowed it to become GPL is because exFat became popular in the past among embedded systems, so as soon as it was available it became very interesting for that side of the community I guess.

Hope this helps

I hope I didn’t offend; that wasn’t my intention. I was just trying to explain Rick’s argument (or, at least, how I’d interpreted it).

Even RISCOS software that’s written in BASIC has to drive the WIMP… BBC BASICs on other platforms are pretty pointless really.

Um. You’d better tell that to the people who write Windows applications in BB4W1, then.

That’s their problem. Pretty pointless from my point of view. It’s not RISCOS software at all. I can’t take my BASIC RISCOS apps, import them into so-called BBC BASIC on a Windows machine, and expect them to work. I can take things I wrote for an A310 and run them on a Pi3b without any modification at all.

I suppose that’s a bit like C. Learning C itself (including the standard libraries) is trivial, but without other libraries it’s pretty useless, and with different libraries it might as well be a different language. BBC BASIC without an environment is trivial and useless, and with a different environment it might as well be a different language.

Is this some sort of “twist and relaunch” game? Anyway…

While your arguments are all correct, some facts need to be added.

The history begins in 1977, and before 1989, while the source code was available, it was not free…
https://en.wikipedia.org/wiki/Berkeley_Software_Distribution

“all versions of BSD used proprietary AT&T Unix code, and were therefore subject to an AT&T software license.”

And of course, this license did have a price.

I remember too the legal battle, that was only fixed in 1994.
In the meantime, the first usable distributions of Linux became available.
I did use for example the Yggdrasil Linux of 1993.

In 1994 Linux 1.0 became available and Red Hat was created.
And just after, came 4.4BSD-Lite Release 2.
Just a few months too late.

Since Internet was not as fast and efficient as today. It took time to realize that this version of BSD was OK from a legal point of view and that the legal dispute was really over. And even if NetBSD was available since 1993, I simply choose Linux. Because of these legal issues, and because of the big buzz from Stallman around GPL (and the fact that with the GPL, the big companies, as ATT, will never try to steal your work). So it was very political.

I remember, as chief editor of Login: the first Linux Expo in France in 1999.
https://lwn.net/1999/features/linux-expo/

And then the second, in 2000, the contracts of millions of Euro signed in middle of geeks, and Microsoft trying to make the buzz with a concurrent event in the same building. Between 1999 and 2000, the commercial future of Linux was settled, while the xBSD were still trying to retrieve the old BSD glory days. And the rest is history.

In 1995, I switched from OS/2 to Windows 95, without success. Then I did buy my first Acorn Computer, a RPC600 and switched to Linux on the PC. I even did not know then that the legal debate around BSD was closed. I knew it just a bit too late…

In 2003, I remember of the eQ R&D (partly my company) coming with its own Linux distribution (still the fastest to install, to date) and eQ R&D Russia coming with pdaXrom, a name you can still find in some embedded Linux distro or in parts of Android. The same dev did also make a lot of work around RISC OS. For example, the MPEG2 player, the port of SDL, work on cross compilation with GCC, etc. A lot of work that some other people claimed to be their later (PN). Bad days, so we just leave the platform.

Our main dev is still around, but do not live in Russia any more:
https://www.linkedin.com/in/alexander-chukov-6a62441b3/