Was 2015 the end of RISC OS?

And does it just seem lively because I’m talking to myself?

I started using RISC OS in 2015 but I have noticed recently that a lot of websites and software haven’t been updated since around that time.

There is no scientific basis for my 2015 claim…I just have a weird feeling that a lot of things seem to have stopped then, or thenabouts.

Just curious if anyone who has been around longer feels that around about 2015 was a particular low point for RISC OS. Is it getting better?

Also, am I the only “new user” who is absurdly enthusiastic about the platform?

I can disprove that hypothesis – my hubersn Software was updated in 2017, and before that in 2011, so no magic 2015 year anywhere in sight!

Not sure if it is getting better or worse. I got into RISC OS when we did not even care about RISC OS but about the “Archimedes”. Probably 1989 or 1990. It was the speed of the ARM and the niceness of the BASIC and the comparably cheap, but powerful software available. Over the years, this completely reversed compared to other systems: other CPUs got a lot more powerful, other application software got a lot more powerful and a lot cheaper, BBC BASIC has been overtaken by every interpreted language on every other OS, and the rest of the world has capable Internet browsers.

So, some would say that the decline began in 1987 with the release of the Archimedes. Or 1989, when RISC OS 2 was released. Or 1991, the A5000. Or 1994, the Risc PC. Or 1996, the StrongARM. Or 1997, the last year you could sensibly browse the Internet with a RISC OS browser. Or 1998, the demise of Acorn. Or 1999, RISC OS 4. Or 2002, the IYONIX and RISC OS 5. Or 2009, the RISC OS-on-BeagleBoard release. Or 2012, the RISC OS-on-RPi release. Or 2015, the last year a significant piece of RISC OS-capable hardware was released (Titanium).

Oh, see: there’s that 2015 year!

You see, we have been declining for a very long time now :-)

but I have noticed recently that a lot of websites and software haven’t been updated since around that time.

You’re close. Time as we know it actually ceased on the 29th of August 2013. Were you not paying attention, or does your simulation have more polygons than mine? Graphics qualities aside – Brexit, Trump, who is going to believe that rubbish? They not better scriptwriters.

about 2015 was a particular low point for RISC OS.

2015 brought us the Pi2. A quad core with a gig onboard for the same tiny price as the original. And, fair enough, RISC OS can’t make it stretch itself, but it’s there (as with the other Pi models) and it works.

If you’re looking for a RISC OS low point, try two decades earlier, and the following decade. As you’re a new user, I’ll just sum it up as one group went 32 bit and one group stayed 26 bit, and for many reasons (real and invented), it all got quite… acrimonious. There’s a reason the RISC OS website (the other one) talks about version 4, version SIX, and makes utterly no reference whatsoever to this version existing. Indeed there’s a reason why they called their last release “6”. If you think it’s stupid, don’t go looking for more info, it only gets stupider.
Anyway, if nominate that as the low period.

In recent times, well, we’re now able to run RISC OS on quite a diverse range of devices from the ass kicking to the Happy Meal priced microscopic board. These machines use modern hardware (USB mice, SD cards, etc) and generally have pretty good display capabilities, as is expected these days.
Would it be blasphemous to suggest that – hardware wise – we’re in a better position now than we ever were in Acorn’s day?
What we lack is video and scripting. One can’t watch kittens on YouTube on RISC OS, and that will hurt us. People interested in RISC OS will leave to use Linux because it’s also free and it has kittens. We don’t have kittens. There’s your problem right there.
Of course, we have BASIC but nobody programs any more, as five seconds looking at any government sponsored “coding is cool” stuff would make abundantly clear. Nobody plays games either unless they can oooh at the realistic hair and even more realistic bullet holes all in 60fps FHD. Nobody watches TV because, hey, they can just torrent on demand. Does RISC OS even have a torrent client? And nobody cares about the operating system because these days they’re all just consumers. It’s like “the e is the internet” on a global scale.
But most of all, we lack kittens.

I was “with Acorn” from the BBC Micro. Compared to so much rubbish of the early eighties (much of it thanks to Clive Sinclair), a machine with a sane dialect of BASIC, built in assembler, good knows how many I/O pins all over the place, and local area effing networking (in a time when many machines had difficulty talking to a tape recorder, Beebs were talking to each other!) was utterly mind blowing.
When I got my A3000, I didn’t know much about the machine other than RISC OS offered a UI (still a rarity in those days), and it came complete with an entirely custom chipset which intrigued me. I was interested in DTP, so having packages available was a bonus (though AcornDTP was awful, it got stolen from school and with the payback I got Ovation and never looked back). Fairly soon I discovered ARM assembler and fell in love with it. It was like “if I was smart enough to design a processor, I’d hope for something like that”. It was quite a step up from 6502, and it kept me sane when college inflicted x86 upon my weary brain. And that was the time when 386 was cutting edge! I did, once, work out the interrupt latency for… what was it? The 8088? It was horrific. Upgraded to an A5000 with a gigabyte SCSI drive, when computers were sold new with half that. These days that’s like “two animé episodes” but back then it was beyond massive. Stuck with the A5000 until CJE did a special on refurbished RiscPCs “for programmers”. Got one of those, an ARM710, RISC OS 3.7, and that was my RISC OS machine until I went over to the dark side. Let’s face it, though, my dark side was mostly a second hand part broken laptop, but it could play DVDs which was better than my DVD player. Why? The PC was persuaded to stop making annoying requirements like forced subtitles or region coded discs. It’d play anything without getting stroppy, and talking directly to a monitor it didn’t have the quality loss of CVBS to a VCR, then RF to an ancient TV. My programming in VB was mostly “to pass the time” and “why not”.
I knew the RISC OS was still around, I knew of the Iyonix and even got to experience it’s appalling efforts at writing small files to USB devices (did that ever get fixed? I got in the habit of creating an archive so it’d only need to sort out writing one file!).
When I was gifted a Beagle xM, I got back involved with RISC OS. I don’t do any of this to make money so I now only program RISC OS. If I’m going to program for fun, it’s going to be a system I like. The problem with the Beagle was in getting decent video out of it. Mostly I was stuck with the s-video, and Jeffrey was brilliant in getting that sorted really quickly. I then moved to a Pi and that machine was my main machine until I finally (years late) upgraded that to a Pi2. I have little interest in looking at the Pi3, I have a quad core with a gigabyte, I don’t see the 3 offering anything extra (to the world of RISC OS, so forget your WiFi and Bluetooth!).

At any rate, no, I don’t think 2015 was the start of anything. Or, was it perhaps? https://www.opendemocracy.net/digitaliberties/andr-staltz/web-began-dying-in-2014-heres-how

Also, am I the only “new user” who is absurdly enthusiastic about the platform?

I have quite a bit experience in talking to RPi owners (technical people who really do cool stuff with their Pi) trying to interest them in RISC OS.

Typically, people expect at least one of the following:

• a browser capable of browsing the whole Internet (including video playback of course)
• remote login via ssh and control it via the command line
• proper multi user environment
• programming language xyz
• WLAN

I think that several of the RISC OS features would still be interesting to many techies (like e.g. being able to actually understand things the OS does, making tinkering and bare metal programming easy…), if say the WLAN and browser problem could be solved.

I started using RISC OS in 2015 but I have noticed recently that a lot of websites and software haven’t been updated since around that time.

I only started a serious exploration of RISC OS in January… From some of the above posts, I’d say 2009 (Beagle release) or 2012 (Pi release) is the year RISC OS was saved!

From what I’ve seen in various old archives, like Stuttgart and the ANS filebase, the largest die-off was approximately 2002.

Encouraging feedback!

a browser capable of browsing the whole Internet

Well…I’m learning how to crawl with this one, then maybe in a few months I’ll be able to take some steps towards this! Definitely the number 1 “issue” with the OS for me…if you can blame the OS for a lack of a suitable program to do a task!

(including video playback of course)

Is there anything in RISC OS that can play videos?

or 2012 (Pi release) is the year RISC OS was saved!

It does make the Pi into a very useful little computer. I wonder if people are put off because they feel the interface looks old-fashioned as well as the missing programs/features outlined above?

Is there anything in RISC OS that can play videos?

MPlayer and KinoAmp, IIRC. I was only able to get MPEG-1 running at a reasonable speed, but that was a while ago (possibly on an original Pi 1).

all got quite… acrimonious

I was aware of some of it and its really quite sad. Similar things killed off other computers as well I guess.

But it will all go away with RISC OS 7. This version will also support WiFi, bluetooth, remote SSH and have a wonderful new version of NetSurf capable of viewing the entire Internet (probably in one go) and be running on a single chip computer that we mount to our eyeballs for ease of access. Naturally it will respond directly to our thoughts and wont have any of this old-fashioned USB nonsense.

Hmm…might have got a little bit carried away there…

MPlayer and KinoAmp, IIRC

And according to the search of the forums I just did here both of them survived the 2015 apocalypse and are still getting updates!

(as a point aside, why do I have to resort to using Google to search these forums instead of the built-in search thingy that always seems to be pants?)

I started using RISC OS in 2015 but I have noticed recently that a lot of websites and software haven’t been updated since around that time.

Some RISC OS people/companies are bad – some notoriously so – at keeping their websites up to date. I wouldn’t use that as an indicator of the amount of life there is in the RISC OS world (or when the decline may or may not have started!)

The continued lack of a decent standards-compliant browser is by far the biggest impediment. Lack of WiFi is #2, but academic in the light of #1.

“Programming language XYZ” is irrelevant – you need bums on seats before you start worrying about what to serve them. Without Facebook, Twitter, Instagram, Netflix, iPlayer, Spotify, banking etc, there will not be bums on seats.

Personally, I think we need to code an OS from scratch, and implement necessary 32-bit ABI/API to support legacy RO apps, but encourage new apps to the modern API. Would have to still be an ARM-based system unless you plan of writing an emulator on top of things. I think the current code-base is stopping progression.

if you can blame the OS for a lack of a suitable program to do a task!

I think we can blame Microsoft for the extreme blurring of the lines between OS and browser.
It’s amazing how much of Win98SE terminally dies if you start pulling out bits that are supposed to be the internet browser. The “customise folder” in (file) explorer gave you the most configuration options by creating a custom HTML page (Folder.htt) for the folder.
Of course, having HTML to describe the behaviour of the filesystem viewer and all of the deep OS stuff mixed up with a web browser, it was quite a “the hell were you thinking!?” scenario.

These days, a browser is just an “accepted” part of any OS. Heck, my e-reader has a built in browser. It’s a bit crap, but it works – even with an e-ink display!

This version will also support WiFi, bluetooth,

Good.

remote SSH

With key exchange, because a simple password just doesn’t cut it these days.

[hmmm, I wonder how that would go with “The Feds” if they make me divulge my site password? I can. I could here (but that would be stupid). But you still won’t get in unless you have the epic big private key that pairs with the one held on the server. Of that, only two copies exist (my two PCs). It’s not on my phone, it’s not on any backup. That’s by intention.]

capable of viewing the entire Internet (probably in one go)

If you’re going to be doing that, can we please have a port of Ghostery built in? That way, I can automagically discard half the internet. And let’s ditch anything from an IP address associated with the countries CN, RU, and IL (these are the ones that routinely try hacking my server).

and wont have any of this old-fashioned USB nonsense

Anybody who’s up with their cyberpunk will know it’s a neck jack which is either like a massive old telephone connector (technically: the GPO Plan 4 plug); or it’s a bunch of phono jacks. Quite often there’s slime involved, because everybody has slimy necks in the future (must be a side effect of dropping the line voltage to brain voltage) and clearly liquid and humidity don’t affect these sorts of electronic signals. Unless that’s the reason the retinal enhancements always seem to glitch in a way that makes you wonder if they managed to cure epilepsy…

why do I have to resort to using Google to search these forums instead of the built-in search thingy that always seems to be pants?

Depends what you’re looking for. Single word searches are the best. Personally, I find Google horrible as it often returns results as a page in the “Recent posts” (which is likely WAY different by the time you get to go there) and it also frequently points me to https://www.riscosopen.co.uk/ which is a valid working alias of this site, but browsers will have varying degrees of nervous breakdown over the SSL certificate (it’s for .org, not .co.uk!).

some notoriously so – at keeping their websites up to date

Yeah… Are RComp ever going to make a page for the mini.m?

Without Facebook, Twitter, Instagram, Netflix, iPlayer, Spotify, banking etc, there will not be bums on seats.

I guess that’s why I can “get by” with NetSurf. It’s YouTube that I wish RISC OS would do. ;-)

Personally, I think we need to code an OS from scratch,

You and who’s army? We’re already way short of developer time for the OS we have, never mind fancy ideas like coming up with a new one! ;-)

The problem with designing a new OS (with a compatibility layer) is that all too often the discussion rapidly stops looking like RISC OS and starts looking like some weird derivation of Linux…

I think the current code-base is stopping progression.

See previous comment. I’d love a more secure version of RISC OS where modules ran as SYS (not SVC), where memory was blocked off so stuff didn’t have the ability to trample all over the RMA, and the billion other pet wishes we all have… but it isn’t going to happen. Maybe if somebody wins the EuroMillions and blows a huge chunk of it we might end up with something; but realistically the code base we have is the code base that works. It’s the GNU Hurd vs Linux question again. The one that is great, and the one that works.

(with a compatibility layer)

Not a compatibility layer, but a drop-in native API/ABI.

stops looking like RISC OS and starts looking like some weird derivation of Linux

No, because you’d need to add RO5 ABI/API support to the kernel so I don’t know why’d you start with the Linux kernel. Personally I’d prefer a microkernel model.

Personally, I think we need to code an OS from scratch,

The first thing any development manager must learn is how to stop inexperienced programmers “fixing it with a new one”.

“This 30 year old codebase that has been used for billions of man hours has a well documented bug… I must write an entirely new codebase!”

so stuff didn’t have the ability to trample all over the RMA

The RMA. The RMA. The RMA. There’s the major problem right there. If only there was some way of visualising the contents of the RMA wait a minute…

Later that same RMA…

Blue things are Modules. Green things are Module Workspace. Red and orange are Other Things… and they’re usually tiny nodes of linked lists. You wonder why important things sometimes get trampled on?

The expanded little red/orange thing is but one of half a billion¹ StrongHelp nodes. Ask yourself why they’re scattered about amongst unimportant things like FileCore, MessageTrans and, you know, actual operating system code.

And oh god don’t get me started on what the fontmanager does.

If only there was some system for putting things like that somewhere else in memory – compartmentalising workspace in some way. These “areas” could be allocated on demand, “dynamically”.

But NO, Dynamic Areas are bad because “address starvation”… as though having large areas of unused address space are a consolation when MyFirstModule™ has just overwritten the root node of the FontManager’s font list or corrupted that StrongHelp “OS” file you’ve been working on for twenty years.

But, you know… beliefs…

¹ Slight exaggeration.

Is this another useful program the world will only see in screenshots? ;-)

You wonder why important things sometimes get trampled on?

I don’t wonder, I despair in my melancholy despondency. Because what can trample is not another errant module (although that’s possible) but rather can quite easily be a user mode application…

Sorry Rick, I’ve edited that post since you replied.

Is this another useful program the world will only see in screenshots? ;-)

All my programs are useful. Come and help yourself to the contents of my freezer while you’re at it! ;-p

can quite easily be a user mode application…

Because Module A needs to allow the app read/write access to one tiny structure, Module B must have its entire workspace corrupted by the same app.

I’m looking at you, MessageTrans.

How about starting with two RMAs, one for Module code and protected workspace, and one for shared memory (with USR mode)? So OS_Module,6 returns blocks from the public RMA, but any module that is slightly worried can use OS_Module,TBD and allocate blocks from the private RMA (which is inaccessible to USR mode).

But this is defensive programming, which has never been part of the Acorn philosophy.

There’s also good arguments for having different areas for different size things, and in particular getting the FontManager and StrongHelp node soup out of the same RMA that holds actual code.

I don’t see any reason why StrongHelp, FontManager, etc. shouldn’t use dynamic areas.

How about starting with two RMAs, one for Module code and protected workspace, and one for shared memory (with USR mode)? So OS_Module,6 returns blocks from the public RMA, but any module that is slightly worried can use OS_Module,TBD and allocate blocks from the private RMA (which is inaccessible to USR mode).

The “protected” workspace would probably have to be readable (+executable) from user mode, since (a) runnable modules will be entered in user mode, and (b) modules may return pointers to static data to user-mode apps.

“This 30 year old codebase that has been used for billions of man hours has a well documented bug… I must write an entirely new codebase!”

This is true, but I believe there are some architectural decisions in RO5 that are bad practice in the 21st century and make it hard to implement newer technologies. Now things are stuck between moving forward or maintaining backwards compatibility. Is there much sense updating an OS if it breaks 90% of you programs?

It is possible to update the RO5 codebase but a lot of work. I feel it’s simpler to create a more modern OS with a shim to support legacy applications.

.. there are some architectural decisions in RO5 that are bad practice in the 21st century and make it hard to implement newer technologies.

Garbage-collected memory management should be in there at a lower level, not something each application has to implement for itself. RISC OS is a single user OS, quite rightly, but a multi-user design has security benefits. It is only the difficulties of getting the scheduling right that makes some GUIs irritating to use.

Come and help yourself to the contents of my freezer while you’re at it!

Are you going to Open Sauce your fridge?

It is possible to update the RO5 codebase but a lot of work. I feel it’s simpler to create a more modern OS with a shim to support legacy applications.

I feel that doing that would quickly lead people to the “why bother” camp and then they’d just run RPCEmu on top of some kind of Linux.

We have what we have and at some point we’re probably going to break backwards compatability with things from the past but if we chuck out the OS and start again I feel we’ll chuck out whatever it is that makes it special to us.

I feel we’ll chuck out whatever it is that makes it special to us.

Exactly. That’s why I said that such discussions often result in “a weird derivation of Linux”.

Think about it (don’t write, just think). Think of all the “modern” stuff that you believe a modern system should offer. Note how much like Linux (or a Linux-Minix hybrid of you were adventurous) is that?

RISC OS is RISC OS. It isn’t perfect by any means, but it is what it is.

Open Sauce

Oh bravo!

I don’t see any reason why StrongHelp, FontManager, etc. shouldn’t use dynamic areas.

Oh, good. I had you wrong.

“protected” workspace would probably have to be readable

Yes, not writable. Whether that makes a significant contribution to stability is debatable. I suspect that (null pointers aside) it’s bad (and unchecked) SWI parameters that are the cause of most app-triggered corruptions. That’s my experience anyway.

Garbage-collected memory management should be in there at a lower level

Write a module.

I am gearing up to write a dynamically-typed object system – a Common Object Representation. It shall be called COR! I have JSON and PSON parsers written, but I’m planning on unifying the resulting object representation. This is something I’ve done elsewhere.

Sadly, although I considered adding garbage collection too, I don’t think it’s workable as it would require every user of COR objects to provide a root scanning interface. Not happening. It will have to be plain old explicit freeing.

I feel that doing that would quickly lead people to the “why bother” camp and then they’d just run RPCEmu on top of some kind of Linux.

As if!

It is possible to update the RO5 codebase but a lot of work. I feel it’s simpler to create a more modern OS with a shim to support legacy applications.

Define “OS”. Is the kernel the OS? Is the ROM image the OS? Is the ROM + disc image the OS?

i.e. how much do you think needs rewriting?