Cooperative Multitasking

Ok you are going off the deep. I have never heard anyone on RISC OS mention this.

A reference to newsnet comp.sys.acorn.* I think

@ DavidS

As it was a question of opinion, I gave an opinion.

Actually it was NOT a question of opinions, it was a fact that is happening outside of RISC OS land. For which many companies are investing a lot of money and many many people (way smarter than you and me together) are working on many many hours.

So who’s reading wrong here?

oh dear this is getting so boring now… even this trick :(

Thank you for your support of furthering RISC OS.

This is frankly a childish trick, and it was another used back in the days, “I DavidS am a RISC OS hero/saviour and you are against it” to put one self in a group of “good” people and me in a group of “bad” people… so yes DavidS enough said dude ;)

And btw are you sure on what you’re trying to say above? Are you really saying I am not doing anything to support RISC OS huh? loool :D

@ DavidS

If you are not greatful for the thanks I have, just know it is genuine, no tricks involved.

In this case then I stand corrected and I apologies for my previous comment, your gratefulness is very much appreciated sir.

I do expect you will get some stuff done. As such I am thankful for the help that you are giving to all RISC OS users.

Quite a lot is being worked on, unfortunately I am incredibly slow due to lack of time, I already work many many hours a week as Software Engineer, so I only have the weekends and with my brain fried after a long week of work.

But I am doing everything I can to support RISC OS and the community, included all the possible crazy projects and all the R&D ideas (like yours for the dynamic linking outside of ELF).

All I am asking you to do is please do not try to shut anyones idea or project, some idea that may sound completely crazy to you may actually open the way to even better approaches. R&D should never ever discouraged. We all have our own opinions and I do work (professionally) on Operating Systems development and have achieved quite impressive results in my career, so I speak for personal experience especially on never try to shut any R&D it doesn’t matter how crazy they may sound.

That’s all I am asking to you DavidS.

I just need to say a couple of things:

I use RISC OS , because I like it. I will stay happy as long as I can continue using it. To be honest, if RISC OS gets to the point that it isn’t RISC OS anymore I probably stop using it.
In order to be able to continue to use it , the user base needs to grow, but lets think in the numbers of hundreds. I would say that 100 NEW users would be enough for several years.

About browsers. I still use NetSurf a lot , and I prefers it compared to firefox and chrome. I kind of likes not to have javascript. BUT , I do use other platforms like Android and iOS to get what I cant get from NetSurf. But having such browsers on RISC OS would need some hard changes to the OS and the question I ask to myself , do I want it? As it is now, big apps prefers lots of ram , preemptive multitasking , multicore and a bunch of other stuff. Running those big apps on RISC OS will never be a pleasent experience. I often compare RISC OS to Amiga (like) systems. Amiga is more like unix (but it isn’t unix). It is way easier to code for. But Amiga community is very different from RISC OS. Most Amiga users use it as a “retro” platform and don’t care if there are any modern apps on it. There is a tiny userbase that uses NG variants like OS4, MorphOS and to some extent AROS. The team that does MorphOS is extremely talented, and realistic. MorphOS has been shown running on X86_64 and can still run PPC and M68K software. MorphOS also has the best browser ( a new one this year) , has been able to use html5 youtube since it was available. I have to repeat myself: RISC OS can’t do all things that for example MorphOS (replace with others if you like) does. This maybe sounds like harsh words, but I don’t think they are.

To sum it up:
RISC OS is good for what it is.
If we change it so it can run all wizbang apps that modern world requires it will not be anything of what it is today.

Thogh the rest of the tools in Open Office are already accounted for in more stable programs,

Isn’t it LibreOffice now, due to some acrimonious forking?

We have freely available software that can match Base, Draw, Writer, Calc, Impress…? We have software that performs the same functions, but not necessarily anywhere near the same level.

I would say why not AbiWord for an open source Word Processor

Two reasons. Firstly it was aimed at being a clone of “classic Word” (sort of pre-2005ish) and secondly it is a dying project (there are no more Windows releases due to nobody maintaining it).

And, if somebody (somebodies) should desire taking on something like LibreOffice, well, there’s a lot more besides a word processor there.

(something we do not have)

Ovation? Okay, it’s a basic desktop publisher, but that’s basically a word processor on drugs, right? https://heyrick.eu/blog/index.php?diary=20160723#ovation

Please not something like Alexa, Siri, et al.

Does anybody here use that sort of thing?
I used to ask Siri dumb questions to see what the replies would be, but the entire concept was actually fundamentally broken due to two primary issues:

1. My iPad Mini did not have GPS, nor do I possess an iPhone to register my WiFi AP with a location. So any question that needed a location to be known would be met with “I don’t know where you are”. Such things as asking what tomorrow’s weather would be. And there was no way of saying, “Siri, I live in XXX”.
2. Every single question is an encapsulated entity. That is to say, there is absolutely no context shared from one question to the next. So one could not say “what’s the surface temperature of the sun”, and then say “what’s that in centigrade”.

As a little test, I gave that to the Google thingy on my phone:

Me: Tell me the surface temperature of the sun.
It: According to Wikipedia, the surface temperature of the sun
    is 5,778 kelvin.
Me: What's that in centigrade?
It: 5779 kelvins is 5504.85 degrees celsius.

So far so good, but:

Me: And the surface temperature of Mars?
It: According to space.com, on average the temperature on Mars
    is about minus 80 degrees Fahrenheit.
Me: What's that in centigrade?
It: According to space.com, on average the temperature on Mars
    is about minus 80 degrees Fahrenheit.
Me: What's that in celsius?
It: According to space.com, on average the temperature on Mars
    is about minus 80 degrees Fahrenheit.
Me: (annoyed) What's -80 F in C?
It: -80 degrees Fahrenheit is -62.222 degrees celsius.
Me: What's the average speed of an unladen swallow?
It: (completely missing the joke) About 24 miles per hour
    according to saratoga.com. In the end, it's concluded
    that the airspeed velocity of an unladen swallow is
    about 24 miles per hour or 11 metres per second.
Me: All your base are belong to us.
It: What you say!!
Me: up up down down left right left right b a
It: The Konami Code. I know it well.
Me: How can I blow up this planet?
It: Here's a matching video.
    [links to Neil deGrasse Tyson: How to Blow Up a Planet]
Me: Will RISC OS ever be popular?
It: (complete gibberish due to thinking I said "will recycle
    waste ever be popular"!)
Me: I meant RISC OS, not recycle waste
It: Here's the top search result
    [links to "Iyonix Issues - APDL"]
Me: Can you port Linux to a 6502 processor?
It: According to 6502.org, compared to the Intel x86
    processor normally used to run Linux, a 6502 is
    very limited. It has a smal fixed stack, a slow
    clock speed, small memory addressing range, and
    no support for memory protection.
Me: So that's a no then?
It: I'm trying my best.

So, Google is a bit better, but still a long way to go.

Personally I do not like the MMO crud,

Me neither, but that’s because I generally suck at games.

That said, I can imagine it being a boon to the sorts of people that don’t get out much (or everybody right now thanks to the unforgiving plague) to join in with various like-minded people to go do stuff as a group activity.
People socialise in different ways. Me? Well, I’m writing this… it’s a form of social interaction.

And then they pannic when the internet goes down in a storm, and does not get restored for a week or two.

I think how soon your connection gets restored depends upon how bad the damage was and how important you are a a customer.
It’s one of the reasons I pay a lot to use Orange’s services. I could pay a lot less with an alternative provider such as Free, however I have personally witnessed a woman go without phone or internet for three months as Free said it was an issue with the phone line, and Orange said it was something inside the house. Orange were only interested in fixing it up until the point where the line enters the property. Beyond that point, it was her/Free’s problem.

[Sadly, I diagnosed the problem immediately just by picking up the phone and hearing loud screeching noises – the €5 ADSL filter was broken; it took arguing with both providers three months before somebody came and sorted that out. Three months of paying for a non-functional service, exchanging boxes, paying call-out fees to get it installed correctly… all for a five euro widget]

So, I pay more. Orange will get me sort-of back online within 24 hours with a loaned 4G AP, but I don’t bother as my phone can do that and I have an 80GB/month cap. Generally a damaged/broken phone line is patched within a few days, to be properly fixed within a few months. My broadband is normally 3.5-4.5 megabit (usually on the lower end of that range). I only lost ~0.2 megabit when a severed line was patched up with domestic cable tied to a nearby tree! The reason I knew the cable had been fixed? My speed had dropped to 2 megabit. It seems Orange resets the rate to that after doing work, and it takes a few days of analysis before the exchange knows what the line is capable of and the rate rises up again.

Anyway, a loss of home broadband is annoying, but there are alternatives. I just stood outside (in the rain), downloaded some stuff from Netflix (at something like 30 megabit via 4G), then went back inside and watched the movies.

Maybe a reminder that it is also important to have local copies of there data on there computers.

Oh, I’m not sold on the cloud. I can see how it can be useful, I do use Google Docs, but I’m also aware that once stuff is uploaded “to the cloud”, it is no longer in your control. Do you really believe that those photos of the drunken office party that the phone automatically synced with Google are encrypted in a way that Google cannot interpret, analyse, scan, and perform biometrics upon? And do we really believe that there’s absolutely no way any of this makes it to the United States where it’s ripe for pilfering due to numerous abuses of the Patriot Act and the like?

Nah. Keep the stuff local. Make your own backups. Then you know where they are and who has seen them.

That contridicts everything else you said in this post, as most of it makes life harder, definitely not better.

It depends. A lot of people believe technology makes life easier, when in reality it is more like swapping one set of problems for a different set of problems.

A nice example is the idea of the paperless office. Is there such a thing?

but I wouldn’t rely on Wikipedia

It’s great for throwaway factoids. For actual research, it’s probably best considered a pointer to where to look for actual answers.

one set of, usually rather good, graphics

There was something on demo in a nearby supermarket. No idea what it was, some cute redhead girl. With frizzy hair. That blinked every so often. And was moving gently due to breathing. If it wasn’t for the almost cartoony colours, it damn near could have been a live video of a random person standing in impressive scenery.

Work your way down the list of things in the Cloverleaf kickstarter and see if you can think of things to add.

Probably more useful than reinventing a wheel, yes.

but it had nothing on the “Acorn Tax”.

Don’t think I’m supporting the excessive pricing of Acorn kit (or Apple), but the main difference between both of those and “PCs” is that much of the work is custom built stuff designed in-house. In Acorn’s case, the processor and video system and memory controller and I/O controller were all custom built, on a completely custom motherboard, in a completely custom box, with custom keyboard and mouse, and a custom operating system to run on it.

A PC, on the other hand… well, weren’t the Archimedes family more or less concurrent with the 386 era? That’d be when companies spat out motherboards almost monthly, put generic parts in them, with a generic (AMI) BIOS, stuck ‘em in a generic beige box with a PS/2 keyboard and maybe (or maybe not) mouse, and shipped it with a third party operating system.
If all that a company is really adding to the equation is the motherboard, then it’s much easier to keep costs down.

But now you can buy a machine that makes a typical RiscPC look like the Electron in comparison1 for pocket money.

Yeah, I keep saying that, but there are people who are too attached to their ancient RiscPCs to let go.

Oh, and the browser – would that be “fast on a Beagle-BM” or “fast on a Pi4”?

“Reasonable” on a processor of around 1GHz or so. But more than speed, competence. It needs to handle the sorts of stuff that people will want to use without crawling, stopping entirely, or crashing.
Safari was usable on a 1GHz dual-A9 in the iPad Mini.
My freebie tablet runs a 1.2GHz (IIRC) quad core processor, and Chrome is usable. Firefox takes a couple of minutes to get going, but it has a lot of content blocking scripts running in the background. It’s not fast, but I can use the web version of Google Docs on it if I’m patient.

To Rick’s list one might also add the facility to upgrade both OS and software without fuss.

Still running 5.23.
It works.
I found 5.25 broke some things in weird ways.

So, I generally prefer to stick with what I know works unless there’s a convincing reason to upgrade.
See also https://heyrick.eu/blog/index.php?diary=20201128 a little over halfway down the page for how my bank forced an upgrade to their app, which is now completely non-functional on my phone…

Security that isn’t equivalent to a bullion depository using an honesty box.

<spits tea clear across the room>
:-)
:-)

The question there is whether Linux on the same hardware can manage that

Sort of yes. Try some browsers on equivalent Android devices. It’s Linux at heart, and ARM, so ought to be a reasonable idea of what ought to be possible.

That would be integrating native access to certain hardware features that have never been looked at before

This. My original Pi1 (and the Zero) have no problem spitting out HD video with overlaid subs using xbmc, osmc, or whatever the hell the thing is called these days. The processor barely does any work (evidenced by the rather laggy UI) and when running flat out the machine rises by a degree or two in temperature.
Now, what spec RISC OS machine is necessary to deal with HD? How much processor does it use?
There’s the answer to that question.

Netflix and TrustZone et al can come later. Once we’re capable of offloading video where it belongs.

That just comes down getting permission to include a few applications as standard that for now are extras.

What opens ODT?

or because a weird news was spread uncontrolled

Probably one of the reasons to unplug. Sometimes it seems to me that mass social media isn’t so much about getting people together as increasing the level of stupid in the population.

I mean, how many people really still think Trump won and it’s all a big conspiracy? Just because a lunatic crybaby shouts it does not make it so, as the courts keep reminding him.

yes people in this forum keep calling this with the wrong term: stability, the correct one in engineering is tolerance to faults

Stability is the correct term if you liken the system to a house of cards. And then shake the table. It’s not so much a “tolerance to faults” as it is “stability”. The end result in both cases? About the same. ;-)

I am not against RISC OS on Linux, let’s see how it goes.

Contentious. I’d rather dislike that, as I have a personal hatred of the filesystem that I view as clumsy and asinine. Oh, sure, there is some pretty impressive technology underpinning the filesystem, but everything hanging of the root directory (including stuff that isn’t even a file, or a device) where it is still case sensitive? Grr…

I am not against RISC OS kernel replacement with a microkernel.

I don’t really know enough to express an opinion here. If the end result looks and feels like RISC OS, and is easy to program (including accessing hardware) then that’s good enough.

then I’ll start a set of video for Youtube to show how easy is to hack RISC OS

Just view my blog from time to time. I made the ROM image writable for a laugh. Just to do it. I did not finish it (needs to clear cache or something as it sometimes crashes) but it’s all done as a BASIC/assembler program. Now, if the ROM is easily remapped as read/write, what sort of nefarious tricks could follow on from that?
Want to rip the entire Wimp slot and working memory from another application? There’s a call for that built right in – Wimp_TransferBlock.

Refer to Charlotte’s honesty box.

and even keep the malware hidden

Depends upon how hidden you like. It’s not hard to have active code somewhere in memory that is neither a program nor a module but has its tentacles extended. It’s also not hard to make apps vanish from the task listing in Switcher. All by carefully using the completely legitimate current APIs, no weird freaky hacks necessary.

offers another layer of compatibility with the rest of the world, so to me it’s welcome, but at the same time I do not want to lose AIF

This is exactly it. ELF has its place and its benefits, but so does the existing formats. It really shouldn’t be an either/or.

build an Alexa that doesn’t spy on people on RISC OS

It will be an interesting project for a year or two. Then it’ll either die out due to lack of compatibility (remember, there are powerful back ends to these “assistants”) or it’ll be popular and somebody will spaff a lot of money to buy it, and suddenly it’ll be spyware.

The enticement is just too great. Not even Linux was immune – Ubuntu’s spyware search. Currently disabled thanks to public shoutiness, but it’s still there. And, of course, the fact that they did it in the first place.

If so I may have to install a news reader, and find a free nntp server to take a look at what has been said in the last decade.

If it holds articles that old, that is. Um… If you can get hold of a PC or smartphone or run some sort of Linux, you might find Google Groups to be a better archive?

I look from time to time. Don’t think I’m missing anything…

May I recommend looking at the various WebKit web browser ports to Amiga OS?

Uh… Why?
SDFS::RISCOSPi.$.!Boot.Resources.!SharedLibs.lib.abi-2/0.libWebKit1/so/1/0/0

there are people who are too attached to their ancient RiscPCs to let go

Let it go! I think that I will have to post a video of me chainsawing my RISC PC.

Archimedes family more or less concurrent with the 386 era

At Colton Software in 1989, we had a damn expensive 16MHz 386. £1000 per MB SRAM. And you thought Archies were costly! Perhaps in comparison to Amiga.

@ DavidS

As you are looking at promoting software development by others. May I recommend looking at the various WebKit web browser ports to Amiga OS?

David I do actually still code on the Amiga and also Atari ST (I am lucky owner of a Falcon), now you actually make a good point here and the reason I mention this is to point out the thing that makes porting to RISC OS most difficult: the Multi-Tasking model.

Now that is also the topics of this thread btw. So far the thing that makes porting to RISC OS most difficult is that RISC OS uses Cooperative Multi-tasking while Amiga OS uses pre-emptive (the exec you’ve mentioned on another thread). This has consequences on the application architecture and the way it is being written for the Amiga, Linux, macOS and Windows.

RISC OS native and very responsive software is almost a piece of art balancing between the application intrinsic data-processing model and algorithms and the “give back control to the WIMP” part. This is also another reason why otter browser and others do not perform well on RISC OS (probably way more than the shared libs model).

To make it easier to port such applications that have no assumption of a multi-task environment like the AmigaOS ones (as well as the Linux, macOS and Windows ones so pretty much everything!) we could make few choices here (I am only listing them, not pushing for any of them in particular):

• Remove the kernel pieces that are right no win the WIMP (TaskManager for example and memory mapping trickeries) and put them in an Exec like module and re-write the WIMP so that it provides only UI features.
  • Phase 1 just put things in a new Module and either keep it unplugged in production or not even in RISC OS distribution until it’s done
  • Phase 2 remove the code from the WIMP and start testing it with the new TaskScheduler separate module
  • Phase 3 improve the new TaskScheduler to support more scheduling algorithms for example:
    • Single Task scheduling (for backward compatibility with very old apps and approaches)
    • Cooperative multitasking (for backward compatibility with old WIMP apps and approaches)
    • Introduce experimental Pre-emptive multitask (not in production until RISC OS kernel is fully reentrant)
• Use the SMP library from Jeffrey to introduce forced task scheduling
  • This could be done enabling SMP also on the main core and use it in conjunction with IRQs, every x ms WIMP has to get back control no matter what (in some ways this is a similar approach to WIMP 2, although if by being in the the RISC OS kernel could help stability by disabling it during non reentrant functions)
• Keep working hard to convert code that is not cooperative to be cooperative like it’s happening right now

The above for the multi-tasking issue, then on the AmigaOS we have few more issue to deal with:

• AmigaOS never really used Virtual memory and/or MMU
  • So that means AmigaOS apps access libraries directly in a very 8bit computers fashion, basically put the library base address in the 68K A6 reg and then calculate the function entry point offset.
  • The above is not too bad when apps are written in C or other languages and source is available, but when in ASM it’s an interesting journey looool :D
  • P.S. I wasn’t one of the user lusting for having AmigaOS feature sin RISC OS ;)

• The final bit the obvious differences between WIMP and Intuition (the AmigaOS UI)

But, besides of the above, sure I can write some tutorial on how to convert Amiga OS C apps to RISC OS, not a biggie for me, I’ll add it to my list of TO DOs for RISC OS :)

@ Stuart

Let it go! I think that I will have to post a video of me chainsawing my RISC PC.

loooool :D give me your Youtube channel I won’t miss this for anything on earth… but yet I will cry when you’ll do that!

Acorn Archimedes A3020 $899.85.

This, this was the mistake… 899 vs 399 of an Amiga 1200, this is also why the Amiga has way more games than an Archie could ever possibly dream of. It’s not in the performance or the capabilities. Archie has always been way better than any Amiga technically speaking, but the price precluded is mass success.

Now karma inverted things! A Rpi is way, way, wayyyyy cheaper than any Amiga (old and new) and so the descendant of the A3010 is basically in every house, while the Amiga is just a nice market.

But, besides of the above, sure I can write some tutorial on how to convert Amiga OS C apps to RISC OS, not a biggie for me, I’ll add it to my list of TO DOs for RISC OS :

Intuition to WIMP? MUI to WIMP? I would like to see the last one. MUI is a great toolkit.

Are there still new updates on that side?

Yes David, still activity.
You can check via google groups

RISC OS is good for what it is.
If we change it so it can run all wizbang apps that modern world requires it will not be anything of what it is today.

Maybe, maybe not.
Maybe it doesn’t need to change fundamentally in the aspects you like.

@ Michael Grunditz

Intuition to WIMP? MUI to WIMP? I would like to see the last one. MUI is a great toolkit.

Yes Intuition to WIMP (that’s the C API), MUI is OOP and not 100% sure if it’s finally OSS, but I totally agree with you that porting MUI to WIMP would be great. What do you think about Zune???

We could start porting the basic new UI objects to toolbox gadgets and then convert the C++ API on top of it… I am thinking… (and no! Not committing to anything until I have something working!)

@Paolo
Zune is ok. I ported SimpleMail to AROS. In Intuition you can have “system friendly” drawing to a window , getting a raster from it. That is something I would like to have in WIMP. Doing a sprite isn’t really connected to a window, just painted over it.

A nice example is the idea of the paperless office. Is there such a thing?

They’ve been making those noises in our hospital for a decade or more.
Every “paperless” system installation included the installation of a number of printers. My standing joke was to greet every “paperless” system project with a comment of “how many printers do we need to buy?”

Then they let a some techies write the spec for a managed print system while another set of techies started writing e-forms and set a “use by date” on the paper version. Show the bean counters the cost of printouts vs. forms. Don’t tell the bean counters that their printers are on the kill list too.

A nice example is the idea of the paperless office. Is there such a thing?

I actually managed to achieve that back in 2003/2004, with a framework I have called back then Virtual Site Server, which was C++ ISAPI on Windows 2000 Plus customisation in Classic ASP (that’s not because of my personal choice).

Anyway on top of that framework I have created all the modules required for a couple of Italian public offices to fully go paper-less. Even regular mails were scanned automatically and with scanners that supported multi-feed. Scanning was transformed in PDF and even added digital signature support and full workflow support similar to outlook email of the time where the document was processed and pushed from one office to the other automatically (pre-defined workflow)

Everything used a M$ SQLServer backend which covered pretty much everything from the data itself to the workflow configuration so it could be scaled up on multiple servers for bigger structures.

The framework was so general purpose that could be used from the above till building just websites or ecommerce sites.

To interface with the scanners back then it used (well had to use) ActiveX and we even sold it to a museum when it was also interfaced with the ticket machine (again via activeX plus local custom drivers.

For the AJAX approach, which did not exists at the time, I had to use Java and Java Applets, this is when I have got in contact with java the first time (it was a nightmare, but it worked in the end)

It did sold a lot back then, after that I moved to Virtual Machine coding…

So I do not see an Acorn Tax, they were among the lower costing.

It got worse towards the end. When I jumped ship in late 1997, Wintel PCs were much better value. (It wasn’t Acorn’s fault, but the explanations mean little when you’re after a computer but have a limited budget.)

I use RISC OS , because I like it. I will stay happy as long as I can continue using it. To be honest, if RISC OS gets to the point that it isn’t RISC OS anymore I probably stop using it.

That’s the huge dilemma at the centre of all of this. What are the fundamental things that make RISC OS what it is, and how many changes can be made until it ceases to be¹? Personally, I favour a “who cares what’s under the bonnet, so long as it drives the same” approach (see my somewhat dewy eyed HelenOS thread²), but obviously not everyone will see things the same way.

¹ Is it truly RISC OS if your app can’t trash the filesystem with a single line of errant code?

² An advantage to flexible and highly modular operating systems is that you can easily replace a boring filesystem handler with one that provides the flexibility to trash the filesystem with a single line of code, should you prefer things that way.

@ Rick

I am not against RISC OS kernel replacement with a microkernel.

I don’t really know enough to express an opinion here. If the end result looks and feels like RISC OS, and is easy to program (including accessing hardware) then that’s good enough.

So again to actually even think of something like that people should first “clean up” the WIMP from the kernel’s pieces. Otherwise things will go terribly wrong unless such micro-kernel is kept single task.

Anyway, assuming that the above will be taken care of, then hypothetically yes it is possible to create a RISC OS micro-kernel based that look identical to the one we use now and behaves very similarly on certain aspects and obviously differently in others.

In the end most apps access an API. As long as such API is respected then the App would be mostly happy, however for code that “miss-behave” or doesn’t respect the rules established for the compatibility layer then things will most likely go wrong (surely for the app in question and possibly for the OS depending on how tolerant it is to such faults).

The only thing that I am not sure of is performance, but that’s a different story…

yes people in this forum keep calling this with the wrong term: stability, the correct one in engineering is tolerance to faults

Stability is the correct term if you liken the system to a house of cards. And then shake the table. It’s not so much a “tolerance to faults” as it is “stability”. The end result in both cases? About the same. ;-)

Sure one could see it this way, no argument on that. But in 90% of the cases people just want the OS to “recover” from an App fault and keep working fine with the others. So in those cases really the term should be tolerance to faults. But again you’re correct when you refer to the fact that the OS can just go wrong by itself and in that case stability is the correct term, as yeah It does appear still unstable in certain conditions (even 5.28 which so far is the most stable I have seen in a long time).

So the only thing I would argue in such a comment (and more than arguing it’s just a question) what would it be the priority on both according to the community?

That’s the huge dilemma at the centre of all of this. What are the fundamental things that make RISC OS what it is, and how many changes can be made until it ceases to be1?

Agreed, if we could figure out even statistically what people in the community think and then see what can be done from there keeping in mind we’d want new users to join so the final product should support also all the trendy things of these days and possibly offer the opportunity to keep following the always changing trends over the years…

As I said previously and in multiple occasions I do not mind what is the path forward as long as it’s not an existing OS just with RISC OS UI on top because that (btw) already exists, has been tried and didn’t really get successful anyway. But again this is my o.5c

Guys question to everyone, would it help if someone collects all the options proposed in a pool and we just ask people to vote for one or multiple options?

I think that I will have to post a video of me chainsawing my RISC PC.

Didn’t… God what was his name… Stan Boland? Guy in charge of Acorn at the end, didn’t he do something silly with a chainsaw?

Open Source?

Source available. It’s a bit messy (APDL owns the rights, I believe…I got permission from Aaron to distribute my builds of Ovation; along with sources in the spirit of the original source release)

directly play Youtube etc. And that is play Youtube usefully, full software renderer

We have a full software renderer. A port of mplayer. It isn’t so useful except for small video sizes. Porting something from somewhere else isn’t going to magically make 720p H.264 (generic video for 2020) suddenly work. We need to do that one properly and toss it over to the GPU. Kind of what it’s there for.

But in 90% of the cases people just want the OS to “recover” from an App fault and keep working fine with the others.

Only 90%? The other 10% are disaster fetishists who are like “Oh my god it’s crashed and taken three hours of work with it, I’ll never write the same thing again. AWESONE !!!!” ?

But yes, application problems should affect the application. And only the application.

we just ask people to vote for one or multiple options?

Yeah, uh, isn’t that the logic that brought us the gift that keeps on giving? ¹

¹ Brexit 💩

We have a full software renderer. A port of mplayer. It isn’t so useful except for small video sizes. Porting something from somewhere else isn’t going to magically make 720p H.264 (generic video for 2020) suddenly work. We need to do that one properly and toss it over to the GPU. Kind of what it’s there for.

720p H.264 plays fine on Titanium, Igepv5 and Rpi 4. Rpi4 will do 1080p as long as the bit rate isn’t too high.

God what was his name…

AFAICR it was Peter Bondar, and I think he took an axe to a Win PC at one of the Acorn shows. The pic appeared in some of the Acorn mags at the time.