The problem with StubsG

Yes indeed. I am an American, and thus accustomed to the ANSI standards, which I do not think have been updated for C in a long time.

October 2018.
And a new version is coming.

@ DavidS

In ANSI? I did not know that, thought that was just ISO.

They adopt the exact same standards these days, so either ANSI C or ISO C means the exact same thing.

Maybe I am looking in the wrong places, though everything I find on C17/18 is that it is ISO. What I am finding with ANSI is that the last update ratified by ANSI was C99.

Actually C99 was already redacted by ISO and ANSI adopted it. The last pure ANSI standard should be C89 while C90, C95 and C99 were all made by ISO and adopted by ANSI. One note is that C89 and C90 are essentially the same language.

All the new ones C11, C17/18 and the very new C2x are all ISO and adopted by ANSI so you can still refer to them as ANSI if you like.

Hope this helps

@ Andrew R.

The reality is that most people here aren’t fussed about people outside of the OS5 space (who aren’t willing to jump on board and update). That’s absolutely fine, but there are enough old users still out there that I can’t ignore them. I think that is why I’m the only one standing up for it. No-one else feels a need to use it because they can justifiably wash their hands of anyone not using 32bit Clibs etc. For me, however, I would look at it differently and say “if my application could run on any of these clibs, why am I forcing anyone to do anything?”. It’s kind of like putting RMensure UtilityModule 5.00 in a !Run file for something that could work on 3.50 otherwise.

Thanks for the clarification of why you need StubG Andrew, and yes it makes total sense from your point of view.

Just one thought: is it really that hard to update SCL on older machines?

The reason of my question is because clearly SCL is moving forward with C18 and more, while StubG (AFAIK) is completely unmaintained right now, so maybe instead of questioning StubG vs modern SCL, wouldn’t it be better to figure out if someone (maybe you) can keep maintaining StubG?

Just a thought…

wouldn’t it be better to figure out if someone (maybe you) can keep maintaining StubG?

I think giving Rick an aneurysm would be bad for the community, overall…

@ Chris Mahoney

I think giving Rick an aneurysm would be bad for the community, overall…

Looool agreed XD let’s save Rick!

I was going to write a technical explanation of just what updating StubsG would involve, but it would be a waste of time me writing it and you reading it. The fact is C code compiled to 32 bit executables with modern compilers can only be guaranteed to work correctly with a contemporary 32 bit SharedCLibrary. If for whatever reason you don’t want to update your old 26 bit SCL, you are also choosing not to run modern 32 bit software.

@ David

I was going to write a technical explanation of just what updating StubsG would involve…

Sure and that would be an interesting reading even for who like many of us already know the issues, but Andrew’s issues are a bit “bigger” than a pure technical perspective. Andrew, on contrary of us, makes his earning from selling RISC OS related products.

Most of the RISC OS Desktop users are in the retro scene right now.

While the (potentially) majority of new RISC OS users are on the RPi side, but a lot (if not the majority) of such users are interested in IoT and embedded applications, retro computers emulation etc… In other words, the RPi market doesn’t necessarily represent a replacement of the old Desktop market for RISC OS.

So it makes sense that Andrew is seeking for a short cut to be able to sell to classic RISC OS users as well as new RISC OS users. This has nothing to do with the technicalities of StubG.

I, like you, am against using StubG, but as Andrew said well, I do not need to earn my living on the RISC OS market, I do software engineering on the cyber security market and mostly on Linux, that’s what pays my salary, so I can comfortably say: Technically StubG is going to be a pain to maintain and to use, moreover 32bit compilers have more and more issues at generating 26bit code, SCL is moving forward and in a really nice way, so why even bother?

If for whatever reason you don’t want to update your old 26 bit SCL, you are also choosing not to run modern 32 bit software.

That is absolutely true and it’s something people in the retro scene need to decide. I personally have moved on and upgraded all my RISC OS 3.xy/4.xy/6, so there is no need for StubG on my Archimedes and RiscPCs, but I can’t speak for everyone.

See? I don’t need saving or explodey-head issues. I just need Druck to say what I said in one sentence rather than two or three paragraphs.

Namely:

If for whatever reason you don’t want to update your old 26 bit SCL, you are also choosing not to run modern 32 bit software.

Exactly this.

that would be an interesting reading even for who like many of us already know the issues

Indeed it would be, and it’s pretty clear where I stand on the topic. ;-)

I can’t help but think that the more the library grows to accommodate updates to C, the more that StubsG will need to handle for itself. At what point do you decide that enough is enough and you’ve actually defeated the point of having a shared library?

On a technical note for regular stubs, there’s room for improvement there too.

For instance, do you need to link in to the extra functions (the other library chunk bits) if they’re not being used?
It still bakes RMEnsure CLib 5.17 in doesn’t it? Really, this ought to adapt to the degree of code being used (C99, 64 bit…) to ensure an appropriate version.

No idea how to make that actually work, so it’s just a wishlist item really.

Most of the RISC OS Desktop users are in the retro scene right now.

Does anybody have any statistics of the diaspora of RISC OS? Are there more people using ancient machines than contemporary?

According to the RISC OS Awards, the majority have five machines of which two are in regular use, the overwhelming majority (90%) use RISC OS 5, and it’s a pretty even split between some sort of Pi and the ARMX6. And, sadly, girls make up 4% of the respondents, though the silver lining is that it’s up 2% compared to last time.
https://www.riscosawards.co.uk/2019/survey.html

In other words, the RPi market doesn’t necessarily represent a replacement of the old Desktop market for RISC OS.

Ah, but why not? Asides from podules and a few applications that never made it into the new world (Aemulor might be useful here), the Pi can do pretty much everything a RiscPC can do, only prettier (bigger display resolution and 16M colours no problem), much faster, using a hell of a lot less power, and with contemporary hardware (plug in a USB flash stick, no quibbles trying to find an IDE drive that works with Acorn’s finicky IDE bus).
My Pi? Sits on the bookshelf. Reachable if I need to poke the reset button. Otherwise it’s a mouse and keyboard and monitor.
That big-ass box on the desk? That’s the PC. Which gets a lot less use exactly because it’s a big-ass box with an equally big-ass power consumption.

For my desktop use of RISC OS, I’m looking at the screen, not the box of magic. As such, a Pi is “like a RiscPC only better in so many ways”.

Note: one can extend the same logic to the other ARM boards, however the Pi offers the ability to put together a functional RISC OS system using salvaged bits (the monitor, mainly) for something like €50-70 (depending on what you want).
Hell, with a PiZero, you might even be able to do it bargain basement for €25-30. But I’d recommend a full sized Pi.
Mine’s à Pi2 (ARMv7 version). Never felt any need to upgrade further as it does all I need it to.

Sure and that would be an interesting reading even for who like many of us already know the issues, but Andrew’s issues are a bit “bigger” than a pure technical perspective. Andrew, on contrary of us, makes his earning from selling RISC OS related products.

Sure, but when you find statements like ‘The reality is that most people here aren’t fussed about people outside of the OS5 space (who aren’t willing to jump on board and update)’, one has to wonder how the non-StubsG-using developers are being portrayed to the customers. The statement itself is demonstrably untrue – as evidenced by the amount of time that many of us spend looking for problems that are reported to us in our software on non-RO5 platforms. But if StubsG is being “sold” as a viable solution to those unwilling to update their SCL to a modern version, what message do they take away about developers not using StubsG? Are we looking at a replay of the vilification of developers by A9home owners?

And why are they unwilling to upgrade, anyway?

@ Steve Fryatt

Sure, but when you find statements like ‘The reality is that most people here aren’t fussed about people outside of the OS5 space (who aren’t willing to jump on board and update)’, one has to wonder how the non-StubsG-using developers are being portrayed to the customers.

Yes the choice of wording sounded a bit harsh, but maybe Andrew was a bit frustrated by the situation, so I am pretty sure there was no intention of any bad portraying. And totally agree with you that the statement is untrue, many of us still care about the old Archies and RiscPCs, guys I still proudly show all my beloved oldies and still use them all :) (just hope Stuart Swales is not reading this or he’ll show up at my door with his chain-sword!!! loool).

But if StubsG is being “sold” as a viable solution to those unwilling to update their SCL to a modern version, what message do they take away about developers not using StubsG?

It should not say anything about developers not using it, c’mon guys I hope we are all past those days of judging people, let’s please create a community where we all can have fun, I think this is a big value that (if we can achieve it) would differentiate RISC OS from all other alienated communities, sometimes the community can do way more than the technology itself.

Are we looking at a replay of the vilification of developers by A9home owners?

Please no more fights, no more vilification, it’s pointless and doesn’t do any good to RISC OS. The past fights and constant arguing only moved people away, that’s really all they have achieved, nothing more.

Anyway just my 0.5c

@ Rick

Does anybody have any statistics of the diaspora of RISC OS? Are there more people using ancient machines than contemporary?

If you try to analyse the amount of public resources on both classic RO and modern RO on the internet you’ll see that there isn’t such a higher grown of resources dedicated to RO 5 compared to RO classic.

2020 seems to have been a good year for RO 5, however the majority of social media have resources dedicated to classic RO and almost nothing to RO 5 (basically even the ROOL twitter account is not that active). The only community around RO 5 that seems to be populated is the Cloverleaf on facebook, but the number of active daily users doesn’t seems that big.

If RO 5 was so massively being used one would expect a grow on public internet resources for it: Youtube, Facebook, twitter, instagram, pinterest etc…

All the professional and semi-professional reviews on Youtube are all for classic RISC OS, with the exception of one or two related to the RPi only and on RPi channels and just as an alternative OS for the RPi. The problem with such resources is that most of the time they present RISC OS as a “weird OS” or an OS that does things in a totally different way than others, so not sure that is helping RO 5 tbh…

Tom of WiFi sheep has made a superb work at presenting how to use RO. I would encourage here everyone with free time to actually start producing more social material for RISC OS, guys even sharing a picture of which game you’re playing today on RISC OS can help, so doesn’t have to be as professional as Tom’s content.

In other words really the world has no idea about RISC OS and if we could do something more close to those new ways of sharing info that probably would help. I still have in my head a review made on youtube from a popular youtuber that stated that RISC OS hasn’t been active for years :(((( Another one reviewing the A3010 said that it wasn’t faster than an Amiga… I mean seriously?!?!? Ok sorry let me stop this here…

So, to conclude this (essay I know I know loool sorry!) one would expect a correspondent grow on internet resources if RO 5 is being used, but such a grown seems not to to be happening while it is happening for classic RO.

On contrary the resurgence of the Amiga for example is quite impressive and the amount of internet resources grow almost daily, not to mention that the Amiga is about to receive the very first binary-translator on a chip project, a quite impressive one and that required some serious effort to produce… Why am I mentioning this on a RO forum, well because the CPU used is an ARM and it binary translate JIT 68030 into ARM code, yup in the chip and you replace the old 68000 with the new ARM chip on a PCB same size of the old 68k… so the Amiga is a bout to become an Archie :D

@ Rick

Ah, but why not? Asides from podules and a few applications that never made it into the new world (Aemulor might be useful here), the Pi can do pretty much everything a RiscPC can do, only prettier (bigger display resolution and 16M colours no problem), much faster, using a hell of a lot less power, and with contemporary hardware (plug in a USB flash stick, no quibbles trying to find an IDE drive that works with Acorn’s finicky IDE bus).

This is a great question! Why not? You are absolutely right, for a RISC OS enthusiast it’s definitely no brainer, Rpi all the way or even better HW! And you forgot to mention thaT the Pi is quieter!!!! :D

BUT (you know it was coming) for a NEW user the issue is RISC OS doesn’t do what they expect an OS to do, so even if there is a big potential Desktop market on the RPi, RISC OS still doesn’t do what we all have listed multiple times and yes some of it is coming with the new browser and the new TCP/IP Stack, so things may change and I can’t wait for this to happen :)

The retro lovers it’s a different story, it’s not about power etc… every time I turn on my Archie or my RiscPC is almost like my father is still here with me, that is something an RPi cannot give me. Every time I play back the old games (included Holed out!!! lol) it’s like is those days again and for those games I want my CRT etc… it’s a different thing and has nothing to do with technology advancement.

But yeah for coding and stuff modern hardware has no rivals and again I am with everyone here asking why is it so hard to update SCL on the old hardware???? :D

The number of RISC OS compatible machines in use is at an overwhelming all time high, the vast majority of these machines having been manufactured in the past nine years. (There are now more Raspberry Pis in existence than all the machines Acorn ever sold.) There simply has to come a point where development breaks away from the past, and takes the most effective route to supporting these machines and attracting new users.

This doesn’t mean that the break has to be at all acrimonious. It just needs to recognize the fact that original Acorn machines are retro hardware which can be readily emulated, and that trying to keep them on board with the latest developments will grow increasingly difficult.

@ Charlotte

There simply has to come a point where development breaks away from the past, and takes the most effective route to supporting these machines and attracting new users.

Totally agreed. I have a strong feeling that what’s happening with RO 5 on the new hardware is that it gets downloaded and tried and when many see it can’t use the whole hardware or it can’t do what a modern OS is “required” to offer they just stop using it.

I may be wrong on that, of course… but if I am wrong then can someone please list all the new stuff done by the millions of RPi users on RISC OS 5? I mean the RPi in 2017 was the 3rd most sold computer of all times in the world (in 2018 they increased their position on that chart), so in theory if many of these users are actively using RO 5 and buying RO 5 software then there should be plenty of resources on the net about this whole new movement of people…

It’s kind of like putting RMensure UtilityModule 5.00 in a !Run file for something that could work on 3.50 otherwise.

It could be worse…

RMEnsure UtilityModule 5.00 Error For f***’s sake, just buy a f***ing Raspberry Pi.

RMEnsure UtilityModule 5.00 Error For f***’s sake, just buy a f***ing Raspberry Pi.

Looool, it’s official I am changing all my RMEnsure UtilityModule 5.00 Error on my !Run files!!!! you are all warned!!! XD

I should add that retro computing definitely has it’s place, and is a potential draw to the world of RISC OS. But are increasingly fraught attempts to retain interoperability between retro and modern hardware really the way to go when emulators are available?

If you try to analyse the amount of public resources on both classic RO and modern RO on the internet you’ll see that there isn’t such a higher grown of resources dedicated to RO 5 compared to RO classic.

I think if you look at the dates, you may well find that a lot of content has to do with the heyday of RISC OS in the latter days of the Acorn era, and the resurgence brought by Select/Adjust.

The market today is a lot smaller than it used to be. In the mid ‘90s Acorn used to rent big halls (like Earl’s Court) which would be full of exhibitors. Nowadays the shows are conference rooms in hotels, a fraction of the size.

That is why available resources are skewed towards “classic” RISC OS. Perhaps a truer picture would emerge if one analysed a little further to discount those things that have not been updated in, say, the last five years.

I’ll give examples of my own: https://heyrick.eu/software/roindex.html and https://heyrick.eu/voyager/

• BudgieMgr – no longer developed, released with sources.
• PCifySite – no longer developed (I don’t even remember this!)
• Teletext – no longer developed, sources released.
• OvHTML – no longer developed (RISC OS and Windows)
• FYEO2 – not mine, I’m just the last one to hack at it, and the JPEG decoding has far too much use of unaligned loads to want to go through and fix it when SwiftJPEG and PrivateEye already do the job.

• QuickVoy – no longer developed
• NewsAgent – not mine, I’m just the last one to hack it. I did look around to see if I could get the sources to Argonet’s configuration module in order to build a stand-alone version of NewsAgent (as I didn’t like how Pluto did things) but somebody else got involved with a different plan (that doesn’t seem to have gone anywhere) and since I’ve pretty much given up on Usenet, I don’t have any need for NewsAgent.
• PreNews/FetchMail/Foxy/NewsFudge/VoyInfo… all connected with Argonet’s suite. I left the UK, and ceased being an Argonet user, in 2002.

My only regret? Losing access to ZFC. I liked that.

At any rate, there’s a lot of stuff kicking around on my site that is “for hysterial raisins” and is being retained simply because it might be useful to somebody. I could clear out all the crap, and I really wouldn’t miss it… but do you remember how many dead links there were in the world the moment Geocities shut down?

The tl;dr version – I don’t agree. I think the large amount of older RISC OS stuff is by virtue of age and a once-much-larger developer base than anything else. Look at what is going on today and I’ll likely tell you the exact opposite.

2020 seems to have been a good year for RO 5

Lots of people stuck at home. What to do? Decisions, decisions… ;-)

Honestly, I think it’s been good since about 2012. We’ve had a good long run riding on the coat-tails of the Pi family. It’s basically been a rebirth.

If RO 5 was so massively being used one would expect a grow on public internet resources for it: Youtube, Facebook, twitter, instagram, pinterest etc…

Yes/no/maybe. A long term complaint is that the RISC OS world has, for a long time shunned this new-fangled technology stuff. It was only 2019 (?) that we had the first streaming camera at a show, after years of people asking. Last year, WiFi Sheep stepped in to make the streaming be more of an event and while there were some hitches, for a first attempt at a long-term live broadcast with insert shows and interviews, it was pretty bloody good.

Now look at certain companies who shall not be named. Another long-term complaint is that everything seems geared towards “announcements at the show”. There’s practically no pre-show hype, and generally the info available online is there as an afterthought (though this is improving due to constant nagging).

The reason why RISC OS doesn’t move in social media is that for all of us, social media is face to face communications, like at shows and such. We’re all rolling like it’s 1996. ;-)

All the professional and semi-professional reviews on Youtube are all for classic RISC OS,

Of course. If you are going to “make money” by doing those sorts of videos, you’re going to review what you can get dirt cheap on eBay. There was one review I watched that had… what was it, an A3010 or something? and was so full of inaccuracies that it shouldn’t have been called a review but more “I have no idea what this is so I’m just going to guess”. Hint – it’s not an Amiga! ;-)

they present RISC OS as a “weird OS” or an OS that does things in a totally different way than others,

Uhh… It is. I think we probably need to communicate what these differences are, and a brief explanation of the logic behind the decisions.

Three I have noticed repeatedly from the few times others have either seen me using RISC OS or tried using it themselves:

• They’re totally floored by the Filer. They are used to some sort of pop-up list of available applications, and a sort of window where the files live… but having it all mixed up like how RISC OS does it blows their mind.
  If you think, even though Windows 3.1 used a very similar interface for applications and files, they were actually separate things; that got even more separated as of W95 where applications hung off the Start menu. I don’t think most typical users of Windows actually know where their application gets installed, it’s just there when it needs started.
  That’s not even getting into the issues of different filing systems. In reality RAM: and ADFS: and SDFS: isn’t that much different to A:, C: and D:.
• They get confused by starting an application and nothing happens. They are used to having a big empty document open in the middle of the screen, and likewise closing the annoying window quits the application. Personally, I absolutely detest this paradigm, but people from over there are used to it.
• Then there are the menus, and the fact that RISC OS has none. Well, it does, if you know the secret incantation. They’re just not wasting lots of screen space…

So, yes, RISC OS is different. Better? Worse? I guess it depends what you’re used to. I grew up with RISC OS, so I’m clearly biased. ;-)

to actually start producing more social material for RISC OS,

I share on my blog, which is heavily slanted towards RISC OS.
I don’t do social media.
Facebook, which I used to have a little-used acocunt with, is an undemocratic perversion.
Never done Twitter, Instagram, or whatever is favourite these days. I have no interest in it. I have no interest in amassing followers (or failing to ☺), nor generously providing personal information for profiling to try to punt more ‘relevant’ advertising that my content blocker would blitz.
The closest I get to social is random stuff on YouTube mostly recorded with my phone. Wednesday, Anna meets snow for the first time (cute!). Today? I think I’ve found what’s wrong with the car (not cute). A running gag is the utter lack of production values, and when I add in stuff using a video editor (on my phone), I tend to lampshade this frequently and/or add captions mocking myself and/or how crappy the video is.

I don’t have a way of recording the RISC OS screen, so I use either a phone or an 8mm video camera (digitised using a Neuros OSD at 640×288). On one video, somebody said in the comments that it looked like it was recorded on a potato. I absolutely loved that. If my ESP32-CAM actually worked (it doesn’t), I have half a mind to fit one into a potato just to make that actually come true.

Given my lack of f***s, obscure sense of humour, constant self-deprecation, and “production values”, I think you might want somebody else to make videos. ;-)

a popular youtuber that stated that RISC OS hasn’t been active for years :((((

Unfortunately YouTube demonstrates quite clearly that any asshole can open their mouth and spout crap. Just look at https://www.youtube.com/user/heyrickblog

almost like my father is still here with me

Ah. Nostalgia. Part of the reason why one of the radio stations I listen to plays stuff from the ‘80s. Harry Styles? Lewis Capaldi? The Weeknd (what happened to the other ’e’?)?
Pah. Give me Depeche Mode, Erasure, Pet Shop Boys, The Cure, Prince, Bowie, Bon Jovi, Bonnie Tyler, and U2 when they still made decent music, Duran Duran, Cyndi Lauper, Dire Straits, and Spandau Ballet. No, I didn’t look that up. I can give you a preferred song from each without stopping to think about it. Mom? She liked “Gold” (mostly the line “you’re indestructible”). I like “Through the Barricades”. It tells a story.
Actually… <tap><tap><tap> Listening to it now.

There are now more Raspberry Pis in existence than all the machines Acorn ever sold.

Uh… That was probably the case in 2012. ;-)

There simply has to come a point where development breaks away from the past […] This doesn’t mean that the break has to be at all acrimonious.

Yes, and yes. But the problem is how to tell users of old machines “For f***’s sake, just buy a f***ing Raspberry Pi.” without making it seem like they’re being picked on because they are still clinging on to the dying embers of Mesolithic hardware.

@ Rick

Thanks for the long and detailed reply, so given that you seems to know well what is going on on the “modern” RISC OS side, I have few questions for you:

1) Do you know what are THE most searched keywords on search engines for RISC OS?

2) Did you have a look at the numbers of replies/views here:

https://www.raspberrypi.org/forums/viewforum.php?f=55

VS the same for other OS running on the Raspberry Pi?

Do you know what are THE most searched keywords on search engines for RISC OS?

No. The Google reports about searches for my site are either this-is-so-obviously-your-homework style questions about ARM, or “how to get extra channels on a Sky Digibox”.
As RISC OS questions don’t figure on my site, I can’t guess the sorts of questions that might be asked other than looking at the sorts of questions people raise here.

the numbers of replies/views here […] VS the same for other OS running on the Raspberry Pi?

Yup. We have one Jeffrey. They have a Linus plus an army of followers and some weirdo bearded bloke in the background.
Not so much David and Goliath as a sole worker ant in a city.

RISC OS definitely needs “deweirding” for newcomers. If someone makes an incorrect assumption about a user interface, they’ve gotten it wrong. If ten people in a row make the same assumption, the user interface has gotten it wrong.

Windows 3.1 used a very similar interface for applications and files

Alas, children born after Windows 3.1 was officially declared obsolete can now legally vote.

But the problem is how to tell users of old machines “For f***’s sake, just buy a f***ing Raspberry Pi.” without making it seem like they’re being picked on because they are still clinging on to the dying embers of Mesolithic hardware.

Given that the hardware traditionalists seem to overlap with the user interface hardliners, can’t some kind of “definitive” edition be declared, after which legacy hardware is officially deprecated? Maybe, given that the RISC OS 5 vs. RISC OS 6 divide is yet another source of baffling weirdness for newcomers, subsequent versions get called RISC OS 7?

…but I fear this thread is turning into every other thread.