Is RISC OS still a participation sport

Or is it an increasingly-proportioned bunch of seniors doing the ARM Coding Classic Exhibition Tour – without caddies mind you, these are all my own legs – to a diminishing-mobility audience?

“Welcome to the RISC OS OPEN here in sunny Florida where the weather’s fine, the alligators friendly and the plus-fours elasticated.”

increasingly-proportioned

Hey now, leave my predilection for Mars out of it.

the ARM Coding Classic Exhibition Tour

My feeling is that if you’re using assembler in 202x and you aren’t doing some seriously low level stuff…

…don’t.

SWI OS_WriteS
=   "You're not the boss of me":=13:=10:=0
MOV PC,R14

😜

I’ve got to admit to writing less and less low level stuff through the years, and moving to higher level languages. I think its a natural progression to want to get more done, the less time you have left. I certainly have no patience to write a non trivial amount of assembler these days, or use languages where you have to do everything from scratch.

Professionally I started with 6502 assembler for weld testers, 68K assembler for the Airbus 320, C for the A330/340, C++ for ECU programmers and 20+ years of other stuff, a few years of C# for cryptography key management, and the last years of mainly Python for testing F1 ECUs (once I’d ported it all from Perl), video conferencing systems and payment card terminals.

Personal projects started with BBC BASIC on the school’s Beeb, then 6502 assembler to get my Electron to run as fast, moving up to ARM assembler and C on the Archimedes. Had a bit of a regression back to ARM assembler while porting all the 26 bit stuff to 32 bit. Converted some RISC OS stuff from C to C++ and even C# although that never ran natively. These days its Python for fun all the way, and some of it such as my stock price scrapper runs on RISC OS so I can import the data directly into Fireworkz.

I wonder if there is even higher level language I can pick up before I retire.

Interesting – it seems as you get older things you did years ago seemed easy – now they take longer:-(

Lots of people like doing puzzles – so might be a good idea to keep doing some 32bit ARM :-)

There are lots of ways to compare and assess programming languages. Some are good for one job, some for another. Of particular interest, IMHO, is the extent that a language provides a mental picture, necessarily oversimplified, of what is going on inside the computer. To what extent does a language help you to think about algorithms or datatypes? Or, for that matter, help you not to think about some things (e.g. memory management).

I know that E.W.Dijkstra said rude things about BASIC, but he had a point. I have always held that too much concentration on BASIC can blind a person to lots of important ideas and solutions, and that the more languages available to be tried on RISC OS the better, particularly from an educational point of view.

In the early years of Acorn sheer economics made assembler and BASIC (and FORTH, for heaven’s sake) the only possibilities. Now we are no longer limited by the platform. What does limit us is that RISC OS lacks the tools available for Linux and Windows, and that very little free software can be built for RISC OS that would not entail a hopeless amount of work. Lua was a special case, because it was originally commissioned by PetroBras, the Brasilian state oil company, to be available on as many platforms as possible. But most programming languages have been developed to be run in a Unix, and that puts them out of reach of RISC OS. This is a sad state of affairs in my view.

Lots of people like doing puzzles – so might be a good idea to keep doing some 32bit ARM :-)

I write Python programs to solve puzzles!

I notice that I left out all my professional RISC OS projects in C and ARM, which occurred during the C++ years – obviously too much fun to count as work.

Gavin; paragraphs please, otherwise its hard to read across the room on the Raspberry Pi powered bedroom TV.

It will never happen but I would like to see mono on RISC OS, because despite its M$ origins the C# language has some of the cleanest and most straightforward syntax and structure of any fully object orientated language – its everything C++ could and should have been (but obviously not the CLR and most of .NET).

It would really bridge the gap for people who have moved on from BASIC to C and really need to escape pointer hell, but find a lot of C++ and STL syntax too nasty. I particularly like single file classes (no mess of header file dependences), a full range of simple to use built in collection classes, and explicitly nullable variables. Not so keen on all array indexes have to be actual int type.

paragraphs please

Of course. Sorry about that.

I have heard good things said about F#, but I have had no experience of .NET stuff myself.

There is a dearth of runtime systems in RISC OS. With the sad death of Novosad we have lost what possibilities of development that Charm may have offered. Though it has to be said that he originally wrote it (for the Motorola 68000) in the spirit of trying to do everything himself, without standing on the shoulders of anybody else. Not a feasible attitude these days.

druck declared

I wonder if there is even higher level language I can pick up before I retire.

Well hurry up, or it’ll probably have to be a ChatGPT derivative.

a ChatGPT derivative

I thought ChatGPT was generally agreed to be automated plagiarism

I wonder if there is even higher level language I can pick up before I retire.

I imagine death will be my next big “life event”.

Probably only time for a bit more PHP!

I may take this a bit further

It is easy to dismiss ChatGPT as just another in a long line of generators. I remember the hype around “The Last One” from DJ AI Systems. It was able to produce BASIC programs from acript but for anything complex the script was more diffcult to write than writing the BASIC.
But AI has moved on. It is brilliant at language translation and plays a mean game of chess. It would not be suprising if it could write good code also.

But AI has moved on. It is brilliant at language translation and plays a mean game of chess. It would not be suprising if it could write good code also.

Artificial it is, Intelligence it falls short of.
These things are good at replicating existing items with a tweak or two to make it subtly different, and good at stitching together items produced elsewhere.
Innovative content? I suspect that requires that slight spark, that creativeness that dances on the edge of mental stability.

Software engineers are well versed in ‘design patterns’ – there is a great deal of copperplate, obvious cruft and necessary verbosity that such pattern-munchers can automate.

As I’ve observed before, the greatest flaw these systems currently have is that they – like all bad programmers – can’t tell the difference between what they know and what they don’t know. Once they get to the point of being able to say “These 500 lines are fine, but I’m a bit unclear how to represent this concept here…” then it’s time to worry.

So that’s about two years.

There are lots of ways to compare and assess programming languages.

Narrow it down thus:

1, What are you actually trying to write?

2, What is available on your chosen platform?

3, What languages do you actually know?

Sadly, on RISC OS, the functional language choice is rather limited. I often use C because it’s better than BASIC but doesn’t contain all the tedious baggage of assembler.

Once upon a time I might have been up for learning something else (like Rust? That’s sort of C++ with even more weird punctuation) but these days I simply don’t care any more. I’m not going to make money from programming and C works well enough for me for what I am doing.

What annoys me, though, is that the ‘C’ supported by the ESP32 compiler seems to be a somewhat nonstandard version with some little things it throws errors over. Was it the var += 1 notation it didn’t like the last time?
Now, normally it wouldn’t bug me that much, but the compiler is such a heap of dung that either on the PC or my smartphone, you’re likely looking at 3-5 minutes for it to throw an error.
That’s the primary reason I don’t do much with the ESP32. The lethargy of the compiler is unbelievable.

But AI has moved on.

It has gotten better at fooling people with supposed “intelligence”, but that just means that better hardware and implementations mean it can do it’s pattern matching wizardry much more rapidly.
There is no “intelligence” whatsoever in Artificial Intelligence, to the point where the name itself is an oxymoron, because there is no understanding. Without understanding, AI will only be able to combine things it has already encountered in different ways, sometimes utterly freakish ways (Dall-E2’s faces, looking right at you!). However it’s simply complicated sets of associations of what goes with what. It doesn’t understand what it is doing, or why, or who for, or…

I wrote a little bit about this nine years ago (eek!) when Microsoft embarrassed itself. https://heyrick.eu/blog/index.php?diary=20160411

How does the brain invent things – like Atomic drives for space craft or interstellar radio?

A job for AI!

There is no “intelligence” whatsoever in Artificial Intelligence, to the point where the name itself is an oxymoron, because there is no understanding.

I’m not so sure. At my request, ChatGPT translated a Dutch text in which the word SMART is an acronym. Not only did it find proper English equivalents that match the same acronym, it also kept the acronym in compositions, such as SMARTly, that don’t exist in Dutch. It really baffled me. I’d say that if it looks and smells like understanding then hm well…

At my request, ChatGPT translated a Dutch text in which the word SMART is an acronym.

Pattern recognition, pure and simple (well not simple, but it has a lot of material for comparison, and it works through it fast)

As previous: “Artificial it is, Intelligence it falls short of.” and Rick:“There is no “intelligence” whatsoever in Artificial Intelligence, to the point where the name itself is an oxymoron, because there is no understanding.”

Intelligence evolved because it enhances survival. Once a creature can move or otherwise interact with its environment it is on the path to evolving imagination – you need to build an internal model of your environment to cope with threats or optimize your relationship with it. When computers can flinch from our annoyance with them you will know that better AI is on its way.

I think people underestimate just how much human ‘intelligence’ IS pattern recognition, and in particular pattern recognition at all layers of that inscrutable engine we call ‘subconscious’.

To a large extent, consciousness is just a confabulatory self-commentary over whatever the lizard-brain spits out. “Oh they’re just like us” we marvel at chimps or crows or virtually any living creature under relatable circumstances. “Isn’t that amazing?!”. No, it isn’t. They’re running nearly the same software on nearly the same hardware.

It’s like an RPC looking at a Beeb and saying “Ooh, you’d almost think it had an Operating System, like me”.

When humanity goes to the stars, it won’t be humans going.

Oh, and when it comes to ChatGPT, we are very much doomed.

I think people underestimate just how much human ‘intelligence’ IS pattern recognition,

Exactly that.

Like a Human looking at a decomposing whale on beach and thinking if I made something like that I could cross the water.