• Re: The RISC OS Renaissance project

    From Bob Latham@3:633/10 to All on Sunday, October 18, 2026 08:54:04
    In article <11b1r4a$tqqq$1@druck.eternal-september.org>,
    druck <news@druck.org.uk> wrote:
    On 17/10/2026 20:15, A Rawnsley - ROdevs wrote:

    * Using modern development tools to deliver new features to
    you sooner
    I assume that means AI, in which case no thanks.

    Is that because you fear an AI take over which will stop anyone
    bothering to programme by hand anymore or is there another reason?

    Bob.


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From David Higton@3:633/10 to All on Sunday, October 18, 2026 10:59:10
    In message <11b1r4a$tqqq$1@druck.eternal-september.org>
    druck <news@druck.org.uk> wrote:

    On 17/10/2026 20:15, A Rawnsley - ROdevs wrote:
    * Using modern development tools to deliver new features to you
    sooner
    I assume that means AI, in which case no thanks.

    You have stopped progressing. You have stopped moving forward.

    This is what seems to happen to many people as they get older: they reach
    a stage where they can no longer cope with progress.

    We should all bear in mind that we are seeing the earliest days of AI.
    Think about any other technological innovation from a long time ago,
    and look what happened. Back in my teens, I had a ride in an 1896
    Daimler motor car. When it was new, it was clearly only recently
    descended from a horse-drawn carriage (take off the shafts, put in a
    tiller steering mechanism, graft on a horribly inefficient petrol engine).
    But look how cars developed over the years since then.

    There are still many people who can't cope with the transition to electric cars. I wonder how they ever coped with the much bigger transition from horse-drawn carriages to motor cars. But they did. Eventually.

    AI is already very helpful, when used properly. Like any tool, you need
    to learn how to use it well; if you don't, your job will be a mess and
    you'll blame the tools.

    And, most importantly, AI will improve from its present state in these
    early days.

    David

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Sniffer@3:633/10 to All on Sunday, October 18, 2026 11:33:03
    On 18/10/2026 10:59, David Higton wrote:
    In message <11b1r4a$tqqq$1@druck.eternal-september.org>
    druck <news@druck.org.uk> wrote:

    On 17/10/2026 20:15, A Rawnsley - ROdevs wrote:
    * Using modern development tools to deliver new features to you
    sooner
    I assume that means AI, in which case no thanks.

    You have stopped progressing. You have stopped moving forward.

    This is what seems to happen to many people as they get older: they reach
    a stage where they can no longer cope with progress.

    We should all bear in mind that we are seeing the earliest days of AI.
    Think about any other technological innovation from a long time ago,
    and look what happened. Back in my teens, I had a ride in an 1896
    Daimler motor car. When it was new, it was clearly only recently
    descended from a horse-drawn carriage (take off the shafts, put in a
    tiller steering mechanism, graft on a horribly inefficient petrol engine). But look how cars developed over the years since then.

    There are still many people who can't cope with the transition to electric cars. I wonder how they ever coped with the much bigger transition from horse-drawn carriages to motor cars. But they did. Eventually.

    AI is already very helpful, when used properly. Like any tool, you need
    to learn how to use it well; if you don't, your job will be a mess and
    you'll blame the tools.

    And, most importantly, AI will improve from its present state in these
    early days.

    David

    That was funny, thanks David.

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Stuart@3:633/10 to All on Sunday, October 18, 2026 14:45:03
    In article <11b29fd$12oth$1@druck.eternal-september.org>,
    druck <news@druck.org.uk> wrote:
    No, I'm continuing to progress hand crafting better high integrity code
    than ever, and have the contracts to prove it, thank you very much.

    The big problem is that there aren't enough coders and time to move RISCOS
    to 64 bit.

    The use of AI, overseen by experienced coders, is making 64 bit RISCOS
    possible in a few years rather than a few centuries.

    It's as simple as that.

    If we had a few hundred, or maybe thousand of you, David, I'm sure it
    could be done in a sensible time scale, but unfortunately there is only
    one of you

    --
    Stuart Winsor

    Tools With A Mission
    sending tools across the world
    http://www.twam.co.uk/

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Stuart@3:633/10 to All on Monday, October 19, 2026 08:00:03
    In article <11b3a6m$1fni5$1@druck.eternal-september.org>,
    druck <news@druck.org.uk> wrote:
    On 18/10/2026 15:45, Stuart wrote:
    In article <11b29fd$12oth$1@druck.eternal-september.org>,
    druck <news@druck.org.uk> wrote:
    No, I'm continuing to progress hand crafting better high integrity code
    than ever, and have the contracts to prove it, thank you very much.

    The big problem is that there aren't enough coders and time to move RISCOS to 64 bit.

    The use of AI, overseen by experienced coders, is making 64 bit RISCOS possible in a few years rather than a few centuries.
    It might be able to vibe code up a C version of a 32 bit asm module
    which might just compile for 64 bits, but AI isn't going to be able to
    solve the fundamental problems which stand between where RISC OS is now,
    and any 64 bit future. That needs real developers to get together and
    come to a consensus, and driving some of what few of them remain by polluting the ecosystem unmaintainable slop is not a great start.

    Perhaps you need to talk to Andrew and find out exactly what is going on,
    how it is being worked on, and what the system and plans are.

    --
    Stuart Winsor

    Tools With A Mission
    sending tools across the world
    http://www.twam.co.uk/

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Theo@3:633/10 to All on Monday, October 19, 2026 19:13:25
    David Higton <dave@davehigton.me.uk> wrote:
    In message <11b1r4a$tqqq$1@druck.eternal-september.org>
    druck <news@druck.org.uk> wrote:

    On 17/10/2026 20:15, A Rawnsley - ROdevs wrote:
    * Using modern development tools to deliver new features to you
    sooner
    I assume that means AI, in which case no thanks.

    You have stopped progressing. You have stopped moving forward.

    This is what seems to happen to many people as they get older: they reach
    a stage where they can no longer cope with progress.

    We should all bear in mind that we are seeing the earliest days of AI.
    Think about any other technological innovation from a long time ago,
    and look what happened. Back in my teens, I had a ride in an 1896
    Daimler motor car. When it was new, it was clearly only recently
    descended from a horse-drawn carriage (take off the shafts, put in a
    tiller steering mechanism, graft on a horribly inefficient petrol engine). But look how cars developed over the years since then.

    There are still many people who can't cope with the transition to electric cars. I wonder how they ever coped with the much bigger transition from horse-drawn carriages to motor cars. But they did. Eventually.

    AI is already very helpful, when used properly. Like any tool, you need
    to learn how to use it well; if you don't, your job will be a mess and
    you'll blame the tools.

    And, most importantly, AI will improve from its present state in these
    early days.

    To be honest, I think is a slightly naive view. Which of course is one the
    AI companies are pushing. I end up in a similar place to druck, but via different reasoning:

    The problem is that the AI business model doesn't work. Currently companies are still in the 'your first hit is free' model of drug-pushing. Later on they'll get you hooked and raise their prices. Unlike most software which
    has economies of scale, in which adding another customer comes at a low marginal cost, AI has negative economies of scale - when they add another customer there's a major extra cost to them, one which is not being covered
    by their $20 or $200 a month flat pricing. They're being kept going by VC
    and other money in the hope of being profitable one day (with various kinds
    of market manipulation along the way). The current situation will not
    last: the VCs are going to want a return on their investment.

    RISC OS is highly exposed to this. There is very little money in the
    market, so even a small increase in costs may turn any AI-based endeavour non-viable. If GPU time is a scarce resource, who will pay more - a bunch
    of RISC OS users or a business who is making money every day?

    Second, RISC OS is very low on developers. The risk is of losing more of
    the developer skills we already have, ok let's call that 'cognitive
    surrender'. Once they are gone, and all we're doing is herding bots all
    day, if we can't afford the bots any more there are no humans left to
    continue development. Plus the bot skills may get worse at RISC OS code
    over time (model collapse). Who will invest effort in training up new developers and who will want to learn skills if they can get a bot to do it?

    Third, this 'sugar rush' of new software explodes the codebases we currently aren't maintaining because there are very few maintainers. The explosion
    has meant an increasing amount of fluff - who asked for a 6809 coprocessor
    in RPCEmu? If AI maintenance is going to be around maybe that's ok, but if
    it becomes more difficult then now we're in a worse situation than before. That plus the potential atrophying of developers leads to a negative cycle.

    This is why I only run open AI models on my own hardware, where I know the hardware and electricity requirements and am not at risk of being cut off. Maybe eventually those will be good enough for this kind of software development, but so far they fall short.

    I make something of an exception for development tools... anything to make human development less horrible is appreciated, even if we try not to look inside and don't assume it's going to remain supported forever.

    So my summary is this is a high risk approach and there needs to be a safety net to catch the pieces if it falls apart. Because it could end up leaving things in a worse situation than it started with.

    Not that I have a better suggestion. So perhaps doing-something is better
    than doing-nothing, even if doing-something could turn negative?

    Theo

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Sebastian Barthel@3:633/10 to All on Tuesday, October 20, 2026 11:32:08
    Am Sun, 18 Oct 2026 14:45:03 +0000 schrubte Meister Stuart:

    In article <11b29fd$12oth$1@druck.eternal-september.org>,
    druck <news@druck.org.uk> wrote:
    No, I'm continuing to progress hand crafting better high integrity code
    than ever, and have the contracts to prove it, thank you very much.

    The big problem is that there aren't enough coders and time to move
    RISCOS to 64 bit.

    The use of AI, overseen by experienced coders, is making 64 bit RISCOS possible in a few years rather than a few centuries. ...

    If we had a few hundred, or maybe thousand of you, David, I'm sure it
    could be done in a sensible time scale, but unfortunately there is only
    one of you

    Evenutally You should give him some of the more important subtasks and problems and he can find a good (!) and non-AI solution for them. I'm
    sure there are some more complex modules that can't be handled by AI and
    then there are numerous others, where AI can be helpful at least.

    Since he seems to be an experienced and succesful programmer this could be
    a challenge that makes him happy - much more than driving his AMG V8 ever could.


    --
    humble humble - mumble mumble - tumble tumble - abra ka da bra - swooosch

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Bob Latham@3:633/10 to All on Tuesday, October 20, 2026 20:30:13
    In article <11b8ckq$39e0e$1@druck.eternal-september.org>,
    druck <news@druck.org.uk> wrote:
    On 20/10/2026 12:32, Sebastian Barthel wrote:
    Since he seems to be an experienced and succesful programmer this could be a challenge that makes him happy - much more than driving his AMG V8 ever could.

    I still want to do it, but I wont be bothering if AI has
    got there first.

    I can understand that.

    To pursue the car analogy, while my car is certainly not the latest
    (no interfering 'safety' features such as lane assist, and good
    old knobs instead of a touch screen), RISC OS is a vehicle from a
    bygone age. You have to remember the pedals are in a different
    order, it has a manual choke, you have to flip a switch each time
    you want the windscreen wipers to move, and worse of all it breaks
    down every time you hit a bump in the road. It needs to be driven
    in a very particular way.

    Blunt, but I get that.

    Taking the car analogy even further, if large parts of RISC OS
    given the AI treatment, it might still superficially look the
    same, but it's like ripping out the compact and melodious internal
    combustion engine and replacing it with some massive heavy whiny
    battery contraption, and I have no interest in ever driving
    something like that.

    We went for years with very poor insecure SMB connections and tragic
    char mapping to the PC world. Then ROOL decided to change that and
    they did a great job.

    ROD also decide to tackle the problem but I'm guessing they didn't
    have a pot (however inadequate) of money to pay for it. So they
    engaged AI and provided us all with another tool and that's also
    great.

    The user differences between them are small but each has it's own
    strengths.

    I'm not saying this is ideal for the future who knows, but in this
    case INHO it has worked out extremely well for the RISCOS community
    who have two excellent secure tools where a few short months ago
    there was none. Not only that but the code in each is not the same
    and that is also an advantage as I see it.

    Just my opinion.

    Bob.


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Sebastian Barthel@3:633/10 to All on Tuesday, October 27, 2026 12:57:08
    Am Tue, 20 Oct 2026 19:41:29 +0100 schrieb der Humanoide druck:

    On 20/10/2026 12:32, Sebastian Barthel wrote:
    Since he seems to be an experienced and succesful programmer this could
    be a challenge that makes him happy - much more than driving his AMG V8
    ever could.
    You've not driven a V8 have you?

    No. Dont need to. But I know how it feels to drive in a BIG old classic
    Land Rover; a machine build 'ages' ago.

    Also, I dont know how it feels to drive a "Hellcat", what probably could
    be the "essence" of pure V8. But this thing seems to try to kill its
    driver if he isnt able to manage it. So no - thanx - I'll keep my feets
    on the pedals of V4 ;)

    Trouble is RISC OS it isn't that an enjoyable thing to do in my spare
    time compared to knocking out home automation stuff in Python on Linux.
    RISC OS is development is an a exercise in frustration which needs time
    to get into again (which I never seem to have), I still want to do it,
    but I wont be bothering if AI has got there first.

    OK. Fair enough.

    Otherwise this is something that can eventually be overcome by recent programming activities. But then again one has to do a sharp cut and not
    to try to keep an 64 Bit version backward compatible in a form, that it
    runs on a RiscPC too ... ;)

    In terms of "automation" I always think that RISC OS has/had some good approach in these field(s) but these are not "promoted". It won't become
    a big-fat "home automation" OS ever, but it could do much more letting
    people do their IO tasks in an easy way.

    To pursue the car analogy,... RISC OS is a vehicle from a bygone age.

    Thats what is is. But also what gives RISC OS its special charm. And if
    it goes the way to 64 bit it could pave the way for a fairly modern old- schoolish OS. The only other OSes that (sort of) does this are EmuTOS and SymbOS. But these are for 16 or 8 Bit computers. RISC OS has been "32
    Bit" from the start on.

    wipers to move, and worse of all it breaks down every time you hit a
    bump in the road. It needs to be driven in a very particular way.

    These "break downs" can (sometimes) be overcome by an emulated preemptive multitasking (that exists since 1996 or so) or by starting the application
    via a task window.

    But: since this is a mechnism very very deep in the old engine, no one
    will change it (cooperative multitask) before a stable 64 bit version
    ever exists. (means: Not in the foreseeable future ... ;) )

    Taking the car analogy even further, if large parts of RISC OS given the
    AI treatment, it might still superficially look the same, but it's like ripping out the compact and melodious internal combustion engine and replacing it with some massive heavy whiny battery contraption, and I
    have no interest in ever driving something like that.

    Thats true.

    And no user wants to have it like this.

    The idea - as I understood - is more to install an actual 980 horsepower
    V8 engine seamless/smooth into an old-school computing system in a way
    thats allows to further run the old applications plus some newer stuff
    that could be borrowed from the open-source community (as the network
    stack, usb or open-gl).

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Joerg Niggemeyer@3:633/10 to All on Sunday, November 01, 2026 08:33:08
    In message <11b8ckq$39e0e$1@druck.eternal-september.org>
    druck <news@druck.org.uk> wrote:

    On 20/10/2026 12:32, Sebastian Barthel wrote:
    Since he seems to be an experienced and succesful programmer this could be >> a challenge that makes him happy - much more than driving his AMG V8 ever
    could.
    You've not driven a V8 have you?

    I saw on your linkedIn you owned two RX8 !?

    IMHO the rotary principle was a fuel burner of much more engineering brilliance.

    My father invented a diesel version of it and knewed Wankel personally.

    Since the ignition system of the early engines inside NSU RO80
    by Bosch was a crap and there was lead inside the fuel, many engines
    died, when not realising that the engine was runniong on chamber only,
    when sparkplug was shortend.

    This was start of nuconverter short nucon, inventing a powerful
    DCDC converter ignition system to overcome poor ignition system of Bosch.
    But too late and not with enough money resources -
    as we know your AMG has not rotary version inside.

    There had been plans of MB and a prototype of S-class coupe
    already inside Wankels
    garage with a 4chamber block - I guess you would like to have one ;-)

    The problem is and that what I learned about:

    The best or good ideas alone are not granting any success - quite
    the contrary, the money rules and even blocks the better ideas.

    If in case you can still sell shit to the market and make more profit
    a lot of companies will simply fight any better competitive
    ideas which may end their story of success.

    So I am not sure if very small companies or dev groups shall not
    try assistance with a new AI tool, if in case otherwise they may
    have simply just no chance at all !?

    I did my last dev a pure analouge LED driver for gamma ray environment
    without any AI, but for sure with PSPICE.

    The basic ideas and fantasy may not come from AI or any other
    tool, but computers may help to develope and this already many years
    ago - now it seems we have a new marketing hype to get a lot of money
    from investors naming it AI ;-O

    In my spare time I also interested in horse power but sitting onside single horse power ;-)



    --


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)