[NTLK] Simulating the Newton: new version!

Thu Oct 1 00:33:55 PDT 2026

Claude seems to be happily generating NewtonScript right now. My local 
one decided that it wanted to add printer support and it is building a 
custom slip for adding networked HP printers because it says that the 
built in one only supports postscript printers... We'll see how it goes.

It has written a heap of test NewtonScript to run all sorts of things 
while developing the port. Here is a sample of one of the test scripts 
it wrote:

// The boot's splash screen (TNotebook::DrawSplashScreen): the screen
// painted black, the ROM's bootLogoBitmap centred above the bottom 140
// lines, and in white under it "Newton 2.1 (717006)" (VersionString), the
// copyright and "Apple Computer, Inc." / "All rights reserved.".  A boot
// script runs before the system is up (gNewtIsAliveAndWell), so the
// splash is still on the display when this looks at it.
//
//   build/host/host/newton --display 320x480 --headless 60 --script 
src/host/demo/splash.ns
//   build/venv/Scripts/python tools/imaging/pgm2png.py build/splash.pgm 
build/splash.png
//
// Prints "splash: <step>: ok" or "FAILED", and "splash: done".

spFailures := 0;
spCheck := func(what, ok) begin
     if not ok then spFailures := spFailures + 1;
     Print("splash: " & what & (if ok then ": ok" else ": FAILED"));
end;

// the white pixels in a band of rows, every other pixel looked at
spWhite := func(top, bottom) begin
     local n := 0;
     for y := top to bottom by 2 do
         for x := 0 to 319 by 2 do
             if ScreenPixel(x, y) > 128 then n := n + 1;
     n
end;

ScreenSnapshot("build/splash.pgm");
call spCheck with ("the screen is black", ScreenPixel(4, 4) = 0 and 
ScreenPixel(315, 475) = 0 and ScreenPixel(160, 300) = 0);
call spCheck with ("the logo is drawn", call spWhite with (60, 330) > 200);
call spCheck with ("the version and copyright are written", call spWhite 
with (340, 382) > 150);
call spCheck with ("nothing below them", call spWhite with (386, 478) = 0);
call spCheck with ("the version string", 
StrEqual(Gestalt(0x01000003).romVersionString, "2.1 (717006)"));
Print("splash: done, " & spFailures & " failed");
HostQuit();

On 1/10/2026 5:26 pm, Vladislav Korotnev wrote:
> That is great. I want to eventually dabble in self hosting something like
> that one day but cannot really afford the budget for it right now. Fingers
> crossed!
>
> For now, farming Codex accounts for trials is the way :P
>
> I wonder how well do models these days work with NewtonScript? I don't
> expect a lot of it was in training data after all...
>
> // Ak.R.
>
> iOS/Mac/Windows & Web developer
> Vaporwave/ambient producer, sound/video engineer
>
> Genjitsu Gadget Lab Member 001
> http://genjit.su
> <http://www.google.com/url?q=http%3A%2F%2Fgenjit.su&sa=D&sntz=1&usg=AFQjCNE8x1vuqw3E4Hjfe9lA59Q23-z36Q>
>
>
> On Thu, Oct 1, 2026, 16:23 Daniel Parnell <me at danielparnell.com> wrote:
>
>> I have a local instance of Qwen 3.8 flash next running another reverse
>> engineering project and while it is not anywhere near as fast as Claude
>> it is getting the job done.
>>
>> On 1/10/2026 5:20 pm, Vladislav Korotnev wrote:
>>> Well, not for many, only until they finally decide it's time to pay back
>>> their trillion dollar debts and jack up the token prices to their actual
>>> costs :P
>>>
>>> The technology is fascinating, but everything that's hosted on some
>>> megacorps computer far away is bound to be a rug pull by definition,
>> sadly.
>>> Unfortunately though at those rates independent compute just won't be a
>>> thing in the next decade.
>>>
>>> // Ak.R.
>>>
>>> iOS/Mac/Windows & Web developer
>>> Vaporwave/ambient producer, sound/video engineer
>>>
>>> Genjitsu Gadget Lab Member 001
>>> http://genjit.su
>>> <
>> http://www.google.com/url?q=http%3A%2F%2Fgenjit.su&sa=D&sntz=1&usg=AFQjCNE8x1vuqw3E4Hjfe9lA59Q23-z36Q
>>>
>>>
>>> On Thu, Oct 1, 2026, 16:07 Greg Toffer <gregtoffer at hotmail.com> wrote:
>>>
>>>> It is frightening, but it’s also the reality today.
>>>> On the other hand, it means that we can now keep this old devices alive
>>>> for many years, since we don’t need dedicated software (or hardware)
>>>> developers to spend time figure them out, to do what AI can do in an
>> hour
>>>> or less.
>>>>
>>>> Daniele
>>>>
>>>> Il giorno 1 ott 2026, alle ore 06:01, Grant Hutchinson <
>>>> newtontalkmessages at gmail.com> ha scritto:
>>>>
>>>>
>>>>> On Sep 30, 2026, at 9:51 PM, Daniel Parnell <me at danielparnell.com>
>>>> wrote:
>>>>> Hi All,
>>>>>
>>>>> I set Claude running last week to reverse engineer the whole Newton
>> 2x00
>>>> ROM to the point where I can run pretty much any Newton software I want.
>>>>
>>>> I’m sorry if I sound like a curmudgeon, but this is frightening.
>>>>
>>>> g.
>>>> ----------------------------------------------------------------------
>>>>
>>>> https://newtontalk.net
>>>> https://bitbang.social/@newtontalk
>>>> https://bsky.app/profile/newtontalk.bsky.social
>>>>
>>>> ----------------------------------------------------------------------
>>>>
>>>> https://newtontalk.net
>>>> https://bitbang.social/@newtontalk
>>>> https://bsky.app/profile/newtontalk.bsky.social
>>> ----------------------------------------------------------------------
>>>
>>> https://newtontalk.net
>>> https://bitbang.social/@newtontalk
>>> https://bsky.app/profile/newtontalk.bsky.social
>> ----------------------------------------------------------------------
>>
>> https://newtontalk.net
>> https://bitbang.social/@newtontalk
>> https://bsky.app/profile/newtontalk.bsky.social
> ----------------------------------------------------------------------
>
> https://newtontalk.net
> https://bitbang.social/@newtontalk
> https://bsky.app/profile/newtontalk.bsky.social