[NTLK] Restarting Newton development?!
griffin at nist.gov
Tue Aug 7 11:45:00 EDT 2018
Depending on how how far this is abstracted, maybe we could leverage
some of the input methods available on other platforms. We'll probably
never directly use, say, Google Handwriting Input (pretty darn good),
but maybe our system could be designed to make very easy to port Android
(or other) input methods. Who knows, developers of some older input apps
may let us access the code or open source it. There is pretty decent
implementation of Graffiti for Android. Another alternative: there are
open source HWR projects on GitHub. If our project is open source, there
might be a viable solution out there.
If I recall my history (please correct me!), the Newton's input method
was one of the obstacles to mainstream acceptance. Now a couple of
decades later mobile computing, and hence input methods, are everywhere.
If there were a way to allow new users to pick an input method close to
what they use now, that could be a benefit.
On 08/06/2018 02:56 PM, Forrest wrote:
> Ah, there’s the rub!
> IMHO, a Newton without HWR is, well...not a Newton.
> And without HWR I’m guessing your only real choice of text input would be a physical keyboard (anyone who’s spent any time using the stylus with the online keyboard knows that is not conducive to writing text of any real length). Another hindrance.
> You’d still have the integration of all the various software entities (note I avoided the use of the word “app”, bowing to a recent discussion on the future of computing) and soups, which is another good Newton feature. But again, for me, that would not be enough.
> Just my thoughts, or two copper coins, as Lord Groundhog used to say.
> Sent from my T-Mobile iPhone 6S Plus
>> On Aug 6, 2018, at 11:22 AM, Steven Frank <stevenf at panic.com> wrote:
>>> On Aug 4, 2018, at 4:38 AM, Simon Bell <simonbell at me.com> wrote:
>>> As for help: I think before diving in to the code we should talk some more about what we want exactly, where to focus our efforts so we can set some realistic goals and milestones. At the moment I have no goals with the project, it’s just an interesting challenge.
>> Well, we've talked before about one or more different directions "restarting Newton development" could go. None are particularly easy, and they all have different pros and cons.
>> 1. Develop a toolkit that runs on modern systems that can build and deploy to original Newton hardware or Einstein. (NTX?)
>> 2. Hoist the NewtonOS 2.x runtime onto a new foundation layer that runs on modern systems. Compatible with existing Newton software and development tools. Einstein does this by partial emulation, so it's very true to the original device, but at a performance penalty. Newton Framework is going the route of native recompilation (WINE-like), which would have great performance but involves a lot more reverse-engineering!
>> 3. Use the open source NewtonScript compiler and runtime that we have. Write our own clean-room implementation of the object store, views, and underlying foundation layer with an eye to the original NewtonOS specification / API, but without concern about necessarily being able to run existing Newton packages (or even the NewtonScript portions from the ROM) as-is. Forward looking, but more of a clean slate. May be possible to bring old Newton software forward if source exists, with some work.
>> 4. Create an entirely new thing... a new work/application environment, inspired by Newton (and Self, etc) concepts (possibly even using NewtonScript as a development language), but with a new visual language (color?? haha) and API with present-day internet in mind. Completely blue-sky. One way to think about it might be: if you had a device with existing smartphone / tablet hardware, but got to write your own OS, what would you make? (The "OS" might in fact just be an app running on iOS or Android, but it helps frame the question.)
>> Probably the biggest problem is that you don't get handwriting recognition out of the box with many of these ideas. That's pretty core to the whole Newton experience. :) But as you go down the list to #4, the idea becomes more abstracted from "Newton, the Apple product" to "Newton, the overall concept and good bits", where maybe HWR is not strictly necessary. But it's probably a less compelling outcome for most people on this list.
>> It seems like the big questions to be answered are:
>> - What hardware would you want to run this thing on, whatever it is, and what are the capabilities of that hardwre?
>> - How important is backward compatibility with existing Newton software?
>> - Does it have to be exactly a Newton, or can it be Newton-like?
>> - Legal issues if any original Apple code is preserved
>> I don't really have a suggestion -- just throwing out ideas for discussion!
More information about the NewtonTalk