I admit that I haven't used Dragon lately. It sounds like it has quite improved. But what I meant is how much do you know about the MATHEMATICS of speech recognition. Fourier Transforms and such? I would love to have a voice/speech theme but I won't be doing it myself. It wasn't a put-down but a question about your skill-set.

Bruce, this looks really really interesting... I've followed your posts about this new system of yours with a lot of interest over the past couple months

Any way I can help out with this? I've got some coding experience and I'm always willing to learn new stuff I was actually about to set off to write my own contacts application, but if I can help out with yours instead I'll hold off on that for the moment

That's what the OpenEinstein people (the Newton emulator) are doing: Keeping the OODB structure,but adding hooks for file systems.

In fairness, Newton's print support, while adequate for the time, is not fabulous. It prints fine to PostScript printers (and to networked Linux printers, with a little hacking) but regular print support is more akin to printscreen functionality (i.e. bitmap based)

Here's a layman's view: The Newton's database structure goes deep. Very deep. So deep in fact that data from one application are always available to others. The famous "Lunch with Bill on Friday" Assistant entry shows this: Write this (or something similar) in the assistant window, and the Newton will go searching for Bills in your contact list; next it'll find next friday in your calendar and determine that lunch is 12:00; before you know it an appointment for lunch on July the 25th pops up in your schedule without you having had to open any of the applications involved.

This has all to do with the "pen-centricity" of NewtonOS. Everything - and I mean everything - was designed with the stylus in mind. E.g.: traditional Newton apps don't have top menus, or any traditional menus at all; key options are available through icons at the bottom of the window (where a hovering pen will not obscure the window's view), while most day-to-day interactions are achieved through pen gestures in the main window (scrubbing for deletion e.g.).

One noteable exception to this is the so-called Newtonworks package, a sort of office suite for Newtons. The reason for this is that it was designed for the eMate, a keyboarded classroom Newton, but made available for the stylus based MessagePads. It was not made for stylus and it shows (even HWR had to be hacked into it).

That's where the Newton's database approach really shines: Newtons have only one contacts pool, which is shared by everything. I once designed a stationary (a sort of page template, but more intelligent) for the Notepad (Newton's version of Xournal, with HWR) that had fields for contacts in it. All I really had to do was draw the fields, the OS did the rest.

Bear in mind that Notepad out of the box is just that: a notepad. It has no database built into it: designing the stationary (which was just like using Dia: drawing shapes on the page) I effectively created a new application. I haven't had that much fun since Borland's Objectvision and wasted days thinking up other stationaries to make life more complicated.

I'd say, go buy a Newton, but I guess that would be stretching it a bit. If you ever can get your paws on one, don't hesitate. They're heaps of fun.

Josiah, It would be great to have you work on part of the system. There are a number of things to do. Tell me what you enjoy most; C++? 3D graphics? Are you able to do debugging? You can reply here or PM me.

Karel,
I think that the "deep integration" can be done by modeling data and using the models to interact with local or online data. That way proprietary formats become public and transparent through the map to a model. As This goes on I hope you will add your ideas.

Everyone, please give more ideas or complaints about the status quo. If we design a great system I will build it, and soon! The design should perhaps be a set of widgets or "views" as I call them. If you don't have ideas right now, keep checking back here because as things develop perhaps you will have a thought.

What would the best freaken system in the world look like?

Hi Bruce,

I just wanted to pop my head into here and say hi.
You have an interesting proposition, however this device cannot currently perform adequate 3d rendering (yet).

I sometimes find it heavy going working with a lot of maemo applications.
Too much software comes from desktop roots and do not really "get" the interaction differences on a touch device.

I have been working on user interaction for a while now and am attempting to create an intuitive touch friendly interface.

I have a demo movie of my maemo work and have released both a binary and source to the test application.

http://www.internettablettalk.com/fo...ad.php?t=21259

I would like to hear more about your software ideas and would be interested to see if anything can come of it.

lcuk, I saw your video when you first released it. In fact, I got part of the idea of scrolling in 3D after watching your video. It would be great to combine our ideas as my work is mostly on letting applications interact by wrapping them via a model.

As for the 3D speeds, I got 10-15 fps on a 770 with openGL and very little optimization. It isn't enough for a game, but it is sufficient for a UI. And I get better than that on the N800.

Bruce,
Do you wanna post that opengl code somewhere?
we could have a look and see whats slowing it down.

It would be good to see what we can actually get working, even in software initially.

if you have been following the threads recently would notice that we are making headway towards obtaining hardware OpenGL support (the PowerVR chip).

We have setup a wiki document which is being passed around staff within Nokia and can be used to justify the need for drivers.

Having real 3d software on this device (no matter how slow for now) and a framework able to use it would be of enormous benefit.

Hey Bruce

I can do some basic debugging, and while I'm relatively good with C++, I wouldn't say it's necessarily my strong point. I can however program decently within the language even though it's not my first choice necessarily. I'm far stronger with Java and I'm semi-decent at Python

I'm fairly strong in 3d modeling and graphics Feel free to PM me an assignment and see if you like my work

Anyway, I'd be glad to help out, like I said in any area that needs work to the best of my abilities

That's great news about the PowerVR chip. I've been hoping for that. As far as the speeding up the code goes, a lot can be achieved through the use of "dirty rectangles" rather than redrawing the entire screen everytime as is done now.

Later this year I am going to Open Source the code. For now let me think about when. If I can get a good set of requirements from this forum and a good start on a usable theme then I will feel more comfortable releasing it earlier.

I have made a tentative release schedule for four or five "releases" to larger and larger groups which I hope will happen rapidly. Currently, until a usable themelet is available, it would be very hard to understand how the system/mapping language works without me writing a book. I am hoping that with a basic theme that people can ask "which part of this code does this? Oh, I see how this works." No group theory/number theory required. (I've scared a few people off in the past my explaining the theoretical details too much. In practice it all makes sense.)