Newton 2.0x Archemedia Open Beta

[JernejL]

2) on my machine the FPS is always 59.9999

it is possibly locked with vsync to your display framerate

4) a minor thing. please do not use rar. it's a proprietary file format. use 7-zip instead

This isn't the place to go evangelizing opensource formats, rar is in reality a good compression format.

EDIT: where is the documentation?

EDIT2: Where is the NewtonContact structure?

Use the wiki, documentation is there:
http://newtondynamics.com/wiki/index.ph ... =Main_Page

if i got right what you mean by NewtonContact structure - it's not a structure but a "handle", there's some documentation on this on the wiki.

[aqnuep]

This is when I feel a bit sorry that Newton is not open-source. If it would be, then I would contribute to the development of it as Newton being a "one-man-show" makes the development process a bit slow. Please, don't take it personal, I know you do all what you can, Julio, just I would be up to helping you if it would be possible because I strongly believe I have the skills, just don't have the opportunity.
So far, the only way I could contribute to the project is by writing the Wiki documentation of the Newton API, which I actually did almost a year ago.
Julio, is there any way to contribute to the development of Newton?

[Stucuk]

If Newton was Open Source there is no guarantee that there would be faster progress(There is only faster progress if the contributors understand the underlying code and don't introduce lots of bugs). There however would be far more bugs as people who didn't fully understand Julio's design would wrongly modify/add stuff and introduce bugs that Julio would proberly have to fix.

Given that the majority of people on this forum don't seem to have the same Physics knowledge as Julio then i doubt many would be able to contribute, and most of the people who could contribute wouldn't as they have other projects(Like whatever they were using Newton with) that take up there time.

[aqnuep]

Open source wouldn't necessarily mean that anybody else than Julio would have commit rights, however, having access to the source code, I could e.g. add a new feature and send the modified sources to Julio, who can review it and commit it if he thinks the feature is nice and working. This is done so with many open source projects and this way the development would be faster as Julio would not have to implement every single feature on his own, but rather design it, make it implemented by others and finally review it. What is the issue with this approach (besides Julio not wanting to expose his "secrets")?

[Stucuk]

You need to have a very good understanding of physics and physics engines to be able to contribute. We arn't talking about a project that every programmer on the planet could contribute to as most wouldn't have the skills to contribute anything. So unless there is actually people in the Newton community that have the physics knowledge (And understanding of physics engine internals) then having it open source wouldn't do anything but let other physics engines nick all of Julios source code or have someone nick the source code and claim its all there own work.

If people truly have the requirements to help, then i don't see why people can't discuss ideas with Julio and give Julio example code (Pseudo code), etc. From there people may gain access to the source code if Julio trusts them.

[Carli]

If Newton was Open Source there is no guarantee that there would be faster progress

OpenSource has NEVER faster progress. The point of OpenSource is that everyone can follow the developement of the project by checking out the VCS and compiling himself.

You can compile OpenSource by yourself and port it to every platform you want. No OS user is really wants to edit every source code.

You need to have a very good understanding of physics and physics engines to be able to contribute

Julio does a lot of work that dont really needs no understanding of physics: Releasing Newton Versions to 5 Platforms is a messy work that I would automatize with nightly builds.

Julio: You should see the clang project. It is a C/C++ Compiler of the LLVM project that generates platform independend byte code. If You would release that bytecode, everyone could compile the newton releases by itself for his own specific platform. That does not open your souce, it's a Bytecode Level that is made for porting and optimizing programs. It also supports Vectorization.

[JernejL]

While i agree that porting to different platforms would most likely work much faster if Julio had someone to assist him, this is purely Julio's choice, and you should ask him directly (perhaps via a PM).

On a side note: this clang project also looks interesting, it aims compatibility with c++ too so Julio might be able to port newton to it.

[rvangaal]

The problem with bytecode (such as Java) is performance and per-system (/OS) specific optimizations. You'd have to compile the bytecode back again to machine code, which is a bit of an extra route you'd probably not want in a physics engine, where a big part of the success depends on performance.

[Stucuk]

OpenSource has NEVER faster progress. The point of OpenSource is that everyone can follow the developement of the project by checking out the VCS and compiling himself.

You can compile OpenSource by yourself and port it to every platform you want. No OS user is really wants to edit every source code.

The point of Open Source is so that anyone can contribute to the project. Its not purely so people can port to other operating systems(Most Open Source Projects have people actualy contributing rather than porting to other OS's).

When it comes to that hand held platform, Julio stated he would port Newton once the platform has matured.

In any case when it comes to porting to another OS, there is absolutely no need to make the project Open Source. If the creator of the project trusts you with the source code, then you could easily do it for the creator.

[Julio Jerez]

Julio: You should see the clang project. It is a C/C++ Compiler of the LLVM project that generates platform independend byte code. If You would release that bytecode, everyone could compile the newton releases by itself for his own specific platform. That does not open your souce, it's a Bytecode Level that is made for porting and optimizing programs. It also supports Vectorization.

wow that looks very interesting for 2.xx I remember xcode had that as a compiel option but I coudl never get it to work,
how do I get that?
will that work for some one wanting to use newton on platforms like PS3 and XBox360?
PS3 will required, special coding for the SPU stuff, but Netwon 3 is addresing that issue, with better mutithreading code layout

I am thinking to going Open source all the way with newton 3.00, I want to see how newton will work on those platform
and the way I see it teh only solution for me will be going opne source.
I am going over a Clean up process, adding comments, legal headers, and separting files into thire own classes.
Newton is build one library that I have writen for very, very long time, and some files are messing, other are not.

[JoeJ]

Open source would be really great!

At the moment i'm keeping an alternate open source physics engine on board, just for the case if anything needed will be impossible to do in newton or it will not be available on a target console platform.

Also, i think it would help to get the attention you desreve.
Doing physics is the hardest challenge in gamedev., and your work is superior to anything else.

So, a hard decission to open such a secret

[Marc]

Wow, open src newton. That'ld be great. The low busfactor http://en.wikipedia.org/wiki/Bus_factor of newton always scared me a little. + having the src probably helps tracking down issues when people can post stracktraces.

[Julio Jerez]

Ok guys newton 2.33 is no ready
I finally have a decent robust fracturing demo that I beleive people can use eassy, at leat I's like to think that, here is a video


note: the slow down is not the engine or the grahics, it is the video capture that somehow slows down a lot when there is lot of activity on the screen.

[Markus]

That's pretty impressive and looks good. Is it correct that you have to specify the pieces in advance, and that the body will shatter completely, regardless of where and with how much force it is hit? What more do you have planned in this regard?

[Julio Jerez]

In this version yes that is correct. But this is the simplest form of shattering that is supported.
However even that simple demo they do break based of the force that his them but they do no brake locally, it is an all or nothing king of effect.
you can see it when sometime the ball does not hit them with enough force and the simple move away but when they collide with other object they break.
I am preparing the demos where they will break locally and hiearchycally depending where they are hit and depending of the impact force.

all of the funtionality for making the effect is part of Newton, not external tools are required. it is all in latest 2.33 version in google code.