IPhysics - Newton/Irrlicht framework

[Nargil]

try changing it to

#include <iostream>
using namespace std;

[olivehehe_03]

I know, I've used iostream before, I tried changing it and I got little further, but still can't compile the Hello World example. I've gotten all the header problems fixed, just got rid of the ".h", but now I'm getting this error...

1>LINK : fatal error LNK1104: cannot open file 'odbc32.lib'

...but I've gone through the source and can't find where that file is linked. Any ideas?

EDIT: It just occurred to me that these problems might be because I'm using Irrlicht 1.3...has anyone else tried IPhysics with 1.3 yet?

[RapchikProgrammer]

Damn, after killing my self at what the hell was the problem i finally found it! The newton dll is corrupt for some reason!!!!! Download the correct dll by either copying it from the newton sdk or by clicking on the link below! And the car demo has some lines commented out, you can either look at my previous post or click on the link below!

https://sourceforge.net/project/showfil ... _id=501255

BTW do copy the map-20kdm2.pk3, from irrlicht media folder to iphysics media folder!

[olivehehe_03]

Thanks for that dll, that fixed the crashes, and I even found out how to fix that linker error I was getting, I'm using VC++ Express Edition so it doesn't come with the platform SDK and can't find that file, so I'm downloading that now. Thanks alot

[Virion]

I am using dev-c++... can you release the library file for gcc compiler? >_<"

[RapchikProgrammer]

You can download the .a file from below! The zip file, iphysics1.5-irr1.2-gcc.zip file is the one you are looking for! I am not very experienced with anything other than visual studio, so correct me if im wrong but static libraries made using codeblocks do run in dev c++ and the gcc compiler?

https://sourceforge.net/project/showfiles.php?group_id=187859&package_id=219487&release_id=501255

[FlyingIsFun1217]

static libraries made using codeblocks do run in dev c++ and the gcc compiler?

I believe they should, as long as they were compiled w/gcc...

Correct ME if I'm wrong

FlyingIsFun1217

[buhatkj]

well i took a quick stab at trying to convert the iphysics code to work with irrlicht 1.3, simply replaced all the matrix calls and so forth that it bombed on with what i thought was appropriate. I got it to compile, but when i tried to recompile and run the helloworld example using my recompiled iphysics lib, i get a buffer overflow at runtime, so SOMETHING is definitely out of wack. not sure what yet. any of you guys have any luck converting this for 1.3? any ideas?

[RapchikProgrammer]

I have iphysics 1.5 for irrlicht 1.3 under works, but there are so many problems in it that it wud be useless to release it! The biggest of all problems is the character controller! Irrlicht 1.3 has different code to control the camera and it has new fast maths, which has a few bugs like when you normalize a zero vector you will get a vector with x,y,z of value #inf.! And some other problems too which took some time to figure out! Neways, ill try to launch that too!

[buhatkj]

vector with x,y,z of value #inf.!

yeh that sounds like an overeager optimizer. irrlicht is real easy to use, and of course optimization can be a good and necessary thing, but also it's good to spend some time making things more robust. i think maybe that osrt of thing could be made to fail more gracefully, or be accounted for otherwise.

[Frosty Topaz]

First off: RP great work all around. I myself was beginning to get very annoyed trying to integrate physics into a project I'm working on (kind of a 3D Soldat) and was about to break down and just do it all myself. So thank you very much.

Secondly: Once 1.5 for irr 1.3 is done I'd recommend taking some time to improve a few things in your code. None of these are major, but they'll help give users a more consistant interface between Irrlicht and IPhysics.

1 - An irr::physics namespace would be nice to prevent any naming collisions in future.
2 - Consistant use of only Irrlicht types (you seem fine for the complex types but don't always use f32, s32, etc.
3 - More comments. Way more. Please.

Once you get 1.5 playing nice with 1.3 I'd be happy to help you work on this stuff. Once again, great work all around.

[RapchikProgrammer]

Thanx for the nice comments frosty topaz!

1 - An irr::physics namespace would be nice to prevent any naming collisions in future.

IPhysics has no namespace, and i dont think anything has any naming collision either, but if you feel something should be named better, do tell me about it!

2 - Consistant use of only Irrlicht types (you seem fine for the complex types but don't always use f32, s32, etc.

I dont get this completely, so please elaborate here! If you mean i sometimes use int and sometimes s32 and sometimes float and sometimes f32, then that was because i quickly released it without cleaning up the code much!

3 - More comments. Way more. Please.

Totally agree with that, i just released this as soon as i can cause someone requested for a quick peek at what i have done till now, thats why i didnt comment it well! Hopefully i will comment it well in the next release!

[Frosty Topaz]

Mostly the namespace would be to prevent any collisions from occuring. Everything seems to be named fine, but namespaces are there mostly to make sure collisions don't happen between libraries. It's a preventative thing.

And, yes that's exactly what I meant about type usage.

[sio2]

I built with VS2005 Pro after much editing of the code.

VS2005 has more compliance with C++ standards, I believe. VS2005 requires stuff like <iostream.h> renaming as <iostream>. Also cout is part of the std namespace - I needed a "using namespace std" in places. VS2005 does not have "iostream.h" whereas VC6 does have, as well as "iostream" (I just checked as I have both installed).

It took me 10 minutes to debug the crash-on-exit. Firstly, CPhysicsCamera::init() has "SCameraKeyMap* km = new SCameraKeyMap" which should be "SCameraKeyMap* km = new SCameraKeyMap [5]". Plus "setKeyMap(km,4);" should be "setKeyMap(km,5);". Second, CPhysics::close() has "delete[] m_log" and it should be "delete m_log" (non-array format).

Also, is the conversion stuff ("NewtonToIrr" et al) necessary? I've not needed it in the demos I've done.

[RapchikProgrammer]

I love you sio2!!!! I never once noticed them, thanks a lot for pointing those out! Most of the exit bugs are out now, problem just exits in a couple of examples thats all!

The IrrToNewton and NewtonToIrr constants help a lot when working with irrlicht and newton in my opinion as the irrlicht newton is the same as one newton unit multiplied by IrrToNewton! Hence a cube of 1,1,1 in irrlicht is really (1,1,1) * IrrToNewton! Its not necessary but it helps to have a sense of uniform scale between irrlicht objects and newton objects!