Hi,

I tried to recompile awayphysics using alchemy. It perfectly compiled and produced the swc file, but it showing error while running the project,

Undefined sym: __Z22btAlignedAllocInternalmi
at Function/<anonymous>()
at Function/<anonymous>()
at awayphysics.collision.shapes::AWPBvhTriangleMeshShape()[F:\RaycastVehicleExperiment\raycast\src\awayphysics\collision\shapes\AWPBvhTriangleMeshShape.as:33]

I tried it from Mac OSX Lion 10.7 and used following command:

g++ -I./ AwayPhysics.c libbulletdynamics.a libbulletcollision.a libbulletmath.a -O3 -Wall -swc -DOSX -o AwayPhysics.swc

adt -version: 3.1.0.4880

If anybody knows about it, please share your thoughts, it would be great learning experience for newbies like me..

Thanks in advance

Out of interest. what did you do to the bullet classes?

First of all, curious about alchemy and the second thing is to experiment around by deriving the bullet classes.

If you knows, can you point out the error here?

Sorry no idea of the error. Plus im not on mac. Im really only just getting into the bullet classes myself, but not had an error yet.
Obvious question does it compile fine without any modification to the bullet classes?

I haven’t made any modifications,  just compiled only.

Downloaded the source from here:
https://github.com/away3d/awayphysics-core-fp11

I’m using flex sdk 4.6. Here I’m attaching the swc file.

Here is some warning and traces i got,

g++ -I./ AwayPhysics.c libbulletdynamics.a libbulletcollision.a libbulletmath.a -O3 -Wall -swc -DOSX -o AwayPhysics.swc
WARNING: Linking two modules of different data layouts!
WARNING: Linking two modules of different target triples!
WARNING: Linking two modules of different data layouts!
WARNING: Linking two modules of different target triples!
WARNING: Linking two modules of different data layouts!
WARNING: Linking two modules of different target triples!
WARNING: While resolving call to function ‘main’ arguments were dropped!
WARNING: While resolving call to function ‘stub_null’ arguments were dropped!
WARNING: While resolving call to function ‘stub_zero’ arguments were dropped!
WARNING: While resolving call to function ‘stub_zero’ arguments were dropped!
WARNING: While resolving call to function ‘stub_zero’ arguments were dropped!
WARNING: While resolving call to function ‘stub_zero’ arguments were dropped!
WARNING: While resolving call to function ‘stub_zero’ arguments were dropped!

2356.achacks.swf, 2727631 bytes written
frame rate: 60
frame count: 1
69 : 4
72 : 2727560
76 : 34
1 : 0
0 : 0
frame rate: 24
frame count: 1
69 : 4
77 : 506
64 : 31
63 : 16
65 : 4
9 : 3
41 : 26
82 : 471
1 : 0
0 : 0
  adding: catalog.xml (deflated 75%)
  adding: library.swf (deflated 77%)

Well the only difference is I dont have the -DOSX I presume that some mac thang.

Its just a case of doing

cd /alechemy/bin/
alc-on

cd ../bulletTestFolder
g++ -I./ AwayPhysics.c libbulletdynamics.a libbulletcollision.a libbulletmath.a -O3 -Wall -swc -o AwayPhysics.swc

As for the errors I get the same apart from
WARNING: Linking two modules of different data layouts!
WARNING: Linking two modules of different target triples!

ps forum hates anything thats not a zip file

I tried using the method you’ve suggested
g++ -I./ AwayPhysics.c libbulletdynamics.a libbulletcollision.a libbulletmath.a -O3 -Wall -swc -o AwayPhysics.swc

But it throws errors related to ‘operator new’ .(for eg:error: ‘operator new’ takes type ‘size_t’ (‘long unsigned int’) as first parameter)

So that i used the solution found here: http://zaalabs.com/2010/04/compiling-zaail-with-alchemy-part-2/

[As per their suggestion i used DOSX. error: ‘operator new’ takes type ‘size_t’ (‘long unsigned int’) as first parameter:To fix this… we need to make a change when g++ gets called, it needs to have the -DOSX flag with it.]

I would like to know, by which platform you’ve successfully compiled it and which flex sdk version you’ve been used?

Flex 4.6.0 win7 64

And yeah that swc doesnt work.

Thanks John, I’ll try it from windows environment.

We had a very rough time trying to get alchemy working to compile Awayphysics some months ago (including getting the same error you are when compiling on OSX). A windows environment with an Intel processor (with AMD alchemy would fail silently) was the only way that we could get it to compile properly.

You are right. It perfectly worked in windows environment with intel processor. I haven’t tried with AMD. Thanks to John.

Does anyone knows about the bullet source code version used by awayphysics?

Anybody knows?

Its in LinearMath folder btscalar.h

#define BT_BULLET_VERSION 278

Yes i got it, Thanks a lot John.