Recent changes to bugs

incorrect behaviour for CEGUI subscribeEvent

Oleg Orel — Mon, 08 Oct 2012 21:09:52 -0000

A problem exists when subscribing (only) for event means that I handled
it in anyway. In other words it doesn't matter what I return from
onKeyDown(for example) True (handled) or False(not handled) CEGUI always
think I handled the event.

Error case is when I want to override some keys in MultilineEditBox,
while I'd like other to keep working as expected they don't work just
because I subscribed for KeyDown.

Investigation showed that:
1) no matter what I return from python handler True/False/None - CEGUI
always get true (I debugged down to the wrapper code)
2) I'm not very strong in py++, but I found (in
code_generators/cegui/hand_made_wrappers.py line 126) the following:

class EventCallback
{
public:

....

bool operator() (const CEGUI::EventArgs &args) const
{
....

else if (dynamic_cast<CEGUI::KeyEventArgs *>((CEGUI::EventArgs
*)&args))
if (mMethod.length() > 0 )
boost::python::call_method<void>(mSubscriber,
mMethod.c_str(),
static_cast<const
CEGUI::KeyEventArgs&>(args) );
else
boost::python::call<void>(mSubscriber,
static_cast<const
CEGUI::KeyEventArgs&>(args) );

....

return true;
}

which rendered to EventSet.pypp.cpp:78 in the generated code.
As you can see it doesn't respect callback's return value because always
returning true and even using boost::python::call with R parametrized
with void...

Reproduced using 1.7.1 rev.1150.
CEGUI 0.7.5
Attached patch (rough, CEGUI native binding is trunk fixed that nicer, but there's a bit different file layout thus no chance to reuse). -p 4

Python-3.x print() SyntaxErrors

tstevense — Wed, 25 Jan 2012 03:10:38 -0000

This file is using python 2.7 syntax for print instead of python 3000 print()

Updates for 1.8

dermont — Sun, 06 Nov 2011 16:57:23 -0000

Main points for each module out with generic updates to 1.8:

ogre
- Move registration code CompositorChain to after Viewport, complains
on module import of no converter for Viewport Listener.

ogreterrain
- patched to allow build on Linux (already on tracker for nearly one year now).
http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=14056

ogrepaging
- updated

ogrertshadersystem
- updated

ogresdksample
- updated ThirdParty/sdksample to include 1.8 directory
- updated environment.py to include above dir and generate_code.py
to copy across files from 1.8 as appropriate
- update environment.py to exlude linking vs Shell/User32 on Linux

ogeral
- Update Singleton to public and variable updates.
http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=14679

ogreoggsound
- updated code generators to exclude OgreOggSoundRecord on Linux
http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=5558&start=45
- added patch for Singleton variable naming (svn code only) - not included in environment.py
http://www.ogre3d.org/addonforums/viewtopic.php?f=19&t=14947

ogrevideoffmpeg
- updated third party code to build against more recent versions of ffmpeg, should still work with older versions(???)
- updated RGB blit code pixel format
- moved inttypes/stdint.h to windows subdir (causes problems on Linux)
- update environment.py/generate_code.py for above

SConstruct
Workaround patch for scons on Linux.
http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=14906

boost
- update to build 1.47 implemented and tested on Linux only.

ois
- update to use 1.3 implemented and tested on Linux only.

demos
- update Demo_Smoke.py to Demo_Smoke_New.py for testing
- update demos/samples Sample_Transparency/Terrain.py testing sdk

Problems:
Still a problem shutting down demos properly and/or problem with SCriptCompilerManager deleting ScriptCompiler on close.

You may want to check the changes to environment.py to ensure they fit in with your build system.

Bullet 2.79 softbody tests

dermont — Thu, 13 Oct 2011 13:43:34 -0000

The following tar file includes some examples from the SoftBody.cpp demo and should cover most of the the following updates.

The btogre module from the following link is required (for DebugDraw support).
http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=14854

1. bt*Shape - getName return string instead of addressof
2. getCollidingObjectsAndPoints -
http://sourceforge.net/tracker/?func=detail&aid=3312809&group_id=186291&atid=916690

myCollisionList = self.dynamicsWorld.getCollidingObjectsAndPoints()
for col in myCollisionList:
i = 0
for pt in col[2]:
b = pt.getPositionWorldOnA()
print i, pt.getDistance() , b.x(),b.y(),b.z()
i+=1
3. btCollisionShape - get/setUserPointer wrapped as get/setUserData as per CollisionObject

4. btSoftBody ctor wrapped as makeSoftBody in main namespace
btSoftBody(btSoftBodyWorldInfo* worldInfo,
int node_count,const btVector3* x,const btScalar* m);

s=4.0
x=[ bullet.btVector3(-s,0,-s),
bullet.btVector3(+s,0,-s),
bullet.btVector3(+s,0,+s),
bullet.btVector3(-s,0,+s)]
m = [0,0,0,1]
psb=bullet.makeSoftBody(pdemo.m_softBodyWorldInfo,4,x,m)

5. btSoftBodyConcaveCollisionAlgorithm and btSoftBodyTriangleCallback now included

6. btAlignedAllocator now wrapped:
- <btSoftBody::SContact,16u>

- btSoftBody::RContact,16u>

- btSoftBody::Anchor,16u>

- btSoftBody::Tetra,16u>

- btSoftBody::Face,16u>

- btSoftBody::Link,16u>

- btSoftBody::Node,16u>

- btSoftBody::Note,16u>

Manually/Ugly Fixes to edit and remove '::_' suffix.
- btSoftBody::ePSolver::_, 16u >

- btSoftBody::eVSolver::_, 16u >

- tPSolverArray
- tVSolverArray

7. btTriangleIndexVertexArray wrapped to accept python lists bullet.makeTriangleIndexVertexArray(), not working yet, don't use.

8. need to update physics/bullet/ __init__.py to include some defines:
import sys

SIMD_2_PI =6.283185307179586232
SIMD_PI = SIMD_2_PI * 0.5
SIMD_HALF_PI = SIMD_2_PI * 0.25
SIMD_RADS_PER_DEG = SIMD_2_PI / 360.0
SIMD_DEGS_PER_RAD = 360.0 / SIMD_2_PI
SIMDSQRT12 = 0.7071067811865475244008443621048490
def btRecipSqrt(x):
return (float((float(1.0)/_bullet_.btSqrt(float(x)))) )
SIMD_EPSILON=sys.float_info.epsilon
SIMD_INFINITY=sys.float_info.max

Updates 2.79 and SoftBody

dermont — Thu, 13 Oct 2011 13:35:40 -0000

See the following link:
http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=14854

Updates to Bullet2.79 and sofybody support - bullet_softbody.patch patched from code_generators/bullet directory. Only tested on Linux.

ogreoggsound fails to compile on Linux

Eliot Eshelman — Mon, 19 Sep 2011 23:59:22 -0000

OgreOggSoundRecorder is only implemented on Windows. It needs to be excluded for Linux builds. Solution here:

http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=5558&p=82664#p82664

Also, the demo Demo_Sound1.py uses the wrong library name when loading OgreOggSound.

Patch is attached.

Ogre Procedural Module

dermont — Sun, 14 Aug 2011 06:31:04 -0000

Patch for Ogre Procedural Module - default branch from:
http://code.google.com/p/ogre-procedural/

http://www.ogre3d.org/addonforums/viewtopic.php?f=3&t=14708

Only tested on Linux and haven't tried patch on fresh python-ogre svn. If there is a problem please let me know.
Also may need to add jpgs from ogre-procedural to demos/procedurial/media. First iteration so probably still needs some work.

"/usr/bin/python" is "/usr/bin/python3" for some platforms

Kibeom Kim — Sat, 02 Jul 2011 04:33:06 -0000

"/usr/bin/python" is "/usr/bin/python3" for some platforms i.e. archlinux.

getCollidingObjectsAndPoints() method crashes Python

Mohammad Akram — Mon, 06 Jun 2011 23:34:17 -0000

The getCollidingObjectsAndPoints() method crashes Python with the following error:

TypeError: No to_python (by-value) converter found for C++ type: class boost::python::list

A test file is attached.

Time Module buggy when using DX9 rendering

Mohican — Fri, 13 May 2011 12:25:22 -0000

Version: 1.7.2
Rendering: Dx9

The module time has function called time.time()
When usign DX9, the function returns the same value all the time.

OpenGL does not have this issue.