The latest SVN bugs thread

You discovered a bug in the engine, and you are sure that it is not a problem of your code? Just post it in here. Please read the bug posting guidelines first.
hendu
Posts: 2600
Joined: Sat Dec 18, 2010 12:53 pm

Re: The latest SVN bugs thread

Post by hendu »

Maximize and restore implemented in https://github.com/clbr/seirr/commit/d7 ... 98c5bc5ad6 .
hybrid
Admin
Posts: 14143
Joined: Wed Apr 19, 2006 9:20 pm
Location: Oldenburg(Oldb), Germany
Contact:

Re: The latest SVN bugs thread

Post by hybrid »

Ah cool, I always wondered what it would take to support the extended WM hints. Is there a patch download link, or can you upload the patch to the tracker?
hendu
Posts: 2600
Joined: Sat Dec 18, 2010 12:53 pm

Re: The latest SVN bugs thread

Post by hendu »

Yes, add ".patch" to the link to get a diff.
hybrid
Admin
Posts: 14143
Joined: Wed Apr 19, 2006 9:20 pm
Location: Oldenburg(Oldb), Germany
Contact:

Re: The latest SVN bugs thread

Post by hybrid »

Thanks, found and applied ;-) Maybe someone can test this, as my Linux environment is too unreliable to make any proper tests.
CuteAlien
Admin
Posts: 9628
Joined: Mon Mar 06, 2006 2:25 pm
Location: Tübingen, Germany
Contact:

Re: The latest SVN bugs thread

Post by CuteAlien »

Great - looks good on my system, minimize and maximize both work now.
IRC: #irrlicht on irc.libera.chat
Code snippet repository: https://github.com/mzeilfelder/irr-playground-micha
Free racer made with Irrlicht: http://www.irrgheist.com/hcraftsource.htm
christianclavet
Posts: 1638
Joined: Mon Apr 30, 2007 3:24 am
Location: Montreal, CANADA
Contact:

Re: The latest SVN bugs thread

Post by christianclavet »

Wow! That was fast! By doing this, you improved a lot the "device" creation on Linux! Thanks!
ent1ty
Competition winner
Posts: 1106
Joined: Sun Nov 08, 2009 11:09 am

Re: The latest SVN bugs thread

Post by ent1ty »

[ogl-es branch]
Sometimes, COGLES2MaterialRenderer::setVertexShaderConstant crashes with a failed core::array assert. Only seems to happen on desktop (archlinux). I'm not actually sure it's not something I'm doing, since I've modified my local copy of Irrlicht in a very slight manner (basically just hendu's suggestion for material type override: http://irrlicht.sourceforge.net/forum/v ... =7&t=45582 , that shouldn't interfere with shader uniforms, right?)

Here is complete call stack:

Code: Select all

#0 0x7ffff6657d67   raise() (/usr/lib/libc.so.6:??)
#1 0x7ffff6659118   abort() (/usr/lib/libc.so.6:??)
#2 0x7ffff6650bdd   __assert_fail_base() (/usr/lib/libc.so.6:??)
#3 0x7ffff6650c92   __assert_fail() (/usr/lib/libc.so.6:??)
#4 0x515436 irr::core::array<irr::video::COGLES2MaterialRenderer::SUniformInfo, irr::core::irrAllocator<irr::video::COGLES2MaterialRenderer::SUniformInfo> >::operator[](this=0x101ff68, index=16877752) (../../include/irrArray.h:316)
#5 0x514bf0 irr::video::COGLES2MaterialRenderer::setPixelShaderConstant(this=0x101ff40, index=16877752, floats=0x7fffffffdee0, count=16) (COGLES2MaterialRenderer.cpp:374)
#6 0x514b72 irr::video::COGLES2MaterialRenderer::setVertexShaderConstant(this=0x101ff40, index=16877752, floats=0x7fffffffdee0, count=16) (COGLES2MaterialRenderer.cpp:364)
#7 0x41d150 ShaderCBDepth::OnSetConstants(this=0x100a700, services=0x101ff48, userData=0) (/home/entity/dev/polygame/jni/ShaderCBDepth.cpp:25)
#8 0x51442a irr::video::COGLES2MaterialRenderer::OnRender(this=0x101ff40, service=0xb96ba0, vtxtype=irr::video::EVT_STANDARD) (COGLES2MaterialRenderer.cpp:152)
#9 0x50dbfb irr::video::COGLES2Driver::setRenderStates3DMode(this=0xb967b0) (COGLES2Driver.cpp:1630)
#10 0x509d65    irr::video::COGLES2Driver::drawVertexPrimitiveList(this=0xb967b0, vertices=0x7ffff7e55010, vertexCount=9216, indexList=0x106a840, primitiveCount=4608, vType=irr::video::EVT_STANDARD, pType=irr::scene::EPT_TRIANGLES, iType=irr::video::EIT_16BIT) (COGLES2Driver.cpp:667)
#11 0x4c4d9a    irr::video::CNullDriver::drawMeshBuffer(this=0xb967b0, mb=0x1045620) (CNullDriver.cpp:1567)
#12 0x69b9af    irr::scene::CMeshSceneNode::render(this=0x103efd0) (CMeshSceneNode.cpp:172)
#13 0x483c39    irr::scene::CSceneManager::drawAll(this=0xfd7a80) (CSceneManager.cpp:1534)
#14 0x407903    EffectRenderer::drawAll(this=0xfe3180) (/home/entity/dev/polygame/jni/EffectRenderer.cpp:42)
#15 0x414235    main(argc=1, argv=0x7fffffffe728) (/home/entity/dev/polygame/jni/main.cpp:87)
And console output:

Code: Select all

Polygame: ../../include/irrArray.h:316: T& irr::core::array<T, TAlloc>::operator[](irr::u32) [with T = irr::video::COGLES2MaterialRenderer::SUniformInfo; TAlloc = irr::core::irrAllocator<irr::video::COGLES2MaterialRenderer::SUniformInfo>; irr::u32 = unsigned int]: Assertion `!(index>=used)' failed.
Code that triggers this behavior:

Code: Select all

 
void ShaderCBDepth::OnSetConstants(video::IMaterialRendererServices* services, s32 userData)
{
    if(!FirstUpdate)
    {
        CamFarID = services->getVertexShaderConstantID("CamFar");
        WorldViewProjMatID = services->getVertexShaderConstantID("WorldViewProjMat");
        WorldViewMatID = services->getVertexShaderConstantID("WorldViewMat");
 
        FirstUpdate = true;
    }
 
    core::matrix4 worldMat = services->getVideoDriver()->getTransform(video::ETS_WORLD);
    core::matrix4 viewMat = services->getVideoDriver()->getTransform(video::ETS_VIEW);
    core::matrix4 projMat = services->getVideoDriver()->getTransform(video::ETS_PROJECTION);
    core::matrix4 worldView = viewMat * worldMat;
    core::matrix4 worldViewProj = projMat * worldView;
 
****services->setVertexShaderConstant(WorldViewProjMatID, worldViewProj.pointer(), 16); /// <== THIS LINE
    services->setVertexShaderConstant(WorldViewMatID, worldView.pointer(), 16);
 
    f32 farDist = Context->Device->getSceneManager()->getActiveCamera()->getFarValue();
    services->setVertexShaderConstant(CamFarID, &farDist, 1);
}
 
Also, I notice method for setting vertex constant just calls the method for pixel constant, is that a good thing?
https://github.com/3ntity/irrlicht-git/ ... r.cpp#L362
irrRenderer 1.0
Height2Normal v. 2.1 - convert height maps to normal maps

Step back! I have a void pointer, and I'm not afraid to use it!
hendu
Posts: 2600
Joined: Sat Dec 18, 2010 12:53 pm

Re: The latest SVN bugs thread

Post by hendu »

The half-life image loader in CNullDriver.cpp has wrong guards. It should have _IRR_COMPILE_WITH_WAL_LOADER_.

If you disable _IRR_COMPILE_WITH_WAL_LOADER_ but have the half life mesh loader on, the link with apps fails currently.
hybrid
Admin
Posts: 14143
Joined: Wed Apr 19, 2006 9:20 pm
Location: Oldenburg(Oldb), Germany
Contact:

Re: The latest SVN bugs thread

Post by hybrid »

Thanks for mentioning, both create*loaders are now under the same define.
hendu
Posts: 2600
Joined: Sat Dec 18, 2010 12:53 pm

Re: The latest SVN bugs thread

Post by hendu »

Numpad keys are broken on linux. They always give home, end, arrow, etc regardless of the numlock state.
stefany
Posts: 58
Joined: Wed Dec 07, 2011 10:50 am

Re: The latest SVN bugs thread

Post by stefany »

On linux, "make install" when a static library is compiled tries to install a dynamic one
zerochen
Posts: 273
Joined: Wed Jan 07, 2009 1:17 am
Location: Germany

Re: The latest SVN bugs thread

Post by zerochen »

@hendu

are you sure that is in the last svn because we fixed that in 4669?

http://sourceforge.net/p/irrlicht/code/4669/

regards
zerochen
hendu
Posts: 2600
Joined: Sat Dec 18, 2010 12:53 pm

Re: The latest SVN bugs thread

Post by hendu »

Yes, using 4908. It's broken in the current code.
CuteAlien
Admin
Posts: 9628
Joined: Mon Mar 06, 2006 2:25 pm
Location: Tübingen, Germany
Contact:

Re: The latest SVN bugs thread

Post by CuteAlien »

Yeah, I can reproduce it here as well.
IRC: #irrlicht on irc.libera.chat
Code snippet repository: https://github.com/mzeilfelder/irr-playground-micha
Free racer made with Irrlicht: http://www.irrgheist.com/hcraftsource.htm
zerochen
Posts: 273
Joined: Wed Jan 07, 2009 1:17 am
Location: Germany

Re: The latest SVN bugs thread

Post by zerochen »

yes you are right.
i guess the state mask is missing in CIrrDeviceLinux::getKeyCode
can anyone test this code?

Code: Select all

 
event.xkey.state &= ~(ControlMask|ShiftMask); // ignore shift-ctrl states for figuring out the key
mp.X11Key = XkbKeycodeToKeysym(XDisplay, event.xkey.keycode, 0, event.xkey.state);
 
XkbKeycodeToKeysym is not very good documented but it should work

regards
zerochen
Post Reply