Irrlicht Engine

Because of the recent discussion about camera rotation and up vectors, and because of my own requirements I wrote a camera animator that controls camera movements by angles.
For a demonstration, I implemented this animator to the Sydney example.

The demonstration animator is controlled by the ...

What happens, when you switch off both light sources, ambient and directional?
And how do you make the shadows?

The first thing:
Which video driver do you use? It seems to me that the software driver as well as the burningsvideo driver seems not to support direct lighting.

The second thing is a little bit tricky and it took me some time to understand this effect:
With the coordinates of your light source ...

Diffuse illumination means that light comes from all space directions on your mesh object. So the brightness of a mesh is completely
independent of normal directions. This is exactly the situation you show in your first image, so your diffuse illumination works pretty well.
To get a brightness that ...

There is an error in your code calculating the texture coordinates:
When you calculate float(x/m_TempSize.Width) and x and m_TempSize.Width are integers, this is an integer division. According to C conventions, the result is the nearest integer below the resultof a float division. In your case this ...

The Maya camera is controlled by its animator CSceneNodeAnimatorCameraMaya that responds to mouse events. A look to the code of the
animateNode method shows that camera position and up vector are overwritten:

pos = translate;
pos.X += nZoom;

pos.rotateXYBy(nRotY, translate);
pos.rotateXZBy ...

I agree completely with you concerning the downward compatibility and I have the following thoughts about this:
If the flag ignoreEmptyTokens is set to true, all is compatible with the old version,
If the flag ignoreEmptyTokens is set to false, the old version didn't work at all (see my post from 10 ...

I think less about the effort to test a flag, but about the user friendliness of the interface. As you mentioned, it is not only hard to find a name, but also to explain its effect to the user.
As I see so far, the only difference between this two split interpretations is an empty string at the end ...

It is the behaviour of split routines I know from other libraries with string manipulation (e.g. Qt), so I expected that this will work in the same way.
I don't know if it makes sense to introduce a new parameter.

I think there was a missunderstanding of the problem, sorry for my bad explanation.
With the new code from trunk I used the following test code:

#include <irrlicht.h>
#include <iostream>

using namespace irr;
using namespace core;

int main(int argc, char *argv[])

{
array<string<c8> > split1 ...

When the flag ignoreEmptyTokens is set to false and the part of the string after the last separator is empty, no empty string is pushed back to the array.
I changed the last three lines of code as follows:

// if ((used - 1) > lastpos)
// changed to:
if (!ignoreEmptyTokens ||(used - 1 ...

The method string::split doesn't work correctly if the flag ignoreEmptyTokens is set to false and there are consecutive separators in a string.

u32 split(container& ret, const T* const c, u32 count=1, bool ignoreEmptyTokens=true, bool keepSeparators=false) const
{
if (!c)
return 0;

const ...