What has changed with NewtonGetEulerAngle?

[pjay]

I've just migrated from Newton 2.24 to Newton 3.12 and I've noticed a change in behaviour with NewtonGetEulerAngle.

I've written a very simplistic & antiquated OpenGL renderer in Purebasic for the purposes of quickly prototyping ideas & it uses the Euler rotation angles for calculating shadow volumes. The trouble i'm having is that these angles are 'waving' about a bit in 3.12, causing the shadows being generated at incorrect angles, whereas it was behaving perfectly in 2.24.

What's the reasoning for NewtonGetEulerAngle now requiring two return pointers (*eulerangles0 / *eulerangles1?)

I've uploaded a small windows example which I hope shows this problem better than I've described it:
https://dl.dropboxusercontent.com/u/6000090/3%20-%20Basic%20Newton%20Dynamics%20-%20Spheres.zip)

Newton 2.24:


Newton 3.12:

[Julio Jerez]

first get Newton 3.13 from GitHub, is not longer the current archive

[pjay]

Thanks for the advice - I downloaded & compiled the latest source but the same behaviour is occurring.

I've updated the zip file with the 3.13 dll.

[Julio Jerez]

the reason why no why function take two pointer to two euler angles is because
for each rotation matrix there are always two different set of eulers angle that yet the same rotation
now the function calculate both pairs?
you can always use the angle returned in the same vector.

can you tell me how are you using NewtonGetEulerAngles.
can you post the code fragment?

[pjay]

I think I may have solved it. I compared the numbers that v2.24 & v3.13 were producing it highlighted that the Pitch and Roll were identical, but the Yaw was different (on the *Eulerangles0 numbers).

In the dgMatrix::CalcPitchYawRoll function, I made the following change:

From:
dgFloat32 yaw0 = -dgSin ( matrix[0][2]);
To:
dgFloat32 yaw0 = -dgAsin ( matrix[0][2]);

This now produces the same Yaw value as v2.24. I'm not sure if it's right or wrong, but it's working as I needed it to

[Julio Jerez]

wow, that was a bad bug.
yes your fix is 100% correct.
I fixed and checked in it.

Please sync again
and thank you very much.