Your patch pass the tests. Good. I think my additionnal tests are enought for now. They prove really well what it happens for this function.
I let you send your patch. I will send the tests when your patch is committed. Can you use j++ instead of ++j. Like this, it will be as the other loops for spherical harmonics part. Nozomi. On 17.10.2012 08:35, Nozomi Kodama wrote: > My bad. > > My patch pass the tests but not yours. Can you check these facts? > > Nozomi Oups, I didn't respect the order while setting the output values. Splitting the loop and starting with the smallest index should make the behavior exactly the same as yours (I think). Something like the attached should fix it, please have a look. Though we may use a test for D3DXSHRotateZ(&out[x], y, 1,25f, out)... as of the function gets smaller ... we may generate our expected values by hand. That way we can run all combinations. Not sure if it is worth the effort. Cheers Rico