I understand that. What I commented on was, as I interpreted it, your confusion that there was a difference between generating the matrix from degrees or radians. There isn't.

Are you a beginner programmer? A +-???E-15 instead of a 0 is an elementary property of double precision floating point numbers. In fact in a documentation of majority of programming languages is written: you can't expect the cos(PI/4) would exactly equal 0, a programmer who need cos(PI/4) to be exactly 0 should use 1 - cos(PI/4)^2 instead. (The reason behind this is a CPU sincos function. cos() is complementary to the sin().)

Self-taught in C++. Just never needed to worry about sin and cos too much before. Beside the fact I never came across it, even in my old C programming lessons.