Currently, it's ultra-generic, in the sense that there's no way for downstream code to figure out the name of individual parts of a multidimensional output. E.g. here's a CSV made from a 2D plot of double_pendulum.osim:
time,/bodyset/rod2[angular_acceleration.X],/bodyset/rod2[linear_velocity.Y] vs. /bodyset/rod1[linear_velocity.Y]/0,/bodyset/rod2[linear_velocity.Y] vs. /bodyset/rod1[linear_velocity.Y]/1,/jointset/pin2[reaction_on_parent.Magnitude]
0,0,0,0,0
0.01,0,-0.103197,-0.0368555,0
0.02,0,-0.206413,-0.0734404,0
0.03,0,-0.309665,-0.109485,0
0.04,0,-0.412968,-0.144723,0
0.05,0,-0.516333,-0.17889,0
0.06,0,-0.619767,-0.211728,0
0.07,0,-0.72327,-0.242986,0
0.08,0,-0.826834,-0.272421,0
Currently, it's ultra-generic, in the sense that there's no way for downstream code to figure out the name of individual parts of a multidimensional output. E.g. here's a CSV made from a 2D plot of
double_pendulum.osim: